Google Reinvests 2005 Profits Into Ecosystem Bets to Counter Microsoft

Acquired

Google successfully neutralized platform dominance from rivals by subsidizing massive, open-source web and mobile products to protect its core advertising revenue flywheel.

AI-assisted summary of Google Part II: Alphabet (Audio) — original source. Check the original for full context.

Source published: 2025-08-26

Detailed notes

## Background Published on August 26, 2025, by the YouTube channel Acquired, this four-hour-plus audio episode dives into the history of Google's strategic evolution into Alphabet. The show notes highlight Google's unprecedented six-year run from 2005 to 2011, during which it built or acquired seven products that surpassed one billion users each.

## TL;DR The episode explores how Google successfully diversified from a pure-play search engine into a multi-product giant spanning Gmail, Maps, YouTube, Chrome, and Android. It unpacks the strategic motivations behind these bets, framing them as vital offensive and defensive weapons in a three-front war against Microsoft, Apple, and Facebook. The hosts also contrast Google's major technical triumphs with cultural missteps like Google+, illustrating how this high-innovation era ultimately paved the way for its Alphabet restructuring and the modern AI revolution.

## Atmosphere & style The presentation is analytical yet conversational, marked by an easygoing, enthusiastic banter between hosts Ben Gilbert and David Rosenthal. The narrative carries a high-energy corporate thriller vibe, particularly when detailing strategic bidding wars, high-stakes platform rivalries, and the intense engineering breakthroughs that defined Google's golden era. It feels deeply researched and retrospective, balancing awe at Google's monumental scaling achievements with a candid, critical look at its systemic failures.

## Key takeaways - [03:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=233s) Google transitioned from a 'pure play' search engine into a diversified powerhouse, but faced initial Wall Street backlash in late 2005 as aggressive reinvestment into products like Gmail and Maps caused flat earnings and declining profitability. - [06:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=397s) Gmail's 2004 launch shifted the consumer paradigm by offering a massive 1 gigabyte of free storage, anticipating that Moore's Law would eventually reduce data storage and search costs to near zero. - [13:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=791s) The engineering foundation of Gmail and early Google web applications popularized Ajax (Asynchronous JavaScript and XML), which eliminated page reloads and enabled desktop-grade software experiences directly in the browser. - [16:56](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1016s) Google developed browser-based apps like Gmail and Google Maps as a defensive strategy against Microsoft, aiming to keep users engaged on the web rather than trapped within Microsoft's desktop operating system ecosystem. - [32:23](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1943s) Google Maps, launched in 2005 after key startup acquisitions like Where 2 Technologies and Keyhole, fueled the Web 2.0 mashup era through its open API and became a multi-billion dollar business via local advertising and licensing. - [42:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2520s) Google Workspace apps like Docs and Sheets split the productivity market with Microsoft Office, capturing absolute user volume and daily usage through free, real-time collaboration while Microsoft retained enterprise revenue. - [01:08:08](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4088s) The acquisition of YouTube for $1.65 billion in 2006 initially caused severe financial bleeding and engineering strain, but eventually evolved into an absolute media powerhouse generating over $50 billion in total annual revenue. - [01:17:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4657s) YouTube's monetization strategy relies on a 50% revenue-sharing model with content creators, driving immense loyalty and establishing the internet's most resilient independent career ecosystem. - [01:30:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5416s) Google acquired DoubleClick for $3.1 billion in 2007 following a fierce bidding war with Microsoft, securing critical programmatic ad exchange infrastructure and access to premium Madison Avenue ad inventory. - [01:52:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6731s) During its golden engineering era in the mid-to-late 2000s, Google consolidated its position as a top tier computer science hub by poaching legendary talent from dying corporate labs like DEC, Bell Labs, and Xerox PARC. - [01:59:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7190s) Google built the Chrome browser to secure its own distribution channel and keep the web platform alive, implementing killer features like the V8 engine, multi-process architecture, sandboxing, and the unified Omnibox. - [02:34:09](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9249s) Google acquired Android for $50 million in 2005 to solve mobile fragmentation and counter Microsoft's mobile operating system licensing model, completely pivoting the product after witnessing the 2007 iPhone reveal. - [02:58:33](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10713s) Android utilized a 'less than free' business model by sharing mobile search ad revenues with carriers and OEMs, successfully commoditizing mobile operating systems and driving Android to a peak global market share of 80%. - [03:23:12](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12192s) Google Plus was a reactionary, top-down command initiative triggered by existential fear of Facebook; its ultimate commercial failure served as an expensive distraction that slowed Google's entry into enterprise Cloud computing. - [03:31:36](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12696s) The failure of Google Plus forced an internal consolidation that unified distinct product 'fiefs' under a single cohesive Google Account login infrastructure, while preserving legacy wins like Google Photos and Google Meet. - [03:35:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12926s) In 2015, Google restructured into a holding company named Alphabet, separating its core lucrative advertising engine from speculative, high-loss 'Other Bets' like Waymo, Verily, and Nest. - [03:39:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13166s) Google concentrated an unmatched collection of global AI talent and corporate research between 2014 and 2016, leading directly to the invention and publication of the transformer architecture in 2017. - [03:44:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13451s) According to Hamilton Helmer's 7 Powers framework, Google's operational strength relies heavily on scale economies for global advertisers, switching costs via user history (Gmail), and deep infrastructural advantages like Borg. - [03:54:09](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14049s) Eric Schmidt's strict product funding filter required a 'core technical insight'—a technical breakthrough that directly drove the user experience—explaining why Google succeeded in search, browsers, and maps, but repeatedly failed in pure social media.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=0s) Turtlenecks and Spoilers: The Summer 2025 Season of Acquired

The episode kicks off with lighthearted banter between hosts Ben Gilbert and David Rosenthal. Ben opens by teasingly asking David if he is intentionally wearing a black turtleneck, wondering aloud if David is trying to dress up like Steve Jobs for a Google-themed episode. David laughs it off, hinting that the outfit is actually going to be one of his "carveouts" and playfully retorts, "You think I dress up like Steve Jobs for a Google episode?" Ben admits he thought it might be a nod to the historic war between Android and Apple, noting that David walked in with a distinct smirk on his face.

After their traditional musical intro, Ben formally welcomes listeners to the summer 2025 season of *Acquired*, introducing it as "the podcast about great companies and the stories and playbooks behind them." He tees up the narrative of Google, explaining how the company built the internet's best search engine in the late 1990s using a breakthrough search algorithm, low-cost commodity hardware, and the lucrative business model of search ads. After going public in 2004, Google curiously diversified into non-search products like Gmail, Google Maps, Google Docs, YouTube, Android, and Chrome. Today, Google boasts 15 products with over half a billion users, and seven products with over two billion users—meaning over 25% of the global human population uses seven distinct Google products.

The hosts acknowledge that alongside these staggering triumphs, Google also launched some "colossal failures." Ben lists examples like Google Wave, Google Buzz, about a dozen different messaging apps, hot air balloons for wireless internet, and the infamous Google Glass. The core theme of the episode is exploring why Google transitioned from a pure search engine into the "innovation factory of the 2000s," how it reorganized into its parent company Alphabet, and how these various initiatives feed back into its original mission to organize the world's information. When Ben notes that the story will conclude at the dawn of the AI era, David jokingly chides him for giving away the ending, to which Ben replies, "Oh, spoilers. Sorry."

- **Google Product Scale:** 15 products have over 500 million users; 7 products have over 2 billion users (representing over 25% of the human population). - **Google Failures Mentioned:** Google Wave, Google Buzz, approximately 6 to 12 messaging apps, Loon (wireless internet hot air balloons), and Google Glass. - **Show Links:** Listeners can sign up for the email list at `acquired.fm/e` to vote on future episode topics and join the community Slack at `acquired.fm/slack`. - **Sponsor:** JP Morgan Payments is introduced as the presenting partner.

---

[03:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=233s) The "Pure Play" Dilemma and Wall Street's Drunken Juggler

David opens the historical narrative by quoting a fictional character, Russ Hanneman, from the HBO TV show *Silicon Valley*: "If you show revenue, people will ask how much and it will never be enough... But if you have no revenue, you can say you're pre-revenue. You're a potential pure play." While Google was undeniably a "cash gushing machine" and didn't suffer from a lack of revenue, it faced a unique dilemma in the mid-2000s. Following its IPO in the fall of 2004, Wall Street viewed Google as a perfect, predictable "pure play" on internet monetization; as internet usage and search volume grew, Google automatically made more money through AdWords.

This investor honeymoon ended abruptly when Google announced its fourth-quarter 2005 earnings. Although full-year 2005 revenue had nearly doubled to $6.1 billion compared to $3.1 billion in 2004, flat earnings and declining profitability shocked the market. Wall Street discovered that Google was aggressively re-investing its massive cash reserves into non-search future bets like Gmail, Maps, the upcoming Google Docs, and a pending $1.6 billion acquisition of YouTube. Wall Street hated this diversification. In January 2006, the stock plummeted 27% as investors felt the company was ruining a perfect business model. David quotes author Steven Levy from his book *In the Plex*, noting that the prevailing market perception was that Google was "tossing balls into the air like a drunken juggler."

- **Financial Metrics (2004 vs. 2005):** 2004 revenue was $3.1 billion; 2005 revenue doubled to $6.1 billion, but flat earnings led to a 27% stock drop in January 2006. - **YouTube Acquisition Value:** $1.6 billion (set for later in 2006).

---

[06:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=397s) The Launch of Gmail and Morally Repugnant Storage

Ben describes the research process for this episode as a nostalgic journey through a "cornucopia of hits of iconic product launches." The first, and arguably most important, non-search product was announced on April 1st, 2004: Gmail. Because of the date, it became the most infamous, non-joke April Fool's Day announcement in tech history. The offering sounded completely unbelievable for 2004: a fully web-based email service accessible anywhere on any device, equipped with built-in Google search, eliminating the need for folders, and featuring a staggering one gigabyte of free storage.

To contextualize how radical 1 GB of storage was, David points out that the dominant webmail services of the era—Microsoft's Hotmail and Yahoo Mail—offered a mere 2 megabytes and 4 megabytes of free storage, respectively. David shares an anecdote from *In the Plex* where Steven Levy interviewed Bill Gates at the *Newsweek* headquarters shortly after Gmail's launch. Gates was visibly offended and bewildered by Gmail, viewing the massive storage allocation as wasteful and doing email wrong. David notes that up until Gmail, the paradigm was to treat email like physical mail—sorting and throwing away messages to conserve space. To Gates, leaving everything sitting permanently on a server was almost "morally repugnant." However, Google's engineers correctly anticipated that because of Moore's Law, the asymptotic cost of storing and searching data would eventually plummet to zero.

- **Storage Capacity Comparison (2004):** Hotmail offered 2 MB; Yahoo Mail offered 4 MB; Gmail launched with 1,000 MB (1 GB).

---

[09:31](https://www.youtube.com/watch?v=QhAftC_zFr8&t=571s) Paul Buchheit and the Usenet Origins of Gmail

Ben shares exclusive insights from a conversation he had with Paul Buchheit, the legendary inventor of Gmail, prior to the recording. The story originates in 1996 when Buchheit was a student at Case Western Reserve University in Cleveland, Ohio. Crucially, Case Western was one of the first campuses in the nation to feature high-speed broadband internet in dorms, allowing Buchheit to temporarily live "15 years in the future." This experience sparked an obsession with the idea that email shouldn't be downloaded to a single machine; bits were becoming free to move, so email information should live in a web browser, accessible at all times. Buchheit built his first webmail prototype during his college years.

In 2001, prior to Google's IPO, Larry Page grew frustrated that engineering management was slowing down the company. Page famously eliminated engineering managers and, alongside engineering head Wayne Rosing, conducted individual meetings with every engineer. Ben highlights this as a testament to early "Googliness" and the sheer caliber of their talent: leadership would simply ask engineers what they wanted to build and task them to "full stack own this product entirely yourself." In Buchheit's meeting, Page and Rosing floated an amorphous email concept to him, motivated partly by a desire to create a "sticky" application that would preserve Google's customer relationships if a competitor ever displaced their search engine. Buchheit, who had originally discovered Google via a Slashdot post in 1999 and applied via `jobs@google.com`, eagerly accepted.

The initial technical foundation for Gmail came from Google’s very first acquisition, a company called Deja News, which held a massive corpus of old Usenet posts. Buchheit had been working on a feature to perform real-time indexing and search on this Usenet archive. He decided to apply that exact same search mechanism to his personal Unix mail directory. This basic search box served as the very first working iteration of Gmail.

- **Key Historical Figures:** Paul Buchheit (Gmail inventor), Larry Page (Google co-founder), Wayne Rosing (early Head of Engineering). - **Google's First Acquisition:** Deja News (the basis for Google Groups and the initial code seed for Gmail search).

---

[13:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=791s) The Birth of Ajax and a Deeply Ironic Microsoft Heritage

While building out his email prototype, Buchheit realized he needed a high-performance web interface. Existing webmail services like Hotmail and Yahoo Mail were notoriously slow because they relied on static HTML; every single action, whether opening an email or saving a draft, required the server to fully reload the entire web page. Buchheit turned to JavaScript, a language invented by Brendan Eich at Netscape in 1995 to add dynamic elements to web browsers. By 2001, JavaScript was widely viewed as a "toy-like," weak language rather than a serious programming tool, largely because it had become a casualty of the brutal browser wars between Netscape and Microsoft's Internet Explorer. Even `google.com` was a static experience where users submitted queries and received an entirely new page of results.

Buchheit experimented with JavaScript to transition web pages into true "web applications." During his experimentation, he uncovered a little-known, obscure JavaScript feature called the `XMLHTTPRequest` API. This feature allowed a web page to automatically fetch new XML data from a server in the background without triggering a page reload. Realizing he had struck gold, Buchheit utilized it to build Gmail, marking the birth of Ajax (Asynchronous JavaScript and XML) and setting the standard for the dynamic "Web 2.0" era.

Ben stops David to deliver a final, deeply ironic twist regarding the origin of the technology. The `XMLHTTPRequest` API was actually first created and implemented by Microsoft within Internet Explorer. Microsoft had built the feature specifically so users could access corporate email via a web browser using a tool called Outlook Web Access. Ben reminisces about logging into his high school email using this exact web access tool, noting that it only functioned properly inside Internet Explorer. The hosts marvel at the profound historical irony that the revolutionary technology Google used to disrupt the webmail market and defeat Microsoft was originally engineered by Microsoft itself for its own mail client.

- **Technical Terms:** Ajax (Asynchronous JavaScript and XML), driven by the `XMLHTTPRequest` API. - **Historical Heritage:** JavaScript was invented by Brendan Eich at Netscape in 1995. The underlying API for Ajax was originally built by Microsoft for Outlook Web Access inside Internet Explorer.

[16:56](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1016s) The Birth of Gmail and the Web as a Platform

When Paul Buchheit built the initial prototype of Gmail, the idea of running a rich application directly in a web browser was a total breakthrough. At the time, users were accustomed to downloading Mac apps from the internet or buying physical software from retail shops like Comp USA. When Larry Page and Sergey Brin saw the prototype, they were "super jazzed" and immediately became the very first beta testers, using it exclusively as their internal mail service. By the time Gmail launched publicly, the entire Google workforce was addicted to it. Though the terminology didn't exist yet, Gmail was effectively functioning "in the cloud"—allowing users to log in and access their data from any network or device instead of relying on a local machine or dedicated physical server.

Larry Page, driven by a pure product-centric philosophy, believed that building a superior web product was inherently good for Google and should be pursued at all costs. However, a major strategic motivation behind Gmail was the existential threat posed by Microsoft. While Google was "printing money" through search and AdWords, their entire business model flowed through Microsoft’s ecosystem. Over 90% of Google's search queries were executed on Windows PCs running Internet Explorer, meaning Google effectively existed "at the pleasure of Microsoft."

- **Core User Habit:** Larry Page evaluated products using the "toothbrush test"—asking whether it was something a user would engage with once or twice a day as part of their daily habit. - **The Google Flywheel:** Users search via Google $\rightarrow$ discover new sites and services $\rightarrow$ spend more time online $\rightarrow$ perform more searches $\rightarrow$ Google monetizes the entire cycle.

[19:31](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1171s) The Strategic Clash with Microsoft

The structural shift toward web applications posed a massive dilemma for Microsoft. Microsoft's traditional business model relied on selling Windows to computer OEMs and layering paid desktop software on top. While web usage wasn't inherently bad for them, the rise of web apps threatened to make the underlying operating system irrelevant, equalizing the playing field for competitors like the Mac or a future beyond desktop computing. Fortuitously for Google, Microsoft was largely distracted by the "albatross" of Longhorn, the development project that would eventually become Windows Vista.

Google executive Eric Schmidt understood this competitive dynamic intimately, having previously served as the CEO of Novell, a company that Microsoft had completely crushed. Google realized that its best defense against Microsoft disadvantaging web applications was to build products so compelling that consumers would "revolt" if access were restricted. Developing Gmail to a launch-ready state took three full years because the engineering required was entirely uncharted territory. Developers had to write complex code using dynamic JavaScript and XML refreshing (Ajax) without the help of modern abstraction layers or development frameworks.

- **Development Timeline:** Gmail spent three years in active development (2001–2004) before its public release and famously remained in "beta" for roughly 10 years.

[21:55](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1315s) The Gimmick That Solved an Infrastructure Crisis

As the 2004 launch approached, Google faced a massive financial hurdle: offering a full gigabyte of free storage per user when competitors were offering a mere 4 megabytes. Even with Google's custom commodity infrastructure providing a 90% cost advantage, the effective cost burden of scaling up to meet unconstrained user demand was terrifying—especially in an era before the public cloud or AWS existed. To throttle server demand, Google devised a restricted invite system, launching on April 1, 2004, with just 1,000 "seed invites" sent to journalists and influential tech figures.

The technical bottleneck accidentally created one of the most brilliant viral growth strategies in internet history. Because invites were capped—usually giving a user just five invites that topped up unpredictably based on server capacity—the service felt highly exclusive. Desperate users begged for access, and Gmail invites routinely traded on eBay for an average price of \$150. This artificial scarcity succeeded because the product quality backed up the hype, turning Gmail from a simple username-reservation landing pad into an indispensable daily tool.

- **Initial Seed:** 1,000 invites sent to influencers and journalists on April 1, 2004. - **eBay Market Value:** Invites commanded an average secondary market price of $150 during the peak of the launch craze. - **Long-term Scale:** Gmail grew from its initial pool of 1,000 beta users to more than 2 billion users over the subsequent 20 years.

[25:25](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1525s) Monetization, AdSense, and the Integration Layer

Figuring out how to monetize Gmail was the second major puzzle for the development team. During the prototyping phase, Paul Buchheit bypassed traditional design barriers by logging into Google's core search ads database and running content-matching algorithms against user inboxes. When he displayed these keyword ads next to emails, early internal testers were initially furious, demanding, "Are you looking at my emails?" Despite the privacy pushback, Larry and Sergey loved the concept. This internal experiment actually predated the launch of AdSense, and the mechanism of content-matching display advertisements directly influenced the framework Google used to build its massive publisher ad network.

The success of Gmail served as an "existence proof" for complex web apps, prompting Larry, Sergey, and Eric to aggressively fund browser-based alternatives to desktop software. Google realized that if applications lived inside a browser, the web itself would become the "point of integration." In this paradigm, hardware OEMs just needed to build machines capable of running a browser, and users no longer had to care about the underlying operating system. Whether Google built these apps or third parties did, accelerating web adoption directly fed Google's core search business. This strategic charge led to a wave of legendary web releases, beginning with Google Maps, which was kicked off in 2003 by a young Associate Product Manager named Brett Taylor.

- **Brett Taylor's Career Pedigree:** APM for Google Maps, founder of FriendFeed, CTO of Facebook, co-CEO of Salesforce, chairman of OpenAI, and former chairman of Twitter.

[33:23](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2003s) The Genesis and Strategy of Google Maps

In 2003, Brett Taylor, a fresh Stanford graduate working as an associate product manager at Google, approached co-founder Larry Page with a warning: "We're missing out here." AOL had just acquired MapQuest for a staggering billion dollars, and word on the street was that Yahoo was gearing up for a massive push to launch Yahoo Maps. Recognizing the threat, Page eagerly approved a web-based map project, operating under Google's broader philosophy that increasing overall web usage inherently funnels more traffic back to their primary search engine—effectively fueling their core "money printer." Beyond pure financial incentives, mapping fit squarely into Google's foundational mission to organize the world's information and make it universally accessible.

At the time, existing services like MapQuest and Yahoo Maps were heavily utilized, but they were vastly different from modern digital maps. They consisted of static web pages without Ajax, built primarily so users could type in an address and print out paper driving directions. The entire business model resembled a "Trojan horse newspaper business," relying on ads printed directly onto those physical sheets of paper. Google saw a clear opportunity to build something superior.

* **Key Names:** Brett Taylor (Associate Product Manager), Larry Page (Google Co-founder), Marissa Mayer. * **Competitor Actions:** AOL bought MapQuest for \$1 billion; Yahoo was actively developing Yahoo Maps. * **The Baseline Concept:** Early 2000s web mapping relied on static pages designed for printing physical directions.

[35:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2100s) Acquiring "Where 2 Technologies" and Pioneering the Web Map

To leapfrog the competition, Google acquired a small Australian startup called Where 2 Technologies, founded by brothers Lars and Jens Rasmussen. The Rasmussons were extraordinary engineers who had created a fully real-time, interactive mapping application—the catch being that it was an installed desktop app. When meeting with them, Page made Google's preferences explicitly clear, stating, "We like the web at Google." In a remarkable feat of engineering, the brothers completely rewrote and rearchitected their entire desktop application into a web app in just three weeks. In doing so, they independently discovered and implemented many of the same JavaScript and Ajax techniques that Google was secretly developing internally for Gmail, which had not yet launched.

This acquisition formed the foundational core of Google Maps. Around the same time, Google snapped up two other critical pieces: ZipDash, which provided traffic data, and Keyhole, a desktop application that eventually became Google Earth. While the speakers briefly debate whether Google Earth was completely folded into Maps, they note that earth.google.com remains a distinct, incredibly powerful 3D web experience today. Interestingly, the speakers joke about how Keyhole originally served as a "Trojan horse downloader" designed primarily to slip the Google Toolbar onto users' Internet Explorer browsers, calling it "the greatest distribution hack for Google search of all time."

* **Where 2 Technologies:** Founded by brothers Lars and Jens Rasmussen; converted a desktop application into a fully interactive web app in three weeks. * **Parallel Acquisitions:** ZipDash (traffic data) and Keyhole (satellite imagery, which became Google Earth). * **Strategic Distribution:** Keyhole historically doubled as a vehicle to drive installations of the Google Toolbar on Internet Explorer.

[37:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2234s) The 2005 Launch, the "Missing Europe" Easter Egg, and Web 2.0 Mashups

Google Maps officially launched in February 2005, causing a massive stir online because of its fluid, dynamic mapping interface. The speakers share a humorous piece of trivia from the very first day of the launch: when users fired up the site, they were greeted by a massive ocean, North America, and the UK floating lonely in the middle of the Atlantic. Everything past the UK—Europe, Asia, and Africa—was completely missing, appearing as empty blue water simply because the team hadn't built it yet. It was the ultimate, literal implementation of a Minimum Viable Product (MVP).

The momentum accelerated dramatically in 2006 when Google released the Google Maps API, a move that truly catalyzed the "Web 2.0" era. Alongside Gmail, the Maps API inspired developers globally to build rich, interactive web applications. Suddenly, software "mashups" became the trend, allowing developers to easily overlay their own data onto Google's maps. This single API became the critical infrastructure that enabled an entire generation of location-dependent startups and mobile-social-local ("MoSoLo") apps to exist.

* **Launch Date:** February 2005. * **The Launch Day Visual:** Only North America and the UK were mapped; the rest of the world appeared as blank ocean. * **API Release:** 2006, sparking the Web 2.0 "mashup" phenomenon. * **Dependent Startups:** Zillow, Uber, DoorDash, Airbnb, Foursquare, and Gowalla.

[39:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2356s) The Mammoth Engineering Challenge and Modern Economics of Maps

Building Google Maps served as a powerful defensive moat against Microsoft by deepening user lock-in to web-based ecosystems. Originally, Google offered the Maps API for free or at incredibly low cost with massive usage limits, adopting a "figure out the business later" founder-driven mentality. For the first five to eight years, Google was essentially the only company willing to pour hundreds of millions—and eventually billions—of dollars into the giant financial hole required to map the physical world from scratch. It was a daunting, uniquely "Googly" engineering problem that involved driving fleets of cars around the globe to capture imagery, cleaning data, crowdsourcing updates, and dynamically blurring out personal information.

Today, those massive investments have yielded incredible strategic value, particularly as core data assets for the AI era and self-driving cars. The platform has grown to over two billion active users. While Google does not publicly break down the exact financial metrics, industry estimates peg Google Maps' annual revenue at anywhere between \$5 billion and \$10 billion. The vast majority of this revenue is driven by sponsored local ads that appear when users open the app, supplemented by a smaller stream of revenue from modern API licensing fees.

* **User Base:** Over 2 billion active users. * **Estimated Revenue:** Well over \$5 billion, potentially up to \$10 billion annually. * **Revenue Streams:** Dominated by sponsored local ads, with the remainder coming from commercial API licensing.

[42:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2520s) Attacking Microsoft's Crown Jewels with Docs and Sheets

Google next took aim directly at Microsoft's most profitable stronghold: Microsoft Office. Together, the applications that make up Google's office suite—now bundled into Workspace and Google Drive—boast over a billion users. While competitors like WordPerfect and Lotus 1-2-3 had tried and failed to break Microsoft's monopoly in the past, Google recognized a structural shift. Instead of trying to out-feature desktop software, Google leveraged Web 2.0 and Ajax to introduce something entirely new: real-time, multi-user online collaboration.

The speakers highlight that these were arguably the first real-time collaborative productivity tools in software history. Google Docs began as an acquisition of a company called Writely, which was founded by Sam Schillace and his co-founders. Writely launched in August 2005 and was bought by Google just six months later in March 2006. Ironically, the Writely team didn't initially set out to build a collaborative tool; they just wanted to see if they could build a word processor using JavaScript and Ajax. However, as the co-founders naturally began editing the code and documents together during development, they realized that collaboration was their "killer feature."

* **Workspace User Base:** Over 1 billion users. * **Historical Context:** Former Novell CEO Eric Schmidt (who became Google's CEO) previously ran WordPerfect, giving him firsthand experience trying to compete with Microsoft Office. * **Writely (Google Docs) Timeline:** Launched August 2005; acquired by Google in March 2006.

[43:55](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2635s) The Open Question of Sheets and the Subsidized War on the Enterprise Agreement

Unlike the Docs team, the creators of what became Google Sheets, led by Jonathan Rochelle, focused on collaboration from day one. They knew they couldn't beat Excel on raw features, so web-based sharing had to be their primary value proposition. At the time, it was an open question whether the browser could even handle two people "partying on the same very basic spreadsheet at the same time." Microsoft had long held an iron grip on the market due to deep feature sets, complex proprietary file formats, and—most importantly—the corporate Enterprise Agreement, which allowed them to bundle Office as an all-in-one platform for IT departments.

An independent startup could never have survived the years it took for web-based productivity tools to gain mainstream enterprise adoption; early on, Google Docs was laughed out of medium and large companies. Google, however, could afford to fully subsidize the massive infrastructure costs because the incremental server load of Docs and Sheets was trivial compared to their core search engine. Google didn't care about making immediate enterprise dollars; their goal was simply to "put the screws to Microsoft" and force them into a costly, slow defensive play to port their desktop software to the web.

* **The Spreadsheet Gamble:** The team focused entirely on web collaboration because matching Excel's desktop feature depth was impossible. * **The Microsoft Moat:** Driven by complex proprietary file formats and B2B Enterprise Agreements. * **The Infrastructure Advantage:** Running Docs and Sheets required massive real-time backend coordination, but it represented a trivial incremental load on Google's existing search infrastructure.

[48:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2915s) Microsoft's Web Dilemma and the Current Market Landscape

Google's strategy forced Microsoft into a challenging engineering and business dilemma. To compete, Microsoft had to build web versions of Word and Excel that maintained pixel-for-pixel, character-for-character "perfect document fidelity" with their desktop counterparts so that printed files matched exactly. One of the speakers shares a personal anecdote from an internship at Microsoft, recalling how incredibly difficult it was just to add headers and footers to the Word web app under those rigid rendering constraints. Furthermore, the web model was culturally antithetical to Microsoft at the time, forcing painful internal debates around pricing, packaging, and whether allowing users to access Word for free online would cannibalize their lucrative desktop licensing.

Today, this tactical move has settled into a fascinating market split. While it is difficult to find perfect apples-to-apples data comparing the two ecosystems, the speakers note that the productivity market has fundamentally divided along lines of volume versus monetization. Google successfully captured the vast majority of absolute users and daily usage through the friction-free nature of Google Workspace, while Microsoft managed to retain the vast majority of the actual enterprise dollars.

* **The Engineering Hurdle:** Microsoft focused heavily on maintaining absolute visual fidelity between desktop and web layouts, slowing down their web development. * **The Business Dilemma:** Web apps forced Microsoft to wrestle with free tiers and usage tiers that threatened their existing business model. * **Current Market Split:** Google owns the majority of user volume and daily usage; Microsoft maintains the vast majority of total revenue.

[50:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3035s) Market Scale and the Revenue Divide

The discussion opens with a stark look at the user counts and revenue generation of productivity suites, highlighting a fascinating disconnect. Google Docs, Sheets, and Slides each boast user bases in the "billionish" range—roughly 500 million to a billion users each. In contrast, Microsoft Office sits at a comparatively modest couple hundred million users worldwide. This discrepancy catches Ben by surprise, prompting an astonished "Whoa. Yeah, that's crazy."

However, David points out that when looking at the actual dollars, the dynamic flips entirely. Microsoft’s productivity and business process segment—which predominantly consists of Office and now includes LinkedIn—generated over $120 billion in revenue in the previous year. Because this is incredibly high-margin revenue, it represents a massive financial stronghold. Meanwhile, Google lumps Workspace into its total cloud segment. The entire Google Cloud umbrella, including all infrastructure as a service and AI infrastructure, brought in less than $50 billion. This means Google's office suite represents only a small fraction of a segment that is already less than half the size of Microsoft's. Ultimately, "Microsoft's still got all the money" while "Google's got all the users," creating a status quo where both giants are perfectly content.

- Google Docs, Sheets, and Slides users: 500 million to 1 billion each - Microsoft Office users: A couple hundred million worldwide - Microsoft Productivity and Business Process revenue: Over $120 billion (high margin) - Total Google Cloud segment revenue (including Workspace and AI infrastructure): Less than $50 billion

[51:58](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3118s) The Long Game of Corporate Careers and Google's Mission

A hilarious and definitive "fun KOD" (kiss of death) or twist of fate anchors the narrative: Sam Schillis, the original founder of Writly who went on to manage Google Docs, Sheets, and Maps, has come full circle and now serves as the deputy CTO of Microsoft. This prompts the timeless observation that "careers are long." Reflecting more broadly on Google's business in the mid-to-late 2000s, the hosts note that Google had fundamentally won search by this point. Even though Bing had not yet launched, it was obvious that the search market was bound to grow exponentially. This absolute dominance meant Google could treat other projects as altruistic vectors to fuel its broader mission. If Larry Page were present, he would argue that a company's true goal isn't merely building the largest business, but fulfilling its core mission—a luxury Google could easily afford given its "money printing machine" in search.

[53:12](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3192s) The Strategic Playbook and the Road to YouTube

Armed with successive web app successes like Maps, Docs, and Spreadsheets, Google developed a clear Web 2.0 acquisition strategy. The company realized it could purchase early-stage web applications, integrate them into its massive infrastructure, turbocharge them, and offer magical user experiences at a fraction of the cost a standalone company would incur. This provided immense strategic value on both offensive and defensive fronts. The hosts joke about what would happen if Google went "really big" with this playbook, eyeing an acquisition target that would be notoriously expensive to run, legally hazardous, and require vast amounts of storage and streaming bandwidth. This foreshadows the entry of YouTube, a startup heavily backed by Sequoia Capital.

Before diving into the YouTube saga, the hosts take a brief interlude to thank their sponsor, Anthropic, and praise its AI assistant, Claude. They draw a humorous parallel between Google's scaling challenges in the 2000s—managing interconnected systems for billions of users—and the exponentially more complex, multi-hour enterprise workflows that modern companies manage using Claude's API and its MCP connector system.

- Web 2.0 playbook: Acquire, turbocharge, lower operational overhead, and leverage for offensive/defensive strategy - Sponsor offer: Half-price Claude Pro for 3 months at `claude.ai/acquired`

[56:25](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3385s) Re-grading History and the TV Ad Dollar Paradigm

Turning their attention to YouTube, whom they dub "the big kahuna," the hosts laughingly acknowledge that their own very early podcast episode covering YouTube is "the most embarrassing thing in Acquired history." To set the record straight, Ben proposes a total update to the "Acquired canon" by reviving their classic grading system specifically for this section.

Setting the historical scene in 2003—before Gmail had even launched—Google began building Google Video. The initial thesis was that video contained immense amounts of information that fit Google's mission, but it was also explicitly driven by the reality of global advertising. At that time, television commanded the overwhelming majority of global ad spend. Ben shares striking historical data from Mary Meeker’s internet trends reports, noting that digital advertising did not eclipse traditional television ad spend until 2017 or 2018. For nearly fifteen years, an enormous monetization gap existed where consumer attention had shifted to digital, but the actual ad dollars remained locked in television.

- Contextual era: 2003, prior to the launch of Gmail - The Ad Flip: Digital advertising did not overtake TV advertising until 2017–2018

[58:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3529s) The Failure of Google Video vs. the Rise of User-Generated Content

Larry Page initially described Google Video, launched in 2005, as a search service for television content. Because user-generated video had not yet matured, Google relied on TV closed-captioning data to make the video searchable, effectively acting as an archiver that pointed users to external sites. However, Google Video was fundamentally the wrong product. At its inception, it didn't even feature an integrated video player; it simply pointed users away from Google to consume content elsewhere. Furthermore, its corporate strategy targeted traditionally produced, mainstream "head" content rather than the long-tail, user-generated content that was bubbling up from the cultural underground.

Concurrently, the mid-2000s saw a massive boom in consumer-led digital video. Thanks to affordable point-and-shoot digital cameras, early dedicated devices like the Flip Cam, and the introduction of user-friendly editing software like iMovie, everyday users suddenly had the power to shoot, edit, and share video content seamlessly. This consumer electronics wave was widely considered the major tech frontier right before the smartphone revolution took over.

- Google Video launch: 2005 (initially lacked an internal video player) - Key consumer tech enablers: Point-and-shoot cameras, Flip Cams, and Apple's iMovie

[01:01:51](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3711s) The PayPal Mafia, Dating Site Roots, and YouTube's Brilliant Playbook

In early 2005, three relatively junior PayPal employees—Chad Hurley, Jawed Karim, and Steve Chen—left the company to found YouTube. David shares two incredible pieces of historical trivia: first, YouTube's original tagline was "Tune in, hook up," functioning as a failed video dating service that resorted to posting Craigslist ads offering to pay attractive women to upload video profiles. Second, Chad Hurley’s father-in-law at the time was none other than Jim Clark, the legendary founder of Silicon Graphics and Netscape, giving the young founders an unparalleled strategic advisor in Silicon Valley.

When the dating angle flopped, they pivoted to a general-purpose video site. YouTube's brilliance relied on a three-fold playbook: 1. **Incredibly simple video uploading:** A frictionless content acquisition model where videos went live instantly without the tedious, 1-to-2-day manual copyright reviews that crippled Google Video. 2. **Seamless consumption:** A highly optimized web-based video player that worked instantly. 3. **Viral distribution via embedding:** The ability to embed a video on any external website, creating an explosive growth loop.

YouTube's internal search functionality was so strong that it rapidly transformed into the second-largest search engine on Earth, trailing only Google itself. This massive aggregation of attention posed a systemic threat to Google; if YouTube chose to expand its search capabilities to other verticals, it could have jeopardized Google’s core business model in the exact same way WhatsApp's user aggregation alarmed Mark Zuckerberg years later.

[01:06:19](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3979s) The Copyright Catch-22 and the Viral Catalyst

A profound structural paradox defined the battle between Google Video and YouTube. As a freshly public corporation, Google had to rigidly observe copyright laws, making it impossible for them to allow unfettered user uploads. YouTube, operating as a scrappy, resource-constrained startup, could hide behind its platform status and adopt a "whatever, upload it" mentality. This structural freedom gave them the ultimate user experience advantage, but it created a classic Catch-22: the exact strategy required to win the users would inevitably cause them to be sued out of existence unless they were absorbed by a giant like Google.

The absolute magic of this model became undeniable when early viral sensations took over. The hosts reminisce about the sheer consumer awe of being sent a link or seeing an embedded video player and realizing you could watch high-quality cultural touchstones completely free. The ultimate catalyst for this era was The Lonely Island's Saturday Night Live digital short, "Lazy Sunday," which users began uploading en masse, cementing YouTube as an indispensable fixture of the modern internet.

[01:08:08](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4088s) The Astronomical Scaling Costs of YouTube's Success

The early trajectory of YouTube shifted dramatically with a single Saturday Night Live skit by The Lonely Island. Uploaded during a brief, volatile phase when YouTube was still an ascendant startup independent of Google, this single piece of content caused a massive 83% spike in traffic. The explosive growth caught the attention of Sequoia Capital, which had essentially incubated the company when its three founders left PayPal. Roelof Botha—who knew the founders from his own days in the PayPal mafia—led Sequoia's rapid initial investment. However, because infrastructure costs began scaling astronomically almost immediately, Sequoia had to quickly lead another funding round just to keep the lights on.

The staggering expenses of running a video platform at this scale break down into three distinct areas: one fixed, upfront cost and two ongoing variable costs. The upfront hurdle is video encoding, a massive expense required to ensure that a single video can play across multiple device types and browsers. The ongoing variables are where the real financial pressure lies. First, storing the massive volume of video files requires immense capital. Second, and most challenging of all, is the network bandwidth. Bandwidth costs money every single time a user hits play, meaning the company’s largest cost driver directly scales with "minutes watched." Without a tightly aligned business model, this dynamic is a financial death sentence unless the operating company possesses its own massive, cheap infrastructure.

- **Traffic Spike:** A single Lonely Island SNL skit increased YouTube traffic by 83%. - **Sequoia Connection:** Roelof Botha made YouTube his first investment upon joining Sequoia due to his PayPal mafia connections with the founders. - **Three Core Infrastructure Costs:** 1. Encoding (One-time, high upfront cost for cross-device/browser compatibility). 2. Storage (Ongoing variable cost). 3. Bandwidth/Networking (The largest variable cost, scaling directly with minutes watched).

---

[01:09:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4192s) In Over Their Heads: The Sale to Google

Within a little over a year of launching, YouTube found itself completely in over its head, facing a triad of overwhelming content, copyright, and infrastructure issues. Paradoxically, this chaos was exactly what the founders wanted—it meant the platform was succeeding beyond their wildest dreams—but the operational realities were unsustainable. The speakers note that if this happened in the modern era of abundant private capital and public cloud infrastructure, a startup could easily raise billions in a Series A round to scale independently. In 2005 and 2006, however, that kind of private capital did not exist. Furthermore, a public market debut was out of the question, particularly due to a massive, looming copyright lawsuit from Viacom.

With virtually no leverage in content negotiations with rights holders and infrastructure costs bleeding them dry, YouTube put itself up for sale. In November 2006—less than 18 months after the product first launched—Google acquired YouTube for $1.65 billion in stock. The speakers emphasize that it was an all-stock deal, a detail that Google’s CFO at the time, Patrick Pichette, famously lamented. Pichette reportedly declared "never again" regarding stock-based acquisitions. Because Google's market cap has since 20xed, paying in cash would have saved Google an extra 20 times multiple on the purchase. Even so, the hosts agree that despite the stock dilution, Google got a "screaming deal" given how unimaginably valuable YouTube has become.

- **Acquisition Date:** November 2006 (less than 18 months post-launch). - **Purchase Price:** $1.65 billion paid entirely in Google stock. - **Viacom Lawsuit:** A major copyright suit that blocked any potential of YouTube going public at the time. - **The CFO's Regret:** Google's market cap 20xed after the deal closed, prompting the CFO to state that it would be the last stock deal they ever do.

---

[01:11:55](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4315s) "Google's First Mistake" and the Penny-Per-View Loss

To piecing together the financial realities of YouTube’s first few years, the hosts draw heavily from data shared by Shishir Mehrotra—the former head of product, CPO, and CTO of YouTube—on Kleiner Perkins' Grit podcast. Immediately following the acquisition, YouTube was generating a modest $30 million in revenue, largely through programmatic advertising tied to the DoubleClick ad exchange. However, the platform was bleeding cash at an alarming run rate of roughly $1 billion a year. The losses were so predictable they amounted to almost exactly "a penny per view." Every time a user loaded a video, Google was effectively flushing a cent down the drain.

This financial hemorrhage terrified Google's financial leadership. For the first couple of years, the CFO was so deeply panicked by the platform's viral growth that they actively explored selling it off to other initial bidders like Yahoo or various legacy media companies, pleading behind closed doors for the platform to stop scaling. Mehrotra noted that internally, the team was "broadly known as Google's first mistake." The hosts look back on their own podcasting history, admitting that when they covered YouTube ten years prior, they also erroneously labeled it a terrible acquisition. Compounding these issues were astronomical music licensing fees; for a long time, YouTube was a primary, yet incredibly expensive, revenue funnel for the music industry, rivaled only later by Spotify.

- **Initial Post-Acquisition Revenue:** $30 million. - **Initial Run Rate Loss:** Approximately $1 billion per year. - **The Financial Metric:** YouTube lost roughly $0.01 (one penny) every single time a video was viewed. - **Competing Bidders:** Yahoo and various traditional media companies had actively tried to buy YouTube alongside Google.

---

[01:13:56](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4436s) The Shift to Browsing, Mobile, and Watch Time

In its earliest era, YouTube wasn't a destination site; users typically discovered videos embedded on external websites. A user would click an embed, watch the clip, and leave. For the first four to five years after the Google acquisition, 90% of the traffic that actually visited YouTube.com came strictly to use the search bar, completely ignoring recommendations. Building consumer habits and engineering sophisticated recommendation algorithms took years. The sheer scale of user consumption was staggering: a 2007 report estimated that YouTube consumed as much bandwidth as the *entire internet* did in the year 2000. By 2014, Mehrotra noted that YouTube single-handedly accounted for 20% of all data bits moving across the internet, underscoring how drastically video outpaces any other media form in bandwidth consumption.

Google's long-term bet, however, focused on the ultimate thesis of digital media: video is what humans fundamentally want to consume, and it can be heavily monetized with ads. Recognizing they had captured "lightning in a bottle," Google leaned into the consumer brand rather than continuing to fund their own competing Google Video project. By 2009, the bet began to pay off; ad revenue tripled in a single year. By 2010 and 2011, YouTube turned profitable. Between 2013 and 2015, the product's North Star shifted entirely toward pure entertainment, aiming to keep users hooked for a solid 15 minutes. This shift was accelerated by the rise of mobile—where YouTube had been an initial launch partner on the iPhone—bringing in "low-intent sessions" where users opened the app simply to be served content rather than searching for a specific link.

- **2007 Bandwidth Metric:** YouTube consumed as much bandwidth as the entire global internet did in the year 2000. - **2014 Bandwidth Metric:** YouTube accounted for 20% of all bits on the internet. - **Financial Turnaround:** Ad revenue tripled in 2009; the platform became profitable around 2010–2011, and reached an estimated $4 billion in revenue by 2012. - **Core Metric Pivot:** YouTube shifted its primary internal metric away from raw video views to overall "watch time."

---

[01:17:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4657s) Creator Monetization, "Logged-Inness," and the Regrade

A crucial factor in YouTube's long-term dominance was its early, radical commitment to creator monetization through revenue sharing. Initially, tech analysts criticized YouTube for giving 50% of its ad revenue away to creators, pointing out that Google Search kept nearly 100% of its ad revenue and Meta platforms paid out close to zero to influencers. While this 50-ish percent split delayed YouTube's profitability by a decade, it created unparalleled incentives for individuals to build lifelong careers and businesses entirely on the platform. The hosts debate the internal ideological battle that ensued at YouTube over how to distribute these views: Camp One favored a subscription model where users curated their own creators, while Camp Two advocated for total reliance on algorithms. Ultimately, Camp Two won, giving rise to a "dark side" where a creator's views are only loosely correlated with their subscriber count.

This operational framework blended Google's two legacy ad models: the first-party AdWords system (where Google owns the media and keeps the revenue) and AdSense (where Google serves ads on third-party publishers and shares the majority of the payout). Furthermore, the rise of mobile introduced a critical structural shift: "logged-inness." Unlike Google Search, which can instantly monetize a user based on immediate intent (e.g., searching for a shovel), YouTube requires users to be logged in so its algorithms can build robust demographic and personalization profiles. This data loop replicates the highly lucrative television advertising ecosystem, which relies heavily on targeted demographics. Concluding this chapter, the hosts prepare to formally "regrade" the 2006 acquisition, noting that YouTube has completely defeated its two historical criticisms: it is now undeniably a destination app, and its massive revenue has thoroughly outrun its infrastructure costs.

> "YouTube is the ultimate instantiation of the internet to me and the power that it can provide to individuals to make a living. It abstracts the need at all to create or run a business. It really just simplifies it down to make content that people watch, you will get money for it."

- **The AdSplit Comparison:** YouTube gave away roughly 50% of ad revenue to content creators, compared to Facebook/Instagram which paid out close to 0% early on. - **The Ideological Conflict:** An internal battle between Camp 1 (user curation/subscriptions) and Camp 2 (algorithmic feeds). Camp 2 proved to be far more effective for user engagement. - **Logged-Inness:** An absolute requirement for YouTube’s advertising flywheel to match the targeted demographic model of traditional television advertising.

[01:23:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5008s) Unpacking YouTube's Financials and Engineering Feats

The hosts begin grading the YouTube acquisition by diving deep into its numbers. In 2024, YouTube generated $36 billion in advertising revenue alone. Because half of that goes directly to creators, Google is left with an $18 billion pool to manage variable costs. Over two decades, YouTube has relentlessly optimized video hosting, bandwidth, compute, encoding, and music licensing. They perform "insane feats of engineering," which include manufacturing their own custom silicon specifically for video encoding.

A fascinating optimization occurs dynamically based on an upload's performance. When a video is first uploaded, YouTube uses standard, less computationally intensive H.264 encoding. Once the video crosses certain view thresholds, it automatically switches to a format that is more expensive to encode but significantly smaller and cheaper to distribute. If a video hits massive scale—such as 5 million views—it is re-encoded yet again to shrink the file size even further, minimizing variable streaming costs on a per-view basis.

- **2024 YouTube Ad Revenue:** $36 billion (with $18 billion kept by Google after creator payouts). - **Total Revenue (with subscriptions):** Over $50 billion, factoring in YouTube Premium, YouTube Music, and NFL Sunday Ticket. - **Media Industry Ranking:** YouTube is the second largest media company by revenue globally, trailing only Disney. It has already eclipsed Netflix ($39 billion in 2024) and its media business is bigger than Disney’s standalone media segment.

[01:25:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5137s) The Ultimate Attention Sink and Regrading the Deal

While Google does not officially release exact usage metrics, the hosts assert that YouTube is almost certainly the largest single property on the internet by total human minutes spent. While Meta's Facebook Blue and WhatsApp claim more total users, YouTube dwarfs them in raw attention. "I think it is the biggest human attention time sink known to man," Ben remarks. To gauge profitability, they turn to a report from research firm MoffettNathansson, which estimates YouTube’s annual operating income at roughly $8 billion.

This sets up the definitive *Acquired* regrading of the deal. Factoring in the initial $1.7 billion purchase price and an estimated $5 billion in losses before reaching break-even within its first five years, Google spent a total of roughly $6.7 billion (and perhaps closer to $5.5–$6 billion) to acquire an asset generating $8 billion in pure profit annually while growing at 10% to 15% per year. Furthermore, the acquisition provided a "triple bottom line": massive revenue, high profits, and crucial strategic insulation against competitors.

- **Estimated Operating Income:** ~$8 billion per year (per MoffettNathansson). - **Standalone Valuation Estimate:** ~$500 billion if publicly traded, or a conservative $200 billion based purely on basic media revenue multiples. - **Grade Revision:** Both hosts enthusiastically upgrade the transaction from its historical "C" grade to a resounding "A+".

[01:27:54](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5274s) How Social Media Became YouTube

The hosts reflect on how Google famously "whiffed" on traditional social networking, only for the entire social landscape to shift directly into Google’s territory. In the mid-2000s and 2010s, social media meant checking in on what personal friends were doing—a space Google missed. However, that era is dead. Social media has bifurcated into private messaging on one end and public user-generated media on the other.

Today, users open platforms like Instagram Reels or TikTok not to see friends, but to watch entertaining videos from strangers. Because user behavior pivoted into video consumption, social media effectively became YouTube. Beyond short-form dominance, this massive repository gives Google an unprecedented strategic advantage: it functions as the largest corpus of video data in the world to train artificial intelligence.

[01:30:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5416s) A Chronology of Testing Culture and Intro to DoubleClick

Shifting gears back toward Google’s core web advertising business, David steers the conversation toward their original *raison d'être*: analyzing history's greatest acquisitions. Before diving into their next big case study, they highlight their sponsor, Statsig, noting how Google pioneered data-driven product development. Google famously optimized its search business by A/B testing 50 different shades of blue for search result hyperlinks, bootstrapping its "did you mean" feature from user query corrections, and testing AI search layouts. Statsig now democratizes these exact product-testing frameworks for modern tech giants and startups alike.

Turning to their next corporate milestone, the hosts note that just months after the October 2006 YouTube deal, Google essentially doubled down. In April 2007, Google agreed to acquire DoubleClick for $3.1 billion in pure cash—a stark contrast to the stock-based YouTube deal. At the time, Google had two primary ad engines: AdWords for search result links, and AdSense, a DIY, self-serve ad network for off-property websites. DoubleClick, however, represented an entirely different beast with a long, volatile history on Madison Avenue.

- **DoubleClick Acquisition Date:** April 2007. - **Purchase Price:** $3.1 billion in cash.

[01:34:02](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5642s) The Tumultuous History of DoubleClick

Founded in New York City in 1995 by Kevin O'Connor and Dwight Merriman, DoubleClick was an early titan of the dot-com boom. Their core business model was two-pronged: building an "ad server" (software enabling advertisers to deploy ads across various websites) and managing a broader network to run paid media. They went public in 1998 as a crown jewel of the internet era. But when the dot-com crash hit, 70% of DoubleClick’s customers didn't just churn—they went entirely bankrupt, as the vast majority of their advertisers were VC-backed startups like Pets.com.

Limping along, DoubleClick sold off its ad network division in 2002 for less than $15 million, leaving only its ad-serving software. By 2004, it was a sleepy, slow-growth firm with a dwindling market cap. Google initially passed on buying them, leading to a 2005 buyout by private equity firms Hellman & Friedman and JMI Management for roughly $1 billion.

- **Founding Year:** 1995 in New York City. - **IPO Year:** 1998. - **Private Equity Buyout (2005):** Purchased for $1 billion (significantly lower than its peak public market value).

[01:36:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5795s) The Birth of the Ad Exchange

Under private equity ownership, new CEO David Rosenblatt and head of product and strategy Neal Mohan—who serves as the CEO of YouTube today—engineered a massive pivot. They invented and launched a completely new paradigm: the "ad exchange." Before this, digital advertising relied on rigid ad networks and basic ad servers. The ad exchange introduced a sophisticated real-time marketplace that could automatically route demand across *competing* ad networks.

Initially designed to monetize "remnant" or unsold page inventory, the exchange utilized a highly complex rules engine. It allowed real-time programmatic bidding that could even displace premium, direct-sold publisher deals if an outside bidder was willing to pay a higher rate. This programmatic mechanism effectively bypassed and disintermediated standard ad networks. While Google's AdSense was a tech-utopian, self-serve system built for small publishers, DoubleClick’s new exchange embraced the complex business realities and relationships of traditional Madison Avenue agencies.

[01:39:20](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5960s) Google's Ad Infrastructure Deficiencies and the DoubleClick Opportunity

By the early 2000s, there was a pressing need to construct "fat pipes for money to flow through," which meant creating direct integrations into the financial systems of major advertising agencies. Because these agencies controlled the marketing budgets for all the world's biggest brands, unlocking these deeply integrated revenue streams was critical. However, Google faced several philosophical and structural roadblocks. For instance, DoubleClick performed advanced enterprise-grade functions like frequency capping to ensure a user wouldn't see the exact same ad 46 times, relying heavily on third-party cookies. Google was philosophically opposed to using third-party cookies at the time, which kept them from executing similar capabilities.

Furthermore, Google lacked deep corporate sales relationships due to their intense corporate obsession with self-serve web pages, where advertisers simply logged in, uploaded creative assets, and transacted on their own. This structurally locked Google out of premium ad inventory. While their search ads were wildly successful, their display network, AdSense, was restricted to the "long tail" of websites, which major advertisers were naturally willing to pay far less to appear on. Recognizing these deficiencies, Google decided to pursue an acquisition of DoubleClick to secure an enterprise-grade solution.

[01:40:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6040s) The Seattle Alarm and the Race Against Microsoft

While the public narrative framed the acquisition as a standard corporate purchase, the reality unfolded as a high-stakes corporate thriller. Tim Armstrong, Google’s longtime head of sales, was looking to meet with the DoubleClick team in late 2006 or early 2007. While emailing them, he mentioned he was going to be in Seattle, and the DoubleClick executives surprisingly replied that they were currently in Seattle as well. This immediately raised a massive red flag for Armstrong. DoubleClick was a New York-based company; there was only one logical reason for its leadership team to be in Seattle, and that was a pending acquisition by Microsoft.

Armstrong immediately sounded the alarm inside Google, contacting CEO Eric Schmidt and co-founder Larry Page. Google's overarching strategic goal was to prevent Microsoft from leveraging its dominance in Windows and Internet Explorer to kneecap Google’s web apps. While Google had neutralized the browser threat through its Web 2.0 strategy, the new emergency was that Microsoft was finally waking up to the threat, moving to build its own search engine, and shifting toward an ad-based business model. Historically, Microsoft’s DNA was firmly rooted in selling software, looking down on advertising as "dirty ads" that they wouldn't dare trade their clean software margins for. However, Microsoft realized that Google was making more money on a per-user basis on any given PC than Microsoft was, forcing Redmond to enter the game.

[01:42:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6169s) The Conference Room Ambush and the Bidding War

The acquisition process quickly devolved into a chaotic scramble. Tim Armstrong was invited to present to DoubleClick, believing the negotiations were still in an early stage. However, an escort in the DoubleClick building accidentally sent Armstrong to the wrong floor. When the elevator doors opened, staff panicked and tried to force the doors closed, telling him to go to a different floor. Sensing something was amiss, Armstrong stepped out anyway, ran down the hallway, and peered into a conference room. Inside, he saw it packed with Microsoft executives, lawyers, and accountants. Realizing they were on the verge of signing a deal, Armstrong begged DoubleClick to halt the process so Google could perform diligence and submit a counter-bid.

A frantic, multi-party bidding war ensued involving Google, Microsoft, Yahoo, and AOL. DoubleClick managed the process using a tailored spreadsheet titled `YMAG.XLS` (representing Yahoo, Microsoft, AOL, and Google). In their pitches, DoubleClick would subtly tweak the numbers in the spreadsheet for each company to demonstrate exactly how much incremental money they would generate by integrating DoubleClick into their respective ad systems.

- Google submitted a formal Letter of Intent (LOI) for $3.1 billion, which included an exclusivity clause preventing DoubleClick from shopping the deal during the diligence period. - The entire Google team relocated to New York, renting a massive war room in a hotel near DoubleClick's offices. - According to an excerpt from the book *Yield*, DoubleClick’s corporate counsel received a last-minute Blackberry message from Microsoft's corporate development team offering to match Google's $3.1 billion bid. - The message included an email from Steve Ballmer stating he was open to a much higher offer, effectively writing a blank check by telling DoubleClick to "simply mark up the paper to meet its needs and then sign it."

[01:45:08](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6308s) The "Hell or High Water" Clause and Closing the Deal

Despite Ballmer's blank check, DoubleClick was locked into the exclusivity window and could not legally respond to Microsoft. A day before the LOI with Google was set to expire, Google delivered an updated term sheet. While the headline purchase price remained unchanged at $3.1 billion, Google added a definitive "hell or high water clause." This meant Google committed to closing the transaction immediately without any further substantive diligence or closing conditions, rendering it "money in the bank." DoubleClick signed the deal immediately. The transaction was an extraordinary win for the private equity firm backing DoubleClick, which successfully turned a leveraged $1 billion investment—consisting of roughly $300 million in equity and $700 million in debt—into a clean $3.1 billion cash sale.

- The acquisition successfully established the "fat money pipes" Google needed to capture massive ad agency budgets. - Because DoubleClick was the undisputed number one player in the space, Microsoft panicked and bought the number two player, a public company called aQuantive, just one month later. - Microsoft paid $6 billion for aQuantive—twice what Google paid for DoubleClick—yet acquiring the second-place player significantly slowed Microsoft's advertising and search timeline, which eventually led to Yahoo partnerships and Bing. - For Google, keeping DoubleClick out of Microsoft's hands made the deal "worth every penny."

[01:46:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6417s) A Pure Business Chess Piece to Defend the Realm

Unlike foundational Google products like Maps, Gmail, or YouTube, the DoubleClick acquisition had absolutely nothing to do with Google's core mission to "organize the world's information and make it universally accessible." This was a cold, calculated move to expand and protect corporate ad revenue. It was a defensive chess piece on the table where ownership of the asset was valuable simply to deny it to competitors. Because of this, Google executives never highlighted DoubleClick on stage when delivering inspiring public speeches about the idealistic future of the company.

Even looking at data from 2024, DoubleClick did not scale into a consumer-dominating force like YouTube. Out of Google's $350 billion in total revenue in 2024, search accounted for $200 billion, while the Google Network segment (where DoubleClick sits) brought in $30 billion. While some argue that AdSense—which was already clearing over $1 billion around the time of the acquisition—would have scaled anyway, DoubleClick was critical to making Google the dominant force in display advertising. Furthermore, the economics of the two segments are drastically different: Google retains roughly 90% of its search revenue after traffic acquisition costs (TAC), but pays out 70% of its Network revenue to publishers. When looking at gross profit, the Network contributes roughly $9 billion compared to search's $175 billion, making DoubleClick relatively inconsequential to the overall financial narrative of modern Google.

[01:49:25](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6565s) The Mid-2000s Search Product Juggernaut and the Rise of AI Parallels

While the advertising battles raged on the sidelines, Google's core search engine was silently "cranking on improvements," driving explosive financial growth between 2003 and 2008. Google began updating its search index far more frequently; whereas users previously searched against data indexed three months prior, the web began to feel truly recent and real-time. During this era, Google rolled out a rapid succession of key verticals and features: * Google Images, Google News, Google Books, and Google Scholar. * Google Suggest, which introduced search autocompletion. * Google Instant (a features since retired), which dynamically updated the entire search results page in real time with every single keystroke. * Personalization features in 2005, which integrated a logged-in user's search history into their results.

In 2007, Google achieved a massive breakthrough with the launch of "universal search." Instead of building entirely isolated search engines for text, images, and video and forcing users to navigate between them manually, universal search deconstructed a single query to understand user intent and blended the media types into one page. Financially, the growth curve was staggering: revenue jumped from $3 billion in 2004 to $6 billion in 2005, and reached $16.5 billion in 2007. This milestone made Google the largest seller of advertisements of any kind in the world, eclipsing all traditional media players individually, even though aggregate digital ad spend wouldn't overtake traditional media as a whole until 2018.

> "Every year for the last 18 years, Google has been the number one seller of advertising of any kind in the world."

This historical dominance underscores exactly what is at stake today in the era of AI. The modern tech landscape faces a multi-trillion-dollar question: can Google maintain its crown as the world's premier advertising destination through this massive technological sea change? There are clear, direct historical corollaries between the current AI wave and the mobile wave Google navigated shortly after this era.

[01:52:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6731s) Sucking in Generational Talents: Google's Golden Engineering Era

Following the mid-2000s boom, Google continued executing technical breakthroughs, achieving true real-time web indexing by 2009 and launching the Knowledge Graph in 2012 to supply snapshot sidebar views of entities with Wikipedia pages. Throughout this timeline, engineers constantly tweaked algorithms to combat search spam. By 2008, Google had firmly solidified its status as the world's preeminent computer science research hub. To call someone a "Google-type engineer" was the ultimate industry compliment. Google had successfully seized the technical mantle from Microsoft, holding onto it tightly before eventually relinquishing that dense concentration of elite talent to companies like Facebook, Stripe, OpenAI, or Anthropic in later eras.

A huge driver of this cultural and technical dominance was Google's aggressive poaching of legendary generational talent from collapsing corporate research labs. When Digital Equipment Corporation (DEC) faltered, Google secured the legendary duo of Jeff Dean and Sanjay Ghemawat. While Jeff Dean frequently transitioned into high-profile executive roles, the two famously operated as a peer-programming partnership. An iconic *New Yorker* article documented their deep career partnership, and from 2002 through the 2010s, nearly every seminal research paper regarding Google's massive infrastructure listed Jeff and Sanjay as either the sole two authors or part of a small five-person cohort. As other pillars like Bell Labs, Xerox PARC, and IBM's labs lost prominence, Google systematically absorbed their heavy hitters, including Bill Coughran and Rob Pike from Bell Labs, transforming the company into a magnet for the world's finest systems and architecture programmers.

[01:54:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6874s) Poking the Giant and Funding Mozilla

The year 2008 required Google to pull off two massive, simultaneous engineering feats that would define the next era of tech: launching their own web browser and their own mobile operating system. Building a browser from scratch back then was an earthshattering engineering undertaking on par with Dave Cutler building Windows NT, or the early work of Jeff Dean and Sanjay Ghemawat. Unlike today's landscape, where companies spin up new browsers using Google’s open-source Chromium engine for free, Google had no blueprint; they had to invent the foundation themselves.

The defensive urgency spiked in February 2008 when Microsoft finally woke up and made a $44 billion bid to buy Yahoo to aggressively enter the search business. Fortunately for Google, Yahoo’s Jerry Yang turned down the acquisition in what is remembered as "one of the worst corporate decisions of all time." Just two years later, Microsoft’s Bing would power Yahoo search anyway for a mere $1 billion, and Yahoo eventually sold itself to Verizon for single-digit billions.

Though spared a combined Microsoft-Yahoo juggernaut, Google had long anticipated this threat. In fact, Larry Page and Sergey Brin wanted to build a web browser as early as 2001. It was Eric Schmidt who put his foot down to avoid poking the bear, famously saying, "I don't want to moon the giant" while Google was still young and vulnerable. Instead of building a browser openly, Google covertly laid the groundwork by becoming the primary financial benefactor of the new Mozilla Foundation—which spun out of Netscape after the AOL acquisition—and its new Firefox browser. They funded Mozilla initially through grants, and later via massive traffic acquisition costs (TAC) to remain the default search engine, establishing a long tradition of buying distribution.

- **Microsoft's Yahoo Bid:** $44 billion offered in February 2008, famously rejected by Jerry Yang. - **The Bing-Yahoo Deal:** In June 2009, Bing took over Yahoo's search backend for $1 billion. - **Eric Schmidt's 2001 Policy:** Refused to build a browser immediately to avoid retaliation from Microsoft.

---

[01:59:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7190s) The "Sleeper Cell" and the Rise of Sundar Pichai

Over a couple of years of funding Mozilla, Google devised a clever talent strategy. Realizing that a nonprofit couldn't offer lucrative stock options, Google began hiring key Firefox engineers directly. They pitched this to Mozilla as a win-win: the engineers would stay well-compensated and protected from poaching by other tech giants, but they would remain dedicated to working on Firefox from inside Google.

In 2004, Google hired a former McKinsey consultant named Sundar Pichai to lead this new "product client group," which focused on installed desktop applications rather than Google's typical web apps. This group sat inside the company for years like a "latent sleeper cell," quietly accumulating browser talent. While the engineers advanced web development by working on tools like Google Gears (for offline capability) and the Google Web Toolkit, they ultimately began losing faith that Firefox could maintain the pace and technical architecture Google required. By 2006, with Microsoft's Bing threat looming, Google officially greenlit the development of its own browser, Chrome. If 90% of searches occurred on Internet Explorer and Microsoft switched the default engine to Bing, Google would have been locked out; they desperately needed their own distribution channel.

- **Sundar Pichai's Hiring:** Joined Google from McKinsey in 2004 to lead the client applications team. - **Google Gears:** A browser extension developed to introduce offline functionality to web applications. - **The Chrome Project Inception:** Official development on Chrome began in 2006.

---

[02:03:58](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7438s) The Six Killer Features of Chrome

When designing the new browser, Google baked in a series of major structural and user-experience breakthroughs. The hosts break these down into six key pillars that set Chrome apart from Internet Explorer and Firefox:

1. **The V8 JavaScript Engine:** A super fast, highly modern virtual machine tailored to run heavy, complex Ajax web applications smoothly. As the self-proclaimed "Ajax company," Google needed to drastically accelerate the web. 2. **Multi-Process Architecture:** Previously, if a web app crashed, it brought down the entire browser window and every open tab. Chrome isolated each tab into its own separate process. 3. **WebKit Rendering Engine:** Borrowed from an Apple innovation, it provided the best rendering architecture available at the time. 4. **Minimal UI Design:** The name "Chrome" historically referred to the real estate around a webpage (toolbars, navigation bars). Google ironically named their browser Chrome because they wanted to feature *minimal* chrome, letting the web applications themselves take center stage. 5. **The Omnibox:** Google merged the separate URL and search bars into a single input field. From a business perspective, if a user didn't type an exact URL, it seamlessly kicked off a Google search, driving massive ad traffic. From a UX perspective, it fixed an accident of history. "Consumers never should have known the phrase HTTP," they notes; they should have just typed "New York Times." 6. **Sandboxing:** Isolating each tab inside its own secure sandbox prevented malicious web code from accessing the underlying PC operating system, resolving a massive contemporary computer security threat.

> "We have found the web-based service delivery model to have significant advantages... [but it is] unable to function offline." — *Larry Page, outlining the browser's philosophy.*

---

[02:08:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7732s) Launching via Comic Book

Google officially launched Chrome in early September 2008, just a week before Lehman Brothers collapsed and the global financial crisis struck. True to Google's quirky culture, they bypassed traditional marketing and hired legendary comic artist Scott McCloud to illustrate a digital comic book that served as an introduction and user manual.

The comic was not targeted at the general public; it was explicitly written for a "weird half-user" who was highly tech-enthusiast but not necessarily a programmer—the classic *Slashdot* reader. It explained advanced concepts like process independence, sandboxing, and the V8 engine. This was a brilliant seeding strategy. These tech-savvy early adopters went home for Thanksgiving a couple of months later and immediately installed Chrome on their families' computers to protect them from malware and phase out Internet Explorer. The viral, grassroots strategy worked spectacularly: within 18 months of launch, Chrome amassed 40 million users.

[02:10:42](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7842s) The Understatement of the Century: Chrome’s Market Dominance

When Google launched Chrome in 2008, it completely rewrote the rules of the browser wars. By 2010, the platform had surged to 70 million users, ballooning to 200 million by 2012. This explosive growth initially decimated Firefox's market share, as Chrome’s launch perfectly coincided with Firefox’s historical peak. After swallowing Firefox’s piece of the pie, Chrome began aggressively eating away at Microsoft’s Internet Explorer (IE). To say this defensive play worked is "the understatement of the century"—it completely liberated Google from the ecosystem overrule of Microsoft.

The shift in market power happened at a staggering velocity. In 2008, Internet Explorer commanded nearly 70% of the browser market, with Firefox holding most of the remainder. Just four years later, in 2012, Chrome and IE were locked in a dead heat with roughly 30% market share each. This period also overlapped with the rapid rise of mobile; Apple's mobile Safari was exploding, and Google's own Android was beginning to scale. By 2014, Chrome became the undisputed leader at 40% market share while IE collapsed to a mere 15%. In modern times, the war is effectively over. According to Cloudflare, Chrome hovers around 70% global market share. Safari sits at roughly 20%—bolstered primarily by its default status on iPhones—leaving a tiny 10% slice of the market for everyone else, including Microsoft's single-digit remnants.

- **2008:** Google Chrome launches against IE's ~70% market share. - **2010:** Chrome reaches 70 million users (and crosses 100 million shortly after). - **2012:** Chrome and IE tie at roughly 30% market share each. - **2014:** Chrome hits 40% market share; IE drops to 15%. - **Modern Era:** Chrome commands ~70% market share, Safari holds ~20%, and Microsoft is left in the single digits.

[02:12:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7977s) Keeping the Web Alive: Incentives and the Chrome Frame

Chrome's wild success wasn't just a triumph of distribution; it was simply a vastly superior product. Its introduction kicked off an incredible era of innovation on the web, forcing Apple into a competitive game of leapfrog where Safari and Chrome constantly traded titles for the world's fastest browser. This rivalry was a dramatic shift from the past. Originally, Steve Jobs had pushed for Safari's creation because he loathed being forced to ship Internet Explorer as the default Mac browser—a legacy requirement from the famous deal where Microsoft invested to save a struggling Apple.

However, as the iPhone era took off, Apple's corporate incentives shifted away from making web applications great. Apple wanted native mobile and desktop applications to thrive, while Microsoft had zero business incentive to push web capabilities forward. Google stepped in to fiercely champion the open web, driving a "good for the world race." Without Chrome pushing the boundaries of what a modern browser could achieve, the web may have stalled out as a viable application platform entirely.

To bypass IT restrictions on locked-down corporate PCs where users couldn't install new software, Google deployed a "delicious" piece of engineering: the Google Chrome Frame. This plugin slipped inside Internet Explorer, completely replacing IE's sluggish JavaScript engine with Chrome's blazing-fast V8 engine and utilizing WebKit for rendering. For millions of corporate workers stuck using ancient versions of IE, Google quietly provided the power to run high-quality web applications right inside Microsoft's own shell.

> "I don't think it is an exaggeration to say that Chrome kept the web alive. As a viable platform for applications."

[02:15:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8128s) The Open-Source Strategy and the Disney World Analogy

The decision to open-source the browser project as Chromium stems deeply from Google's foundational culture of wanting to contribute to the ecosystem and "not to be like too evil about it." Strategically, keeping it open-source was highly logical because Google ultimately monetizes through search engine volume rather than software licenses. While a competitor building a better browser using Chromium might theoretically force Google to pay traffic acquisition costs through revenue sharing, in practice, fragmentation was a massive win as long as Microsoft didn't control the gateway. With an incremental gross margin on search revenue sitting near 98%, slipping down to an 87% gross margin from light revenue sharing was a perfectly acceptable worst-case scenario to neutralize an existential platform threat.

To conceptualize Google's broader strategy with the web, a comparison can be made to Walt Disney’s expansion from California to Florida. When Walt Disney built Disneyland, he realized he had to play by the frustrating rules of local city governments. When he bought a massive plot of land for Walt Disney World, he purposefully established an independent government district around the park so Disney could make the rules.

Similarly, in the PC era, Microsoft owned Windows and dominated the technology landscape. Google, operating a free search engine, didn't own the underlying platform and existed at Microsoft's pleasure. By heavily investing in web standards and Chrome, Google engaged in an ecosystem-building exercise. They stepped in as the self-appointed stewards of the open web, putting a finger on the scale to ensure their "neighborhood" remained a healthy, thriving environment for their core business.

[02:20:12](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8412s) Counterfactuals, DOJ Realities, and Independent Hits

A quick look at what might have happened without Chrome reveals just how critical the browser was to Google's survival. If Microsoft had launched Bing in 2009 into a landscape where they still commanded 70% of the browser market, they could have defaulted millions of users to Bing for a critical four-year window before mobile reached maturity. Even though users preferred Google, defaults are incredibly powerful levers—which explains why Google pays Apple upwards of $20 billion a year to maintain its default search status. Without Chrome acting as a critical distribution shield, Bing could have become a truly formidable threat.

This reality heavily colors the modern Department of Justice antitrust proceedings regarding a potential divestiture of Chrome. If the DOJ forces Google to sell Chrome and bans search traffic acquisition payments, Chrome loses its entire underlying business model. Whether owned by Google, a traditional search company, or a modern AI company, a browser fundamentally relies on driving traffic to a search backend to monetize.

Ultimately, Chrome's rise proved that Google could create runaway successes without merely relying on legacy distribution. While they did cross-promote Chrome on their search homepage, its dominance—much like Gmail, Google Docs, and Google Sheets—was driven by viral, word-of-mouth adoption because it was a genuinely fantastic, free product. Of course, these massive undertakings were heavily subsidized in their infancy by the "old money printing machine in the basement of Google," affectionately referred to as Uncle AdWords. This streak of legendary product wins set the stage for Sundar Pichai's eventual ascension to CEO and prepared Google for its biggest operating system play yet: Android, which now powers over three billion active devices worldwide.

[02:24:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8689s) Sponsor Spotlight: V0 by Vercel

As Google's history shows, web development paradigms are constantly shifting, and Vercel has emerged as the infrastructure backbone for modern web and AI development. Powering major digital platforms like PayPal, Ramp, Under Armour, Notion, Runway, and Cursor, Vercel is pushing boundaries even further with V0, their cutting-edge AI app builder.

Operating through a clean, familiar chatbot interface similar to Gemini, Claude, or ChatGPT, V0 allows anyone to program, design, iterate, and instantly deploy full-stack web applications purely through text prompts. No coding knowledge or wireframing skills are required. Marketers can instantly spin up a product landing page, creators can build independent content hubs, and small businesses can generate operational homepages complete with contact forms.

- **User Base:** V0 has quickly scaled to over 4 million users, including product managers, founders, and students. - **Production-Ready:** Every application generated by V0 runs directly on Vercel's enterprise-grade cloud, ensuring secure, automatic scaling out of the box. - **Ecosystem Philosophy:** Vercel acts as "customer zero" for its own products, leveraging its native AI cloud infrastructure to power its own consumer-facing AI developer tools.

[02:26:47](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8807s) The Lineage of 165 University Avenue

The physical spaces occupied by Google throughout its history have achieved a legendary status in Silicon Valley. While the company's journey famously began in Susan Wojcicki's Mountain View garage before eventually expanding into the sprawling Googleplex—the former Silicon Graphics (SGI) campus—there was a critical intermediate step. For a couple of years, Google operated out of a building located at 165 University Avenue in downtown Palo Alto. This exact location later served as the birthplace for PayPal, earning it a reputation as an exceptionally lucky building within the tech ecosystem.

When Google moved out of the University Avenue office in August 1999, the space was immediately occupied by Danger, a startup founded by former Apple engineer Andy Rubin. Rubin had previously left Apple with a group of self-described rebels to build General Magic, a legendary but ill-fated 1990s startup that essentially attempted to create the iPhone 15 years too early. Following the collapse of General Magic, Rubin established Danger with the initial, surprisingly modest ambition of building a wireless version of the CueCat scanner—a cat-shaped, computer-plugged device used for scanning barcodes. Rubin figured that since General Magic had been too far ahead of its time, a simpler barcode-scanning concept would be safer. However, Danger's very first employee, Hiroshi Lockheimer, convinced Rubin to think bigger and revisit the ambitious mobile concepts of General Magic.

- **165 University Avenue:** The historic Palo Alto office space that housed Google, Danger, and later PayPal. - **General Magic:** A famous 1990s Silicon Valley failure that attempted to build smartphone technology long before the market was ready. - **The CueCat Vision:** Andy Rubin's original hardware concept for Danger before being persuaded to pivot back to mobile operating systems.

[02:29:42](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8982s) The Rise of the Sidekick and the Pivot to Android

The decision to revisit mobile software led Danger to develop the Sidekick, a revolutionary device launched in close partnership with T-Mobile. The Sidekick became a massive cultural phenomenon, heavily embraced by celebrities and even serving as a central plot point in an episode of *Entourage*. Alongside the BlackBerry, it represented the earliest realization of a cell phone where the primary use case shifted entirely from voice calls to messaging and rich applications. Microsoft would eventually acquire Danger in 2008, but by that time, Rubin had already moved on. He left the company in 2003 to start a brand-new venture called Android.

In its absolute infancy, Android was not designed for cellular phones at all. Instead, Rubin’s vision was to build a powerful, cross-platform, open-source operating system for point-and-shoot digital cameras, which were owned by hundreds of millions of consumers at the time. Rubin viewed digital cameras as a potential Trojan horse to establish a dominant operating system. The team quickly realized, however, that the market was moving in the opposite direction—phones were rapidly absorbing the functionality of cameras, rather than cameras evolving into phones. Because the software they were writing was highly versatile, Rubin pivoted the startup's delivery vector from digital cameras to smartphones.

- **Hiroshi Lockheimer:** Danger’s first employee, who later became a core architect of Android and Chrome at Google. - **The Sidekick:** A messaging-focused smartphone that pioneered the shift away from voice-centric mobile usage. - **Android's Camera Origins:** The operating system was initially pitched as an open-source platform for digital cameras before pivoting to phones.

[02:31:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9100s) Disrupting a Stagnant Smartphone Market

When Android pivoted to smartphones, the existing market landscape was strictly divided. On one side were fully integrated, full-stack players like Nokia and BlackBerry, who manufactured their own hardware and wrote their own proprietary software. BlackBerry dominated the enterprise sector with this model. On the other side were original equipment manufacturers (OEMs) that built hardware but licensed their software from third parties like Palm or, most notably, Microsoft with Windows Mobile. Under Microsoft's desktop-style licensing model, OEMs paid a single-digit dollar fee per device to use Windows Mobile and built their custom phone interfaces on top of it.

This status quo created an environment where everyone involved was "fat and happy." Carriers and OEMs were making immense profit margins because the phones of the era were relatively basic and cheap to manufacture, yet consumers were paying upwards of $100 a month on long-term carrier contracts. Because the devices lacked advanced capabilities, they consumed very little cellular data, keeping network infrastructure costs low for carriers. Into this stagnant market stepped Andy Rubin, pitching a disruptive proposition to carriers and hardware manufacturers: stop paying Microsoft or Palm for software, and accept a highly powerful, open-source operating system for free.

- **Windows Mobile Licensing:** Microsoft's mobile strategy relied on charging OEMs single-digit fees per phone, mimicking their successful PC desktop business model. - **Carrier Economics:** Pre-iPhone smartphone contracts cost consumers roughly $100 per month while consuming negligible amounts of data. - **The Free Software Counter-Signal:** For an independent startup, offering software for free initially made Android look desperate to major telecom companies.

[02:34:09](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9249s) Google's $50 Million "Kick in the Pants"

Despite Rubin's compelling pitch, Android faced immense friction as an independent startup. Major telecommunications giants like AT&T and Verizon refused to gamble billions of dollars of infrastructure on a rinky-dink startup with only eight employees and a $10 million valuation. By 2005, Rubin had managed to convince Taiwanese manufacturer HTC to collaborate on a hardware prototype, but the company was rapidly running out of money. While hunting for a critical investment round, Rubin met with Larry Page. Page immediately bypassed the idea of a venture round and offered to buy the company outright. In July 2005, Google acquired Android for $50 million.

While the $50 million price tag is often celebrated as one of the greatest steals in tech history, it obscures the reality that Google subsequently poured billions of dollars into its development. Android was effectively built from the ground up as an in-house Google project, utilizing the startup primarily as a "kick in the pants" to force immediate action. Google’s leadership team—Eric Schmidt, Larry Page, and Sergey Brin—fully recognized that they were dangerously late to the mobile transition. BlackBerry was seeing massive consumer adoption, and Windows Mobile proved that a vibrant software market existed. Furthermore, Google could see from its own internal search and map traffic that mobile data demand was real. Managing over 350 different mobile versions of Google Maps for a fragmented sea of feature phones had become a logistical nightmare, confirming that a unified mobile operating system district was desperately needed.

- **The Acquisition:** Google purchased Android in July 2005 for $50 million, absorbing its eight-person team. - **Fragmentation Strain:** Google was forced to maintain 350 independent versions of Google Maps to support the fragmented mobile market of 2005. - **The Alternative Timeline:** Without the acquisition of Android, the hosts argue the non-Apple smartphone market would have likely been entirely consolidated by Microsoft's licensing model.

[02:39:04](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9544s) The iPhone Keynote and the Death of "Sooner"

Following the acquisition in the summer of 2005, Google split its Android development into two parallel paths. The near-term project was a non-touchscreen, BlackBerry-like device with a physical keyboard, code-named "Sooner." The long-term, advanced research project was code-named "Dream"—a highly ambitious touchscreen smartphone. By the summer of 2006, Eric Schmidt had joined Apple’s board of directors, gaining an up-close look at the secret development of the iPhone. When Steve Jobs publicly unveiled the iPhone in January 2007, he even invited Schmidt on stage to celebrate the close alliance between the two companies.

At the time of the keynote, Android had not yet been announced, and the public was entirely unaware of its existence. Jobs knew that Google was working on a mobile project, but it is believed Schmidt had only disclosed the BlackBerry-style "Sooner" prototype, keeping the touchscreen "Dream" project hidden. During the presentation, Schmidt awkwardly joked that Apple and Google were so close they should merge into a single company called "AppleGoo." Watching old footage of the event reveals a visible tension; Jobs smiles politely for the audience but clearly displays what can only be described as "the ick."

The first iPhone shipped in July 2007 without an App Store or support for third-party developers, meaning it relied entirely on built-in applications. Of the small handful of default apps on the home screen, two of the most critical—Maps and YouTube—were powered by Google's back-end APIs, though the interfaces and icons were designed by Apple. The original YouTube icon, featuring a nostalgic wood-grained, old-school CRT television, was deeply despised by the internal YouTube team at Google. They recognized that Apple omitted their official logo and felt that an old-school TV icon was completely antithetical to YouTube's mission of inventing the future of digital video. Behind the scenes, the Google mobile teams building the infrastructure for these iOS apps were led by Vic Gundotra. Meanwhile, the core Android team watched the iPhone keynote in awe, realizing instantly that their physical-keyboard "Sooner" prototype was entirely obsolete, prompting them to throw it straight into the trash.

- **Project Sooner vs. Project Dream:** Google's dual-track strategy featuring a physical-keyboard design (Sooner) and a touchscreen design (Dream). - **"AppleGoo":** Eric Schmidt's onstage joke during the 2007 iPhone keynote regarding a hypothetical merger between Apple and Google. - **Google's iOS Presence:** Despite the brewing corporate rivalry, Google's services made up a significant portion of the original iPhone's native applications, managed on the back-end by Vic Gundotra.

[02:43:32](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9812s) Steve Jobs Goes Thermonuclear

The moment Steve Jobs discovered what Google's Android team was secretly developing, he went "ballistic"—or, to use his own historic choice of words, "thermonuclear." The timeline is slightly complicated by Eric Schmidt remaining on the Apple board until 2009, but Jobs' fury leaked to the press during an internal Apple all-hands meeting. He bluntly told employees: "We did not enter the search business. They entered the phone business. Make no mistake, Google wants to kill the iPhone. We won't let them." To this day, Apple has largely refrained from competing in Google’s core search business, content to simply pocket a massive "spiff" of revenue from the traffic they route to Google, allowing them to maintain their privacy-focused branding without getting their hands dirty with data sharing.

According to Walter Isaacson's biography, Jobs vowed to spend his last dying breath and every penny of Apple's $40 billion bank account to destroy Android, calling it a "stolen product." Despite the rage, a reasonable counterargument exists that both tech giants took inspiration from one another. While Apple popularized multi-touch, precursor companies had experimented with it earlier, and Android independently arrived at similar technical conclusions. However, multi-touch became the ultimate legal battleground. Jobs threatened massive patent lawsuits, forcing Google to cripple early Android models. For the first few years, Android phones completely lacked intuitive features like pinch-to-zoom or swipe UI navigation gestures, relying instead on four prominent physical navigation buttons at the bottom of the device.

- **Apple's Cash Reserves:** Steve Jobs claimed he would spend all $40 billion of Apple's cash to fight Android. - **Android Constraints:** Due to patent threats, early Android versions lacked pinch-to-zoom and swipe navigation, mandating four physical buttons. - **Android Innovations:** Features that Apple later copied from early Android include the App Marketplace (Play Store), a pull-down notification center, and the ability to drag and rearrange home screen apps.

---

[02:47:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10034s) The Launch of the HTC Dream (T-Mobile G1)

In November 2007, roughly ten months after the iPhone's reveal, Google announced the formation of the Open Handset Alliance—a confusing mega-partnership with HTC, Motorola, Samsung, LG, T-Mobile, Sprint, Qualcomm, Intel, Broadcom, and Texas Instruments. The industry spent a year wondering if Google was actually going to build a physical phone. The answer arrived in the momentous month of September 2008—the exact same month Chrome launched and Lehman Brothers collapsed. Google unveiled the T-Mobile G1, manufactured by HTC and internally named the "HTC Dream." One of the hosts recalls coding a Java application for the G1 during a college capstone project alongside a teammate who later went on to found the startup Daily Booth.

The G1 was a fascinating, somewhat clunky hardware hybrid. It paired a touchscreen with physical navigation buttons and featured a slide-out horizontal QWERTY keyboard reminiscent of the T-Mobile Sidekick. Crucially, it boasted multitasking and third-party applications out of the box. The launch event itself was held in a commercial kitchen in New York City and was so haphazard that video footage of it is nearly impossible to find today. In a hilariously iconic tech moment, Larry Page and Sergey Brin showed up and literally rollerbladed onto the stage, navigating around T-Mobile and HTC executives clad in formal suits. Despite the chaotic rollout, the G1 was a respectable success, selling over a million units in the US and capturing a 6% smartphone market share, placing it on par with the entirety of Palm.

- **iPhone Market Dominance:** While the G1 sold 1 million units over its lifetime, Apple sold 11 million iPhones in 2008 and 20 million in 2009, claiming over 50% of the early smartphone market. - **iPhone's Early Flaws:** The original iPhone lacked copy-paste, multitasking, wallpaper customization, alternative app distribution, 3G data speeds (initially running on the incredibly slow AT&T Edge network), and was entirely locked to AT&T in the US.

---

[02:51:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10317s) Holiday 2009: "Droid Does" and the Turning Point

Everything changed during the holiday season of 2009 with the launch of the Motorola Droid. When the hosts previously interviewed former Microsoft CEO Steve Ballmer, he specifically cited Christmas 2009 as the exact moment Android won the market. AT&T had spent two years pummeling Verizon by stealing their highest-paying, data-hungry smartphone subscribers. Desperate to change the game, Verizon decided to go all-in on Android, making the Motorola Droid their flagship device. The hardware featured a massive screen, a slide-out keyboard, a 5-megapixel camera, and a removable battery. But its absolute killer feature was on the software side: it was the first device to ship with free, turn-by-turn Google Maps navigation.

Overnight, Google completely "sherlocked" and obsolete an entire consumer electronics category. Up until that point, drivers paid hundreds of dollars for standalone hardware units from Garmin, TomTom, or Magellan, on top of recurring monthly subscription fees just to get live directions. The iPhone's version of Google Maps required users to dangerously and manually tap the screen to advance to the next driving step. Verizon backed this superior product with an aggressive, unforgettable marketing blitz. They licensed the "Droid" name from Lucasfilm—crediting them at the bottom of every commercial—and ran dark, edgy ads that directly mocked Apple's cheerful, white-background aesthetic. The ads ran through a checklist of things the iPhone couldn't do before a static-filled cut proclaimed: "Droid Does."

- **Sales Milestones:** The original Droid sold 250,000 units on its opening weekend and hit 1 million units faster than the original iPhone. - **Accolades:** *Time* magazine named the Motorola Droid its Product of the Year for 2009. - **The Droid Lineup:** Verizon's multi-year marketing muscle birthed subsequent hits like the Droid Incredible, Droid X, Droid 2, Droid Bionic, and Droid Max, firmly establishing Android's US user base.

---

[02:58:33](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10713s) Google's "Less Than Free" Masterstroke

The massive success of the Droid effectively closed the window for competitors like Microsoft. Handset manufacturers were historically wary of Microsoft because Windows Mobile required a $5 to $10 licensing fee per device, and they had watched Microsoft suck all the profit out of PC hardware makers like Compaq. Google countered with a brilliant alternative: Android was completely open-source, highly customizable (as seen in Samsung's heavily modified operating systems), and completely free.

Even wilder was a strategy that wasn't widely publicized at the time. Venture capitalist Bill Gurley exposed the reality in an incredible blog post after hearing whispers from industry insiders. Google wasn't just giving Android away for free; they were making it "less than free" by actively paying carriers and original equipment manufacturers (OEMs). To incentivize them to choose Android over Windows or proprietary software, Google offered a lucrative revenue-share model, kicking back a percentage of the highly profitable ad revenue generated by searches originating on those phones.

The "Less Than Free" Business Model and Navigating Platform Shifts

The speakers analyze the profound nature of the mobile platform shift, framing it as an even larger transition than the shift from desktop PC to the web. Google successfully navigated this paradigm through what was dubbed the "less than free" business model. In this environment, Google aggressively gave Android away for free to carriers and Original Equipment Manufacturers (OEMs), essentially paying them through traffic acquisition cost (TAC) rev-shares to adopt a platform that ensured the underlying mobile ecosystem would not be fully captured by a competitor like Apple. This strategic move effectively neutralized contemporary competitors like Palm, BlackBerry, and Microsoft, who simply could not compete with a free, high-value operating system.

Historically, dominant tech giants have struggled to survive successive platform shifts; IBM ruled mainframes but lost the PC era, while Microsoft dominated PCs but lost the web. Google managed a rare feat by staying intensely relevant across both the web and mobile eras. While they did not generate staggering direct profits from selling hardware or operating systems, they ensured that their core profit engine remained intact. In the early years of the smartphone boom, the most prominent element of the Android touchscreen UI was a "giant freaking Google search bar" pinned right at the top of the screen.

- **Historical Platform Shifts:** IBM (Mainframes to PC loss), Microsoft (PC dominance to Web era loss), Google (Web dominance to Mobile survival). - **Core Strategy:** Open-sourcing Android acted similarly to Chromium, keeping the underlying layer open so Apple's mobile Safari could not become a restrictive "new Internet Explorer". - **Legacy Tech Lifespans:** Dominant companies can maintain their "money machines" for a long time even after losing platform relevance, noted by how IBM famously generated more revenue than Microsoft for much longer than popular memory suggests.

Android's Tidal Wave and the Default Search Mandate

The conversation shifts into what the hosts call the "hit parade," tracking Android's explosive global market growth from late 2009 onward. When the Motorola Droid launched during the 2009 holiday season, Android held a meager 5% to 6% of the global smartphone market share. Within a single year, that number skyrocketed to 30%, with Google announcing that over 200,000 Android devices were shipping daily. By 2011, market share hit 50%, and by the close of 2013, it crested at an astonishing 80%—a figure higher than Android's contemporary share of roughly 72%, against Apple's modern 27.99%.

This massive footprint was achieved through a clever "network of networks" approach reminiscent of the Visa network. Google did not need to manufacture every phone; instead, they leveraged multiple major manufacturers who handled their own marketing. While any company could technically use the bare Android Open Source Project (AOSP) without Google services, Google offered an "offer you can't refuse": to get the official app store where all the apps lived, along with native apps like Gmail and Google Maps, manufacturers had to set Google as the default search engine. The hosts joke that this requirement was the ultimate, non-negotiable payment for the ecosystem.

- **Android Global Market Share Growth:** - Holiday 2009: 5% – 6% - Holiday 2010: 30% (with 200,000+ devices shipping daily) - 2011: 50% - 2013: 80% (Peak market share) - Present Day (2026 context): ~72% (vs. Apple at ~27.99%) - **Competitive Push:** Android forced Apple to innovate faster, driving the implementation of cheaper iPhones, larger screens, and significantly better cameras to counter early Android standouts like the Droid's 5-megapixel camera, which made the original iPhones' 2-megapixel shooters look "crappy".

The Samsung Tension and the Origin of the Pixel Strategy

A key moment of friction emerged in the early to mid-2010s when Samsung realized they could utilize Android's flexible platform to essentially copy the premium feel of the iPhone. As Samsung's Galaxy devices began shipping in massive volumes, Samsung started stripping out core Google services in favor of their own proprietary software. Recognizing this as "a bridge too far," Google responded by launching the Pixel program.

Though sold to consumers, the Pixel was fundamentally designed as a "reference device" for the broader non-Samsung OEM market. It was meant to showcase high-quality premium designs, exceptional camera integration, and clean software features for other manufacturers to emulate. The speakers compare this directly to the Microsoft Surface strategy, noting how Steve Ballmer was incredibly adamant about building first-party hardware to show third-party OEMs exactly how Windows hardware should be executed.

- **Precursor Hardware Program:** Google previously ran the Nexus program before pivoting to the more premium, reference-focused Pixel strategy. - **Strategic Goal:** Counterbalance Samsung's ecosystem leverage and prevent the fragmentation or erasure of core Google services on mobile hardware.

The Napkin Math of Android's True Financial Value

The hosts dive deep into financial disclosures to answer a fundamental question: what is the actual business value of Google owning Android versus not owning it? To calculate this, they look at two primary components: direct revenue generated from the Google Play Store and the massive traffic acquisition cost savings realized by controlling their own distribution network. By analyzing Google's annual reports, they break down a total of $55 billion spent on traffic acquisition costs (TAC) to strip away obfuscations and isolate the mobile carrier and OEM payouts.

Through their financial breakdown, the hosts deduct roughly $21 billion from the total TAC to account for AdSense and DoubleClick publisher payouts (based on a standard 70/30 split on $30 billion gross network revenue). This leaves $34 billion explicitly dedicated to direct search traffic acquisition. Knowing that Google pays Apple $20 billion for Safari search placement, they deduce that the remaining $14 billion is distributed to non-Apple partners like Mozilla Firefox, web portals, mobile carriers, and OEMs.

- **Estimated Android Carrier/OEM Payouts:** Approximately $10 billion of the remaining TAC goes directly to Android carriers and OEMs after factoring out roughly $4 billion for Firefox and legacy web portals. - **The Value Split:** Google pays carriers and OEMs about half of what they pay Apple, but spread across billions more devices, meaning the per-device revenue share is significantly smaller (estimated at a quarter of Apple's rate). - **Play Store Financials (2019 Lawsuit Data):** - Revenue: $11.2 billion - Gross Profit: $8.5 billion - Operating Income: $7.2 billion - **Core Search Ads Comparison (2019):** ~$100 billion in revenue, ~$85 billion gross profit, and ~$30 billion operating income. - **The Ultimate Conclusion:** Android's primary value is not direct profitability or cost-saving, but rather "derisking their future" and insulating their multi-hundred-billion-dollar core search engine from platform gatekeepers.

Sundar Pichai's Rise and the Social Blindspot of Orkut

As the Android ecosystem matured, a leadership transition occurred when Andy Rubin departed Google. Sundar Pichai, who had steadily been gathering structural responsibilities ranging from application clients to Chrome, took over the newly unified Chrome and Android teams in 2013. The speakers note how Pichai would frequently take the stage expressing immense pride in Google’s "two open platforms." With the global active Android device count eventually soaring past 3 billion, Google felt incredibly confident by the 2010–2011 era, proving that massive individual product failures did not matter if a company's core hits were world-changing, decades-long platforms.

However, a massive competitive storm was brewing in an adjacent category: social. While the industry remembers the eventual failure of Google+, Google's fascination with social networking actually dates back much further. In January 2004, prior to Gmail, the Google IPO, or Facebook's launch on the Harvard campus, a Turkish engineer named Orkut Büyükkökten utilized his 20% time to launch a social network within the company called Orkut. While it failed to gain traction in the United States, it became an absolute cultural phenomenon internationally.

- **Scale of Android:** Over 3 billion active devices worldwide against a global population of roughly 7 billion people. - **Orkut Peak Metrics:** Reached approximately 300 million users at its peak and became the single dominant social network in both Brazil and India. - **Early Social Initiatives:** Google attempted "Open Social" in 2007 to counter Facebook's growing closed ecosystem, but the project faltered because Facebook refused to participate.

[03:16:27](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11787s) The "Googly" Decentralization of the Late 2000s

By 2009, Google had expanded into a massive but highly siloed operation. The company was concurrently launching monumental platforms like Chrome and Android, while also building experimental collaborative tools like Google Wave. The sheer decentralization allowed entirely separate teams to work out of different buildings, sometimes even competing or ignoring adjacent existential company battles, such as Android's brewing fight with Apple. When researching the era and questioning product leaders about the ultimate through-line or strategic vision holding everything together, the common consensus was simply that it was "googly"—people organically built what they found cool and beneficial for the open web.

While this chaotic independence felt like a lack of coordination from the bottom up, it was actually a highly effective top-level strategy. The thin layer of core guidance from executives was deliberately kept tight at the top and never over-communicated downward. This approach shielded individual teams from worrying about corporate mechanics, allowing them to focus strictly on building great products people loved without constantly calculating how a feature would serve the overarching advertising business.

- **Open Social (2007) & Google Wave (2009):** Early, decentralized web projects launched during Google's rapid expansion era. - **Product Autonomy:** The structural silos acted as a feature rather than a bug, driving early product velocity.

[03:18:04](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11884s) The Social Threat and Palace Intrigue

The organic product era stumbled as Google Wave failed because users simply didn't know what to do with it, and Google Buzz launched in 2010 only to immediately implode in a massive privacy debacle. Following the Buzz failure, Urs Hölzle—a deeply influential senior vice president and the creator of Google’s distributed infrastructure—penned a watershed internal memo. Mirroring Bill Gates’ famous 1995 internet memo, Hölzle warned of a fundamental sea change: the web was organizing around people rather than pages, posing a massive threat that required an immediate, decisive deployment of personnel.

While the public narrative frames the subsequent creation of Google+ purely as a direct war against Facebook, the move was deeply tied to internal palace intrigue. Google had become dangerously fragmented into competitive fiefdoms led by big personalities. Android was fighting an isolated existential mobile battle, Chrome was beginning to build its own operating system in direct friction with Android, and YouTube and Search operated on completely separate islands. Gmail was the only property that truly owned user identity because it required a login, while YouTube maintained an entirely independent username and password system. Sensing this fragmented mess, Larry Page decided to step back in to get the company on track, using Google+ as a forced mechanism to galvanize, recentralize authority, and unify the corporation under a single banner.

- **The Sea Change Memo (2010):** Written by Urs Hölzle, warning that Google must pivot to a people-centric social web. - **Corporate Fragmentation:** Competed fiefdoms included Android, Chrome, Search, YouTube, and Gmail. - **Larry Page's Return:** Officially announced as returning to the CEO seat in January 2011 (taking over that April), immediately moving his office into the building dedicated to the new social project. - **The Facebook Walled Garden:** Facebook posed an existential threat because Google search could not index its content, and unlike previous partners like AOL, Facebook refused third-party ad servers after hiring Sheryl Sandberg to build an in-house closed-loop monetization system.

[03:22:44](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12164s) The Top-Down Command of Google+

Google+ was engineered during a high-pressure, one-year sprint that was explicitly un-Googly. It was entirely top-down, command-and-control, and completely devoid of organic consensus or a core technical insight. The initiative was spearheaded by Vic Gondotra, the former frontman and master of ceremonies for Google I/O who had led developer relations. Given immense institutional authority by Larry Page, Sergey Brin, and Eric Schmidt, Gondotra aggressively pushed the social mandate across the company, famously telling the press that the project represented "the next generation of Google. It is Google plus one."

The rollout deeply poisoned Google’s internal culture. Entire engineering headcounts were forcibly stripped from existing teams to staff the Google+ division, and corporate OKRs and company-wide bonuses were explicitly tied to its success. Legacy product teams rebelled against the mandatory integrations; an intern at the time recalled a Picasa team member telling him to "f off" when asked about merging their photo repository with the platform. In an obsessive push to make everything "plus-oneable," Google forced the feature onto mobile display ad units via AdMob and infamously converted standard YouTube comment sections into public Google+ posts, nearly killing their existing cash cows in the process.

- **Vic Gondotra:** Appointed as the strategic leader of the Google+ initiative with absolute corporate backing. - **Forced OKRs:** Employee financial bonuses were directly coupled to the adoption and performance of Google+. - **Product Integrations:** Features like Google Hangouts, Google Photos, and "Sparks" originated here, but were forced onto users without natural product-market fit.

[03:26:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12413s) Desktop Circles and the Longhorn Effect

The fundamental product mistake of Google+ was trying to replicate Facebook at a time when Mark Zuckerberg had already realized the traditional social network model was dying. While Google poured immense resources into a desktop-first interface featuring "Circles"—a highly rigid, computer science way of manually sorting friends into overlapping groups that real users hated—Facebook was busy acquiring Instagram and WhatsApp. Zuckerberg recognized that social behavior had bifurcated into public media and private messaging, leaving Google aggressively scaling a product format that the market was actively moving away from.

Ultimately, Google+ became the company's "Windows Longhorn" or "Windows Vista"—a catastrophic corporate distraction that siphoned the best talent and resources away from critical projects. During these years, executive chairman Eric Schmidt co-taught a class at Stanford's Graduate School of Business. Right after Facebook acquired WhatsApp, Schmidt walked into the classroom and lamented, "God damn it, we missed it. We totally missed it." Beyond entirely missing the mobile messaging wave, the distraction severely crippled Google’s entry into enterprise Cloud computing, leaving them strategically behind Amazon and Microsoft for years to follow. Most damaging of all, the heavy bureaucracy introduced during the Plus era permanently slowed Google's product velocity, shifting public perception toward a corporation that invents fascinating technology but struggles to ship cool consumer products.

- **The Desktop Interface:** Google+ launched with a desktop-only user interface centered around manual "Circles" management. - **Strategic Blunders:** Resources were diverted away from mobile messaging and Cloud infrastructure. - **The Post-Plus Lull:** Between the launch of Google+ and the modern emergence of Gemini AI, Google failed to launch a single breakthrough consumer service.

[03:31:36](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12696s) The Ghost of Google+ and the Post-Crisis Unification

The grand, decade-long experiment of Google+ left behind a fascinating strategic legacy. In retrospect, the aggressive corporate push into social media was triggered by an existential panic over Facebook. The hosts recall a 2013 strategy memo penned by Neil Mohan, which explicitly warned of the risk that Facebook could become "the starting point of the internet." Google recognized that social data was the future and mobilized entirely to capture it. Facebook was equally terrified, viewing Google as an incoming juggernaut at a time when Facebook was newly public and navigating its own turbulent transitions. Ironically, despite both tech giants treating this as the ultimate corporate battle, Google’s social effort was largely a "nothing burger."

The entire saga mirrors the ending of the Coen brothers' movie *Burn After Reading*, where a series of chaotic, high-stakes events unfold, leaving the characters to wonder if any of it actually mattered. Google+ quickly became a footnote in history, and Google remained perfectly financially stable without a traditional social network. Yet, the crisis was not wasted. Out of the "smoking crater" of Google+, the company managed to rescue two massive, enduring products: Hangouts (which eventually evolved into Google Meet) and Google Photos, the latter boasting over a billion users today.

More importantly, the panic forced an era of unprecedented internal consolidation. Before Google+, the company operated like a collection of separate "fiefs," where individual leaders amassed independent power and built fragmented services. Google+ forced the implementation of a single, unified Google Account across the ecosystem, creating a cohesive visual design and centralized login infrastructure that became foundational for the company's future. Furthermore, while Google ostensibly lost the social war, the definition of social media eventually shifted to look like messaging or YouTube. Since YouTube emerged as the winning paradigm for user-generated content, the joke is that Google could have done absolutely nothing, skipped the social panic entirely, and just watched their "money printer go burr."

- **Key Strategy Memo (2013):** Neil Mohan warned that Facebook threatened to become the primary gateway to the internet. - **Surviving Google+ Spin-offs:** Google Photos (now a billion-user product) and Hangouts (which transitioned into Google Meet). - **The Shutdown Logistics:** Vic Gundotra left Google in 2014, and Google+ was officially shuttered in 2019 following a major security breach. - **The "Googly" Redirect Loop:** After Google+ closed, links redirected to Google Currents. When Currents was also discontinued, all old links permanently redirected to a standard Google Workspace blog post announcing the Currents shutdown, effectively erasing a decade of user-posted articles and complicating research for the episode.

---

[03:35:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12926s) The Bridge to Alphabet and the Rise of Sundar Pichai

By 2015, it was clear that Google needed to transition into a new corporate era. In August of that year, the company announced a massive structural reinvention, transforming into a holding company called Alphabet. Larry Page stepped up to become the CEO of Alphabet, while Sundar Pichai was named the CEO of Google—which remained by far the largest and primary revenue-generating operating unit. Interestingly, despite years of unifying products, they chose not to break out YouTube or any other major consumer product into its own distinct subsidiary; everything core remained housed tightly within Google.

Alphabet was designed to separate the core, highly profitable search business from the speculative "Other Bets." They broke out Google X (which included Waymo at the time before it spun out on its own), the smart-home company Nest, Google Fiber, and two distinct health ventures, Calico and Verily. Additionally, Alphabet absorbed the corporate investing arms, Google Ventures and Capital G.

The selection of Sundar Pichai as Google's CEO served as an essential "healing" mechanism after the turbulent political friction of the Google+ era. Pichai was a deeply respected executive with immense credibility dating back to his early work spearheading Chrome and Android—two foundational platform pillars that successfully pushed Google into the mobile age and shielded it from existential platform threats. Intriguingly, Pichai achieved this despite never actually working directly within the core search or advertising divisions. His diplomatic, consensus-building personality made him the ultimate "peacemaker among big egos" at a time when internal unity was desperately required.

- **Alphabet Core Structure (August 2015):** Larry Page became CEO of Alphabet; Sundar Pichai became CEO of Google. - **The Portfolio of "Other Bets":** Nest, Google Fiber, Calico, Verily, Google X, Google Ventures, and Capital G.

---

[03:37:47](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13067s) Financial Scale and the Immutable Business Model

At the close of 2015, the sheer financial scale of the company was staggering. Alphabet recorded $75 billion in total revenue, with a dominant $52 billion originating from first-party properties like Google Search, AdWords, Gmail, and Maps. In contrast, the DoubleClick and AdSense network ecosystem brought in a smaller, lower-margin slice of $15 billion. YouTube had also crossed into profitability by this time. Overall, Google brought in roughly $23 billion in operating income, easily absorbing the $3.5 billion loss generated by the speculative Other Bets portfolio.

The fascinating takeaway for the hosts is that while these 2015 figures were massive, the underlying business model remained completely identical to what it was in 2002. Over the subsequent ten years leading into 2026, the business would scale an additional 5x, yet its primary engine remained search ads. The core product didn't change; rather, the original seed of search advertising successfully captured what turned out to be the largest, highest-intent marketplace in human history.

- **2015 Revenue Breakdown:** $75 billion total revenue ($52 billion from first-party Google sites, $15 billion from network ads). - **Profitability Metrics:** $23 billion in operating income from core Google operations, offset by a $3.5 billion loss in Other Bets.

---

[03:39:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13166s) The AI Teaser: When Google Owned the Talent

In a compelling closing teaser for the next era, the conversation pivots to an extraordinary convergence of talent. Between 2015 and 2016, a breathtaking percentage of the world's top artificial intelligence pioneers were simultaneously logged into the corporate directory as standard Google employees. The roster included Alex Krizhevsky (co-creator of AlexNet), his PhD advisor and "godfather of AI" Geoff Hinton, and co-author Ilya Sutskever (who later co-founded OpenAI). It also included Anthropic co-founder Dario Amodei, former Tesla Chief AI Scientist Andrej Karpathy, alongside prominent researchers like Chris Olah, Noam Shazeer, and Ian Goodfellow.

Furthermore, following the 2014 acquisition of DeepMind, Google employed Demis Hassabis, Shane Legg, and Mustafa Suleyman (who now runs AI at Microsoft), alongside Stanford's Andrew Ng, Quoc Le, Oriol Vinyals, and the collective authors of the historic transformer paper. Because Google invented and published the transformer architecture in June of 2017, they created the baseline mechanism powering every modern Large Language Model today. An insider tracking this ecosystem recently joked to partners that "the 'T' in ChatGPT stands for transformer, and we invented that."

During this exact window, Ilya Sutskever posed a radical hypothesis to his peers on the Google Brain team: what if they built an unimaginably massive neural network and trained it on the entire internet? Google was uniquely positioned to execute this because their core search index had already mapped the entire web, providing a massive, unmatched data cache. Through this historical lens, Google's frantic product expansions over the previous decade can be reinterpreted not as fragmented failures, but as a massive dragnet capturing the data, infrastructure, and elite talent required to build the future of AI. This aligned perfectly with a prophetic quote from Larry Page in the year 2000, who remarked that an ultimate search engine would possess artificial intelligence, understanding everything on the web to give you exactly what you wanted.

- **DeepMind Founders (Acquired 2014):** Demis Hassabis, Shane Legg, and Mustafa Suleyman. - **The Transformer Milestone:** Invented and published by Google researchers in June 2017. - **Larry Page Quote (Year 2000):** "Artificial intelligence would be the ultimate version of Google... We have enough disk space to store like a 100 copies of the whole web." At the time, Google ran on roughly 6,000 computers with an index that would measure 70 miles high if printed.

---

[03:44:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13451s) Economic Analysis: Google's Seven Powers

The episode transitions into a structured business analysis using Hamilton Helmer's *7 Powers* framework, which identifies the strategic attributes allowing a business to sustain differential returns over its competitors. Google represents a unique analytical challenge because the consumer user experience is completely detached from the actual economic transaction; to understand Google's power, one must look through the eyes of the advertiser choosing where to deploy their next marginal dollar.

Evaluating **Counterpositioning**—a power rarely held by large incumbents—the hosts highlight Android as a flawless historical example. Google aggressively counterpositioned against Microsoft’s paid operating system licensing model by offering Android as a "less-than-free" open-source alternative, monetizing downstream via search ads. Microsoft could not replicate this because they lacked a highly profitable search engine capable of subsidizing an entire mobile operating system at scale.

When examining **Scale Economies**, Google operates as a definitive one-stop shop for global advertisers. Instead of spending operational headcount managing disparate ad campaigns across fragmenting platforms, an advertiser can seamlessly access search, display, and video through a single interface. On top of massive infrastructure advantages, Google's ad marketplace thrives on auction mechanics: the denser the pool of competing advertisers and high-intent users, the more efficiently individual auctions find their maximal, profit-generating price.

Turning to **Network Economies**, the dynamics are apparent across multiple verticals. YouTube functions as a classic two-sided network where more creators attract more viewers, which in turn draws deeper monetization. Android enjoys a parallel two-sided network effect between mobile application developers and smartphone consumers. Finally, **Switching Costs** manifest strongly in consumer-facing applications like Gmail, where a user clinging to twenty years of searchable, free email history is practically locked into the ecosystem. However, the hosts note that switching costs are far less pronounced for core advertising buyers; brands do not continuously buy Google keywords because it is logistically painful to leave, but rather because Google commands the ultimate high-intent search traffic on the web.

- **Counterpositioning Power:** Android disrupted paid operating systems by utilizing a free, ad-subsidized business model. - **Network Economies Examples:** YouTube (Creators vs. Viewers) and Android (Developers vs. Users). - **Switching Costs Nuance:** High for consumer applications (e.g., 20 years of data in Gmail), but lower for advertisers who stay primarily due to Google's unparalleled user intent pool rather than structural friction.

[03:47:55](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13675s) Switching Costs, Brand Halos, and Internal Infrastructure

The discussion opens with a nod to the immense switching costs embedded in modern tech ecosystems. Once an algorithm, like YouTube’s, is fully dialed into a user’s personal interests, leaving the platform becomes an incredibly hard sell. Transitioning to the historical context of Google’s golden era, the speakers reminisce about the sheer corporate halo that surrounded the brand during its initial product blitzes. Back then, any new release from Google—simply by virtue of bearing the name—would have users "chomping at the bit" to get an invite. One speaker laughs about how desperate they were to acquire Google Wave invites; even though the product ultimately crashed and burned, the initial mystique and dazzle of the Google brand were intoxicating. This powerful brand equity ironically held Google back during the onset of the AI revolution, making them reticent to slap their storied name on early models until they were practically "shoved off the cliff."

Shifting focus to cornered resources, the conversation zeroes in on Google’s infrastructure and data legacy heading into the AI age. While their massive physical infrastructure acts as a scale economy allowing them to run products at rock-bottom costs, it also functions as a profound process power. During their heyday, this gave Google the unique ability to launch experimental products far cheaper than anyone else. More importantly, Google has historically engineered internal software systems that completely outclass anything available in the outside world. Even though they frequently publish open-source equivalents of their stack, they keep the genuine "secret sauce" behind closed doors. For instance, rather than migrating heavily to Kubernetes, Google still runs internally on Borg, a system that departed engineers deeply miss.

* The YouTube content catalog stands as a premier cornered resource for training next-generation AI. * Google relies heavily on its proprietary internal system, Borg, rather than its open-source offspring, Kubernetes.

[03:50:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13800s) Playbook: Platform Realities and the Acquisition Cycle

The conversation moves into the "Playbook" portion of the episode, starting with a fundamental question: Is Google actually a platform company? Looking back, Google desperately wanted to achieve platform status, but they approached it indirectly. Instead of owning the literal platform of the web, they advanced its core capabilities to benefit their primary money maker. Even with Android under their belt as a target development platform, Google doesn't operate like a pure-play platform business such as Apple or Microsoft—companies that sell hardware or software and build an ecosystem to bolster those direct sales. Google is fundamentally a shadow platform or an ecosystem company where, at the end of the day, their "bread is buttered" by advertising.

Another brilliant playbook dynamic from the 2010s was Google's relentless stream of smaller, strategic acquisitions. Beyond blockbuster deals like YouTube, Android, DoubleClick, and AdMob, Google scooped up dozens of smaller tools that formed the backbone of Google Groups, Spreadsheets, Docs, Blogger, and Maps. This buying spree fed into a unique, cyclical talent strategy. Programmers would join Google, collaborate with brilliant minds, leave to form a brand-new startup, and eventually get reacquired back into the mothership a few years later. From an outside perspective, this talent loop looked chaotic, but it was a calculated long-term play. Backed by the massive "money printer" that was AdWords, Google could easily afford to fund these indirect, generous strategies to ensure the wider web bloomed.

* **Key historical acquisitions mentioned:** Applied Semantics (AdSense tech), Urchin (Google Analytics), Dodgeball, FeedBurner, reCAPTCHA, Slide, Jambool, Like.com, Widevine, Admeld, Punchd, Zagat, Sparrow, and Wavy. * Pure-play platform companies like Apple and Microsoft monetize the platform layer directly, whereas Google uses its ecosystem to protect and feed its core ad business.

[03:52:45](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13965s) The Cultural Divide: Google Glass vs. Meta Ray-Bans

The speakers transition into an illuminating cultural comparison sparked by a deep dive into old Google I/O keynotes. Looking back at the early promotional content for Google Glass—a product that has since become the butt of tech jokes—they realize that its core features are nearly identical to modern successes like the Meta Ray-Bans. Both utilize side gestures and take photos, though Google Glass was technically more advanced by attempting to run basic, text-based applications. The defining difference between the two projects isn't the technology or the timing; it is the fundamental cultural approach of the companies behind them.

"Google's made you look like a cyborg. Meta's is for normal people," one speaker points out, calling it the ultimate metaphor for the ideological divide between the two tech giants. Google is historically staffed by wacky academics who struggle to see why a jarring, futuristic aesthetic might alienate everyday consumers. In contrast, Facebook's product culture is explicitly built around social dynamics and trying to be cool, which led them to secure a direct design partnership with Luxottica to blend their technology into standard, fashionable eyewear.

* Google's engineering-first mindset resulted in an advanced but socially awkward "cyborg" look for Glass. * Meta focused on fashion and consumer acceptance by partnering with eyewear conglomerate Luxottica.

[03:54:09](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14049s) Cross-Era Successes and the Billion-User Product Metric

Reflecting on internal culture, the hosts discuss how Google successfully motivated engineers simply by challenging them to solve incredibly complex technical problems. For years, teams could focus entirely on building elegant, useful tools without ever having to worry about an immediate business model. This created a corporate structure featuring a razor-thin layer of elite strategic thinkers at the top, supported by a massive foundation of talent single-mindedly focused on making great products. This strategy paved the way for Android, which they hail as "the mother of all wins." Spanning a business model across entirely different technology eras is an incredibly rare feat in tech, yet Google managed to move from desktop search to mobile dominance seamlessly.

This historical run culminates in an astonishing statistic: Google currently possesses eight distinct products that have crossed the billion-user mark. When comparing this to the rest of Big Tech, Google's consumer footprint is unmatched. Meta sits in second place with four core billion-user apps, while Apple effectively has only one independent consumer product of that scale if you exclude default utility apps that come pre-installed for free on the hardware.

* **Google's 8 billion-user products:** Search, Android, Chrome, YouTube, Gmail, Maps, Drive, and Photos (Nine if counting the Play Store separately). * **Meta's 4 billion-user products:** The Facebook "Blue app," WhatsApp, Instagram, and Messenger. Meta claims Meta AI is a fifth, but the hosts dismiss this since it is deeply embedded across existing apps. * **Apple and Microsoft comparison:** Under a strict definition excluding default free apps, Apple has one (iPhone). Microsoft possesses two (Windows and LinkedIn), while Amazon has zero products with over a billion users.

[03:58:24](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14304s) The Core Technical Insight and Carveouts

For the final element of the quintessence, one speaker reveals a hidden thesis: almost every single wildly successful, organic Google product is built upon a breakthrough, academic-level "core technology insight." When looking at Google’s biggest wins, the technical breakthrough *is* the user experience. PageRank defined Search; elegant auction mechanics fueled AdWords; Ajax transformed Gmail; and massive engineering scaling powered YouTube and Chrome. Even Android, which relied heavily on partnerships and distribution, was a massive technical feat of wrestling Linux onto a battery-powered pocket device. By contrast, when Google tries to build products driven purely by a social user experience rather than a foundational technical breakthrough—such as Google+ or Google Wave—they consistently fail. This rule was so ironclad that former CEO Eric Schmidt would openly refuse to fund projects if the product managers couldn't clearly state their core technical insight.

The episode winds down with the traditional "carveouts" segment. After teasing the audience about a long-awaited personal update regarding a recent gaming console purchase, the speaker holds the suspense for one more minute to share a family anecdote from a recent trip to New York City for their Radio City event. Staying the week with his daughters, they visited the immersive Bluey experience hosted at the CAMP store. He thoroughly praises the activation, noting that it completely lived up to the hype by perfectly recreating the show's iconic house in a physical space, allowing kids to roam free before delivering a magical character moment at the very end.

* Eric Schmidt used the presence of a "core technical insight" as a strict litmus test for funding internal projects. * Products like Google Photos succeeded because of long-term investments in underlying AI features, whereas Google struggles to build purely social apps like Instagram.

[04:03:47](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14627s) Choosing the Steam Deck over the Switch

After giving a high recommendation to a Bluey-themed experience in New York, the conversation shifts to personal tech purchases. Ben reveals that he ultimately bought a Valve Steam Deck. Although his hectic schedule with events at Radio City and preparing for this massive podcast episode left him with very little time to actually play it over the past month, he remains thoroughly impressed.

The decision came down to a simple realization about his family dynamics. Ben explains that he desperately wanted his older daughter to be ready to play *Mario Kart* with him, but "she's just not" there yet. Because the purchase was purely for his own personal enjoyment rather than a shared family activity, the Steam Deck won out over a Nintendo console. He strongly endorses the device, praising Valve for abstracting complex PC gaming mechanics into a seamless console-like experience. For someone who loves PC-style games but hasn't been a traditional PC gamer in years due to the friction of building a machine or finding a place to put a massive rig, the Steam Deck offers the perfect solution: "buy the damn thing, turn it on, buy the games, play them." Valve removes all the headaches regarding drivers and hardware specifications.

- Ben still plans to buy a "Switch 2" at some point, likely within the next year. - The Steam Deck successfully delivers handheld console simplicity for PC titles.

---

[04:05:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14722s) Ben's Carveouts: AI Assistants and Pocket Cameras

Moving into the "carveouts" segment, Ben shares three recommendations. His first favorite is the AI assistant Claude by Anthropic. Swearing to God that the endorsement is completely unrelated to their official sponsorship of the show, Ben notes that using AI has fundamentally transformed how he prepares for episodes, stating that he "cannot imagine going back" and finds himself using it all day.

His second carveout focuses on a camera debate, sparked by a Twitter exchange with Sahil Mohan Desai (referred to as Shil Monot) that morning when Ben was supposed to be preparing for the episode. Ben recently purchased a Fujifilm X100VI—the internet's darling camera—which he enjoyed carrying around his neck in New York for its nostalgic, 35mm-equivalent shooting style and film simulation profiles. However, his actual carveout recommendation goes to an older piece of gear: the Sony RX100 VII. While it is a 2019 model that desperately needs a USB-C upgrade to fix its annoying charging situation, its ultra-compact size allows it to fit directly into a pocket. Paired with a smartphone, it serves as the ultimate space-constrained setup, offering a massive zoom lens and superior image quality over standard phone cameras.

- **Claude:** A total game-changer for podcast episode preparation and daily workflows. - **Fujifilm X100VI:** Fun and artistic, but ultimately large enough to require hanging around the neck. - **Sony RX100 VII:** A highly practical, pocket-sized point-and-shoot camera with an incredible zoom. - **Adobe Lightroom Denoise:** An AI-powered production feature that pairs beautifully with point-and-shoot camera files to clear up image noise.

---

[04:07:44](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14864s) Premium Cashmere and Season Closings

Ben’s third and final carveout is a high-end clothing brand called Kerosimi, started by an Acquired listener who sent him a sample garment. Ben wears the ultra-nice cashmere shirt during the recording session, noting that it feels great on both cool and warm days. It fits perfectly into his ideal "capsule wardrobe" philosophy, which typically consists of a simple, no-thought-required, long-sleeve dark crew neck that looks sharp and pairs with anything. His co-host complements the look, joking that after staring at Ben for seven and a half hours of recording, he was thinking the entire time, "God, Ben is looking good."

The hosts then wrap up the massive episode by thanking their core partners for the season and acknowledging the extensive list of industry insiders, executives, and creators who provided invaluable insights behind the scenes.

- **Kerosimi:** A premium clothing company producing high-quality cashmere apparel. - **Season Partners Mentioned:** JP Morgan Payments (`[jpmorgan.com/acquired](https://jpmorgan.com/acquired)`), Anthropic (`claude.ai/acquired`), Statsig (`[statsig.com/acquired](https://statsig.com/acquired)`), and Vercel (`[vercel.com/acquired](https://vercel.com/acquired)`). - **Episode Contributors (Group 1):** Hiroshi Lockheimer, Tim Armstrong, Sam Schillace, Hunter Walk, Nick Fox, Shona Brown, and Clay Bavor. - **Episode Contributors (Group 2):** Max Ross, Greg Corrado, Demis Hassabis, Arvin Navaratnam (Worldly Partners), Paul Buchheit (creator of Gmail), Bill Coughran, Jonathan Rochelle, Bradley Horowitz, John Hanke, Ben Eidelson, Easar Lipkovitz, and Ben Ling.

---

[04:10:04](https://www.youtube.com/watch?v=QhAftC_zFr8&t=15004s) Cross-Recommendations and Signing Off

In the final minutes, the hosts direct listeners to their deep catalog for the full historical context of the tech industry's biggest rivalries. They encourage anyone who hasn't done so to listen to their first Google episode regarding the origin of the search engine, as well as their comprehensive three-part Microsoft series featuring an interview with Steve Ballmer, which provides "the other side of the story." They also highlight their massive Meta episode as highly relevant context for the themes discussed.

For listeners hungry for more founder-focused interviews, they recommend their secondary feed, ACQ2. They highlight a recent interview with industry legends Bret Taylor and Clay Bavor focused on the current landscape and future trajectory of AI. The show concludes with a call to action for fans to join the conversation in the Acquired Slack channel and sign up for the email newsletter to vote on upcoming topics for the fall season. The duo signs off with their signature outro music, leaving listeners with their classic closing refrain: "Who got the truth? Is it you? Is it you? Is it you?"

- **Acquired Slack Community:** Accessible at `acquired.fm/slack`. - **ACQ2 Recommendation:** Feature interview with Bret Taylor and Clay Bavor regarding AI. - **Fall Episode Planning:** Listeners on the email list can actively participate in voting for upcoming topics.

[Source](https://www.youtube.com/watch?v=QhAftC_zFr8)

中文简报

Google将2005年利润再投资于生态布局以抗衡微软 Google通过补贴大规模的开源网页及移动端产品，成功化解了竞争对手的平台主导地位，以此保护其核心广告收入飞轮。本期节目探讨了Google如何成功从单一的搜索引擎转型为涵盖Gmail、Maps、YouTube、Chrome和Android的多产品巨头。节目剖析了这些押注背后的战略动机，将其视作在对抗微软、Apple和Facebook的三线战争中至关重要的攻防武器。主持人还将Google重大的技术胜利与Google+等文化层面的失误进行了对比，阐述了这一高度创新的时代最终如何为其重组为Alphabet以及现代AI革命奠定基础。 Google从“单一业务”搜索引擎转型为多元化巨头，但由于对Gmail和Maps等产品的激进再投资导致收益停滞和盈利能力下滑，在2005年末遭到了华尔街的初步抵制。 (3:53) Gmail于2004年发布，通过提供高达1GB的免费存储空间颠覆了消费者认知模式，这基于其预判摩尔定律最终会将数据存储和搜索成本降至接近于零。 (6:37) Gmail和Google早期Web应用的工程底座推广普及了Ajax（异步JavaScript与XML），消除了页面重新加载的需求，并直接在浏览器中实现了桌面级软件体验。 (13:11) Google开发Gmail和Google Maps等基于浏览器的应用是针对微软的防御策略，旨在让用户活跃在网络上，而不是被困在微软的桌面操作系统生态中。 (16:56) Google在完成对Where 2 Technologies和Keyhole等关键初创公司的收购后于2005年推出Google Maps，通过开放API推动了Web 2.0混搭时代，并通过本地广告和授权许可发展成为价值数十亿美元的业务。 (32:23)

Reading summary

Acquired hosts Ben Gilbert and David Rosenthal argue that Google's post-IPO expansion was a strategic campaign to neutralize Microsoft by shifting computing to the open web and securing distribution channels. Google succeeded whenever it anchored products in deep technical breakthroughs, turning browser-based software, digital advertising, video, and mobile operating systems into massive conduits for its core search engine. Web Applications as an Operating System Neutralizer (6:37) Google's non-search expansion began in earnest on April 1, 2004, with Gmail, which offered 1 gigabyte of free storage—dwarfing Microsoft Hotmail's 2 megabytes and Yahoo Mail's 4 megabytes. Engineer Paul Buchheit utilized the XMLHTTPRequest API to eliminate static page reloads, effectively inventing Ajax and ushering in dynamic web applications. Former Novell CEO Eric Schmidt recognized that rich web applications threatened Microsoft's core business model by making the underlying desktop operating system irrelevant to users and developers. This web-first playbook accelerated through targeted acquisitions and heavy infrastructure spending. Google acquired Where 2 Technologies, whose founders ported their desktop mapping tool to a browser-based application in three weeks, creating Google Maps alongside Keyhole and ZipDash. Google poured billions into driving camera cars, gathering geospatial data, and offering low-cost APIs. The hosts note that Google Maps now serves over 2 billion active users and generates an estimated $5 billion to $10 billion annually, primarily through local search ads. Monetizing Global Attention via YouTube and DoubleClick (56:25) In 2006, Google acquired YouTube for $1.7 billion. Initially, the platform was bleeding roughly $1 billion annually—losing approximately one penny per view due to crushing video encoding, storage, and bandwidth costs against just $30 million in revenue. YouTube sustained these losses while giving away roughly 50% of ad revenue to creators. According to MoffettNathansson estimates cited by the hosts, YouTube has since scaled into the internet's largest aggregate attention sink, generating approximately $8 billion in annual operating income. To capture agency budgets outside its self-serve AdWords and AdSense networks, Google bought DoubleClick in April 2007 for $3.1 billion in cash, incorporating a "hell or high water" clause to block an impending counter-bid from Microsoft. Under David Rosenblatt and Neal Mohan, DoubleClick created the real-time programmatic ad exchange, connecting competing ad networks and routing unsold inventory. The deal established Google's agency relationships and forced Microsoft into a hasty $6 billion purchase of second-place aQuantive. Chrome and Android as Distribution Shields (119:50) Google realized that relying on third-party gateways left search vulnerable. In 2004, Google hired Sundar Pichai to oversee client applications, eventually greenlighting the Chrome browser in 2006 to ensure Microsoft could not lock Google out by defaulting Internet Explorer to Bing. Chrome debuted in September 2008 alongside an explanatory Scott McCloud comic book detailing its sandboxing and V8 engine, securing 40 million users within 18 months and resetting web standards. Simultaneously, Google disrupted mobile operating systems through its 2005 acquisition of Andy Rubin's Android. While Microsoft charged original equipment manufacturers $5 to $10 per device for Windows Mobile, Google gave Android away for free and shared search ad revenues with carriers and hardware makers. Android's global smartphone market share surged from roughly 5% in 2009 to 80% by late 2013. In exchange for the Google Play Store and bundled proprietary apps, Google mandated default search placement, eliminating existential mobile distribution risks. Technical Insight Versus Top-Down Social Culture (202:44) The hosts identify a structural rule across Google's history: products succeeded when driven by a core, academic-grade technical breakthrough that directly shaped the user experience—such as PageRank in Search, dynamic rendering in Gmail, or high-performance browser architecture in Chrome. When Google attempted top-down, socially engineered initiatives devoid of a core technical insight, the results faltered. This friction culminated in the 2011 launch of Google+, an executive-mandated initiative led by Vic Gondotra. Google tied bonuses and corporate metrics to platform adoption and forced uninvited integrations across YouTube comments and Picasa, generating internal cultural rebellion. In 2015, Google restructured into Alphabet under Larry Page, establishing Pichai as Google CEO to stabilize core units, while segregating speculative "Other Bets" like Waymo, Nest, and Calico. The hosts observe that Google's historical advantages in infrastructure, talent, and distribution now face critical tests. While Google gathered the pioneers of modern deep learning and published the 2017 transformer paper that underpins modern artificial intelligence, Department of Justice antitrust scrutiny targeting default search distribution agreements and a potential Chrome divestiture threatens the structural traffic loop that financed Google's expansion.

精炼阅读摘要

Acquired节目回顾了Google从单一搜索引擎扩展为综合科技巨头的历程，指出其扩张的核心逻辑在于通过Web应用瓦解传统操作系统壁垒，并借助底层核心技术突破驱动多项超级产品。 Web应用崛起打破操作系统壁垒 (6:37) 在2004年推出Gmail之前，主流网络邮箱仅提供极小存储（如Hotmail的2 MB与Yahoo Mail的4 MB），而Google不仅提供1 GB的免费容量，还利用XMLHTTPRequest API开创了无需整页刷新的Ajax架构。这一突破直接确立了Web 2.0时代的交互标准。随后，Google收购Where 2 Technologies并在三周内将其桌面地图重构为Web版Google Maps，进一步夯实了浏览器端应用的能力。这一系列举措对依赖Windows操作系统和桌面软件收费的微软构成了直接挑战。Google管理层（包括前Novell首席执行官Eric Schmidt）深知操作系统的控制力，因而策略性地将浏览器打造成集成层。只要用户主要通过浏览器处理事务，底层操作系统便失去排他性，进而直接保护并反哺了Google核心的搜索与广告生态。重估YouTube与双重广告飞轮 (56:25) 在内容与变现层面，Google通过并购补齐了关键拼图。2006年收购的YouTube初期面临惊人的网络带宽与存储开销，平台一度以每播放一次亏损约一美分的速度运行，被内部戏称为“Google的首个错误”。但通过给予创作者约50%的广告分成，YouTube构建起深厚的创作者生态，逐渐成长为全球占用用户时长最多的平台之一。根据MoffettNathansson估算，YouTube年营业利润已达约80亿美元，节目主播据此彻底修正了早年对该收购的负面评价。在广告技术端，Google于2007年以31亿美元现金抢先击败微软收购DoubleClick。借助DoubleClick构建的实时广告交易平台（Ad Exchange），Google成功切入麦迪逊大道的品牌广告与传统代理商预算，使其核心广告引擎从早期的AdWords与AdSense，跨越式升级为覆盖全网展示与长尾流量的庞大商业网络。分发保卫战：Chrome与Android (119:50) 面对微软通过IE默认搜索引擎推广Bing的潜在封锁，Google在2008年推出了自主浏览器Chrome，由Sundar Pichai负责的客户端团队推进。凭借V8引擎、沙盒机制与进程独立等性能优势，Chrome迅速侵蚀IE份额，为Google牢牢锁定了桌面搜索入口。在移动端，Andy Rubin创立的Android原本面临诺基亚、黑莓和Windows Mobile的割据局面。Google不仅将Android开源并免费提供给硬件厂商，甚至通过搜索广告收入分成实施了所谓的“比免费还划算”策略。这种联盟模式促使Android全球智能手机市场份额从2009年底的5%至6%飙升到2013年底的约80%。作为绑定条件，厂商必须将Google设为默认搜索引擎并预装应用商店等核心服务。主播团队推算，在每年约550亿美元的流量获取成本中，剔除向苹果支付的200亿美元及发布商分成后，约有100亿美元回流至Android运营商与OEM厂商，从而以极高性价比筑牢了移动端分发防线。技术突破基因与管理机制的演进 (215:26) Google早期的成功普遍遵循“以核心技术突破定义产品体验”的规律，例如PageRank之于搜索、Ajax之于Gmail、以及工程规模化之于Chrome与YouTube。然而，当缺乏底层技术洞察、纯粹依靠自上而下的行政命令强推社交产品Google+时，公司不仅遭遇内部团队反弹，也因强行绑定产品而险些伤及自身业务。这一对比印证了Google在缺乏核心技术突破时难以凭纯体验驱动社交产品的局限。 2015年Alphabet重组成立控股架构，将成熟的Google业务与Waymo、Nest、Calico等“Other Bets”拆分，并由深具声望的Sundar Pichai接任Google首席执行官，修补了前期高压留下的内部裂痕。与此同时，2015至2016年间Google汇聚了全球顶尖的AI研究力量，并在2017年发表了奠定现代大语言模型基石的Transformer架构，但在商业化应用落地与品牌保护之间仍面临平衡难题。 Google凭借强大的搜索现金流与深厚的基础设施，成功完成了从Web应用、视频流媒体到移动操作系统的跨周期跃迁。然而，当前面临的监管反垄断压力（如针对Chrome剥离的调查）以及人工智能范式转变对传统搜索商业模式的冲击，使其未来的分发与变现机制依旧充满变数。

Detailed notes — zh

## Background 本节目由知名商业播客《Acquired》的主持人 Ben Gilbert 和 David Rosenthal 于 2025 年 8 月 26 日发布在 YouTube 平台。根据页面简介，节目重点复盘了 Google 在 2004 年至 2011 年期间，如何从一家“纯搜索”公司裂变式打造出 Gmail、Maps、Drive/Docs、YouTube、Chrome、Android 和 Photos 等 7 款十亿级用户产品，并深入剖析其与微软、苹果、Facebook 的三线战争、Google+ 的重大挫折，以及最终重组成 Alphabet 并奠定现代 AI 革命技术地基的完整商业历程。 ## TL;DR 节目全面梳理了 Google 从 IPO 走向 Alphabet 集团的黄金创新十年，揭示了其如何将核心搜索广告（AdWords）作为超级印钞机，利用超低边际成本与底层学术级技术突破向 Web 平台全面扩张。面对微软的桌面封锁、苹果的移动生态以及 Facebook 的社交围墙花园威胁，Google 通过收购并壮大 YouTube、推行“低于免费”开源战略打造 Android，成功掌控了 Web 与移动端的核心入口。尽管自上而下的 Google+ 豪赌因违背产品基因和移动规律惨遭失败并险些重创文化，但 Google 最终完成母公司架构重组，并在 2015 年前后汇聚全球顶尖 AI 先驱阵容，催生了以 Transformer 架构为代表的现代人工智能基石。 ## Atmosphere & style 整期节目展现出《Acquired》标志性的硬核商业考古风格，长达四小时的对谈信息密度极高，叙事兼具宏大的科技史诗感与微观的工程洞察。两位主持人默契配合、思维碰撞激烈，既有拆解 Google 精妙战略飞轮时的由衷惊叹，也有在复盘 Google+ 灾难和错失移动即时通讯时的扼腕叹息与犀利审视。听感不仅是一场穿梭于硅谷黄金年代商业战争的技术复盘，更让听众仿佛置身于科技巨头生死决断的会议室内，感受到工程师理想主义与残酷商业现实碰撞出的强烈张力。 ## Bulletin - [00:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=0s) 在 2005 至 2011 年的六年间，Google 打造或收购了 Gmail、Maps、Docs、YouTube、Chrome、Android 和 Photos 等 7 款十亿级用户产品，至今没有任何科技公司的十亿级产品总数超过其一半。 - [02:58:43](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10723s) Android 凭借开源免费及将搜索广告收入向运营商和 OEM 分成的“低于免费”策略击溃 Windows Mobile，在 2013 年底将全球市场份额推升至 80% 以上。 - [01:11:17](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4277s) YouTube 从最初每次播放烧钱一美分、年亏损近 10 亿美元的困境中逆转，通过将核心指标转向“观看时长”并推行 55% 创作者分成，成长为年收入超 500 亿美元的内容平台。 - [03:28:10](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12490s) 自上而下强推的 Google+ 成为公司的“Windows Vista 时刻”，不仅未能在社交领域破局，还导致 Google 错失了移动通讯与公有云的关键战略发展窗口。 - [03:46:06](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13566s) 2015 年前后 Google 汇聚了 Hinton、Sutskever、Amodei、Hassabis 等 AI 奠基团队，并于 2017 年提出 Transformer 架构，完成了向大模型时代的底层技术铺垫。 ## Key takeaways - [28:03](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1683s) Google 跨产品扩张的核心逻辑是借助 Web 削弱微软操作系统的核心地位，将浏览器转化为新的集成中心，从而反哺核心搜索业务。 - [06:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=409s) Gmail 首次以 1GB 免费空间和异步刷新（Ajax）技术颠覆传统 Webmail，确立了以极低边际成本与全量云端存储为特征的 Web 2.0 应用范式。 - [37:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2234s) Google Maps 转向 Ajax 架构并免费开放 API，不仅引爆了 Web 2.0 混搭生态，还直接为 Uber、Airbnb、DoorDash 等独角兽奠定了底层技术基石。 - [45:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2735s) 面对 Microsoft Office 的专有格式和企业协议壁垒，Google Docs/Sheets 放弃堆叠复杂功能，转向主打无冲突多人协同与极低的使用门槛。 - [01:01:02](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3662s) YouTube 凭借免审查上传、免安装 Flash 播放器和支持外链嵌入的三大支柱崛起，并在收购后通过算法推荐替代订阅流，构建起独特的双边创作者生态。 - [02:58:43](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10723s) Google 采用 AOSP 开源但捆绑 GMS 的策略主导移动操作系统，通过 TAC 分成让依赖授权费的传统系统在经济模型上面临逆向定位困境。 - [03:06:38](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11198s) Google 每年向苹果 Safari 支付数百亿美元 TAC 以锁定默认入口，而 Android 生态则以更低的单机成本防御了搜索流量入口免遭第三方封锁。 - [03:18:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11902s) Google+ 强制将员工考核与产品形态绑定，不仅激起创作者与内部员工反弹，其桌面端“圈子”思维也与移动化、即时化的真实社交演进方向背道而驰。 - [03:42:08](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13328s) 重组为 Alphabet 确立了控股架构，不仅由 Sundar Pichai 带领 Google 走出 Google+ 创伤，也为 Other Bets 及前沿 AI 研发提供了独立运营机制。 - [03:46:06](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13566s) Ilya Sutskever 等人在 Google Brain 提出将全网搜索索引数据输入超大神经网络训练的假说，奠定了 Google 在大模型革命中的核心技术起点。 - [03:50:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13852s) 从《7 Powers》维度看，Google 的核心壁垒在于 YouTube 的双边网络效应与数据资产、Android 的逆向定位以及全渠道广告投放的边际规模效应。 - [03:58:27](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14307s) Google 产品的成功规律在于底层必须有学术期刊级的核心技术突破支撑，缺乏底层硬核技术创新、纯靠交互构想的产品往往在内部难以存活。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=0s) 开场：黑高领毛衣与节目的序幕节目在一阵轻松打趣中开场。Ben Gilbert 看着搭档 David Rosenthal 身穿黑色高领毛衣走进录音室，嘴角带着坏笑，忍不住调侃他是不是故意在这期关于 Google 的节目里穿得像史蒂夫·乔布斯（Steve Jobs）一样，以此影射 Android 与苹果 iPhone 之间的旷世大战。David 笑着否认，解释说这件衣服纯粹会成为他本期节目后面的推荐环节（Carveout）之一。玩笑过后，节目正式引入 Acquired 2025 年夏季赛季的主题。Ben 和 David 回顾了 Google 在 20 世纪 90 年代末的崛起：凭借突破性的 PageRank 搜索算法、基于廉价商用硬件搭建的高性价比服务器集群，以及人类商业史上最强大的商业模式——搜索广告（Search Ads / AdWords），Google 打造出了一台无与伦比的“印钞机”，并在 2004 年成功上市。然而，奇怪的事情紧接着发生了：上市后的 Google 竟然一头扎进了与搜索毫不相干的广阔领域。他们开发了基于浏览器的颠覆性邮件服务 Gmail、体验大幅超越时代的 Google Maps、支持实时协同编辑的 Google Docs、收购了 YouTube、自研了操作系统 Android，还推出了自己的浏览器 Chrome。时至今日，Google 拥有 15 款用户量突破 5 亿的产品，其中更有 7 款拥有超过 20 亿用户——这意味着地球上超过 25% 的人口每天都在使用这 7 款产品中的每一款。与此同时，他们也留下了无数轰轰烈烈的“惨败史”：为了对抗 Facebook 而推出的社交网络 Google+、昙花一现的 Google Wave 和 Google Buzz、反反复复推出的十几个即时通讯应用，以及试图提供空中 Wi-Fi 的热气球项目（Project Loon）和饱受争议的 Google Glass。 Ben 指出，Google 绝大部分营收从始至终都来自于网页搜索广告，那么他们为什么还要涉足这形形色色的领域？本期节目的核心，就是解构 Google 在 2000 年代作为“创新工厂”的历程、其重组为母公司 Alphabet 的背后逻辑，以及所有这些产品如何服务于其深层商业闭环与“组织全球信息”的宏伟使命，直至 AI 时代的序幕拉开。 ```mermaid flowchart TD accTitle: Google 从纯搜索走向多元巨头的业务演进 accDescr: 展示 Google 从核心搜索广告出发，逐步延展到多款数十亿级用户产品与 Alphabet 重组的逻辑演化 subgraph Core["核心印钞机 (1998-2004)"] Search["Google 网页搜索"] AdWords["AdWords 搜索广告模式"] Search --> AdWords end subgraph Expansion["全方位多产品扩张 (2004-2011)"] Gmail["Gmail 网页邮箱"] Maps["Google Maps 地图"] Docs["Docs 实时协同办公"] YT["YouTube 视频平台"] Chrome["Chrome 浏览器"] Android["Android 移动操作系统"] end subgraph Destination["战略终局与组织重构"] Alphabet["Alphabet 集团架构"] AI["通往现代 AI 时代"] end AdWords -->|"资金充沛二次下注"| Expansion Expansion -->|"抵御外部威胁 / 组织信息"| Alphabet Alphabet --> AI ``` --- ### [03:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=233s) 华尔街的疑虑：“醉酒杂耍者”与纯业务执念在正式拆解产品之前，David 引用了 HBO 知名科技喜剧《硅谷》（Silicon Valley）中亿万富翁角色 Russ Hanneman 的一句经典台词： > “如果你公布了营收，大家就会问有多少，而且永远都不会觉得够。原本能涨 100 倍、1000 倍的明星公司，突然之间就变成了只涨 2 倍的平庸之辈。但如果你还没有营收，你就可以说自己是‘预营收（Pre-revenue）’阶段，你是一个极具潜力的‘纯粹概念股（Pure Play）’。关键不在于你赚了多少钱，而在于你值多少钱——谁最值钱？那些一直在亏钱的公司！” David 解释道，Google 在 2004 年至 2006 年间面临的恰恰是“纯概念股（Pure Play）”的困境。2004 年秋季 IPO 之后，Google 股价在短短两个月内翻倍，华尔街对其推崇备至。在投资者眼中，Google 的逻辑无比纯粹完美：互联网发展越快，人们搜索次数越多；搜索次数越多，AdWords 带来的广告收入就越丰厚。这是一门近乎完美的纯粹搜索业务。转折点发生在 2005 年第四季度财报发布时：Google 在 2005 全年实现了 61 亿美元营收，较 2004 年的 31 亿美元翻了一番，但单季度利润却几乎持平，盈利能力明显下滑。原因在于管理层开始将大笔真金白银重新押注于看似无关的新项目——Gmail、Maps、即将问世的 Google Docs，以及在 2006 年斥资 16 亿美元收购 YouTube。 2006 年 1 月，由于华尔街极其厌恶这种分散注意力和利润的投入，Google 股价重挫 27%。科技记者 Steven Levy 在其名著《In the Plex》中尖锐地记录了当时外界的嘲讽：在华尔街眼里，Google 离开搜索业务向外乱撞的举动，就像是一个“喝醉酒的杂耍艺人（tossing balls into the air like a drunken juggler）”，他们正在把一门完美的好生意搞得一团糟。 - **2004 年全年营收**：31 亿美元。 - **2005 年全年营收**：61 亿美元（同比接近翻倍）。 - **2006 年 1 月财报后股价表现**：单月暴跌 27%。 - **核心分歧**：华尔街希望 Google 坚守“高利润率搜索中介”定位，而管理层则坚决投入下一代互联网基础平台。 --- ### [06:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=409s) 愚人节的震撼：Gmail 的颠覆性降临引发这轮创新狂潮且定下历史基调的第一个非搜索重量级产品，是 2004 年 4 月 1 日愚人节发布的 Gmail。在当时，这不仅是一个极具争议的产品，更被公认为科技史上最著名的“假戏真做”愚人节公告。在此之前，人们熟悉的免费网页邮箱是微软旗下的 Hotmail、雅虎邮箱（Yahoo Mail）以及 AOL。这些服务普遍只提供极小的存储空间——Hotmail 仅有 2 MB，Yahoo Mail 也不过 4 MB。用户必须反复清理收件箱、下载备份到本地或者逐条删除，像对待实体纸质信件一样费心归档。而 Google 在愚人节当天宣布：推出完全运行在浏览器中的网页邮箱，内置 Google 搜索功能，无需建立繁杂的文件夹，提供高达 1 GB 的免费存储空间，用户永远无需再删除任何邮件。由于这与行业现实相差数百倍，几乎所有人都以为这只是个玩笑。 Steven Levy 曾在书中提及一个生动的细节：在 Gmail 发布不久后，他在《新闻周刊》（Newsweek）纽约总部采访比尔·盖茨（Bill Gates）。当谈到 Gmail 竟然白送 1 GB 空间时，盖茨感到难以置信，甚至感到被冒犯。盖茨认为给用户这么大的空间完全是在浪费资源，“这在道德上是不可理喻的（morally repugnant）”，把所有未清理的邮件直接堆在服务器上简直是对计算资源的亵渎。然而 Google 的底层技术直觉是极其前瞻的。早在 2001 年该项目启动时，团队便根据摩尔定律与互联网普及率做出推论：数据传输、存储和搜索的边际成本必将无限趋近于零。随着存储成本的断崖式下跌，全世界传递的电子邮件量必然会呈指数级爆炸，因此“无限容量加全文搜索”才是未来无可逆转的标准形态。 --- ### [09:31](https://www.youtube.com/watch?v=QhAftC_zFr8&t=571s) 保罗·布赫海特与 1996 年的灵感起源为了彻底理清 Gmail 的诞生脉络，Ben 透露自己在录音前专门采访了 Gmail 之父——保罗·布赫海特（Paul Buchheit）。虽然外界通常认为 Gmail 的立项是在 1999 年至 2001 年间，但这个想法的火种早在 1996 年便已埋下。当时保罗还是克利夫兰凯斯西储大学（Case Western Reserve University）的一名学生。David 补充指出，凯斯西储大学在全美极具技术前瞻性，是全国最早在全校宿舍和整个园区铺设宽带互联网的高校之一。在全美绝大部分网民还在忍受拨号上网的龟速与按分钟计费的时代，保罗已经提前沉浸在了全天候、高速宽带连接的“未来世界”中，也正是这种超前的网络环境，激发了他构建完全基于 Web 的现代邮件系统的初始野心。 ### [10:20](https://www.youtube.com/watch?v=QhAftC_zFr8&t=620s) 保罗·布赫海特与 Gmail 原型的诞生早在 1996 年读大学期间，保罗·布赫海特（Paul Buchheit）就产生了一种强烈的直觉：必须把邮件下载并保存在本地电脑上的模式实在太糟糕了。信息理应触手可及，而数据传输的成本正在变得近乎免费。他在大学时代就痴迷于一个想法——电子邮件应该直接存在于网页和浏览器中，无需任何下载，并为此写出了一个简易的 Webmail 原型。1999 年，保罗在极客社区 Slashdot 上看到了 Google 的招聘信息，随即便向 `jobs@google.com` 发送了简历并成功入职——通过电子邮件被录用，冥冥之中注定了他日后的方向。到了 2001 年（Google 尚处于 Pre-IPO 阶段），拉里·佩奇（Larry Page）嫌公司推进速度太慢，做出了一个极为激进的决定：裁撤掉所有工程经理。随后，拉里与工程主管韦恩·罗辛（Wayne Rosing）亲自逐一找每位工程师面谈，探讨他们可以独立负责的项目。这一幕展现了早期“Googliness”文化的精髓以及当时招揽的人才质量——管理层直接询问工程师：“你有什么想法？我们这里有几个方向，你能否一个人全栈包揽一个完整产品？”由于他们了解保罗对 Web 端邮件的兴趣，便向他抛出了这个宽泛的概念。这一方面是出于产品构想，另一方面也是战略防御：Google 正在寻找能增强用户黏性的产品（正如雅虎凭借 Yahoo Mail 即使在搜索落后之后仍能长期锁住用户一样）。保罗接手项目后，最初的代码种子其实来源于 Google 的第一起收购案——Deja News（一个庞大的旧版 Usenet 帖子语料库，后来演化为 Google Groups）。当时保罗负责处理其实时索引系统以支持全语料检索。他灵机一动，将这套索引机制直接挂接到自己个人的 Unix 邮件目录上。最初形态的 Gmail，本质上就是一个用来检索他私人收件箱的搜索框，界面就像在搜索刚刚收购的 Usenet 归档帖子一样。 ```mermaid flowchart LR accTitle: Gmail 原型的演化逻辑 accDescr: 展示保罗·布赫海特如何将 Deja News 检索技术转变为第一代 Gmail 原型 subgraph S1["灵感与技术积累"] A["1996年大学设想: 免下载的纯网页邮件"] B["Deja News 收购: Usenet 实时索引技术"] end subgraph S2["原型孵化"] C["挂接个人 Unix 邮箱目录"] D["第一代 Gmail 原型: 邮件即时检索框"] end A --> D B --> C --> D ``` --- ### [13:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=791s) JavaScript 的复兴与 Ajax 的意外源头有了一个能快速检索的底层后端后，保罗面临的下一道难关是前端交互界面。当时市面上已经存在 Hotmail 和 Yahoo Mail 等基于网页的邮件服务，但体验极度糟糕：用户只要执行一个操作——打开邮件、写草稿、移动文件夹——整个网页就必须重新从服务器加载刷新一次，反应迟钝且割裂。为了让网页能够表现得更像流畅的桌面软件，保罗开始琢磨 JavaScript 的潜力。JavaScript 是布兰登·艾克（Brendan Eich）在 1995 年于网景（Netscape）发明的，初衷是让浏览器支持动态逻辑而不仅展示静态 HTML 文档。然而，在随后微软 Internet Explorer 绞杀 Netscape 的浏览器大战中，JavaScript 几乎成了牺牲品；到了 2001 年，它长期被主流开发者视为只能做做简单页面动画的“玩具语言”。即便是当时的 google.com 也是纯静态的：用户在搜索框敲入关键词，服务端处理后直接返回一个全新的静态结果页。保罗在深入钻研时，意外挖出了一个鲜为人知的接口：`XMLHttpRequest`。这个功能允许网页在后台静默向服务器请求并拉取 XML 数据，完全不需要刷新或重新渲染整个页面。保罗意识到这是无价之宝——正是这一发现，催生了后来席卷 Web 2.0 时代的革命性技术理念：Ajax（Asynchronous JavaScript and XML，异步 JavaScript 与 XML）。历史在此刻充满了讽刺意味：`XMLHttpRequest` 这个让 Google 彻底颠覆桌面软件的核心利器，最初竟然是由微软为了 Internet Explorer 自行研发并内置进去的，而微软当年开发它的唯一目的，是为了让自家 Outlook 网页版（Outlook Web Access, OWA）能勉强在 IE 里实现局部动态刷新。 - **Ajax 带来的质变**：在此之前，网络上几乎全是需要反复整页刷新的静态页面；Gmail 成为了全球首个获得大规模普及、树立 Web 2.0 标杆的 Ajax 网页应用。 - **划时代的用户体验**：用户无需再去电脑专卖店（如 CompUSA）购买光盘盒装软件安装在 PC 上，只需打开浏览器，就能获得不逊于原生桌面应用的响应速度。 --- ### [16:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=994s) 云端应用与来自微软平台的生存危机当保罗将工作原型展示给拉里·佩奇与谢尔盖·布林（Sergey Brin）时，两位联合创始人极其兴奋，并立即成为了 Gmail 最早的种子测试用户。他们迅速停用了其他邮件工具，在公司内部全天候依赖 Gmail 处理日常工作。很快，全 Google 的员工都切换并沉迷于这项服务。在“云计算”一词尚未普及的年代，Gmail 已经实现了真正意义上的云端体验：所有邮件安全存放在远端服务器上，不占用本地硬盘空间，用户可以在任何设备、任何网络环境下无缝登录访问。对拉里而言，做 Gmail 的初衷非常纯粹：拉里骨子里是一个极度重视体验的产品人，他坚信只要在 Web 端打造出体验超越以往的产品，就对 Google 有利。但在这层产品热情背后，还隐藏着一层更为沉重的战略梦魇——微软的全面威胁。彼时，Google 凭借搜索与 AdWords 缔造了商业史上最赚钱的印钞机，但它的根基脆弱不堪，因为所有流量和收益都建立在微软搭建的地基之上： - 超过 90% 的 Google 搜索查询来自于运行微软 Windows 操作系统的个人电脑； - 其中超过 90% 又是通过微软自家的 Internet Explorer 浏览器发起的。 ```mermaid flowchart TD accTitle: Google 在微软生态下的双层依赖危机 accDescr: 展示 Google 搜索业务受制于微软操作系统与浏览器的层级关系 subgraph MS["微软掌控的底层平台"] OS["底层操作系统: Microsoft Windows (占>90%)"] Browser["上层入口网关: Internet Explorer (占>90%)"] end subgraph GG["Google 业务层"] Search["核心印钞机: Google 搜索 & AdWords"] end OS --> Browser --> Search ``` Google 拥有 Web 上最杀手级的应用，却完全“苟活于微软的慈悲之下”。更要命的是，两家公司的底层商业飞轮存在不可调和的矛盾： - **Google 的飞轮**：用户使用 Google 搜索 $\to$ 发现更多优质网站与服务 $\to$ 花更多时间泡在网上 $\to$ 产生更多搜索需求 $\to$ 发现更多内容……Google 坐地通过网络广告将这整个流量繁荣期货币化。 - **微软的飞轮**：依赖 OEM 厂商销售预装 Windows 的 PC 硬件授权，再向用户兜售安装在本地的高价盒装软件（如 Office）。对微软而言，它完全不在意 Web 体验好坏，因为当时微软无法直接从公共互联网中变现；而如果下一代计算平台变成 Web，全世界的开发者都转向开发免安装的 Web 应用而非 Windows 原生应用，Windows 的护城河就会彻底崩塌。面对这种不可避免的替代风险，网页端应用的发展对微软构成了直指命门的生存威胁，这也拉开了两家巨头长达数年的全方位战争序幕。 ### [20:18](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1218s) 微软的盲区与 Google 的战略防线在 2000 年代初，微软内部的大多数人还没意识到 Web 作为应用平台的巨大威胁。此时的微软正被代号为 "Longhorn" 的系统拖得焦头烂额——这个最终演变成 Windows Vista 的项目成了一只沉重的战略信天翁，极大地分散了微软的注意力。在微软看来，只要牢牢控制着 Internet Explorer（IE），他们就仍然掌握着全局。然而 Google 对此看得一清二楚，尤其是 CEO 埃里克·施密特（Eric Schmidt）。施密特在加盟 Google 之前是 Novell 的首席执行官，而 Novell 当年正是被微软以极其残酷的手段无情碾压并摧毁的。施密特对微软的垄断杀伤力有着刻骨铭心的体会。因此，Google 内部对 Gmail 充满狂热的原因绝不仅仅是一款邮箱工具：他们迫切需要在普通用户心中建立起强大的杠杆。唯有让大众彻底爱上富 Web 应用（Rich Web Applications），未来一旦微软企图利用系统或浏览器特权去遏制 Google、打压 Web 应用的发展，庞大的用户群体才会起而反抗，成为保护 Google 免受微软绞杀的终极防线。 ```mermaid flowchart TD accTitle: Google's Defensive Flywheel Against Microsoft accDescr: Diagram showing how consumer adoption of Web apps breaks Microsoft's OS monopoly. subgraph MS ["微软传统护城河"] IE["控制 IE 浏览器"] OS["掌控 Windows 操作系统"] end subgraph Google ["Google 防御反击战略"] Gmail["打造极致 Web 应用 (Gmail)"] Love["用户产生极高粘性与喜爱"] Leverage["形成用户杠杆 & 倒逼开放 Web"] end OS -.->|"Longhorn/Vista 牵制注意力"| IE Gmail --> Love Love --> Leverage Leverage -->|"瓦解控制力"| IE ``` --- ### [21:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1276s) 长达三年的研发孤旅与“邀请制”神来之笔 Gmail 的研发从 2001 年一路持续到 2003 年，耗时整整三年才具备公开发布的雏形，随后甚至在“Beta”测试状态中停留了接近十年。在当时，这样的开发周期极为漫长，主要原因在于一切都是平地起高楼：当时业内几乎没有关于深层 JavaScript 开发的知识积累，更遑论后来被称为 AJAX 的 XML 动态无刷新异步通信技术。如今开发者习以为常的现代化抽象层和前端框架在当时完全不存在，每一个动态刷新的交互细节都要纯靠手工敲出来。到了 2004 年准备上线时，Google 遇到了一个极其棘手的工程与经济难题：他们承诺为每位用户提供高达 1GB 的免费存储空间。彼时市场上竞品（如 Hotmail、Yahoo Mail）通常只提供微薄的 4MB 空间。即使 Google 凭借卓越的自研廉价通用硬件集群，能够将基础设施成本压缩到同行的十分之一，但在完全没有 AWS 等公有云、必须自建数据中心的年代，1GB 打完一折后依然相当于 100MB 的实际成本，远超传统竞品的 4MB。如果任由公众疯狂涌入注册，服务器成本和算力需求瞬间就会被冲垮。为了控制基础设施负载并避免巨额亏损，Google 祭出了堪称神来之笔的“邀请制”（Invite System）： - **千人种子起步**：2004 年 4 月 1 日发布当天，Google 仅向 1,000 名种子用户发放了 Gmail 的邀请码。这些种子名额主要发放给科技圈内有巨大声量的关键人物（即今天所称的 Influencer）以及核心记者。 - **稀缺配额机制**：每位用户最初只能拿到极少量的邀请名额（大约 5 个）。一旦分完，后台补充邀请码的时间和数量完全动态不可预测，有时突然补 5 个，有时只给 3 个，完全取决于 Google 当天机房服务器能够挤出多少剩余算力。 - **病毒裂变与 eBay 炒卖**：这一被迫控制成本的技术限制，意外演变成了科技史上最具光环的饥饿营销。能用上 Gmail 迅速变成了一种象征品味的特权标志。在 eBay 上，一个 Gmail 邀请码的平均转让拍卖价格甚至被炒到了 150 美元。 - **牙刷测试（Toothbrush Test）**：正如拉里·佩奇（Larry Page）所言，Gmail 能够引发轰动不仅靠营销，关键在于产品过硬，通过了所谓的“牙刷测试”——它不是那种注册完就弃用的新鲜玩具，而是如同每天刷牙一两次一样，成为用户绝对离不开的日常生活习惯。 --- ### [25:24](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1524s) 变现探索：广告争议与 AdSense 的思想火花解决了成本雪崩的风险，团队紧接着面临商业化的问题：虽然更繁荣的 Web 理论上能带动搜索与间接收入，但邮箱本身如何直接变现？在 Gmail 原型开发阶段，保罗·布赫海特（Paul Buchheit）直接接入了 Google 现成的广告数据库。当时 Google 已经积累了庞大的关键词搜索广告库，保罗尝试根据用户的收件箱邮件正文进行“内容上下文匹配”（Content Matching），直接在邮件侧边栏展示相关的文本广告。出乎意料的是，原本专为搜索结果设计的广告，在结合邮件内容后依然表现出了极强的相关性与实用价值。然而，当这个功能在 Google 内部测试时引发了轩然大波： - **内部员工激烈抗议**：当时将 Gmail 作为日常办公邮箱的 Google 员工们愤怒质问：“你们在偷看我的私人邮件吗？！”日后外界在隐私层面对 Gmail 产生的所有道德与法律顾虑，在 Google 内部早已经过了一轮激烈的提前预演。 - **创始人一锤定音**：拉里·佩奇和谢尔盖·布林（Sergey Brin）却对此拍案叫绝，敏锐地意识到“这显然就是终极商业答案”。 - **AdSense 的先导雏形**：在时间线上，这项邮件文本匹配实验甚至早于 AdSense 的正式发布。当时 Google 仅有针对搜索结果页关键词广告的 AdWords，尚无面向第三方网站发布商的展示广告网络。Gmail 的这次探索直接启发了第一代 AdSense——将基于内容上下文的广告匹配逻辑，从收件箱平移到了海量第三方网站的内容页面中。 2004 年 4 月公开发布后，Gmail 彻底引爆了互联网。在接下来的二十年里，它从最初仅 1,000 名公测受众成长为一个坐拥超过 20 亿活跃用户的全球超级应用。即便如今许多高端用户使用 Superhuman 等高级第三方客户端，底层也依然首选 Gmail 的后端架构。 --- ### [28:03](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1683s) 战略觉醒：Web 作为万物集成的终极平台 Gmail 的空前成功让拉里、谢尔盖和施密特看清了未来的战略蓝图。他们得出了一个震动公司发展轨迹的结论：“我们必须疯狂复制这个模式，把所有能想到的东西全部搬进浏览器！” 这个战略布局全面打通了 Google 的业务引擎： 1. **做大 Web 生态盘子**：促使用户在浏览器中投入的时间成倍增加。网民只要停留在 Web 上，就会产生更多的搜索需求，直接推高 Google 核心搜索业务的营收；不仅如此，部分像 Gmail 一样的应用还能在内部直接实现自我变现。 2. **构筑对抗微软的“马奇诺防线”**：推动全球公众迅速爱上并深度依赖 Web 应用。只要人们的数字化生活迁移到浏览器内，微软通过 Windows 和 IE 对产业施加的控制力就会被彻底瓦解。 3. **重新定义“集成点”（Point of Integration）**：正如科技分析师本·汤普森（Ben Thompson）的理论，Google 意识到 Web 可以取代操作系统成为软件生态的核心集成点。硬件制造商、OEM 厂商、独立应用开发者以及终端用户，不再需要围绕具体的本地操作系统打转。用户只需要一个浏览器，OEM 只需要一个能打开浏览器的轻量系统，而所有的逻辑和价值均在云端与 Web 之中流转。对于 Google 的商业模式而言，最美妙之处在于其天然的外部性：无论是 Google 亲自下场开发并变现的应用（如 Gmail、Google Maps、Google Docs 以及后来的 YouTube），还是外界第三方所建立的任何成功 Web 服务，对 Google 而言都是巨大的胜利。只要能为 Web 的全面普及推波助澜，Google 就能源源不断地收获这股历史顺风带来的庞大红利。这直接催生了接下来 Google 在 Web 产品线上的全面大爆发。 ### [30:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1811s) 赞助商插播：摩根大通的嵌入式金融与无缝支付体验在继续深挖 Google 的产品矩阵之前，节目插播了一段关于摩根大通支付（J.P. Morgan Payments）的赞助介绍。Ben 和 David 提到，杰米·戴蒙（Jamie Dimon）在上个月与他们的访谈中也谈及了这笔长期投入。现如今，最顶尖的科技公司往往会把支付基础设施变得对终端用户完全“隐形”——无论是网约车打车、在电商平台购物还是管理订阅服务，最好的体验就是整个支付流程完全内嵌、毫无割裂感，用户根本不需要跳转到第三方独立页面。这种嵌入式金融（Embedded Finance）方案由摩根大通强大的底层 API 驱动。看似极度简化的前端体验，背后承载着极为复杂的资金流管理、合规、全球安全与清算网络。这正如 Google 所展现的产品逻辑：最简单、用户黏性最强的产品，底层往往隐藏着巨大的工程复杂度。凭借这套兼具弹性和安全性的全球级基础设施，摩根大通支付在 2024 年 Tear Sheet's Big Theory Awards 中斩获了最佳综合嵌入式金融平台奖。 ### [32:10](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1930s) 从 Gmail 到 Google Maps：全面拓展 Web 应用版图 Gmail 的巨大成功向全世界提供了一个确凿的“存在证明”：基于 AJAX 的 Web 应用不仅可行，甚至能在体验上碾压传统桌面软件。这一突破瞬间点燃了 Google 内部的野心：“我们现在真的可以构建真正的 Web 应用了，那就彻底放开手脚大干一场吧！”紧随 Gmail 之后，Google 瞄准的下一个重磅阵地就是地图（Maps）、文档（Docs）和表格（Spreadsheets）。在当时，业内绝大多数人根本不相信这些重型桌面级应用能搬到浏览器上运行。要做到这一点，需要极致的技术攻坚和超前的产品远见。Google Maps 的故事最早始于 2003 年，甚至早于 Gmail 的正式发布。当时刚刚从斯坦福大学毕业、作为 Google 首批 APM（助理产品经理）加入公司的布雷特·泰勒（Bret Taylor）——也就是后来联合创立 FriendFeed、出任 Facebook CTO、Salesforce 联席 CEO、Twitter 董事会主席以及 OpenAI 董事长的那位传奇人物——敏锐地嗅到了战略危机。布雷特跑去找联合创始人拉里·佩奇（Larry Page），直言 Google 正在错失重大机遇：AOL 刚刚斥资近 10 亿美元收购了 MapQuest，而小道消息透露 Yahoo 也即将大举发力推出 Yahoo Maps。拉里的第一反应非常纯粹而典型：“这是个 Web 产品吗？”得到肯定的答复后，拉里当即拍板：“去干吧！” ```mermaid flowchart TD accTitle: Google Maps Strategic Rationale accDescr: 解释推动 Google 开发 Maps 的双重动机：间接商业飞轮与直接公司使命。 subgraph Rationale["战略驱动力"] direction TB Mission["崇高使命： 组织全球信息，使其人人可用"] Flywheel["商业飞轮： 推动 Web 使用率 ➔ 搜索量增加 ➔ 广告印钞机加速"] end subgraph Product["核心标的：Google Maps"] direction TB Maps["Google Maps 项目启动 (2003)"] end Rationale --> Product ``` 对 Google 而言，布局地图存在双重合理性： - **间接商业飞轮**：推动更多人频繁使用 Web，从而带动 Google 搜索使用量，最终让背后的广告“印钞机”加速运转。 - **核心公司使命**：“组织全球信息，并使其普遍可访问、有用”。地图所包含的地理空间信息，恰恰处于这一使命的正中央。 ### [34:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2056s) 收购 Where 2 与地图交互架构的技术颠覆在当时，MapQuest 和即将登场的 Yahoo Maps 虽说体量巨大，几乎每个家庭和网民都在使用，但它们的本质完全是静态网页。那时的地图并不是用来在屏幕上实时交互漫游的，唯一的目的就是查询驾车路线，然后把路线指引**用打印机打印在纸上**带出门。这些服务的商业模式本质上就像“特洛伊木马报纸”——在打印出来的路线纸页边缘塞满横幅广告。布雷特·泰勒、拉里·佩奇以及玛丽莎·梅耶尔（Marissa Mayer）意识到，这绝不是地图该有的样子。于是，Google 在澳大利亚悉尼相中了一家名为 Where 2 Technologies 的小型初创公司。这家公司由拉尔斯·拉斯穆森（Lars Rasmussen）和延斯·拉斯穆森（Jens Rasmussen）两兄弟创立。拉斯穆森兄弟是无可挑剔的技术天才，他们已经开发出了一款极其惊艳的实时可交互平移地图，但唯一的问题是：它是一款必须安装在本地的 Windows 桌面客户端软件。在双方碰面时，拉里·佩奇提出了一个明确要求：“这正是我想要的，但必须跑在 Web 上。在 Google，我们热爱 Web（We like the web at Google）。”拉斯穆森兄弟展现出了神级工程能力：在短短三周之内，他们将整个底层软件架构全盘推翻重写，彻底改造为 Web 应用程序。在几乎没有参考内部方案的情况下，两兄弟凭借一己之力独立探索并实现了大量 JavaScript 与 AJAX 异步通信技术——而这与 Google 内部开发保密中的 Gmail 所使用的核心技术完全不谋而合。 Google 随即火速将 Where 2 收购囊中，奠定了 Google Maps 的内核骨架。与此几乎同时，Google 还连续收购了另外两家公司： - **ZipDash**：提供实时交通流量数据支持。 - **Keyhole**：拥有卫星三维地球技术，后来发展为独立的 Google Earth。 David 顺带提到了一个有趣的细节：早期的 Keyhole / Google Earth 曾是桌面端安装程序，后来很多人以为它被彻底合并进了地图，但如今访问 `earth.google.com` 依然能体验到完全运行在现代浏览器中、极为强悍的独立 3D 地球。在当时，Google Earth 的客户端分发实际上扮演了另一个神级特洛伊木马：在安装时捆绑推广预装在 Internet Explorer 上的 Google Toolbar，构成了那个时代最具破坏力的搜索分发黑客手段。 ### [37:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2234s) 2005 震撼上线与 2006 掀起 Web 2.0 巨浪的 Maps API 2005 年 2 月，Google Maps 正式向公众发布，整个互联网瞬间为之沸腾。无需刷新页面、用鼠标随意拖拽平移的动态网页地图，彻底颠覆了全球用户的认知。David 在这里分享了一个鲜为人知的首发日彩蛋： - 2005 年上线当天打开 `maps.google.com` 时，屏幕中央只能看到一片巨大的大西洋和北美大陆。 - 往大西洋中间看去，孤零零地漂浮着英国（UK）。 - 英国往东则是一片茫茫大洋——欧洲大陆、亚洲、非洲全都无影无踪，完全是一片空白海面，因为工程师团队当时根本还没来得及录入渲染这些区域的数据。所谓的产品 MVP（最小可行性产品），直接把半个地球给“淹没”了。紧接着在 2006 年，Google 作出了一个深远改变互联网历史的动作：向全球开发者免费开放 **Google Maps API**。这一举措真正引爆了 Web 2.0 时代的黄金爆发期： - **混搭（Mashups）潮起**：全世界的开发者开始疯狂利用 Google Maps API 与自身的数据进行交叉融合，构建出前所未见的新型网站。 - **孕育独角兽集群**：从房产巨头 Zillow，到后来的出行打车巨头 Uber、外卖平台 DoorDash、短租平台 Airbnb，这些改变现实世界的公司若没有底层成熟的 Google Maps API 作为地基，根本无从谈起。 - **催生“SoLoMo”风潮**：带动了当年轰轰烈烈的社交本地移动化（Social-Local-Mobile）大浪潮，催生了诸如 Foursquare、Gowalla 等整整一代基于地理位置服务（LBS）的创新产品。从战略防御角度审视，Google Maps 与 Maps API 构筑了抵御微软桌面霸权的铁壁铜墙。Google 不仅亲自打造出了日后服务数亿乃至数十亿用户的杀手级 Web 原生应用，更通过 API 生态让成千上万的初创企业将命脉深深扎根在 Web 端，使得 Web 对传统操作系统平台的锁定效应坚不可摧。在最初的 5 到 8 年时间里，Google Maps API 长期维持免费或极高调用阈值的超低收费模式。创建覆盖全球高精度地理数据的地图是一个吞噬海量资金与算力的工程深渊，在当时只有拥有极致狂热与使命驱动的 Google，才愿意不惜一切代价向这个无底洞中大笔砸钱，最终奠定了全球数字地图领域的绝对垄断地位。 ### [40:38](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2438s) Google Maps：疯狂的“Googly”豪赌与数十亿美元的战略资产在构建 Google Maps 的过程中，Google 做出了一个极其典型、充满创始人风格的“Googly”豪赌。他们意识到不能永远依赖第三方授权数据，于是决定从零开始自建底层地图：砸下数亿美元乃至数十亿美元，派出专门的车队穿梭于大街小巷拍摄街景，动态捕获海量地理图像，同时还要解决保护个人隐私、为人脸与车牌打码等极具挑战的工程难题。为了确保地理信息的即时更新，Google 还建立了一套众包更新体系，调动全球用户的力量维护地图。这种看似疯狂、耗资巨大且令人生畏的工程挑战，Google 硬是凭借工程师文化全盘接了下来。这项投资的战略价值不可估量。它不仅为后来的 AI 时代以及以 Waymo 为代表的自动驾驶汽车奠定了至关重要的数据基石，而且在商业上也取得了巨大成功。如今，Google Maps 拥有超过 20 亿的活跃用户。尽管 Google 从未在财报中单独拆分其营收，但外界普遍估计 Maps 每年能带来 50 亿至 100 亿美元的收入。其收入主要来自搜索推荐等赞助广告——就像在 Google 搜索中展示商户推广一样，当用户打开地图探索周边时随处可见——较小的一部分则来自面向开发者的 API 授权费用。 - **超 20 亿月活跃用户**：Google 旗下又一个跻身十亿级用户殿堂的超级产品。 - **预估 50 亿至 10 亿美元年营收**：以本地商家广告为主，B 端 API 接口调用授权为辅。 - **AI 与自动驾驶底座**：街景与动态地理空间数据成为 Google 发展具身智能与 Waymo 无人驾驶的核心壁垒。 --- ### [42:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2520s) 进军办公套件：直击微软帝国的心脏在确立了地图的领先地位后，Google 将目光投向了由 Google Docs、Google Sheets 以及后续的 Google Drive 组成的在线办公套件（即后来的 Google Workspace）。如果将整个套件与云端硬盘算在一起，其用户规模同样突破了 10 亿。但这不仅仅是一套工具的发布，更是 Google 第一次将手术刀直接切向微软最核心的利润命脉——Microsoft Office。在科技行业历史上，试图在生产力工具领域挑战微软的先驱数不胜数，从早期的 WordPerfect、Lotus 1-2-3 到 Lotus Notes，几乎全部铩羽而归。具有讽刺意味的是，当年收购并运营 WordPerfect 的公司正是 Novell，而时任 Novell 的 CEO 恰恰就是后来执掌 Google 的 Eric Schmidt。Eric 对正面迎战微软 Office 造成的惨烈代价再清楚不过。但 Google 并没有沿着传统桌面软件的老路与微软进行功能堆叠，而是凭借 Web 2.0 与 AJAX 技术开辟了全新的竞争维度。他们敏锐地察觉到，Web 应用第一次让一件前所未有的事情成为可能：**实时的多人多端在线协同**。 ```mermaid flowchart TD accTitle: Microsoft Office 与 Google Docs 竞争范式对比 accDescr: 对比传统桌面 Office 的护城河与基于 Web 的 Google Docs 协同颠覆模式 subgraph DesktopMoat["微软桌面护城河"] F["海量功能壁垒 Feature Wall"] P["专有文件格式 .doc /.xls 格式锁定"] EA["企业协议大包销售 Enterprise Agreement"] F --> P --> EA end subgraph WebDisruption["Google 网页端降维打击"] AJAX["Web 2.0 & AJAX 技术"] RealTime["实时多人在线协同"] FreeAccess["免安装 / 链接即用 / 零边际成本"] AJAX --> RealTime --> FreeAccess end WebDisruption -.->|"颠覆传统工作流"| DesktopMoat ``` --- ### [43:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2614s) Writely 与 Google Sheets：在浏览器中发明“实时协作” Google Docs 的前身是一家名为 Writely 的初创公司，由 Sam Schillace 和两位搭档共同创立。Sam 认为，Writely 很有可能是软件工程史上首款支持多人实时协作的软件，在 Web 出现之前这种架构根本无法成立。耐人寻味的是，团队最初只是想探索 JavaScript 和 AJAX 究竟能做什么，带着“在浏览器里做一个文字处理器会怎样”的极客心态写出了原型。他们起初压根没把协同作为卖点，但随着创始团队在开发中顺理成章地一起在线敲字，他们猛然意识到：“天哪，多人实时编辑才是颠覆性的杀手级特性（killer feature）！”Writely 于 2005 年 8 月推出，仅仅过了 6 个月，Google 就在 2006 年 3 月果断将其收购。 Google 电子表格（Sheets）的诞生逻辑则略有不同。负责孵化该项目的 Jonathan Rochelle（其创办的 2Web Technologies 同样被 Google 收购）坦言，起初大家甚至不知道在浏览器中运行电子表格是否可行。但基于 Gmail 的成功，Google 坚信可以在浏览器里实现实时计算。团队深知如果去比拼计算公式和复杂宏功能，绝无可能击败深耕数十年的 Excel。因此 Sheets 立项伊始便放弃了功能复刻路线，全力押注于分享与协同，聚焦于一个核心命题：能否让多个人在同一个基础表格上毫无冲突地同时狂欢输入？ - **Writely 极速收购**：2005 年 8 月上线，2006 年 3 月并入 Google，从发布到被收购仅约半年窗口期。 - **差异化竞争策略**：避开 Excel 几十年来构建的功能高墙，专注浏览器端多人即时交互的独特体验。 --- ### [45:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2735s) 为什么只有 Google 能挑战微软的 Office 霸权？在 Google 之前，任何独立公司想在生产力领域与微软对抗都是死路一条，这源于微软牢不可破的三重壁垒： 1. **数十年的功能积累之墙（Feature Wall）**：企业用户依赖无数微小而特定的边缘功能； 2. **专有文件格式的网络效应**：华尔街或咨询公司用复杂的.xls 搭建了庞大财务模型，离开 Excel 便无法正常流转与计算； 3. **企业协议（Enterprise Agreement, EA）绑销**：微软将全套办公软件直接打包卖给企业 IT 部门。企业并不需要每款软件都做到业界最优，只要平台大包全包，IT 主管便乐意全盘买单，外部独立软件几乎毫无切入机会。面对这套成熟的商业闭环，一家初创公司哪怕做出了在线文档，也无法承受支撑高并发实时协同所需的昂贵服务器基础架构，更无法突破缺乏商业变现途径的困境。但 Google 却拥有全球独一无二的解法： - **基础设施边际成本趋零**：支持 Docs 和 Sheets 多人协作的服务器负载，与 Google 每天承载的超大规模搜索流量相比九牛一毛，现有算力中心完全可以直接消化； - **不对称竞争与战略牵制**：Google 根本不需要靠办公软件直接向企业收软件许可费。微软靠 Office 赚取巨额利润，而 Google 的目标只有两个：让所有人习惯停留在 Web 上，以及用极低的成本给微软“上眼药”，让微软陷入战略焦虑与手忙脚乱。 --- ### [48:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2896s) 战略牵制与降维打击：从“像素级保真”到主导全球使用量 Google 的这一记突袭让微软陷入了被动防御。当时业界不断逼问微软：“你们针对 Office 的 Web 战略是什么？Word 和 Excel 什么时候能在浏览器里实现协同？”微软被迫将自己的王冠之珠搬上网页端，但这直接与其商业模式产生了剧烈冲突——如果用户可以在网页上免费用 Word，谁还会花大钱购买桌面许可证？复杂的许可和打包定价让用户困惑不已。更致命的是软件基因的束缚。主播 Ben Gilbert 分享了自己当年在微软实习的亲身经历：当时他在 Word 网页版团队实习，负责为网页版添加页眉和页脚。微软的核心工程要求是“代码移植与完美文档保真度（document fidelity）”——网页端必须与 Windows 桌面版完全一致，网页显示与打印机打出来的必须做到逐字符、逐像素（pixel for pixel, character for character）完全对齐。这种沉重的历史包袱构成了极其严苛的工程枷锁，导致其产品迭代缓慢且臃肿。而 Google Docs 完全没有这些顾虑。它无需安装、无需激活许可证，扔一个网页链接就能让所有人立即查看并协同编辑。时至今日，尽管两者的真实活跃用户数统计口径存在重合，但市场格局已经非常清晰：**Google 赢得了生产力软件领域绝大多数的用户量和使用时长，而微软依然守住了绝大部分的企业采购资金**。对 Google 而言，这已经是一场彻头彻尾的战略大胜。 ### [50:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3035s) 用户归 Google，金钱归微软：Docs 与 Office 的两重天地如果单纯对比用户规模，Google Docs、Sheets 和 Slides 展现出了极其惊人的统治力。这套在线办公套件中的每一个独立产品，用户量基本都在 5 亿到 10 亿之间；而放眼全球，微软 Office 的用户数实际上只有几亿。然而，商业世界的现实却呈现出截然相反的戏剧性反差。微软包含 Office 以及 LinkedIn 的生产力与业务流程部门（Productivity and Business Processes），上一财年斩获了超过 1200 亿美元的营收，而且全都是极其高毛利的纯利润。反观 Google，其办公套件 Workspace 甚至没有单独列出，而是被打包合并在 Google Cloud 部门中。涵盖全部 IaaS 基础设施、AI 算力以及 Workspace 的整个 Google Cloud 业务，上一年的总营收还不足 500 亿美元，其中办公产品的贡献只占很小一部分。正如两位主播笑谈的那样，这是一场奇妙的妥协与平衡：“微软赚走了所有的钱，Google 赢得了所有的用户，两边皆大欢喜。”在这段宿怨中，最具戏剧性的注脚当属 Writely（Google Docs 前身）的联合创始人萨姆·希拉奇（Sam Schillace）。他在卖掉公司后一度执掌 Google 的 Docs、Sheets 乃至 Maps 团队，而如今他的身份则是微软的副首席技术官（Deputy CTO），令人不禁感叹科技圈职业生涯的奇妙轮转。 - **微软生产力与业务流程部门年营收**：超过 1200 亿美元，依托高利润的企业级 Office 订阅。 - **Google Cloud 整体年营收**：不足 500 亿美元，包含了整个云基础设施、AI 算力和 Workspace。 - **用户体量悬殊**：Docs 各套件用户达到 5 亿至 10 亿级，远超 Office 的数亿体量。 ### [52:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3134s) 搜索王座之上的野心：用极低边际成本重构 Web 2.0 到了 2000 年代中后期，Google 在搜索领域的大局已定。尽管此时微软的 Bing 甚至尚未问世，但所有人都清楚，搜索市场只会变得越来越庞大，宛如一台永不停歇的印钞机。对于身处巅峰的 Google 而言，如果拉里·佩奇（Larry Page）坐在现场，他或许会直言不讳：一家企业的终极目标并不单纯是把商业规模做到最大，而是为了践行自己的使命——整合全球信息。凭借 Maps、Docs 和 Spreadsheets 的辉煌战绩，Google 逐步验证并确立了一套凶悍的战略打法：敏锐发掘具备潜力的 Web 2.0 应用，将其收入囊中，注入 Google 庞大的技术中台进行火箭式加速，最终为全球用户提供近乎魔幻的免费网络体验。更致命的是，Google 拥有无人能及的基础设施规模效益，他们承载并运营这些复杂服务的边际成本，仅仅是任何独立创业团队或竞争对手的一个零头。在巩固核心防御、拓展进攻阵地的同时，管理层萌生了一个更宏大的念头：既然这套模式行之有效，为什么不玩一把更大的？比如去涉足一个需要海量存储空间、吞吐天量带宽、甚至伴随着堆积如山版权官司的吞金巨兽——网络视频。 ```mermaid flowchart TD accTitle: Google Web 2.0 并购与赋能闭环 accDescr: 展示 Google 如何通过收购外部新兴应用，注入自身底层基建，达成商业与战略双赢的闭环机制。 subgraph Acquisition["外部发现与收购"] A["发掘 Web 2.0 明星原型 (如 Keyhole, Writely, YouTube)"] end subgraph ScaleEngine["Google 核心赋能"] B["注入极低成本的 海量算力与存储基建"] C["打通广告变现体系与 全网分发入口"] end subgraph StrategicOutcome["战略与商业回报"] D["为数十亿用户提供免费极致体验"] E["强化护城河，瓦解微软等桌面壁垒"] end A --> B B --> C C --> D D --> E E -.->|"源源不断的搜索利润反哺"| A ``` ### [54:20](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3260s) 中场赞助：Anthropic 与 Claude 的企业级智能进化在回顾 Google 从纯搜索引擎向平台型巨头演进的过程中，系统复杂度的爆炸式增长清晰可见——数以十亿计的用户、错综复杂的跨系统联动以及无休止的数据流转。然而放在当下的 AI 时代，当年的技术难题甚至显得有些“小巧可爱”。现代企业在拥抱大模型时所面临的系统规模与互联挑战，复杂度直接被放大了十倍以上。正是在这一背景下，Anthropic 的 AI 助手 Claude 成为了企业级开发与知识中枢的核心支柱。与一般的对话工具不同，Claude 专注于处理极其复杂的长线工程任务，例如耗费资深工程师数周时间的庞大代码库重构，或是数千页跨国合规文档的深度提炼与跨文本比对。凭借更低的幻觉率、持久的长上下文记忆能力以及创新的 MCP（Model Context Protocol）连接器标准，Claude 能够无缝集成到企业已有的 Jira、GitHub、HubSpot 与 Square 工作流中。目前 GitLab 等一线科技公司已深度采用 Claude 改造产研流水线。节目为 Acquired 听众提供了专属福利：访问 claude.ai/acquired 即可享受 Claude Pro 连续 3 个月半价试用。 ### [56:25](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3385s) 电视广告的肥肉与 Google Video 的迷途开局正式聊到网络视频这一庞然大物时，两位主播调侃起了播客早期的“黑历史”——早年间他们对 YouTube 的评分存在严重误判，今天必须在“Acquired 正典”中为其彻底平反重新打分。回溯历史，Google 进军视频领域的构想早在 2003 年便已萌芽，当时连 Gmail 都尚未正式上线。拉里·佩奇等人将手伸向视频，逻辑非常直接：视频包含着海量的信息，完全符合“整合全球信息”的宏伟愿景；而更为根本的经济驱动力在于，当时全球最丰厚、体量最庞大的广告资金池不在互联网，而在传统电视广告（TV Ads）手中。 ```mermaid flowchart LR accTitle: 电视广告与数字广告主导权的历史交替 accDescr: 展示从2003年电视广告占绝对主导，到2018年数字广告终于实现历史性反超的时间线与拐点。 A["2003年：Google Video立项 电视广告统治全球预算"] --> B["2000年代中期：玛丽·米克预警 用户注意力转移但广告变现滞后"] B --> C["2017-2018年：历史性拐点 数字广告总额首次超越电视广告"] C --> D["新生态主导者： YouTube / Meta / TikTok"] ``` 互联网女皇玛丽·米克（Mary Meeker）在当年的《互联网趋势报告》中曾年复一年地指出：全球用户的注意力早已大规模向数字网络倾斜，但广告预算的转移却严重滞后。这种割裂一直持续到 2017 至 2018 年前后，全网数字广告的总盘子才终于历史性地超越了传统电视。而在 2005 年，Google Video 作为由广告商业团队主导的项目正式亮相，却走上了一条令人啼笑皆非的弯路： - **电视节目字幕检索**：当时计算机视觉与自动语音转录技术尚不成熟，Google Video 早期本质上是一个“电视节目的搜索引擎”。团队通过抓取电视节目的隐藏字幕（Closed Captioning）来实现内容关键词检索。 - **缺乏原生播放器**：最初上线的版本甚至无法直接在网页上观看视频，它延续了传统搜索的流量中介逻辑——告诉你哪个台的哪个节目在什么时间段包含这段内容，或者直接把你重定向到第三方外部网站去收看。 - **错判内容重心**：它将宝全部押在了电视台制作的头部传统影视内容上，试图去索引“下一次在电视上播放的时间”，完全忽视了长尾的、由草根网民自发拍摄制作的用户生成内容（UGC）。而此时，Flip 便携录像机等平民数码拍摄设备的出现，正悄然引爆一场普通人自制视频的平民数字革命。 ### [01:01:02](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3662s) 便携相机的普及与 YouTube 的诞生在智能手机时代来临之前，数码卡片相机（Digital Point-and-Shoot Cameras）和类似 Flip Cam 的专用掌上摄像机迎来了爆炸式普及。当时大众普遍认为数码卡片机会成为继个人电脑之后的下一个核心消费电子载体。这一硬件普及浪潮让普通大众在 2000 年代中期第一次拥有了随时随地拍摄视频的工具，同时配合刚刚兴起的苹果 iMovie 等桌面剪辑软件，让个人视频创作与剪辑彻底飞入寻常百姓家。正是在这个用户原创内容（UGC）蓄势待发的大背景下，2005 年初，三位当时在 PayPal 相对年轻资浅的员工——查德·赫利（Chad Hurley）、贾韦德·卡里姆（Jawed Karim）与陈士骏（Steve Chen）离开了已并入 eBay 的 PayPal 团队，联手创立了 YouTube。关于创立早期的历史，节目中提到了两则冷知识： 1. **最初的定位与口号**：YouTube 最初并不是全品类视频网站，而是一个视频约会平台，标语是“Tune In, Hook Up”（上线配对）。他们甚至还在旧金山湾区的 Craigslist 上投放广告，出资招募有吸引力的女性上传视频个人资料，结果几乎无人问津。在遭遇冷场后，团队迅速转型为允许任何人上传任何视频的通用平台。 2. **隐秘而强大的人脉背景**：出任 CEO 的查德·赫利，当时的岳父是硅谷元老级风投传奇吉姆·克拉克（Jim Clark，Silicon Graphics 与 Netscape 的创始人）。三位创始人不仅自带“PayPal 黑手党”的光环，更背靠克拉克这位能够指点江山的行业导师。 ### [01:03:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3830s) YouTube 的三重飞轮与潜在威胁 YouTube 之所以能以惊人的速度脱颖而出，核心在于极其巧妙的三大支柱模式： 1. **极致简化的内容供给（Killer Content Acquisition）**：任何人在任何时间都可以上传任何格式的内容。视频一经服务器转码完毕立即上线，不设任何前置版权审查。相比之下，Google 自家的 Google Video 要求人工耗费一到两天逐一审查、核实版权并批准后才能发布，这种流程在 UGC 时代根本无法规模化运作。 2. **零门槛的内容消费（Killer Content Consumption）**：基于 Web 端 Flash 技术打造了稳定出色的视频播放器，用户打开网页就能即时点播，体验流畅，解决了当时在网页上看视频频遭崩溃的痛点。 3. **病毒式的分发模型（Killer Growth and Distribution）**：外链嵌入（Embed）机制。用户可以将视频播放框无缝嵌入到 MySpace、个人博客等外部网页，实现跨站传播。此外，YouTube 自诞生起就内嵌了强大的视频搜索引擎。短时间内，YouTube 内部的搜索量就飙升到了仅次于 Google 搜索的世界第二大搜索引擎。这让 Google 感受到巨大的潜在战略威胁：就像马克·扎克伯格（Mark Zuckerberg）后来对 WhatsApp 垄断即时通信心存忌惮一样，即便 YouTube 当时不打算做全网网页搜索，但当海量用户的注意力、使用习惯与搜索行为全部聚合在 YouTube 平台时，它随时具备向外延展搜索边界的能力。 ```mermaid flowchart TD accTitle: YouTube 三重飞轮模型 accDescr: 展示 YouTube 依靠超低门槛供给、即时网页播放与外链嵌入形成的用户与流量飞轮 subgraph Acquisition["1. 内容供给端"] A["任何用户随时上传"] --> B["后台自动快速转码"] B --> C["即刻发布（免前置版权审查）"] end subgraph Consumption["2. 内容消费端"] C --> D["极佳的 Web 网页播放器"] D --> E["站内快速视频搜索"] end subgraph Distribution["3. 病毒式传播端"] D --> F["无缝跨站外链嵌入 (Embeds)"] F --> G["爆炸式流量与注意力聚拢"] G --> A end ``` ### [01:06:19](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3979s) 烧钱黑洞与 Google 的 16.5 亿美元收购作为创业初期的初创公司，YouTube 拥有 Google 这种上市公司所不具备的“法律豁免心态”——把版权责任推给避风港原则，先野蛮生长再说；但与此同时，这种狂飙突进也带来了巨大的生存危机。这一时期，大量受版权保护的热门电视内容被用户传上网站。典型代表便是《周六夜现场》（SNL）由 The Lonely Island 制作的短剧《慵懒星期天》（Lazy Sunday），仅这一个免广告、免费观看的盗版短剧视频，就让当时独立运营的 YouTube 网站总流量直接暴涨了 83%。红杉资本（Sequoia Capital）合伙人鲁洛夫·博塔（Roelof Botha，同为 PayPal 前同事）在 YouTube 创立之初便迅速领投，并在随后的轮次中继续追投，但资金消耗速度令人咋舌。 YouTube 面临着三座难以逾越的高山： - **高昂的单次处理成本**：视频上传后需要针对各类设备与浏览器进行昂贵的多规格格式转码（Encoding）。 - **巨大的可变存储开销**：海量用户上传的视频必须永久保存在磁盘上。 - **毁灭性的带宽流量账单**：网络带宽成本完全随着播放分钟数（Minutes Watched）线性放大，在没有跑通视频广告商业模式之前，播放量越大，亏损越严重。伴随而来的还有以维亚康姆（Viacom）为首的传统传媒巨头提起的毁灭性版权诉讼。如果放在现代，YouTube 或许可以在一级市场融资数百亿美元，并借助 AWS 等公有云基础设施以独立公司形式运转并最终上市；但在 2005 至 2006 年间，私有资本规模有限，且身负巨额诉讼根本不可能进行 IPO。面对版权方无休止的施压与即将耗尽的现金流，YouTube 在上线不到 18 个月后挂牌出售。而坐拥庞大自建低成本服务器基础设施、亟需阻截潜在竞争对手的 Google，在 2006 年 11 月以 16.5 亿美元的全股票对价正式将 YouTube 收入囊中。这笔全股票交易在日后让卖家收益暴增，甚至传闻让 Google 随后的首席财务官帕特里克·皮谢特（Patrick Pichette）一度感叹“再也不做这种全股票并购了”。 - **三位创始人**：查德·赫利（Chad Hurley）、贾韦德·卡里姆（Jawed Karim）、陈士骏（Steve Chen），均出身自 PayPal 早期工程与设计团队。 - **早期关键投资人**：红杉资本合伙人鲁洛夫·博塔（Roelof Botha），该项目为他加入红杉后的首批核心投资之一。 - **流量催化剂**：SNL 短剧《Lazy Sunday》上传后使 YouTube 单周/短期全网流量暴增 83%。 - **收购时间与对价**：2006 年 11 月，产品上线不足 18 个月，Google 支付 16.5 亿美元股票完成收购。 ### [01:11:17](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4277s) 每播放一次就烧掉一分钱：“谷歌的第一个大错误” 即使谷歌当时是以价值 16.5 亿美元的股票完成了对 YouTube 的收购，而在此后多年里谷歌的市值飙升了 20 倍以上（意味着如果当年付全额现金，谷歌能多赚整整 20 倍的账面回报），这笔交易依然被公认为科技史上最划算的买卖之一。然而在收购初期，公司内部和华尔街的感受却完全相反。前 YouTube 产品负责人兼实质上的 CPO/CTO 希希尔·梅赫罗特拉（Shishir Mehrotra）在 Kleiner Perkins 的播客节目中披露了大量早期内幕数据：收购刚完成时，YouTube 每年的营收仅有大约 3000 万美元（主要依靠 DoubleClick 广告网络做程序化广告变现），但每年的亏损年化额度（Run Rate）却高达近 10 亿美元。最触目惊心的一个统计是：YouTube 当时的亏损额几乎严丝合缝地等于“每次播放亏一分钱”。这意味着用户每在 YouTube 上刷新播放一次视频，谷歌就等于直接往马桶里扔掉一美分。面对这种几何级数膨胀的流量，当时的谷歌首席财务官（CFO）每天都活在恐慌之中，甚至在内部哀求：“求求你们，千万别让它再增长了！”然而覆水难收，用户增长根本停不下来。财务团队甚至秘密研究能否把 YouTube 倒手卖给当年同样参与竞购的雅虎或其他传媒巨头。在很长一段时间里，YouTube 在谷歌内部被普遍称为“谷歌的第一个大错误”；就连 Acquired 节目十年前做回顾时，也曾公开判定这是一笔极其糟糕的收购。 - **天价音乐版权**：除了服务器带宽支出，昂贵的音乐版权授权协议也是吞噬现金的无底洞，YouTube 在极早期就成为了全球唱片业最重要的收入来源之一，体量堪比后来的 Spotify。 - **纯粹主义投资者的不满**：谷歌长期以来是以搜索与文字广告为核心的高利润“纯血互联网公司”，而 YouTube 这种重资产、重带宽、高亏损的视频流媒体，让追求高毛利的公开市场投资者极为排斥。 --- ### [01:13:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4420s) 吞噬互联网 20% 流量的怪物：从搜索框到算法推荐早期的 YouTube 在产品逻辑上与今天截然不同。用户最初发现 YouTube 视频几乎全靠外部网站的内嵌播放器（Embed），比如 MySpace 博客或论坛；看完后顺手点几下就离开，下一次访问依然来自其他网页的跳转，极少有人直接以 YouTube.com 作为上网冲浪的起点。在被收购后的头四五年里，即便用户直接打开 YouTube 主页，也有超过 90% 的流量是奔着搜索框去的，人们只在有明确目标时来搜索特定视频，对于旁边的任何推荐内容都视若无睹。要培养用户的浏览习惯，并打磨出足够精准的推荐算法，需要耗费数年的技术积累。与此同时，视频对全球网络基础设施的侵蚀达到了惊人的地步。2007 年的一份行业报告估算，仅 YouTube 一家网站当年消耗的带宽总量，就已经相当于 2000 年整个人类互联网的带宽总和；而根据希希尔·梅赫罗特拉给出的数据，到了 2014 年，全球互联网上流转的所有数据比特中，有整整 20% 都来自 YouTube。然而，谷歌顶住巨额亏损坚持下来的长远逻辑，正如分析师玛丽·米克（Mary Meeker）在互联网趋势报告中所指出的：人类天生渴望消费视频，而注意力的聚集地就是广告变现的金矿。谷歌清醒地意识到自己内部研发的 Google Video 已经落后，与其自己重新造轮子，不如直接吞并已经抓住时代机遇、占据用户心智的 YouTube。 ```mermaid flowchart TD accTitle: YouTube 商业化扭亏为盈与基础设施演进路径 accDescr: 展示 YouTube 从早期高带宽亏损到移动化转型、推荐算法成熟并实现盈利的演进过程。 subgraph Phase1["2006-2008：巨亏瓶颈期"] A["外链内嵌为主 90% 站内访问依赖搜索"] --> B["带宽成本爆炸 每次播放亏损 1 美分"] B --> C["年亏损近 10 亿美元 内部被称为'谷歌的错误'"] end subgraph Phase2["2009-2012：商业化拐点"] D["2009 年广告收入翻三倍"] --> E["2010-2011 年实现扭亏为盈"] E --> F["2012 年年营收约 40 亿美元 达到大致收支平衡"] end subgraph Phase3["2013 以后：飞轮成型"] G["核心指标转向 Watch Time"] --> H["移动端普及与个性化推荐"] H --> I["55% 创作者分成生态"] end Phase1 --> Phase2 Phase2 --> Phase3 ``` - **盈利拐点（2009–2012）**：2009 年 YouTube 广告收入同比暴增三倍，成为业务起飞的起点；2010 至 2011 年间公司正式转亏为盈；到 2012 年，全平台年营收已达到约 40 亿美元，实现了稳固的微利与规模平衡。 --- ### [01:16:31](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4591s) 15 分钟的娱乐使命：移动化、创作者分成与算法的胜利到了 2013 至 2015 年间，YouTube 迎来了一场根本性的产品蜕变。团队确立了一个全新的北极星指标：“确保用户只要打开 YouTube，就能开开心心地被娱乐 15 分钟”，而产品的全部工程目标就是扫平通往这 15 分钟的一切障碍。推动这一变革的核心催化剂是智能手机的全面普及——作为初代 iPhone 的预装合作方，移动端为 YouTube 带来了海量的“低意图访问”（Low-intent Sessions）。用户不再是为了找特定教程或特定片段而来，只是在闲暇时顺手点开 App 找乐子；同时，手机端天然的高账号登录率，让算法能搜集精准的用户行为，驱动个性化推荐飞速迭代。在内部考核上，团队也果断废弃了单纯的“播放次数”（Views），全面转向“观看时长”（Watch Time）。更深远的是其商业模式的颠覆。YouTube 极早推出了创作者分成计划，将约 50%（实际多为 55%）的广告收入直接分给视频作者。尽管在 Acquired 早期的节目中，主播曾批评这种重度分成是比谷歌搜索广告或 Facebook/Instagram 差得多的商业模式（Meta 平台上的网红几乎拿不到官方广告分润），但正是这种慷慨的利益捆绑，让无数普通人得以全职在互联网上建立事业，彻底改变了全球内容产业的生态。 > “在 YouTube 内部，围绕分发机制曾爆发过长达数年的路线之争：第一派主张‘关注订阅为王’，认为最强的互动来自于用户自主挑选关注的博主；第二派则坚信‘算法至上’。不幸的是，最终被证明正确的正是算法派。绝大多数时候，机器推算出的下一条视频，就是比订阅列表里的内容更能让你一直看下去。” 这种算法对订阅关系的架空，构成了今天 YouTube 创作者经济复杂的一面：一个人哪怕拥有数百万订阅者，播放量依然完全取决于算法是否推荐。但无论如何，只要一个人拥有一台摄像设备并产出被算法认可的内容，就能跨过一切传统中介直接获取商业回报，这正是互联网赋能个人所能达到的极致形态。 ### [01:20:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4828s) 广告分成文化与 YouTube 飞轮的“登录态”命门在讨论 YouTube 的商业模式前，必须对比 Google 的两项核心广告支柱：首先是搜索广告，广告主向 Google 付费，除了支付给分发渠道的流量获取成本（TAC, Traffic Acquisition Costs）外，Google 几乎能够保留 100% 的广告收入；其次是以 AdSense 和 Google 内容联盟网络为代表的展示广告，Google 会将约 70% 的绝大部分收入分给外部网站所有者与发布商。YouTube 的商业架构恰好处于这两者之间——它既是一个第一方平台，又必须像内容网络一样将大量收入分给创作者。正是因为 Google 早已在 AdSense 体系中习惯了与内容创作者进行巨额分成的商业模式，他们才能毫不犹豫地推行与视频创作者五五分成的机制；如果 Google 当初仅仅是一家纯粹的搜索引擎公司，这种 50% 的让利分成恐怕会在公司内部引发巨大阻力。与此同时，移动端与桌面端呈现出一个关键差异：移动端用户处于“登录态（Logged-in）”的比例显著更高。对于传统的 Google 搜索而言，用户是否登录对其有效性和商业化并没有根本性的影响——用户如果搜索“铁锹”，系统就会展示铁锹的广告，底层的极强搜索意图直接构成了转化下漏斗，Google 压根不需要掌握你详尽的人口统计学特征；Gmail 等产品的登录状态固然很好，但对搜索业务的商业提升幅度有限。然而，YouTube 却完全不同：整个 YouTube 的推荐飞轮与商业化运转，完全依赖于“登录态”。无论是精准分发视频内容还是进行广告投放，YouTube 都与传统电视广告的运作逻辑高度契合——它极其依赖受众的人口统计学数据（例如雪佛兰赞助橄榄球赛事的逻辑），必须了解具体是哪一类群体正在屏幕前观看。 ```mermaid flowchart TD accTitle: Google 核心业务商业模式与登录态对比 accDescr: 对比搜索广告与 YouTube 在收入分成及登录态依赖度上的核心差异 subgraph Search["Google 搜索模式"] S1["强搜索意图 (如: 搜'铁锹')"] --> S2["高转化广告"] S2 --> S3["获得近 100% 广告收入"] S4["登录态: 可有可无 (Nice to have)"] end subgraph YouTube["YouTube 模式"] Y1["依赖登录态与画像数据"] --> Y2["电视级受众精准定向"] Y2 --> Y3["50% 分成给创作者 / 50% 留存平台"] Y4["登录态: 核心命门 (Essential)"] end ``` ### [01:22:45](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4965s) 破除历史质疑：从烧钱黑洞到超 500 亿美元超级巨兽两位主播开始对 Google 收购 YouTube 这一历史交易进行重新打分评级。回溯到 2016 年的节目，当时他们对 YouTube 抱有两个核心质疑：其一，怀疑它根本不是一个终点型网站（Destination Site）；其二，平台只能留下 50% 的总流水，却要背负极其可怕且看不到尽头的底层基础设施成本。如今事实彻底粉碎了当年的论调：YouTube 不仅是终点站，更是世界上最强大的“目的地 App”——用户每晚睡前打开它，把时间全盘托付给其推荐算法，沉浸在订阅与非订阅的精选视频之中。在财务与成本控制方面，YouTube 的工程团队展现了长达二十年的极致优化： - 2024 年单单广告收入就达到了 360 亿美元，即使扣除分给创作者的一半，留给 YouTube 自身的也有 180 亿美元。 - 为压低视频托管、带宽、转码计算及音乐版权等可变成本，YouTube 甚至自主研发了专用于视频转码的定制芯片（Custom Silicon）。 - 引入极为精细的阶梯转码策略：新上传的视频默认采用通用的 H.264 编码；当视频播放量上升到一定台阶后，系统会调用计算成本更高但传输体积更小的格式重新编码；若播放量突破 500 万次等更高门槛，则会再次进行高压缩比编码，以此最大化降低每次播放的边际带宽支出。 - 如果将 YouTube Premium、YouTube Music 以及 NFL Sunday Ticket（NFL 周日超级转播套餐）等订阅收入合并计算，YouTube 2024 年的总营收已经突破 500 亿美元大关。这个体量意味着 YouTube 已经超越了 2024 年总营收为 390 亿美元的 Netflix，成为仅次于迪士尼的全球第二大媒体实体；由于迪士尼的总收入中还包含主题乐园、邮轮与周边衍生品等大量非媒体业务，YouTube 实际上早已超越了迪士尼纯媒体业务的规模，且整体营收有望在短期内超越迪士尼整个集团。在用户使用时长维度上，虽然 Facebook 主应用与 WhatsApp 的总用户规模更大，但 YouTube 几乎毫无悬念地是整个人类互联网上总使用时长最长的超级应用，堪称人类有史以来最大的“注意力时间黑洞”。 ### [01:26:12](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5172s) 重估收购：社交演变为视频时代的战略奇迹（A+ 评级）根据知名研究机构 MoffettNathanson 今年的测算报告，YouTube 每年的营业利润（Operating Income）大约在 80 亿美元左右。回顾 Google 的初始投资成本：16.5 亿美元的收购对价，加上后续几年间每年的亏损（单年亏损从未显著超过 10 亿美元，且在约 5 年内就实现了盈亏平衡），Google 的累计总投入大约在 55 亿到 67 亿美元之间。以大约 60 亿美元的总沉没资本，换取一个如今每年源源不断产出 80 亿美元净利润、且营收每年依然保持 10% 到 15% 复合增长的超级巨兽，其投资回报率堪称商业史上的神迹。更深层次的价值在于 Google 战略维度的“三重底线”：不仅收获了财务营收与利润，更完成了对竞争对手的战略阻截，以及对“整合全球信息”使命的延续。最具戏剧性的转折在于社交媒体的演化路径：Google 本身在传统社交网络上彻底溃败，但随着时间推移，整个社交网络形态发生了剧烈裂变——传统的“熟人社交网络”已经瓦解，分裂为“私密通讯”与“公开媒体”。无论是在 TikTok 还是 Instagram Reels 上，用户刷到的绝大多数都是陌生人制作的趣味视频。曾经击败 Google 的社交巨头们，最终全盘转型成了“类 YouTube”的算法推荐内容媒体，兜兜转转回到了 Google 依靠 YouTube 占领的腹地。不仅如此，YouTube 还坐拥全网最庞大的高质量视频语料库，这成为了 AI 大模型时代不可复制的核心训练资产。MoffettNathanson 评估认为，如果 YouTube 作为一家独立公司公开上市，其市场估值可达 5000 亿美元；即便剔除未来战略溢价并采取传统媒体公司的保守倍数，其估值也至少在 2000 亿美元。综合其现金流与防御壁垒，两位主播毫无争议地推翻了当年的评分： - 商业评级：正式将 Google 对 YouTube 的收购评级从当年的 **C** 上调至当之无愧的 **A+**（甚至可以说是 **A++**）。 - 战略总结：尽管不能简单地将其视为“将 60 亿变成 5000 亿”的投机，但这毫无疑问属于商业历史上最伟大、最成功的收购案之一。 ### [01:30:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5400s) 告别 YouTube，重返 Google 的核心广告业务在复盘 YouTube 这笔交易时，主播们指出最初支付的 16.5 亿美元主要是用 Google 的股票支付的。虽然动用自身股票支付伴随着巨大的机会成本，但从结果来看，无论怎么算，这依然是一笔令人难以置信的“血赚交易”（"screaming deal"）。在打趣地“修正了历史记录”之后，对话转向了 Google 帝国真正的基本盘——网络广告业务，同时也回归到了播客《Acquired》最纯粹的创办初衷：探讨商业史上最伟大的并购案例。紧随 YouTube 之后的正是另一场载入史册的重磅收购：DoubleClick。在深入 DoubleClick 之前，节目插入了赞助商 Statsig 的介绍。早在 2004 年至 2005 年期间，Google 之所以能在搜索领域建立起无人能及的壁垒，很大程度上归功于其先驱性的数据驱动产品文化。Google 将这种严谨推向了极致：他们曾测试过 50 种不同深浅的蓝色链接，以找出点击率最优的一款；利用用户主动修正拼写的数据训练出了著名的“您是不是要找……”（Did you mean）纠错功能；甚至近期还将用户随机分组放入 A/B 测试中体验 AI 搜索。Statsig 正是将这种以往只有 Google 等科技巨擘才拥有的底层测试框架、功能标记（feature flags）和数据分析工具，开放给包括 OpenAI、Figma、Notion 和 Anthropic 在内的现代科技企业。 ### [01:32:55](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5575s) DoubleClick 的前世：从互联网泡沫之星到被私募贱卖如果说 2006 年 10 月以 16.5 亿美元收购 YouTube 已经是一笔天文数字，那么仅在数月之后的 2007 年 4 月，Google 再次豪掷 31 亿美元现金（全额现金，而非股票）买下 DoubleClick，金额直接翻倍。这标志着 Google 全面进军展示广告（Display Advertising）领域。当时 Google 内部拥有两条主要的广告业务线：一是针对搜索关键词结果展示的 AdWords；二是在第三方网站网络分发文本广告的 AdSense。与 YouTube 这种急速蹿红的初创黑马不同，DoubleClick 的故事要深厚得多，这在阿里·帕帕罗（Ari Paparo）的新书《Yield》中有详尽记录： - **创立与上市**：DoubleClick 成立于 1995 年（甚至早于 Google 诞生），由凯文·奥康纳（Kevin O'Connor）和德怀特·梅里曼（Dwight Merriman）在纽约创立。其业务由两部分构成：一是帮助广告主跨网站投放广告的软件系统——广告服务器（Ad Server）；二是汇聚媒体资源的广告网络（Ad Network）。1998 年它作为互联网泡沫初期的明星公司成功上市。 - **泡沫破裂与业务剥离**：2000 年互联网泡沫破裂后，DoubleClick 遭受了毁灭性打击，多达 70% 的客户不仅流失，而且直接破产关门。由于当时品牌广告预算尚未大规模迁入互联网，其客户大多是像 Pets.com 这样互相烧钱互投的 VC 初创公司。2002 年，苟延残喘的 DoubleClick 以区区不到 1500 万美元（以“M”为单位）的白菜价抛售了其广告网络业务，只保留了核心的广告服务器软件。 - **私募接盘**：到了 2004 年，DoubleClick 变成了一家增长缓慢、市值缩水的“沉睡型软件公司”。虽然其广告服务器依旧被业界广泛使用，但线上营销盘子总体有限。在寻求出售时，Google 曾与其会面评估但最终放弃。2005 年，私募巨头 Hellman & Friedman 与 JMI Management 联手以约 10 亿美元将其私有化收购——这个价格只有其 IPO 峰值的一半。 ### [01:36:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5795s) 私募操刀与 Ad Exchange 的诞生：程序化展示广告革命在私募机构接管后，大卫·罗森布拉特（David Rosenblatt）出任 CEO，而他麾下的产品与战略负责人，正是后来主政 YouTube 的现任 CEO 尼尔·莫汉（Neal Mohan）。许多行业观察家后来一致认为，Google 收购 DoubleClick 最宝贵的资产之一，就是得到了尼尔·莫汉这位顶尖商业天才。 ```mermaid flowchart TD accTitle: 程序化广告交易平台演进流程 accDescr: 展现从传统媒体与广告网络到广告交易平台（Ad Exchange）的实时竞价演变 subgraph OldModel["传统分发模式"] P1["大型发布商 (如 New York Times)"] -->|"直接销售"| D1["预留广告位 (例如 Ford)"] P1 -->|"长尾剩余流量"| AN["单一广告网络 (Ad Network)"] end subgraph ExchangeModel["DoubleClick 创新：Ad Exchange 模式"] P2["发布商 / 核心网页"] --> ADX["广告交易平台 (Ad Exchange)"] ADX -->|"实时竞价 (RTB)"| B1["广告网络 A"] ADX -->|"动态加价比对"| B2["代理商交易台 (Trading Desk)"] ADX -->|"甚至压过直销单"| B3["外部程序化竞价方"] end ``` 在私募资本支持下，DoubleClick 迎来了一场彻底蜕变：他们开发出了颠覆整个数字营销行业的全新产品——广告交易平台（Ad Exchange）。在此之前，数字广告逻辑极其单一（仅有广告服务器与广告网络两端），而 Ad Exchange 的划时代构想在于：在各大广告网络之间实现跨平台的流量与需求智能路由。 - **从吃“残羹冷炙”到重构规则**：Ad Exchange 诞生之初原本只用于处理所谓的未售出“剩余流量”（Remnant Inventory），即网站未卖出的空余展示位，通过引入外部实时竞价（RTB）机制来榨取额外变现价值。 - **颠覆传统的实时竞价（RTB）**：技术的底层逻辑极为精密复杂。它支持毫秒级的实时竞价，甚至允许外部出价方直接与媒体的直销订单（Direct-sold ads）进行竞争。例如，即便《纽约时报》与福特汽车签有固定的排期合同，如果有外部广告主愿意在特定曝光上开出更高的实时竞标价，也可以瞬间将原广告替代。 - **向上截流与底层化**：这一机制使 Ad Exchange 跃升到了所有传统广告网络的前端，形成了一层彻底的“去中介化”（disintermediation）。它成了各大媒体、代理商交易台（Agency Trading Desks）以及规则引擎必须对接的最底层基础设施。原先用来卖长尾流量的工具，最终反客为主，成为了主流品牌广告主和大型发布商采购数字媒体的最核心渠道。与 DoubleClick 这套契合麦迪逊大道（Madison Avenue）复杂商业关系、多层代理利益与高级博弈的成熟系统相比，Google 当时的 AdSense 显得过于简陋——它主要服务于小型个人站长，依靠自助式的纯技术流逻辑运作，宛如工程师乌托邦式的玩具，根本无法满足主流品牌巨头和各大传媒集团复杂的现实交易需求。 ### [01:39:18](https://www.youtube.com/watch?v=QhAftC_zFr8&t=5958s) Google 的展示广告困境与对 DoubleClick 的渴望进入 2000 年代初，麦迪逊大道的广告行业已经完成了从传统《广告狂人》时代向数字化的深刻演变。广告代理商手中握着大品牌最丰厚的预算，并通过直连财务系统的“吸金粗水管”（fat pipes）控制着庞大的资金流动。然而，Google 此时在广告生态中却面临数个棘手难题。首先，DoubleClick 之所以能完成频次控制（frequency capping，避免用户把同一则广告看 46 遍）等高级定向功能，依赖的是第三方 Cookie（third-party cookies）；而当时的 Google 在理念上极其排斥第三方 Cookie，导致自身无法实现此类技术。其次，Google 极度痴迷于自助式服务（self-serve）网页系统，缺乏与顶级广告主的线下销售关系，致使其实际上被隔绝在最优质的广告版位之外。在 AdSense 体系下，Google 广告只能被投放到尾部网站（long tail of websites），广告主只愿意为这些低质版位支付微薄的费用。为了打通通往大企业广告预算的通道并补齐企业级技术，Google 开始将目光投向 DoubleClick。 - **核心矛盾**：Google 依靠自助式长尾投放，无法进入顶级品牌预算和高端媒体资源库；DoubleClick 则深嵌于各大广告代理公司的财务与投放系统。 - **技术阻碍**：频次控制等展示广告高级功能依赖第三方 Cookie，这在当时违背了 Google 的产品价值观。 --- ### [01:40:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6040s) 警报拉响：西雅图的偶遇与微软的威胁表面上看似是 Google 谋划收购 DoubleClick，但背后的真实诱因远比公开故事刺激得多。在为主播录制做背景调研时，Google 前销售负责人 Tim Armstrong 透露了一段内幕：2006 年底至 2007 年初，Tim Armstrong 打算与 DoubleClick 的管理层接洽，碰巧他因私事要去西雅图，在与对方发邮件沟通时，对方竟然回复：“你也在西雅图？巧了，我们现在就在西雅图，不如就在这里碰面吧！”听到这句话，Tim Armstrong 瞬间在 Google 内部拉响了最高警报，直报 Eric Schmidt 和 Larry Page——因为 DoubleClick 是一家总部位于纽约的公司，他们专程飞到西雅图的唯一可能，就是正与微软（Microsoft）闭门商谈出售事宜。这一局面正是 Google 最忌惮的噩梦。此前 Google 通过推动 Web 2.0 和网络应用策略，成功瓦解了微软利用 Windows 或 Internet Explorer 卡脖子的威胁。但如今微软终于醒悟，准备做十年前就该做的事情：认真对待搜索与广告业务。长期以来，微软的软件 DNA 根深蒂固，内部信奉“我们靠卖软件赚钱，绝不为了脏兮兮的广告牺牲软件销售”，MSN 等业务做广告仅仅是因为其媒体属性。然而微软此时震惊地发现，Google 从单个 PC 用户身上榨取的收入居然超过了微软自身。这促使微软决心全面下场争夺数字广告王座。 ```mermaid flowchart TD accTitle: Microsoft vs Google DoubleClick Bidding Race accDescr: 微软与 Google 争夺 DoubleClick 的收购竞标过程 subgraph Trigger["突发线索"] A["纽约的 DoubleClick 高管现身西雅图"] --> B["Tim Armstrong 察觉异样并向 Schmidt 和 Page 报警"] end subgraph Stakes["战略威胁"] B --> C["微软觉醒：意图获取展示广告第一名"] C --> D["微软准备构建独立搜索与数字广告帝国"] end subgraph Bidding["竞购战 (YMAG)"] D --> E["竞标方展示：Yahoo / MSFT / AOL / Google"] E --> F["Google 递交 31 亿美元 LOI 与排他协议"] F --> G["Ballmer 开出空白支票承诺加价匹配"] G --> H["Google 祭出无条件 Hell or High Water 条款终结交易"] end ``` --- ### [01:42:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6169s) 走错楼层撞破底牌，与价值 31 亿美元的决胜条款随后展开的竞购堪称商业谍战。在前往 DoubleClick 参与谈判时，Tim Armstrong 以为这还只是一场早期摸底沟通，结果在护送人员带领下误乘电梯到了错误楼层。电梯门打开的瞬间，带路员工惊恐失色，拼命按关门键试图让他去别的楼层。Tim Armstrong 敏锐察觉不对，立刻冲出电梯跑向走廊，赫然看到一间会议室里坐满了微软的高管、律师以及审计会计团队，正在准备敲定最终收购协议。Tim Armstrong 当即极力劝阻 DoubleClick 签约，为 Google 争取到了进场尽调并提出竞争性报价的时间。随后的战局陷入混战，DoubleClick 甚至制作了一份名为 `YMAG.XLS` 的复杂测算表格（对应 Yahoo、Microsoft、AOL、Google），向各家分别演示一旦将 DoubleClick 接进各自的广告系统能带来多少新增收入。经过数轮角逐，Google 正式递交了 31 亿美元的意向书（LOI），并附带尽调期间不得另寻买家（No-shop）的排他条款。谈判进入白热化阶段，Google 团队全员进驻纽约 DoubleClick 办公室附近的豪华酒店。在纪实商业书《Yield》中记载了一个极具戏剧性的场景：DoubleClick 的法务在查看黑莓手机后举给当时的 CEO David Rosenblatt 看——微软企业发展部门发来邮件表示愿意匹配任何报价，并附上了 Steve Ballmer 的直接邮件。Ballmer 明确表示，如果平价匹配不够，DoubleClick 可以直接在合同上自行填入想要的金额并签字，微软将火速副署盖章，无需复杂谈判。Ballmer 实际上开出了一张“空白支票”，只为拿下 DoubleClick。在排他期即将结束、LOI 还有一天到期的关键时刻，Google 递出了一份条款修改：总金额仍维持在 31 亿美元，但加入了一条关键的“不顾一切（Hell or high water）条款”——Google 承诺彻底放弃任何实质性的后续尽职调查或附加撤单条件，只要签字即刻视同落袋为安。面对这份毫无变数的确定性协议，DoubleClick 最终落笔签约。持有 DoubleClick 的私募股权机构（最初以 3 亿美元股权和 7 亿美元债务杠杆完成 10 亿美元收购）成功以 31 亿美元出局，斩获惊人回报。 - **核心出价**：31 亿美元现金，附带无条件闭眼完成交易的 Hell or high water 条款。 - **PE 回报**：私募财团以 3 亿美元的实际股权撬动 10 亿美元杠杆收购，最终以 31 亿美元卖给 Google 套现离场。 - **Ballmer 的空白支票**：微软高层直接承诺允许 DoubleClick 任意改写合同金额，微软将不经谈判直接签字。 --- ### [01:45:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6357s) 防御性棋局：遏制微软与 DoubleClick 的真实价值这笔收购对 Google 而言意义非凡，成功将广告代理商的资金通道与展示广告核心管道纳入囊中。更关键的战略意义在于，DoubleClick 是行业无可争议的头把交椅。眼见 DoubleClick 被夺走，痛失阵地的微软在短短一个月后迅速转向第二大展示广告平台 aQuantive，并以两倍于 Google 的价格——整整 60 亿美元将其吞下。尽管微软买下了第二名，但这一挫折极大地延缓了其在展示广告、雅虎搜索合作以及未来推出 Bing 的节奏。对 Google 来说，哪怕仅仅是为了把行业领头羊从微软手里抢下来、卡死对手的命脉，这 31 亿美元花得也物超所值。主持人们深刻剖析道：与 Google 地图、Gmail、YouTube 这些符合“组织全球信息并使其人人可访问”宏大愿景的产品截然不同，收购 DoubleClick 没有任何崇高的理想主义色彩。这纯粹是一场出于商业防御和利益扩张的冷酷商业棋局——Google 拥有广告交易网络（Network），再通过吞并交易所（Exchange）形成系统性垄断优势。Google 高管在公开演讲中极少登台描绘 DoubleClick 的未来，因为这完全是一笔“只为稳固护城河、不讲浪漫故事”的买卖。甚至放眼今天，DoubleClick 也并未像 YouTube 那样演变成统治全球大众消费者的超级产品。以 2024 年的数据来看，Alphabet 全年总营收达到 3500 亿美元左右，其中约 2000 亿美元来自核心的 Google 搜索广告，而包括 DoubleClick、AdSense 在内的 Google 网络展示业务（Google Network）仅占约 300 亿美元。事实上，即便没有 DoubleClick，原先由 AdSense 贡献的相当一部分展示广告网络收入也本就存在。然而在当时那个决定命运的节点，它成功封死了微软在数字广告腹地向 Google 发起致命冲击的可能。 ### [01:48:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6514s) DoubleClick 的战略地位与搜索印钞机的对比两位主持人梳理了收购 DoubleClick 的真正意义与规模对比。尽管 DoubleClick 让 Google 稳固了展示广告（display ads）领域的统治地位——如果没有这笔收购，Google 很难单凭自身达到那种统治力——但放到 Google 庞大的帝国版图中看，它远不能和 YouTube 相提并论。彼时 AdSense 已经在创造数十亿美元的营收，即便没有 DoubleClick 也会继续扩张。这种对比在财务结构上体现得淋漓尽致。Google 从搜索业务中获得大约 2000 亿美元的收入，扣除流量获取成本（TAC, Traffic Acquisition Costs）后，自己能留下约 90%；而在展示广告网络（Google Network）中，Google 需要向发布商支付高达 70% 的分成，总营收也仅约 300 亿美元。从毛利润（Gross Profit）的角度来看，这相当于 1750 亿美元与区区 90 亿美元之间的悬殊差距。因此，尽管 DoubleClick 业务庞大，但在 Google 真正的利润引擎面前，“它对全局并没有那么举足轻重”。 ```mermaid flowchart TD accTitle: Search versus Network Gross Profit Comparison accDescr: A comparison between Google Search and Google Network revenues and profit retention. subgraph Search["核心搜索业务 (Search)"] S_Rev["约 $2,000 亿总营收"] --> S_Keep["保留约 90% 毛利"] S_Keep --> S_Profit["约 $1,750 亿毛利润"] end subgraph Network["展示广告网络 (Google Network)"] N_Rev["约 $300 亿总营收"] --> N_Payout["70% 分成支付给发布商"] N_Payout --> N_Profit["约 $90 亿毛利润"] end ``` ### [01:49:25](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6565s) 2003–2008：搜索技术大跃进与全球头号广告商的诞生将视线从外围业务拉回核心基本盘，在 2003 年至 2008 年间，Google 搜索经历了一系列关键的技术迭代。早期的网页索引更新极其缓慢，用户搜索到的往往是三个月前的静态缓存，而这一时期 Google 大幅提高了索引抓取频率，使搜索体验首次迈向准实时化。伴随这一进程，Google 相继推出了 Google 图片（Images）、Google 新闻（News）、Google 图书（Books）与 Google 学术（Scholar）。随后推出的搜索建议（Google Suggest，即自动补全）以及划时代的 Google Instant（每敲击一个字母就实时触发全新搜索并无缝刷新结果页，虽然现已被弃用，但在当时惊艳全场），彻底重构了交互范式。 2005 年，Google 开始基于登录用户的搜索历史引入个性化推荐机制；2007 年，Google 正式推出“通用搜索”（Universal Search），告别以往将网页、图片、地图、视频拆成不同引擎由用户手动选择的旧模式，转而利用算法解构查询意图，自动将多元内容聚合呈现。技术突破直接转化为恐怖的营收增长：Google 营收从 2004 年的 30 亿美元飙升至 2005 年的 60 亿美元，到了 2007 年更是冲上 165 亿美元。也是在 2007 年，Google 正式加冕为**全球最大的广告销售商**——不仅是数字广告领域第一，而是横扫电视、报刊等所有传统媒介的总冠军。 - 持续霸榜纪录：自 2007 年起的整整 18 年里，Google 始终保持着全球全品类广告收入第一的王座。 - 媒介转折点：数字广告总盘子直到 2018 年才在整体规模上超越传统媒体，但 Google 凭借惊人的集中度提前 11 年登顶全球单体广告之冠。 - 后续重要演进：2009 年 Google 全面实现网页实时索引（real-time indexing），2012 年正式上线知识图谱（Knowledge Graph），在搜索实体事物时右侧自动生成结构化信息卡片，算法层面也持续针对垃圾外链与垃圾网站（anti-spam）进行清洗。 ### [01:52:42](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6762s) 计算机科学圣殿：顶尖系统架构师的群聚效应在产品与算法演进的背后，Google 在 2008 年前后已经牢固确立了全球计算机科学顶级研究机构的地位。在那个时代，称赞一个人是顶尖程序员，最极致的赞美莫过于“那是一位 Google 级别的工程师（Google type engineer）”。Google 接过了微软曾经的王冠，吸纳了整个行业最顶级的技术大脑。随着 DEC、贝尔实验室（Bell Labs）、施乐帕克研究中心（Xerox PARC）以及 IBM 研究院等老牌巨头实验室的式微与分流，一代传奇系统架构师纷纷涌入 Google。其中最为传奇的搭档便是来自 DEC 的 Jeff Dean 与 Sanjay Ghemawat。两位主持人特别致歉上一期低估了 Sanjay 的贡献，指出《纽约客》曾专门撰文记录两人非凡的技术友谊与长年“结对编程”（pair programming）的佳话。从 2002 年到 2010 年代，几乎所有关于 Google 底层海量分布式架构的核心论文，作者名单里都有 Jeff 和 Sanjay 的名字。此外，贝尔实验室的 Unix 与 Plan 9 先驱 Bill Coughran、Rob Pike 也齐聚于此。正是这批计算机系统领域的顶尖人才，奠定了 Google 突破性的技术底座，也为 2008 年即将打响的两场硬仗——打造自主浏览器（Chrome）与自主移动操作系统（Android）——备足了核心工程弹药。 ### [01:55:02](https://www.youtube.com/watch?v=QhAftC_zFr8&t=6902s) 微软醒来求购雅虎，Chrome 箭在弦上 2008 年 2 月，悬在 Google 头顶多年的达摩克利斯之剑终于落下：微软正式吹响大举进军搜索战场的号角，向雅虎（Yahoo）开出了高达 440 亿美元的收购要约。然而，杨致远（Jerry Yang）做出了商业史上最灾难性的拒绝决策之一。短短两年后的 2009 年 6 月，微软推出 Bing 搜索引擎，仅以 10 亿美元级别的对价便接管了雅虎搜索的底层驱动；再往后，雅虎最终以几十亿美元的低价贱卖给 Verizon。虽然微软的收购落空为 Google 赢得了宝贵的喘息之机，但 Google 深知正面碰撞无可避免，而他们早已备好了秘密武器——Chrome 浏览器。今天用 Chromium 内核打包浏览器相对容易，但当年从零构建浏览器引擎是一项堪比 Dave Cutler 操刀 Windows NT 的浩瀚工程。事实上，早在 2001 年，Larry Page 和 Sergey Brin 就极力主张开发独立浏览器，因为整个 Google 的命运都悬在微软的 Internet Explorer 之上。但彼时新任 CEO Eric Schmidt 严厉制止了这一冲动：“不行，我们现在绝不能去捅马蜂窝（poke the bear），现在的 Google 太年轻、太脆弱了。”这正是两位年轻创始人需要成熟职业经理人掌舵的关键时刻。 ### [01:57:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7073s) 扶植 Mozilla 与 Firefox：流量获取成本（TAC）的前身 2001 年前后，埃里克·施密特（Eric Schmidt）曾有名言警告公司切勿在时机未成熟时去“挑衅巨人”（don't want to moon the giant，指微软）。然而，Google 并没有坐以待毙，而是未雨绸缪地成为新成立的 Mozilla 基金会及其火狐浏览器（Firefox）的核心资助方。Mozilla 源于网景（Netscape）被 AOL 收购后拆分出来的非营利机构。起初 Google 主要是通过捐款和研究资助来支持 Mozilla，但随着 Firefox 正式发布并推广，Google 的支持模式转变成了支付“流量获取成本”（Traffic Acquisition Costs，简称 TAC），以此换取 Firefox 将 Google 设为默认搜索引擎。这种合作模式正是当今科技巨头商业机制的原型——正如 Google 如今每年向苹果支付约 200 亿美元以保持 Safari 默认搜索地位一样，这套机制早在 Firefox 身上就已臻于成熟。实际上，TAC 的起源甚至早于 Mozilla，此前 Google 就曾付费给各类软件开发商捆绑 Google 工具栏（Google Toolbar）。随着时间推移，支付模式逐渐转为基于在浏览器内发起搜索所产生收入的分成（Rev Share）。Google 拥有悠久的付费购买搜索分发渠道的历史，而 Mozilla Firefox 成了当时最重要的承载平台。 ```mermaid flowchart LR accTitle: Google TAC and Talent Acquisition Flow accDescr: Diagram showing how Google funded Mozilla and later brought core talent in-house to form the Client Group. subgraph MozillaPartnership["Mozilla 合作阶段"] A["Google 资助与 TAC 搜索分成"] --> B["Firefox 获得资金 / Google 成为默认搜索"] end subgraph TalentAbsorption["人才吸纳阶段"] B --> C["以非营利机构无法发期权为由"] C --> D["吸纳核心 Firefox 工程师入职 Google"] end subgraph InternalTeam["内部孵化"] D --> E["客户端产品组 Client Group (负责人：Sundar Pichai)"] E --> F["潜伏研发 Chrome"] end ``` ### [01:59:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7190s) 挖角核心人才与“潜伏分队”：桑达尔·皮查伊登场在资助 Mozilla 数年并成为其重要源码贡献者之后，Google 决定直接将 Mozilla 的核心 Firefox 工程师挖角至 Google 内部。Google 对外的说辞十分冠冕堂皇：这本质上依然是在资助 Mozilla 和 Firefox——毕竟 Mozilla 是一家非营利组织，无法为这些顶级工程师提供股权激励，随时可能被其他科技公司挖走；不如让他们直接成为 Google 员工，一边领 Google 的薪水并享受股票期权，一边继续做原本的浏览器开发工作。这批从 Mozilla 过来的工程师构成了 Google 内部全新的“客户端产品组”（Client Group）。与当时 Google 全力投入的 Web 应用（Web Apps）不同，该小组专注开发安装在个人电脑上的本地客户端程序。2004 年，Google 从麦肯锡（McKinsey）挖来桑达尔·皮查伊（Sundar Pichai）担任该团队的掌门人。施密特在《In the Plex》一书中曾评价拉里·佩奇（Larry Page）和谢尔盖·布林（Sergey Brin）此举极其聪明：既然这帮人能造出出色的 Firefox，他们当然也有能力为 Google 打造另一款顶尖的浏览器。这支团队在 Google 内部隐蔽地运转了数年，就像一支随时准备激活的“潜伏小组”（sleeper cell），静待微软的威胁彻底显现。在此期间，该团队打造了用于离线功能的浏览器插件 Google Gears，并推出了支持更复杂 Web 应用开发的 Google Web Toolkit（GWT）。然而，团队逐渐对 Firefox 的研发节奏和质量把控失去信心，同时也产生了一系列与 Mozilla 不同的架构技术设想。 - **防范微软垄断**：如果当时 90% 的 Google 搜索都发生在微软的 Internet Explorer（IE）上，一旦微软推出自己的搜索引擎（后来的 Bing），微软绝对不会为了 Google 的分成而保留 Google 为默认引擎。 - **搜索拍卖的流动性优势**：微软做 Bing 忽视了一个核心规律——搜索领域容不下第二名。拍卖机制流动性最高的那一方注定赢家通吃，Google 凭借庞大的广告主池和流动性，单次搜索流量的变现价值永远高于第二名。 - **立项时间节点**：2006 年，Google 董事会终于拍板，正式立项启动自有浏览器 Chrome 的研发。 ### [02:03:58](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7438s) 重新定义浏览器：V8 引擎、多进程架构与极简 UI 面对 IE 对 JavaScript 和 Ajax 等前沿 Web 技术支持迟缓的局面，Google 在 Chrome 中注入了数个颠覆行业格局的杀手级创新： 1. **超高速 V8 JavaScript 引擎**：作为“Ajax 公司”，Google 必须让复杂庞大的 Web 应用运行得飞快且稳定。V8 虚拟机带来了革命性的执行性能提升。 2. **多进程架构（Multi-Process Architecture）与防崩溃设计**：在 Chrome 问世之前，主流浏览器的所有标签页都运行在同一个进程内，一旦某个网页应用崩溃，会导致整个浏览器及其打开的所有标签页瞬间全部闪退。拉里·佩奇当时坦言，基于 Web 的软件交付模式虽有巨大优势，却受限于浏览器的缓慢、不稳定和无法离线。Chrome 将每个标签页作为独立的操作系统进程运行，哪怕某个标签页崩溃，也仅影响该页本身。在 Web 应用才兴起约四年的时代，这彻底颠覆了桌面软件的使用体验。 3. **借力 WebKit 渲染引擎**：团队采纳了当时表现最优异、由苹果主导开源的 WebKit 作为底层渲染引擎。 4. **极简主义与“Chrome”的反讽命名**：在软件工程界面术语中，“Chrome”原本指代浏览器内容展示区周围的所有框架元素（包括工具栏、导航栏、边框等占用空间的部件）。Google 将浏览器命名为 Chrome 带有反讽意味，其核心设计理念恰恰是“极简 Chrome”（minimal Chrome），尽可能压缩框架的存在感，让网页内容和 Web 应用本身占据绝对主角。 ### [02:06:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7588s) 合二为一的地址栏：Omnibox 与商业变现闭环 Chrome 的另一项关键创新是“通用地址栏”（Omnibox）。在此之前的浏览器界面经历过一段尴尬的“青春期”：左侧是一个输入 URL 的常规地址栏，右侧则是一个独立的搜索框。当时普通公众尚未普遍意识到，右上角搜索框里默认的搜索引擎背后是巨额的商业付费赞助（无论是 Google、Bing 还是 Yahoo）。 ``` 传统浏览器界面： [ URL 地址栏 (直达网址) ] [ 搜索框 (商业赞助引擎) ] Chrome Omnibox 界面： [ 搜索或输入网址 (二合一 Omnibox) ] ``` Google 敏锐地洞察到，从用户体验和自身核心商业模式两方面来看，双输入框都是低效冗余的。用户无需在心里区分“我在输入一个特定网址”还是“我想搜索一个关键词”，如果输入的不是标准 URL，浏览器就应当默认发起搜索。这种将地址与搜索合二为一的设计不仅带来了极致清爽的视觉与操作体验，更为 Google 搜索结果页面直接引流了巨额的展现量（Page Views），从而创造了爆发式的商业变现机会。 ### [02:07:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7660s) 多功能地址栏与沙箱架构：重新定义浏览器安全与体验将地址栏与搜索框合二为一的“多功能地址栏”（Omnibox），不仅完美契合了 Google 的广告与搜索商业模型，从用户体验的角度来看也是一次深刻的纠偏。在设计者看来，URL（统一资源定位符）从一开始会暴露给普通大众，本质上是技术实现细节的一种泄漏，堪称计算机历史上的一个“意外”——用户根本不应该被逼着去理解并键入 `https://` 这类晦涩的协议前缀，他们想去《纽约时报》，理应只需直接输入“New York Times”即可，这正如当年美国在线（AOL）试图用“AOL Keywords”做的事情一样。Omnibox 顺应了这一直觉：用户固然可以在这个框里输入网址，但它真正的核心使命是成为发起 Google 搜索的起点。除了交互层面的创新，Chrome 在底层工程架构上的最大突破是进程隔离与“沙箱机制”（Sandboxing）。在此之前，浏览网页对用户的 PC 来说几乎等同于一场安全冒险，恶意代码极其容易通过浏览器漏洞击穿操作系统。而在 Chrome 中，每一个标签页都在独立的沙箱环境中运行，彼此隔离，标签页内部运行的任何恶意程序都无法逃逸到系统或其他标签页中，这一设计彻底重塑了现代计算机安全与浏览器架构的标准。 ```mermaid flowchart TD accTitle: Chrome 多进程沙箱架构 accDescr: 展示 Chrome 浏览器如何通过多功能地址栏统一入口，并将每个标签页隔离在独立沙箱中以保障安全与稳定性 subgraph UserInterface ["统一入口"] Omnibox["多功能地址栏 (Omnibox) 统一 URL 与 Google 搜索入口"] end subgraph BrowserCore ["浏览器主进程 (Browser Kernel)"] UI_Proc["UI 与网络管理进程"] end subgraph SandboxedTabs ["标签页独立沙箱 (Sandboxed Processes)"] Tab1["标签页 A V8 引擎 + WebKit 渲染 独立沙箱环境"] Tab2["标签页 B V8 引擎 + WebKit 渲染 独立沙箱环境"] Tab3["标签页 C 崩溃 / 恶意脚本不影响宿主机"] end Omnibox --> UI_Proc UI_Proc --> Tab1 UI_Proc --> Tab2 UI_Proc --> Tab3 ``` ### [02:08:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7732s) 2008 年的雷曼破产之周与那本惊艳的 Chrome 漫画 Google 于 2006 年正式立项开发 Chrome，并在 2008 年 9 月初将其推向世界——这个发布时间点极具历史戏剧性，正好赶在雷曼兄弟（Lehman Brothers）倒闭、全球金融危机彻底爆发的一周之前。更有趣的是 Google 别出心裁的宣发策略：他们并没有发布枯燥的传统新闻稿，而是邀请了著名漫画家斯科特·麦克劳德（Scott McCloud），精心绘制了一本长篇数字漫画书，作为 Chrome 的产品介绍与用户手册。这本漫画的目标受众群体定位极其微妙：它不是写给对技术一无所知的普通大众的，却也不要求读者必须是硬核程序员。它精准瞄准了那些能够理解独立进程、沙箱机制、V8 引擎以及 JavaScript 加速的技术极客与 Slashdot 读者。正是这群关键的“晶核用户”（seed crystal users），构成了 Chrome 病毒式传播的始发点：几个月后的感恩节假期，正是这些技术爱好者回到家中，主动在父母和亲友的电脑上卸载 Internet Explorer 并换上 Chrome，理由简单而有力：“为了你的电脑不中病毒、不泄露财务信息，你必须立刻停止使用 IE。” - **立项与推出时间**：2006 年正式启动研发，2008 年 9 月初正式发布。 - **发布宣发载体**：由漫画大师 Scott McCloud 操刀的数字技术漫画（Comic Book）。 - **早期传播路径**：技术发烧友在感恩节回家帮亲友装机，以“安全性”为由彻底取代 IE。 ### [02:10:37](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7837s) 浏览器版图的大溃败：从 IE 垄断到 Chrome 统御 Chrome 随后的增长轨迹完全验证了技术颠覆的力量。上线仅 18 个月，Chrome 的用户数就达到了 4000 万；到 2010 年突破 7000 万；2012 年更是跃升至 2 亿。它的横空出世首先重创了原本势头正盛的 Firefox，让后者的市场份额在登顶后迅速回落；紧接着，Chrome 开始全方位蚕食巨无霸 Internet Explorer 的领地。当 Chrome 在 2008 年问世时，微软的 IE 牢牢占据着全网近 70% 的绝对垄断份额，Firefox 瓜分了剩下的绝大部分。但仅仅四年后的 2012 年，IE 的份额便从 70% 暴跌至 30%，与 Chrome 以 30% 左右的份额并驾齐驱；到了 2014 年，Chrome 稳居 40% 的统治地位，而 IE 进一步跌入 15% 的谷底，战局彻底终结。如今，IE 已经彻底退出了历史舞台，根据 Cloudflare 等机构的数据统计，Chrome 独占全网约 70% 的浏览器市场份额；即便算上所有默认搭载 Safari 的 iPhone 和苹果设备，Safari 在移动端加桌面端的总份额也仅有约 20%，微软自家的浏览器份额则只剩下可怜的个位数百分比。Google 彻底扭转了曾受制于微软操作系统的被动局面。 | 年份 | Chrome 发展里程碑 / 市场份额 | 竞争对手动态 (IE / Safari / Firefox) | |:--- |:--- |:--- | | **2008** | 正式发布；以安全和速度为核心卖点 | IE 占据近 70% 绝对垄断份额；Firefox 达到巅峰 | | **2010** | 用户数突破 7000 万，迅速突破 1 亿大关 | Firefox 份额开始被 Chrome 严重分流 | | **2012** | 用户数达 2 亿，市场份额升至约 30% | IE 份额暴跌至 30%，两者打成平手；移动端 Safari 借 iPhone 崛起 | | **2014** | 市场份额攀升至 40%，确立无可争议的霸主地位 | IE 降至 15%，名存实亡 | | **现今** | 占据全球约 70% 的统治性市场份额 | Safari 约占 20%（主要依靠苹果移动端）；微软仅剩个位数残存份额 | ### [02:12:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=7977s) 挽救 Web 平台，以及特洛伊木马“Chrome Frame” 从更宏大的科技发展史来看，“Chrome 拯救了 Web（万维网）”绝非一句夸大之词。在微软凭借操作系统垄断将 Web 视为本地应用程序威胁而故意放任 IE 停滞不前时，苹果也因 iPhone 和 App Store 的爆发，将重心转向原生应用（Native Apps），缺乏推动 Web 成为强大应用平台的根本商业动机。正是 Google 为了自身搜索与云端服务生态的生存，不遗余力地突破浏览器性能上限，才倒逼苹果在 WebKit 和 Safari 上展开激烈的性能竞速，使 Web 得以作为成熟的应用承载平台继续存活。在推翻 IE 统治的过程中，Google 还使出了一记极具创意的杀招——“Google Chrome Frame”插件。当时全美乃至全球数以亿计的企事业单位员工因为公司 IT 部门的设备锁定策略，无法自行安装第三方浏览器，这也是 IE 迟迟保留大量份额的核心原因。Google 便为这些被迫使用 IE 的用户打造了 Chrome Frame 插件：它直接嵌入 IE 内部，悄然将原本陈旧的渲染引擎和 JavaScript 引擎替换为 Chrome 的 V8 和 WebKit。借助这一特洛伊木马，即便身处古老的 IE 界面中，用户也能丝滑运行现代高品质的 Web 应用。 ### [02:15:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8128s) Chromium 为何开源：文化基因与高边疆商业算盘关于 Google 为何将 Chromium 完全开源，表面上看这深深根植于 Google 工程师崇尚开源、“不作恶”（Don't be evil）的企业文化。然而在崇高情怀的背后，同样存在着一层极度精明且成立的商业逻辑： Google 的真正商业飞轮来自搜索与广告变现，只要用户的每一次搜索需求最终由 Google 满足，浏览器本身是否闭源并不重要。即便有人基于 Chromium 二次开发并夺得市场份额，甚至 Google 需要向其支付一定的流量获取成本（TAC, Traffic Acquisition Costs），对于 Google 而言也绝对是一笔划算的买卖。在扣除流量获取成本之前，Google 搜索业务的边际毛利率高达 98% 左右；即便为了渠道分润让渡一部分利益，整体仍能保持在 87% 左右的极高毛利率水平。更重要的是，通过 Chromium 的全面开源与标准化，Google 彻底瓦解了微软一家独大并掐死 Web 入口的“生存级风险”（existential risk），将 Web 规则的制定权永久锁定在了有利于开放生态与自身搜索业务的一侧。 ### [02:17:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8248s) 浏览器与迪斯尼世界：平台间接控制的哲学对于 AI 浏览器时代的商业变局，核心博弈点在于：未来的 AI 浏览器是否还愿意接受 Google 的付费成为默认搜索引擎？在这场宏大的战略博弈中，浏览器的所有者必须自愿接受 Google 的巨额支票，这一商业闭环才能成立。为了解释 Google 打造自身护城河的深层逻辑，节目中提出了一个精妙的类比——迪斯尼乐园与华特迪士尼世界（Walt Disney World）： - **加利福尼亚的迪斯尼乐园**：虽然高度精心制作、追求作者风格与完美体验，但它受制于安纳海姆市政府的条条框框，缺乏对周边环境与底层规则的绝对掌控权。 - **佛罗里达的华特迪士尼世界**：迪斯尼决定买下大片荒地，甚至在周边建立了自己的管理区（芦苇溪改善区），从而自己制定市政规则、全权主导底层设施。 ```mermaid flowchart TD accTitle: 平台控制力演进图 accDescr: 对比微软在PC时代的绝对所有权与Google在Web时代通过Chrome维持间接控制的生态模式 subgraph PC_Era["PC时代：微软模式"] A["Windows 操作系统"] --> B["完全拥有与统治平台"] end subgraph Web_Era["Web时代：Google模式"] C["开放 Web（无单一所有者）"] --> D["构建 Chrome 与标准组织"] D --> E["间接引导生态与保护底层环境"] end ``` 从 1998 年起，Google 最大的威胁始终是微软，这并非因为 Bing 本身或广告系统的直接竞争，而是因为底层平台控制权的潜在颠覆。在 PC 时代，微软是精神与现实上的平台霸主，牢牢掌握 Windows。步入互联网时代，虽然 Google 成为了 Web 时代的代表性平台公司，但没有任何一家公司能单独“拥有”Web。Google 并不直接向用户收取基础设施费用，而是依靠网络生态的繁荣来赚取搜索广告收益。为了避免寄人篱下、完全受制于微软的鼻息，Google 必须扮演开放 Web 生态的管家，通过推出 Chrome、主导 Web 标准组织，对天平施加关键影响力，以确保自己赖以生存的家园（Web 平台）始终健康且不受恶意操纵。 ### [02:20:12](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8412s) 默认设置的威力与 Chrome 的监管分拆悖论 Chrome 的巨大成功让负责该项目的桑达尔·皮查伊（Sundar Pichai）最终问鼎 Google CEO 之位。如果复盘历史推演“要是 Google 当初没有推出 Chrome 会怎样”： - 微软于 2009 年推出 Bing，当时仍独占浏览器市场 70% 的份额。 - 移动端浪潮直到 2012 年左右才真正爆发。 - 倘若没有 Chrome，微软完全可以在长达三到四年的时间里将 70% 桌面用户的默认搜索引擎强行切换为 Bing。尽管习惯强大的用户会手动改回 Google，但“默认设置的威力（The power of defaults）”是极其可怕的——这也正是 Google 每年愿意向苹果支付 200 亿美元以保住 Safari 默认搜索位置的原因。没有 Chrome，Bing 很可能会成长为 Google 致命的竞争对手。因为无论商业形态如何演变，浏览器的核心变现路径几乎只有一条：**靠搜索引擎（或未来的 AI 公司）为流量入口买单**。这种商业现实直接映射到了当下美国司法部（DOJ）对 Google 反垄断诉讼的矛盾之中：如果监管机构强制要求 Google 剥离 Chrome，Chrome 唯一的商业化生路依然是向 Google 或第三方收取流量分发费；若监管层既不允许 Google 拥有 Chrome，又禁止 Google 付费购买浏览器的默认入口，那么独立出来的 Chrome 将完全丧失作为商业体存活的可能。 ### [02:22:13](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8533s) 独立成林的产品神话与全球最大操作系统 Chrome 取得巨大成功后，Google 甚至萌生了用 Chrome OS 正面硬刚微软 Windows 的想法。尽管 Chromebook 依靠极高的性价比在教育市场攻城略地，但在通用 PC 操作系统领域，Windows 至今依然维持着约 70% 的绝对垄断统治（即便科技圈普遍使用 Mac，苹果的 PC 份额也仅在 15% 上下）。然而在移动领域，Google 打造了人类历史上规模最大的操作系统——**Android**，如今活跃设备数已突破 30 亿台大关。回顾 Google 的核心产品组合，一个惊人的事实在于：它们都不是单纯依靠原有产品的流量灌顶而成功的“搭便车”产物。 - 很多大公司推出新功能组件时，仅靠既有垄断渠道硬推，这种扩张算不上真正令人敬佩。 - Google 虽然会在搜索主页给予适度曝光，但无论是 Chrome、Gmail，还是 Google Docs 与表格，无一不是凭借无可匹敌的产品体验、极佳的口耳相传以及早期的免费策略迅速风靡全网（如同感恩节回家帮全家人电脑安装 Chrome 的科技极客们带动的病毒式传播）。 - 当然，在这些业务具备独立造血能力之前，它们背后都站着 Google 地下室里那台源源不断吐出巨额现金的“印钞机”——AdWords（Uncle AdWords）。 ### [02:24:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8690s) 赞助商插播：Vercel 与 v0 在深入展开 Android 的波澜壮阔历史之前，节目感谢了本季度的核心赞助商 Vercel。Vercel 已成为现代 Web 与 AI 开发的基础设施底座，支撑着 PayPal、Ramp、Under Armour、Notion、Runway 以及 Cursor 等诸多知名企业。本期重点介绍了其革命性的 AI 生产力应用构建工具 **v0**： - **全栈 AI 开发与部署**：界面类似于 Claude、ChatGPT 或 Gemini，但用户输入自然语言 Prompt 后，v0 能直接生成、设计、迭代并部署一个完整可用的全栈 Web 应用或功能型网站。 - **人人皆可创造**：践行皮克斯经典动画《料理鼠王》中“人人皆可烹饪（Everybody can cook）”的理念，非技术人员无需编写任何代码或设计线框图，市场人员可一键搭建落地页，小微企业可秒级生成包含联系表单的官网。 - **惊人的发展态势**：v0 当前已拥有超过 400 万用户，从产品经理、独立开发者到教师学生均在其上构建真实生产级应用。 - **自研闭环的范例**：生成的每个 v0 应用都能直接无缝部署在 Vercel 自家的基础设施上，生动诠释了 Vercel 作为自身产品“零号客户（Customer Zero）”、以自己的 AI 云赋能自家 AI 产品的商业闭环。 ### [02:26:57](https://www.youtube.com/watch?v=QhAftC_zFr8&t=8817s) 从车库到大学路：安迪·鲁宾与 Danger 的起源 Google 的办公场所一向充满传奇色彩：最早是苏珊·沃西基（Susan Wojcicki）位于门洛帕克的车库，后来是山景城前 SGI（Silicon Graphics）园区改造的 Googleplex。但在这两处之间，Google 曾在帕洛阿尔托市中心大学路 165 号（165 University Avenue）度过过一段岁月——这栋楼堪称硅谷的风水宝地，后来还孕育了 PayPal。1999 年 8 月 Google 搬离此处后，搬进来的正是安迪·鲁宾（Andy Rubin）创立的移动设备公司 Danger。安迪此前是苹果公司的工程师，后来跟随一批“反叛者”离开苹果加入传奇初创公司 General Magic；那家在 90 年代初折戟的公司，本质上是在技术条件还不成熟时，提前 15 年试图造出 iPhone。 General Magic 分崩离析后，安迪创办了 Danger。最初他的构想甚至称不上宏大：他打算做一款无线版的“CueCat”条形码扫描猫形硬件。Danger 的第一号员工、日后在 Google 掌管 Android 和 Chrome 多年的广桥博史（Hiroshi Lockheimer）劝他：“几年过去了，也许我们应该重新审视 General Magic 当年的构想。”在广桥的推动下，Danger 最终造出了惊艳硅谷的 Sidekick（侧滑全键盘手机），并与 T-Mobile 合作推向市场。Sidekick 与黑莓（BlackBerry）一同开创了“手机的核心用途不再是打电话，而是发信息与移动富媒体”的时代。这款设备迅速引爆名人圈，甚至成为美剧《明星伙伴》（Entourage）中的标志性道具。 - **大学路 165 号的传承**：从 Google 到 Danger 再到 PayPal，这栋建筑见证了移动计算与互联网基础设施的数度接力。 - **广桥博史的关键介入**：作为 Danger 的一号员工，广桥扭转了安迪研发条形码扫描器的弯路，促成 Sidekick 诞生，日后他更成为 Android 体系的核心高管。 - **Danger 的最终归宿**：微软在 2008 年收购了 Danger，但安迪·鲁宾早在 2003 年便已抽身离开，投身下一段冒险。 --- ### [02:30:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9026s) 创立 Android：从数码相机系统到智能手机的突围 2003 年安迪·鲁宾创办了 Android，但其最原始的商业构想令人啼笑皆非——他们并不是要做手机系统，而是想为当时风靡全球的“傻瓜数码相机”（Point-and-Shoot Cameras）打造一套跨平台、开源的操作系统。安迪当时敏锐地看到数以亿计的消费者都在购买数码相机，希望将相机系统作为特洛伊木马打入硬件世界。然而团队很快意识到技术演进的真正方向：不是相机演变成手机，而是手机正在彻底吞噬相机。万幸的是，他们写出的底层软件架构同样完美适配手机，安迪果断带领 Android 调头转向智能手机市场。当时的智能手机市场虽然规模有限，但格局十分固化。市场上主要分为两派：一派是软硬件全栈自研的巨头，包括如日中天的诺基亚（Symbian）以及在企业市场所向披靡的黑莓；另一派则是硬件代工厂商（OEM），它们向外购买操作系统，主要是从 Palm 或微软购买 Windows Mobile 授权。微软沿用了桌面操作系统的打法，向硬件厂商每台设备收取个位数美元的授权费。硬件厂商造出来的手机功能贫弱，因而物料成本极低；运营商则通过合约套餐向用户每月收取上百美元的高额费用，同时因为设备能力羸弱，数据流量消耗极小。整个产业链上的所有既得利益者——微软、硬件 OEM 和电信运营商——过得“又胖又快乐”，根本没有打破现状的动力。 ```mermaid flowchart TD accTitle: Android's Strategic Pivot and Market Landscape accDescr: Diagram illustrating Android pivoting from digital camera OS to mobile OS, confronting the 2005 smartphone landscape. subgraph Pivot["Android 早期战略转型"] A["最初构想：数码相机开源操作系统"] -->|发现手机正在吞噬相机| B["战略转向：智能手机操作系统"] end subgraph OldWorld["2005 年固化的移动市场"] C["全栈自研阵营 (Nokia / BlackBerry)"] D["授权分发阵营 (Windows Mobile / Palm OS)"] E["运营商垄断与代工厂 (低成本设备 + 高额月费)"] end B -.->|"免费开源模式挑战"| OldWorld ``` --- ### [02:33:33](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9213s) 停滞的生态与拉里·佩奇的 5000 万美元闪电收购在这个令史蒂夫·乔布斯（Steve Jobs）深感厌恶的沉闷泥潭中，安迪·鲁宾带着 Android 杀入市场。他四处游说手机制造商和各大运营商：“不要再给微软或 Palm 付授权费了，我给你们提供一款完全免费的操作系统；它是开源的，支持第三方应用开发，能让硬件变得无与伦比的强大。”然而现实给了安迪当头一棒：没有一家巨头愿意理会他。对于 AT&T、Verizon 这些手握数十亿美元命脉的电信运营商来说，去押注一家只有 8 名员工、估值仅 1000 万美元的草台班子简直是天方夜谭；更深层的原因在于，运营商才是真正的付费客户，用户糟糕的体验对产业链的既得利益而言微不足道。熬到 2005 年，Android 已经苦撑了两年，虽然说服了台湾代工厂 HTC（宏达电）合作打造出一款原型机，但公司资金几乎告罄。在安迪四处筹集新一轮融资时，他见到了拉里·佩奇（Larry Page）。拉里听完后当场拍板：“别融什么新一轮了，我直接把你买下来怎么样？”2005 年 7 月，Google 出资 5000 万美元闪电收购了 Android。尽管这笔交易日后常被赞誉为科技史上最超值的收购之一，但这只是一种认知错觉——收购只是给了一记推力，Google 随后在内部投入了数十亿美元的研发才真正做成了 Android。而拉里之所以如此果决，是因为 Google 领导层清醒地意识到自己在移动浪潮中已经迟到了：不仅黑莓和 Windows Mobile 证明了智能手机的崛起，当时与 Google 关系极为亲密的苹果也已暗流涌动，距离 iPhone 的震撼问世只剩下 18 个月。 - **超前与受挫**：Android 早期试图以免费开源击穿运营商和代工厂的旧体系，但因团队体量过小、生态固化而屡屡碰壁。 - **HTC 制造的原型机**：在被收购前，HTC 与 Android 合作打造了最初的工程测试样机，奠定了双方日后的长期合作基础。 - **Google 的移动焦虑**：2005 年中期，埃里克·施密特、拉里·佩奇与谢尔盖·布林一致认为 Google 必须立即拥有自己的移动战略入场券，从而促成了这场 5000 万美元的闪电交易。 ### [02:36:54](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9414s) 移动端需求的显现与 Android 的战略先机 Google 当时已经非常清楚地观察到了消费级移动设备的强劲需求。他们针对早期的移动设备推出了专门适配的 Google.com 网页版，后台监测到的访问流量正在飞速攀升。更具决定性意义的是地图服务——移动端上的 Google Maps 展现出了极其强大的“杀手级应用”特质。然而，由于当时智能手机硬件和系统极度碎片化，Google 竟然需要不可思议地同时维护多达 350 个不同版本的移动端 Google Maps，以适配市面上五花八门的机型。这让管理层清醒地意识到：在桌面端围绕核心搜索业务建立起的“护城河”固然坚固，但移动端这片全新的土地同样急需建立自己的根据地。回过头来看，Google 在 2005 年收购 Android 是一次千钧一发的抢跑。此时距离第一代 iPhone 发布仅剩 18 个月，如果 Google 没有在 2005 年夏天果断出手拿下 Android 并启动自研，而是拖到 2007 年 1 月再从零起步，就彻底丧失了在移动时代立足的先机。届时，整个移动操作系统市场很可能会重演桌面时代的剧本：高端市场由苹果走垂直一体化整合路线，而惊慌失措的手机 OEM 厂商则别无选择，只能购买并授权微软销售的移动操作系统。安迪·鲁宾（Andy Rubin）早早为 Android 构想的商业模式恰好与 Google 的战略不谋而合——操作系统完全免费开放给 OEM 厂商和电信运营商。戏剧性的是，当年 Android 作为一家独立初创公司向硬件厂商免费提供系统时，对方反而充满疑虑，认为这是走投无路时的“示弱信号”；而一旦 Android 归于 Google 旗下并由其免费开放，厂商们立刻深信不疑，认为 Google 背后一定有一套宏大而成熟的商业蓝图。 ```mermaid flowchart TD accTitle: Android Acquisition and Market Landscape accDescr: Diagram showing how Google acquiring Android prevented Microsoft from dominating the OEM mobile OS market. subgraph Google["Google 视角与策略"] A["发现移动端巨大需求 (350个 Maps 适配版本)"] --> B["2005年收购 Android 并免费开放"] B --> C["阻截微软复制 PC 垄断"] end subgraph Market["移动 OS 潜在市场格局"] D["垂直一体化阵营: 苹果 (Apple)"] E["开放授权阵营: 若无 Google 则是微软 (Microsoft)"] end B -.-> E ``` --- ### [02:39:44](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9584s) 两个原型、iPhone 震撼发布与尴尬的“Apple-Goo” 2005 年夏天 Android 团队并入 Google 后，内部最初规划了双轨并行的两条产品线： - **“Sooner”原型**：面向近期的过渡设备，形态类似当时风靡商务市场的黑莓（BlackBerry），配备全键盘、没有触摸屏。 - **“Dream”原型**：属于长期的前沿研究项目，是一款搭载触摸屏的高端智能手机。 2006 年夏天，时任 Google CEO 埃里克·施密特（Eric Schmidt）受邀加入苹果公司董事会。不久之后的 2007 年 1 月，史蒂夫·乔布斯在科技史上最伟大的发布会上正式揭开了初代 iPhone 的面纱。令人啼笑皆非的是，施密特本人作为特邀嘉宾走上了乔布斯的发布会舞台，而当时外界对 Google 秘密研发的 Android 尚一无所知。虽然乔布斯知晓 Google 在做类似黑莓的全键盘项目（Sooner），但他对触摸屏的“Dream”项目毫不知情。在台上仅约三分钟的演讲中，施密特试图表现得轻松幽默，甚至开玩笑说苹果和 Google 的关系紧密到应该合并，新公司干脆叫“Apple-Goo”，并调侃称 iPhone 让他们“无需真正合并两家公司就能融为一体”。舞台边转播镜头切到乔布斯脸上，乔布斯虽然努力保持风度微笑配合，但神情中写满了肉眼可见的尴尬与不适（“the ick”）。在初代 iPhone 上，两家公司的合作确实深入骨髓： - 当时 iPhone 既没有 App Store，也不支持第三方原生应用，主屏上总共仅有十余个系统应用。 - 其中仅有的两个第三方服务应用——YouTube 和 Google Maps，全部来自 Google（界面与图标由苹果设计，底层由 Google 移动团队提供 API 与后端支持）。 - YouTube 团队对苹果设计的带有木纹边框的复古 CRT 电视机图标极为厌恶。他们认为这既没有露出 YouTube 品牌，也完全违背了 YouTube 代表“未来视频形态”的理念。 - 负责为初代 iPhone 搭建这些 Google 服务后端的 Google 移动业务主管，正是日后主持打造 Google+ 的维克·贡多特拉（Vic Gundotra）。 --- ### [02:43:20](https://www.youtube.com/watch?v=QhAftC_zFr8&t=9800s) 转向 Dream 与乔布斯的“全核打击”狂怒 iPhone 发布会当天，Android 团队在台下目瞪口呆。第二天，团队毫不犹豫地将所有的“Sooner”键盘机原型“直接扔进了垃圾桶”。电容式多点触摸屏无可争议地代表了移动设备的未来，“Dream”项目立刻被转为最高优先级的“主力突击队”。当乔布斯意识到 Google 正在利用施密特在苹果董事会期间获取的行业认知、全面转向打造对标 iPhone 的触摸屏操作系统时，彻底陷入了狂怒。后来在一次苹果内部全员大会上，乔布斯的愤怒言辞被泄露给媒体： > “我们从未涉足搜索业务，他们却闯进了手机业务。别被骗了，Google 就是想置 iPhone 于死地。我们绝不会让他们得逞。” 在沃尔特·艾萨克森（Walter Isaacson）执笔的《史蒂夫·乔布斯传》中，乔布斯更是留下了载入科技史册的决绝宣言： > “我要摧毁 Android，因为这是一件偷来的产品。为了这事，我不惜发动核战争（Thermonuclear War）。就算耗尽我最后一口气，耗尽苹果在银行里的全部 400 亿美元现金，我也要纠正这个错误。” 尽管火药味十足，乔布斯直到 2009 年才正式将埃里克·施密特请出苹果董事会。对于“抄袭”的指责，外界也有许多反思声音：当时多点触控技术在学界和先驱公司中早已存在，技术演进在那个历史节点交汇并催生相似方案在所难免。然而，由于苹果手握多项严密的多点触控手势专利，乔布斯以此严厉威胁起诉 Google。这种法律压迫直接导致早期 Android 系统在数年内不敢公开支持“双指缩放”（Pinch to Zoom）和系统级滑动手势；早期几乎所有 Android 手机底部都极其刻板地保留了 4 颗物理导航按键，这正是规避苹果触控专利围剿的直接产物。 ### [02:46:45](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10005s) 差异化与反击：Android 的早期优势与开放手机联盟在围绕 Android 是否“抄袭” iPhone 的争论中，如果站在 Google 的立场审视，就会发现两者的发展路径截然不同。当 Android 最终面世时，它不仅带来了 Android Market（即后来的 Google Play 商店），还带来了滑动下拉通知中心（Notification Center）——能够一站式聚合所有应用程序发来的通知；此外，它还支持在主屏幕上自由拖拽重排应用图标。这些创新设计后来都被苹果反向借鉴或直接吸纳，甚至连下拉通知栏的功能苹果都花了好几年才补齐。正如节目中所调侃的毕加索名言：“伟大的艺术家偷取灵感（Great artists steal）”，两家巨头在移动交互领域实际上是互相“借鉴”。 2007 年 11 月，也就是乔布斯揭晓初代 iPhone 大约 10 个月后、iPhone 正式发售 5 个月后，Google 抛出了一枚重磅炸弹：正式宣布成立“开放手机联盟”（Open Handset Alliance，简称 OHA）。该联盟集结了一众产业链巨头，包括硬件制造商宏达电（HTC）、摩托罗拉（Motorola）、三星（Samsung）、LG；移动运营商 T-Mobile、Sprint；以及芯片与半导体大厂高通（Qualcomm）、英特尔（Intel）、博通（Broadcom）和德州仪器（Texas Instruments）。这一消息在当时引发了全行业的困惑与猜测——外界普遍看不懂这到底意味着什么，大家纷纷议论：“Google 这是要自己做手机，还是不做手机？”随后整整近一年的时间里，市面上几乎没有出现任何实质性的产品动静。 ```mermaid flowchart TD accTitle: 开放手机联盟生态架构 accDescr: 展示 Google 发起成立的开放手机联盟涵盖的各领域核心巨头伙伴 subgraph Google["软件与平台核心"] G["Google (Android OS)"] end subgraph OEMs["硬件制造伙伴"] H1["HTC"] H2["Motorola"] H3["Samsung"] H4["LG"] end subgraph Carriers["运营商伙伴"] C1["T-Mobile"] C2["Sprint"] end subgraph Silicon["芯片与半导体"] S1["Qualcomm"] S2["Intel"] S3["Broadcom"] S4["TI"] end G --> OEMs G --> Carriers G --> Silicon ``` --- ### [02:48:02](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10082s) T-Mobile G1（HTC Dream）：踩着旱冰鞋登场的极客首秀转折点出现在 2008 年 9 月。那是一个风云变幻的月份——雷曼兄弟宣告破产、全球金融危机爆发，而 Google 不仅推出了 Chrome 浏览器，还联合 T-Mobile 正式发布了首款商用 Android 手机：T-Mobile G1（Google 手机或称 Gphone）。这款设备正是由 Andy Rubin 最早的代工原型伙伴 HTC 制造，其在 HTC 内部的正式产品代号为“HTC Dream”（梦想）。G1 配备了支持触控的屏幕与正面物理导航键，同时还拥有类似于当年 Sidekick 的横向侧滑全键盘（QWERTY）。与当时的 iPhone 相比，G1 原生支持多任务处理（Multitasking），并允许运行第三方应用程序。这场发布会的氛围极其奇特且充满戏剧性。发布地点选在纽约市的一家商业后厨（Commercial Kitchen）举行，现场组织得颇为仓促随意，以至于如今在网上都很难找到完整的官方视频，只剩下一些零星的片段与照片。当 HTC 和 T-Mobile 的一众高管西装革履、神情严肃地站在台上时，Google 联合创始人拉里·佩奇（Larry Page）和谢尔盖·布林（Sergey Brin）竟然脚踩着双排旱冰鞋（Rollerblades），一路滑进了发布大楼，并穿着旱冰鞋直接滑到了舞台中央。尽管发布方式略显滑稽随意，G1/Dream 最终依然取得了相当体面的市场表现：在美国本土卖出了 100 多万部，仅凭这一款机型就夺得了全美 6% 的智能手机市场份额，几乎能与当时整个 Palm 公司平分秋色。然而，彼时的智能手机整体盘子仍然不大，真正的移动浪潮还要等到 2011 年才彻底爆发；更严峻的是，苹果正以不可思议的速度抢占市场——在 G1 整个生命周期卖出 100 万部的同时，iPhone 在 2008 年卖出 1100 万部，2009 年更是达到 2000 万部，几乎一夜之间垄断了全美 50% 以上的智能机市场。 - **主持人的大学往事**：主播在俄亥俄州立大学（Ohio State）读大四做毕业设计项目时，四人团队中有一位同学拥有一部 T-Mobile G1，他们便针对它写了一个 Java 移动应用；该同学毕业后创立了著名的图片社交网络 DailyBooth。 - **早期 iPhone 的产品软肋**： - 缺乏文字复制粘贴（Copy-Paste）功能； - 无法实现多任务后台运行； - 界面几乎没有任何自定义空间，甚至早期连壁纸都无法更换，只能使用默认纯黑底色； - 缺少当时备受商务与极客用户青睐的实体物理键盘； - 最致命的痛点是运营商独占：全美仅限 AT&T 网络发售，且初代仅支持极慢的 2G EDGE 网络，网络带宽完全跟不上设备的交互潜力。 --- ### [02:52:19](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10339s) 2009 假日季的转折点：Motorola Droid 与 Google 导航空降智能手机霸权争夺的决定性时刻发生在 2009 年的圣诞假日季——摩托罗拉 Droid（Motorola Droid）的横空出世彻底扭转了乾坤。微软前 CEO 史蒂夫·鲍尔默（Steve Ballmer）在接受节目专访时曾无比笃定地表示，自己永远忘不了 2009 年的那个圣诞节，因为那正是 Android 彻底拿下市场支配地位的关键节点。当时整个美国公众对“Android”这个底层的移动操作系统缺乏感知，但对“Droid”这个名字却几乎人尽皆知；后来更是延续为对三星“Galaxy”的狂热认知。Google 非常务实地退居幕后，将品牌主导权完全让渡给硬件厂商与运营商。背后最大的推手是美国电信巨头 Verizon。在 iPhone 独家签约 AT&T 的整整两年时间里，Verizon 遭受了惨烈打击。AT&T 不仅从 Verizon 挖走了数以百万计的入网用户，更可怕的是夺走了 ARPU 值最高、最愿意为大额数据流量套餐买单的“高净值黄金用户”。被逼入绝境的 Verizon 痛定思痛，决定破釜沉舟：全力押注 Android 阵营，向摩托罗拉采购并定制一款能够正面抗衡 iPhone 的旗舰王牌，并砸下天量营销资源。Motorola Droid 搭载了在当时堪称巨大的高清屏幕、侧滑全键盘、500 万像素摄像头以及可更换电池。然而，Droid 身上最强大、最具摧毁力的“杀手级应用”来自 Google 的软件生态：它是全球第一款预装 **Google Maps 逐向语音导航（Turn-by-turn Navigation）** 的手机。在此之前，车载导航是一个极其庞大的独立消费电子品类，消费者需要花费数百美元专门购买 Garmin、TomTom 或 Magellan 导航仪吸在挡风玻璃上，并且每年还要额外缴纳月费或订阅费来更新地图数据。Droid 与 Google Maps 的结合在一夜之间将这一整个百亿美元级行业彻底宣判死刑（Sherlocked）——手机自带的原生免费语音导航体验远超专用硬件，再也没有人愿意为独立 GPS 硬件和按月订阅付费。更讽刺的是，同一时期的 iPhone 上的 Google Maps 根本无法实现自动语音连续导航，用户在驾驶过程中必须每转一个弯就低头用手指手动点击屏幕进入“下一步”，在实际行车中体验极为危险且笨拙。 - **杀手级整合**：Google Maps 的免费逐向语音导航彻底淘汰了 TomTom、Garmin 等车载便携式 GPS 硬件品类。 - **品牌与广告轰炸**：Verizon 特地向卢卡斯影业（Lucasfilm）购买了《星球大战》中“Droid”一词的官方商标授权，并在每支广告末尾的小字中注明。 - **现象级营销“Droid Does”**：全美铺天盖地播放“Droid Does”系列电视广告，广告前 80% 到 90% 的篇幅都在尖锐列举 iPhone 无法实现的痛点（不能多任务、没有实体键盘、不支持自由定制），随后以机器人的机械冷酷声效打出标语“Droid Does”，成为一代美国消费者的集体记忆。 ### [02:56:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10610s) 摩托罗拉 Droid 与“Droid Does”反击战 Verizon 倾注全力打造的摩托罗拉 Droid 营销战役极具攻击性与反叛色彩。广告开头故意模仿苹果经典的纯白背景、轻快欢快的背景音乐以及淡入淡出的精致字体，列举着“我不能多任务处理”、“我没有可更换电池”等限制；然而在广告的最后五秒，画面突然切入强烈的刺耳静电噪音与黑酷粗犷的视觉风格，伴随着经典冷峻的机械声宣告：“Droid Does”（Droid 能做到）。Verizon 的首席营销官（CMO）直言，这场铺天盖地的营销就是要“唤醒整个市场”。市场反响呈爆发之势。第一代摩托罗拉 Droid 在发售首个周末就售出 25 万部，其突破 100 万部销量的速度甚至超过了当年的初代 iPhone。《时代》周刊更是将 Droid 评为 2009 年度最佳产品。背后核心的驱动力在于 Verizon 庞大的用户群中积压了太多对“真正智能手机”的渴求——即便撇开谷歌地图语音导航等功能不谈，光是能在 Verizon 顶级网络上用上一部成熟的智能手机，就足以引爆市场。随后几年，Verizon 趁热打铁推出了一整套“Droid 系列”矩阵（HTC Droid Incredible、Droid X、Droid 2、Droid Bionic、Droid Max 等），彻底奠定了 Android 在美国市场的庞大基本盘。 ```mermaid flowchart TD accTitle: Android 商业生态合作体系 accDescr: 展示 Google 如何通过开源操作系统、捆绑 GMS 与分成机制掌控移动生态 subgraph GoogleStrat["Google 战略层"] A["AOSP 开源系统 完全免费开放"] B["GMS 核心套件 Maps / Gmail / Play 商店"] C["预设默认搜索引擎 Google Search 顶部常驻"] end subgraph Partners["硬件与渠道生态"] D["OEM 手机厂商 (HTC / 三星 / 摩托罗拉)"] E["电信运营商 (Verizon 等)"] end subgraph ValueFlow["价值与利益流动"] F["Less than Free 搜索流量分成 (TAC)"] G["海量移动设备落地 掌控底层生态控制权"] end A --> D B --> D D --> E B --> C C --> F F --> D F --> E E --> G D --> G ``` ### [02:58:43](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10723s) “低于免费”的商业模式与微软的溃败在非苹果阵营（硬件制造商与操作系统分离的模式）中，Google 彻底甩开了所有竞争对手。此前，微软原本握有巨大的胜算，但其商业模式却存在致命缺陷：微软向每部手机的运营商和厂商索取 5 到 10 美元的 Windows Mobile 系统授权费；而 Google 不仅系统完全免费开源、允许厂商自由定制修改源代码（如三星后来的定制系统），还反复提醒手机厂商“我们不是微软，绝不会像微软当年对待 PC 厂商那样吸干产业链的所有利润”。更为颠覆的是，Google 实际上推行的是“低于免费”（Less than Free）的商业模式。传奇风险投资家比尔·格利（Bill Gurley）曾撰写过一篇轰动业界的博文，揭露 Google 不仅不收系统费，还会将手机上产生的部分搜索广告收入以分成（TAC，流量获取成本）的形式返还给运营商和手机厂商。比尔·格利敏锐地预言：微软、Palm 和黑莓已经毫无生还可能，没有任何商业模式能够击败“不仅白送、甚至倒贴钱给你用”的降维打击。对 Google 而言，付出一小部分搜索收入分成，换取的是避免底层平台被微软或苹果彻底封锁的战略主动权，确保自己无需受制于可能将流量随意导向别处的移动 Safari 或其他垄断浏览器。 - **授权费用对比**：微软每台设备收取 5~10 美元授权费；Android 开源免费，还额外提供搜索收益分成。 - **比尔·格利的论断**：面对“低于免费”（Less than Free）的商业模型，传统收取软件授权费的操作系统（Windows Mobile、PalmOS、BlackBerry OS）在经济逻辑上被直接宣判死刑。 - **Google 的战略算盘**：宁可把几个百分点的搜索收入当作 TAC 分给运营商与 OEM，也绝不能让移动 Safari 成为下一个掌控流量命脉的 IE 浏览器。 ### [03:01:36](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10896s) 跨越移动平台代际迁移的商业奇迹在科技商业史的长河中，几乎没有任何一家巨头能够成功跨越重大计算平台的代际迁移并在新时代继续称霸。科技行业八十年来的铁律是：颠覆往往来自外部，每个巨头通常只能充分享受属于自己的黄金时代，随后便逐渐边缘化。IBM 统治了大型机时代，但在 PC 时代跌落神坛；微软横扫了 PC 时代，却在 Web 时代失去绝对统治力；唯独 Google 在 Web 时代登顶之后，面对智能手机这场规模更庞大、破坏力更强的平台剧变，不仅没有掉队，反而将自己的核心摇钱树——搜索业务——无缝迁移并进一步发扬光大。苹果虽在移动时代封神，但其在 PC 时代从未真正赢得过统治地位；能够在连续两个重大技术范式转移中同时成为无可争议赢家的，唯有 Google。 Android 手机的普及为 Google 带来了前所未有的用户触达。即便早期 Google 并不能直接从开源系统本身或硬件销售中获取巨额利润（应用商店分成也远不及苹果庞大），但每一部出厂的 Android 设备主屏幕上最显眼的位置，永远横亘着那个巨大的 Google 搜索框。Android 从 2009 年底起如海啸般席卷全球： - 2009 年底（摩托罗拉 Droid 发布时）：Android 的全球智能手机市场份额仅为 5%~6% 左右； - 2010 年底：全球份额飙升至 30%，日均激活设备超过 20 万台； - 2011 年：市场份额达到 50%； - 2013 年底：全球市场占有率突破 80% 的历史巅峰（甚至高于如今约 72% 的水准，同期苹果约为 28%）。 Google 建立了一套类似 Visa“网络之网络”的多边杠杆体系。厂商固然可以仅仅下载开源的 AOSP（Android Open Source Project）去研发完全独立的系统，但如果不接入 Google 移动服务（GMS），手机就无法预装拥有海量软件生态的官方应用商店（Google Play），也无法搭载官方的原生 Gmail 和 Google Maps。一旦厂商选择接受这个“无法拒绝的条件”，就必须默认集成 Google 搜索，而 Google 随后奉上的搜索广告分成更让厂商彻底绑定在这一生态链条上。正是这种极为强大的生态扩张，反向倒逼苹果走出舒适区，促使 iPhone 在随后的岁月中不得不推出更大屏幕、更多元价格区间以及更强摄影能力的机型来迎战 Android 阵营。 ### [03:06:38](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11198s) 硬件对抗、三星异心与 Pixel 的战略防御早年智能手机的相机性能与今天不可同日而语。初代 iPhone 的摄像头仅有 200 万像素，成像质量相当粗糙；而摩托罗拉 Droid 在发布时便主打 500 万像素摄像头，正是这种激烈的正面对抗推动了整个智能手机摄影技术的演进。然而，进入 2010 年代初中期，Android 生态内部悄然出现了巨大的裂痕——三星凭借 Android 极具弹性的开源平台，深度模仿 iPhone 的高端路线并推出了 Galaxy 系列，其出货量迅速呈现爆发式增长。随着羽翼丰满，三星开始在部分设备上剥离 Google 的核心服务，转而强推自家的软件与服务体系。这一举动触碰了 Google 的底线。为了遏制三星一家独大并防止生态脱轨，Google 在 Nexus 之后推出了 Pixel 系列。Pixel 不仅是一台面向普通消费者的旗舰手机，它在战略上更像是一部“参考样机”（Reference Design）——如同微软当年强推 Surface 来教导 PC 厂商一样，Google 借 Pixel 向除三星之外的整个 OEM 市场展示：顶级硬件、卓越相机以及深度整合的系统体验应该如何打造，以此鼓励其他硬件厂商跟进复制，维持 Android 阵营内部的生态平衡。 --- ### [03:08:18](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11298s) 拆解 Android 商业账本：流量获取成本（TAC）与核心护城河评估 Android 的商业价值，核心并非单看它赚了多少钱，而在于两笔账的叠加：其一是 Google Play 商店的直接变现，其二则是 Google 为保障核心搜索流量入口所节省的天价“买路钱”。许多人误以为 Android 作为自有平台无需支付流量成本，但深入拆解财报会发现，Google 依然需要向运营商和手机厂商支付庞大的分成，关键在于这笔成本与向苹果支付的费用相比究竟如何。 ```mermaid flowchart TD accTitle: Google 流量获取成本 (TAC) 构成拆解 accDescr: 展示总计 550 亿美元 TAC 如何细分至 AdSense 分成、苹果 Safari 以及 Android 与浏览器合作方 A["总流量获取成本 (TAC) 约 550 亿美元"] --> B["广告联盟媒体分成 (AdSense/DoubleClick) 约 210 亿美元"] A --> C["搜索核心流量获取 约 340 亿美元"] C --> D["苹果 Safari 默认引擎 约 200 亿美元"] C --> E["非苹果渠道流量采买 约 140 亿美元"] E --> F["Firefox 及其他网页入口 约 40 亿美元"] E --> G["Android 运营商与 OEM 厂商 约 100 亿美元"] ``` 以 550 亿美元的总流量获取成本（TAC）为例，这一数据由两部分构成：一是付给展示 AdSense 和 DoubleClick 广告的媒体发布商的分成；二是购买搜索引擎流量入口的支出。在 Google 网络约 300 亿美元的年营收中，按三七分成计算，付给外部发布商的成本约为 210 亿美元，这意味着纯粹流向搜索入口的买量费用为 340 亿美元。扣除每年向苹果支付的 200 亿美元 Safari 默认引擎费后，剩下的 140 亿美元由 Firefox 等浏览器以及 Android 阵营的运营商与 OEM 分享。 - **非苹果渠道分配**：在剩余的 140 亿美元中，扣除 Firefox 及其他网页合作渠道的约 40 亿美元，实际流向 Android OEM 厂商和移动运营商的资金约为 100 亿美元。 - **成本效益对比**：这 100 亿美元虽然是向苹果支付金额的一半，但它覆盖的 Android 设备规模远超 iPhone，实际每台设备的收入分成比例低得多（估计仅为给苹果分成的四分之一左右）。 - **Play 商店盈利能力**：反垄断诉讼披露的数据显示，2019 年 Google Play 商店营收达到 112 亿美元，毛利为 85 亿美元，营业利润达 70 亿美元。 - **战略防御本质**：与 Google 搜索及广告核心业务近千亿美元的收入和约 300 亿美元的营业利润相比，Play 商店的利润只是锦上添花。Android 最大的价值在于“为公司的未来消除风险”，确保搜索入口不被第三方操作系统切断，成功将 Google 在 Web 时代的垄断地位平移至移动互联网时代。 --- ### [03:13:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11614s) 桑达尔掌舵双平台与 Android 的全球统治伴随着 Andy Rubin 的淡出，Sundar Pichai 逐步接管了 Google 旗下的核心产品线。他从早期的应用客户端到 Chrome 浏览器，再到 2013 年正式接管合并后的 Chrome 与 Android 部门，集齐了 Google 最核心的两大开放平台。无论何时走上演讲台，桑达尔都对这两个触达全球数十亿人的开放平台倍感自豪。此时的 Android 已发展成庞然大物，全球月活跃设备数突破 30 亿部——在一个人口仅 70 多亿的星球上，这意味着极其惊人的渗透率。回顾历史，移动操作系统的窗口期转瞬即逝，微软因为行动迟缓彻底掉队，而 Google 凭借敏锐的嗅觉在窗口关闭前拿下了 Android。到 2010 至 2011 年间，Google 内部士气高涨，即便中间偶有试验性产品折戟，但只要手握搜索、Chrome 和 Android 这类具备几十年生命力的世界级超级平台，科技巨头的基本盘便坚如磐石。 --- ### [03:14:46](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11686s) 社交阴影下的盲区：早于 Facebook 的 Orkut 在接连取得平台级胜利后，Google 环顾四周，科技行业唯一能够撼动其地位的庞大板块只剩下一个领域——社交网络（Social）。外界往往以为 Google 是社交领域的绝缘体，并对即将来临的 Google+ 溃败记忆犹新，但实际上 Google 很早便涉足该领域。早在 2007 年，Google 就尝试推出开放社交平台 OpenSocial，但由于 Facebook 的强势缺席，无法调动真正社交关系链的联盟最终沦为泡影。然而鲜为人知的是，Google 早在扎克伯格创办 Facebook 之前，就已经拥有了一款现象级的社交产品。2004 年 1 月——在 Gmail 发布、Google IPO 以及 Facebook 在哈佛大学宿舍上线之前——一位名叫 Orkut Büyükkökten 的土耳其工程师利用 Google 著名的“20% 自由时间”，打造了一款同名社交网络 Orkut。尽管这款产品未能在美国本土掀起波澜，但在国际市场上却大放异彩，其用户量巅峰时期达到 3 亿，并曾长期稳坐巴西和印度第一大社交网络的宝座，成为了 Google 进军社交赛道最早也最奇特的一段注脚。 ### [03:16:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11782s) 去中心化的繁荣与社交探索的屡战屡败 2007 年的 OpenSocial 和 2009 年的 Google Wave 构成了 Google 早期进军社交领域的典型写照。2009 年的 Google 规模已极其庞大，各个业务部门如同林立的孤岛：Android 团队躲在另一栋大楼里与苹果展开生死决战，而与此同时另一拨人却在做天马行空的 Google Wave。这种高度去中心化的组织架构在早期之所以行之有效，是因为各团队只专注于把自己负责的产品做到极致。当主播们在调研中询问各产品线的负责人“当时 Google 的整体贯穿战略究竟是什么”时，得到的回答往往出奇一致：“我们只是觉得这很酷、很 Googly，对 Web 生态有好处。”各条业务线埋头做产品，根本不需要去操心怎样帮广告业务赚钱。这种将顶层战略极度精简、且无需向下层层灌输的机制，反而是推动 Chrome、Android 等产品顺利落地的关键“特性（feature）”，而非缺陷。然而，缺乏统一方向的自由探索在社交领域屡屡碰壁。Google Wave 尽管发布会上演示视频惊艳四座，但由于定位模糊、没人搞得懂该怎么用而迅速夭折；紧接着在 2010 年推出的 Google Buzz，更是一上线就引发了灾难性的隐私泄露丑闻，旋即被草草关停。 ```mermaid flowchart TD accTitle: Google 2010 年前后的部门割裂格局 accDescr: 展示 Google 内部各个团队各自为政、缺乏统一身份与协调的孤岛状态 subgraph Business["各自为政的业务藩镇"] Android["Android 团队 独立孤岛 / 对抗 Apple"] Chrome["Chrome 团队 开发 ChromeOS / 职能重叠"] Search["Search 团队 核心命脉 / 拒绝外部干预"] YouTube["YouTube 独立账号密码体系"] Gmail["Gmail 公司唯一真正掌控的账号体系"] end Crisis["Urs Hölzle 备忘录：社交是人际网络的范式转移"] --> Centralize["拉里·佩奇重返 CEO 借社交危机重新集中公司权力"] Centralize -.-> Business ``` --- ### [03:18:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=11902s) 内部藩镇割据与 Urs Hölzle 的“危机备忘录” 就在 Buzz 遭遇惨败后，Google 极具声望的技术元老、分布式基础设施的奠基人 Urs Hölzle 起草了一份震撼全公司的备忘录。这份备忘录的性质堪比比尔·盖茨在 1995 年写下的那份著名的“互联网浪潮”（Internet Tidal Wave）备忘录。Urs 在信中疾呼：互联网正经历剧烈的范式转移，整个网络正从以网页和应用为中心转变为以“人”为中心；面对社交网络的崛起，Google 必须立即调集精兵强将，采取果断且大规模的战略应对。从更深层的公司政治来看，这不仅是一场对抗外部对手的战役，更是一场借题发挥的“方便的危机”（convenient crisis）。彼时的 Google 内部四分五裂，各路大将画地为牢： - Android 像个封闭海岛独自作战； - Chrome 也在做操作系统，与 Android 边界模糊，而皮柴（Sundar Pichai）当时尚未将二者统一； - 掌控核心算法与变现命脉的搜索团队神圣不可侵犯，外人休想染指； - YouTube 沿用自己原有的独立用户名和密码体系； - 整个公司唯有 Gmail 拥有真正意义上的用户登录与身份账号系统。眼看内部陷入彻底的混乱与内耗，尚未正式重掌 CEO 大权的拉里·佩奇（Larry Page）敏锐察觉到了失控的危险。他需要一个全公司级的统一目标来打破部门壁垒、把散落的权力重新收拢到最高管理层手中。拉里选中的借口与抓手正是社交——即便这一转型注定会带来血雨腥风。2010 年 5 月，Google 召集了内部排名前 50 的核心高管闭门开会；到了 2011 年 1 月，官方正式宣布拉里·佩奇将于当年 4 月回归重任 CEO，而拉里甚至直接将自己的办公室搬进了未来的 Google+ 大楼。 --- ### [03:21:14](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12074s) 围墙花园的致命威胁与 Google+ 的自上而下式豪赌在拉里和管理层眼中，防御 Facebook 是一场不折不扣的生死之战。此前 Chrome 是为了防御微软，Android 是为了防御苹果和微软，而如今由扎克伯格统领的 Facebook 正在构筑一个密不透风的“围墙花园”。Facebook 内部的数据完全屏蔽 Google 搜索爬虫的抓取，同时还挖走了雪莉·桑德伯格（Sheryl Sandberg）自建闭环的广告变现系统，不允许任何外部广告服务器介入。这与当年的 AOL 截然不同——当年 Google 还能帮 AOL 提供变现支持，而 Facebook 则是全盘自营；坊间更盛传 Facebook 打算自研搜索引擎。一旦用户注意力被彻底分流拦截，Google 的根基将被连根拔起。为了打赢这场战争，Google 彻底抛弃了以往那种依靠工程师灵感、自下而上、寻求共识的文化，展开了一场长达一年的高压攻坚战。这场战役由时任高管维克·贡多特拉（Vic Gundotra）挂帅： - **挂帅人选**：维克此前负责开发者关系并在 Google I/O 上担任当家主持，极具演讲号召力，深受佩奇等创始人器重并被赋予了空前的组织特权。 - **全站绑定定位**：Google+ 不仅是一个独立社交网络，更是被强行嵌入所有产品底层的黏合剂。维克当时对媒体高调宣称：“这是下一代 Google，是 Google+1。” - **强硬的人员与目标抽调**：公司从搜索、地图、YouTube 等王牌团队中强行抽调大批骨干并入 Google+；各业务团队当季的 OKR 被直接绑定上了“完成与 Google+ 的集成指标”。这种自上而下的命令与控制风格，在崇尚自由平等的 Google 内部引发了剧烈反弹。后来成为 TechCrunch 执行主编、当时还是 Google 实习生的丹尼·克里奇顿（Danny Crichton）曾撰文回忆：由于到处充斥着生硬的强行集成，等到他入职时，Google 内部的企业文化已经变得极度有毒（poisonous），团队之间的抵触与怨气在私下里迅速蔓延。 ### [03:25:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12322s) 强推“+1”与险些杀死的摇钱树在全力推进 Google+ 的狂热期，Google 内部的抵触情绪正在四处蔓延。一位被 Google 收购的代码仓库团队负责人，在被问到何时把产品接入 Google+ 时，甚至直接让对方“滚蛋（f*** off）”，而他绝非个例。当时整个公司的员工年终奖金都与 Google+ 的成败直接挂钩，高层甚至走火入魔到在 AdMob 的移动横幅广告底部硬塞进一个“+1”按钮——主播们不禁失笑：“到底有谁会想给一条广告点赞（+1）？”这完全是 Google 试图对标 Facebook “Like” 按钮的病态产物，他们坚信任何 Google 产品都必须能被“+1”。更致命的是，Google 甚至将触手伸向了 YouTube，强制将 YouTube 的视频评论区全面改版为 Google+ 的帖子。这一激进举动激起了创作者和用户的滔天愤怒，无异于亲手扼杀自己好不容易培育出来的“下金蛋的母鸡”。从纯产品设计的角度来看，Google+ 并非毫无新意，它带来了包括 Hangouts、Google Photos、Sparks 等一系列极具前瞻性的功能构想，对社交网络底层架构做了极其认真的重构。然而致命的症结在于：根本没有任何真实用户希望重新思考社交网络。这场浩大的战役纯粹是 Google 自上而下的意志，而非用户驱动的需求。他们试图给一个根本没有验证出产品市场匹配度（Product Market Fit, PMF）的产品强行加注火箭燃料，妄图通过自身庞大的生态流量强推其走向规模化。 ```mermaid flowchart TD accTitle: Google Plus Ecosystem Forcing Architecture accDescr: 展现 Google 将 Google+ 强行缝合到原有核心资产所造成的反弹与破坏 subgraph CoreAssets["原有健康业务 (摇钱树)"] YT["YouTube 社区与评论"] AdM["AdMob 移动广告网络"] Acq["被收购的独立团队与代码库"] end subgraph ForcingLayer["自上而下的 KPI 强制推进"] Bonus["全公司奖金与 G+ 绑定"] PlusOne["全量嵌入 '+1' 按钮"] end subgraph Fallout["产生后果"] UserAnger["用户与创作者强烈抵制"] DevBacklash["内部团队反弹与文化受损"] NoPMF["缺乏真实 PMF，纯靠输血"] end Bonus --> CoreAssets PlusOne --> YT PlusOne --> AdM YT --> UserAnger Acq --> DevBacklash CoreAssets --> NoPMF ``` --- ### [03:26:50](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12410s) 错位的战场：桌面端“圈子”与分化的社交未来主播们指出，Google+ 最大的战略失误甚至不仅是“在已经有 Facebook 的世界上再造一个 Facebook”，而是它诞生的那一刻，传统的 Facebook 模式其实已经开始走向黄昏。在 2011 年 6 月 Google+ 正式上线并于 2012 至 2013 年间疯狂发力时，马克·扎克伯格早就敏锐地意识到社交的演变方向变了。扎克伯格当时在做的事情是火速收购 Instagram、重金买下 WhatsApp，将 Meta 的底座重新解构为两个截然不同的赛道：一边是公开的 UGC 媒体平台（如 Instagram、YouTube），另一边则是绝对封闭的私密即时通讯（Private Messaging）。而这期间的 Google 却在背道而驰：他们居然搞出了一个“桌面优先（Desktop First）”、甚至拥有专属桌面端交互界面的产品，逼迫用户在电脑屏幕上把好友手动拖拽归类到一个个“圈子（Circles）”里。这种产品形态带有极其浓厚且僵化的纯计算机科学思维：“我的好友处在重叠或不重叠的集合中，我需要极其严谨地给他们打标签，以便我以确定性的方式决定跟谁分享什么。”然而现实世界里根本没有普通人愿意花时间去打理这些社交集合，这一脱离移动互联网大势的沉重架构，注定了它的悲剧收场。 - **社交赛道的分化**：公开内容流（Public Media / UGC）与私密通讯（Private Messaging）各自演进，传统静态关系链的桌面社交形态彻底过时。 - **工程师思维的局限**：“圈子（Circles）”功能完全是把图论与集合论强加给终端用户的产物，严重缺乏对人类真实社交习惯的共情。 --- ### [03:28:10](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12490s) 谷歌的“Vista时刻”：错失通讯与云计算的重大代价主播明确将 Google+ 时代定性为 Google 的“Windows Longhorn / Windows Vista 时刻”。在微软的发展史上，Vista 的难产与泥潭牵扯了全公司最顶尖的工程资源和管理精力，直接导致微软未能及时遏制 Google 的崛起，并在开发者群体中丧失了号召力。那么，Google+ 又给 Google 带来了怎样惨痛的机会成本？表面上看，当时的 Android、YouTube、Chrome 和核心搜索业务依然强劲，但实际上它在两个至关重要的战略高地留下了不可挽回的真空：第一个重大代价是**移动即时通讯（Messaging）**。在非 Google+ 的平行宇宙里，像 WhatsApp 这样的移动通讯巨头完全可能被 Google 收入囊中。主播回忆起自己在斯坦福商学院（GSB）攻读 MBA 时的经历：当时 Eric Schmidt 刚卸任 CEO 成为执行董事长，在商学院联合授课，那门课是全校体验最好的王牌课程之一。就在 Facebook 宣布收购 WhatsApp 的那个学期，Schmidt 走进教室时满脸懊恼与沮丧，脱口而出：“该死的，我们错过了，我们彻底错过了（God damn it, we missed it. We totally missed it）。” Google 正是因为将注意力全部锁死在 Google+ 上，才对眼皮底下的移动通讯革命视而不见。第二个更为沉重的代价是**云计算（Cloud）**。早在“Z-Quake（扎克伯格地震）”备忘录发出时，Google 最应该全力以赴重兵投入的本该是云基础设施，但其高层精力却被社交战役全面稀释。在很长一段时间里，Google 在云计算战略上迷失了方向，直接导致其日后的 Google Cloud 在市场份额上被 Amazon AWS 和微软 Azure 远远甩在身后。更深远的负面影响在于企业文化和创新敏捷度的崩塌：从 Google+ 时代之后，顶尖人才因内部政治与文化恶化而流失，Google 从一家执行力爆棚、持续推出颠覆性产品的创新机器，彻底沦为一家庞大、缓慢且充斥着官僚主义的大公司。在随后的十多年里，直到 Gemini 问世之前，Google 几乎再也没有推出过任何具有划时代意义的全新消费级超级应用。 --- ### [03:31:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12712s) 虚惊一场的生存危机与意外的遗产回看当年的恐惧，整个硅谷当时都被笼罩在“生存危机”的焦虑中。2013 年，Neal Mohan（后出任 YouTube CEO）曾在一份公司战略备忘录中极其严肃地预警：“Facebook 存在成为整个互联网入口的巨大风险。”无论是 Google 还是刚上市不久、焦头烂额应对移动转型的 Facebook，双方都以为这是一场你死我活的世纪决战。但历史开了一个巨大的玩笑：Google 在社交战场上一败涂地，但它的核心搜索与广告现金流根本毫发无损；而 Facebook 最终也走向了完全不同的业务轨道。整场战役荒诞得就像科恩兄弟的黑色幽默电影《阅后即焚》（Burn After Reading）的结尾——经历了一系列鸡飞狗跳的混乱和惨烈交火后，所有人面面相觑：“等等，刚才发生的这一切，到底有什么意义？” 戏剧性的是，社交领域最终演化出的两大终局形态——私密即时通讯和以视频为核心的公开 UGC 媒体，而后者正是 Google 旗下的 YouTube。Google 真正的王牌一直都在自己的后院，它本该什么都不做，静静地看着 YouTube 的“印钞机轰鸣运转”。不过，Google+ 的废墟中也诞生了两笔极具价值的遗产： 1. **孵化出两款亿级长青产品**：从 Google+ 剥离出的 Hangouts（后演变为 Google Meet）以及成长为拥有超过十亿用户的 **Google Photos**。 2. **重塑了统一的底层基建与公司架构**：正所谓“绝不要浪费一场危机”，Google+ 彻底打破了早期各业务团队占山为王、各自为政的“诸侯割据”局面，逼迫全公司实现了统一的 UI 设计语言，并建立了至关重要的底层基础设施——**Google 账号统一认证体系（Google Accounts）**，为 Google 转型为一个高度协同的整体奠定了基石。 --- ### [03:34:18](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12858s) 黯然落幕：Vic Gundotra 离职与 Google+ 的终结这场闹剧最终以掌舵者的退场划上句号。作为 Google+ 运动的绝对总指挥，Vic Gundotra 最终于 2014 年黯然离开 Google。此后，Google+ 在内部被逐步边缘化、拆解。直到 2019 年，Google 官方发布公告，借由发现了一处可能泄露用户隐私的大型安全漏洞作为体面的台阶，正式宣布彻底关闭面向消费者的 Google+ 服务，这一耗费了公司无数顶级资源和工程心血的宏大项目终于成为了科技史上一道沉重的注脚。 ### [03:34:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12874s) Google+ 的彻底落幕与荒谬的重定向链接在回顾完 Google 试图涉足社交网络的惨痛教训后，主播们谈到了关闭 Google+ 的后续荒谬现象。Google 产品线频繁更名、关停是出了名的，其间甚至推出过诸如 Currents 这样名字相似但定位模糊的过渡产品。当 Google 最终决定关闭 Google+ 时，造成了极其严重的网络信息断层：当年许多深度用户在 Google+ 上撰写了大量长篇文章与思考帖，如今这些内容彻底化为乌有，甚至连播客做本期节目的历史考证都受到阻碍。更为滑稽的是，现在用户在互联网任何地方点击以 `plus.google.com/...` 开头的历史链接，都会被统一重定向到 Google Currents；然而，Google Currents 此后也已经被官方关停。最终，点击任何曾经的 Google+ 链接，只会把你带到一篇告知用户“Currents 现已关停”的 Google Workspace 官方博客说明页面。这种层层关停后留下的荒诞死胡同，被两位主播戏称为“最典型的 Google 作风（The most Googley thing）”。 ### [03:35:28](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12928s) 2015 重组：Alphabet 诞生与 Sundar Pichai 接棒时间来到 2015 年 8 月，Google 迎来了彻底的重组与新纪元。公司宣布将自身重组为控股母公司 Alphabet。拉里·佩奇（Larry Page）出任 Alphabet 的 CEO，而桑达尔·皮查伊（Sundar Pichai）则晋升为核心子公司 Google 的 CEO。在新的架构中，Google 仍然是迄今为止规模最大、最核心的运营实体，且包含搜索、YouTube、Android、Chrome 等全部打通的核心消费级产品，并未拆散。与此同时，其他处于前沿探索阶段的项目则被归类为“Other Bets（其他赌注）”，独立于 Google 运营。 ```mermaid flowchart TD accTitle: Alphabet 2015 架构示意图 accDescr: 展示 2015 年重组后 Alphabet 控股公司下辖的 Google 核心业务与 Other Bets 其他赌注业务结构 subgraph Alphabet["Alphabet 控股公司 (CEO: Larry Page)"] subgraph GoogleEntity["核心实体: Google (CEO: Sundar Pichai)"] G1["Search / AdWords"] G2["Gmail / Maps / Docs"] G3["Chrome / Android"] G4["YouTube"] end subgraph OtherBets["Other Bets (其他赌注)"] OB1["Google X / Waymo"] OB2["Nest / Google Fiber"] OB3["Calico / Verily (生命健康)"] OB4["GV / CapitalG (风险投资)"] end end Alphabet --> GoogleEntity Alphabet --> OtherBets ``` 关于为什么是皮查伊出任 Google 的掌舵人，主播们给出了深刻的战略与文化解读： - **组织治愈与声望**：Google 刚经历了 Google+ 强行捆绑全公司产品带来的内部创伤与士气低落，迫切需要一位受人爱戴的领导者来进行“组织疗愈”。 - **过硬的产品功勋**：皮查伊虽从未直接执掌过搜索或广告业务，但他亲自领导打造了 Chrome 和 Android 这两大核心平台。正是这两大平台成功将 Google 稳固推入了移动互联网时代，化解了最大的生存威胁，并持续为搜索飞轮提供无限动力。 - **卓越的和解型领导风格**：在顶尖天才扎堆、高层膨胀自负的 Google 文化中，皮查伊展现出了极强的协调平衡能力与谦和气质，是能在各大元老和强硬派之间调和纷争的“和事佬（peacemaker）”。 ### [03:37:47](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13067s) 2015 财务全景：依然是全球最强悍的搜索广告机器从财务数据来看，截至 2015 年底重组之际，Google 已经长成一个庞然巨物： - **年营业收入**：达到 750 亿美元。 - **收入结构**：第一方网站（Google 自身站点，包括搜索、AdWords、Gmail、Maps）贡献了高达 520 亿美元；由 DoubleClick 和 AdSense 构成的外部网络广告贡献了约 150 亿美元（但这部分收入利润率偏低）；此时的 YouTube 已经实现了盈利。 - **利润表现**：核心 Google 业务贡献了约 230 亿美元的营业利润，而 Other Bets 业务线当时每年亏损约 35 亿美元。两位主播指出，回顾这些数字会让人产生双重震撼：一方面，2015 年的 Google 规模已经庞大到令人发指；但另一方面，在接下来的十年里，其业务在**完全相同的商业模式**下居然又实现了 5 倍的惊人扩张。自 2002 年敲定拍卖竞价搜索广告以来，Google 最底层的变现机器从未发生过本质变化。在这十年间打造的所有千亿级重磅产品，本质上都是在给这颗最初的搜索广告种子保驾护航，使之顺利穿透平台迁徙，成长为人类商业史上最巨大的市场。 ### [03:39:26](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13166s) 黄金一代齐聚：Google 的 AI 人才梦幻阵容在进入最终商业分析前，节目给出了一个震撼人心的历史注脚：在 2015 到 2016 年 Alphabet 刚成立的短短 12 个月内，当今人工智能革命中最具决定性意义的核心先驱，居然几乎全都是 Google 的在册员工。这些名字构成了推动未来数十年技术变革的人类 AI 顶峰群星： - **深度学习开创者**：AlexNet 的作者 Alex Krizhevsky，其导师、“深度学习教父” Geoffrey Hinton，以及共同作者 Ilya Sutskever（后来成为 OpenAI 的联合创始人与首席科学家）。 - **当今 AI 领袖**：Dario Amodei（后来创立 Anthropic）、Andrej Karpathy（曾任特斯拉人工智能总监）、Chris Olah、Noam Shazeer（Character.ai 创始人）、生成对抗网络（GAN）发明者 Ian Goodfellow。 - **DeepMind 创始人团队**（Google 于 2014 年收购）：Demis Hassabis、Shane Legg 以及 Mustafa Suleyman（现微软 AI 业务负责人）。 - **学术与工业界大师**：吴恩达（Andrew Ng）、Quoc Le、Oriol Vinyals。 - **奠定 LLM 时代基石的先驱**：开创性提出 Transformer 架构的全部论文作者——Google 在 2017 年 6 月发表了《Attention Is All You Need》，而这一架构正是今天几乎所有主流大语言模型（LLM）的底层核心。甚至在当时，Ilya Sutskever 还在 Google Brain 团队中向同事们抛出了那个划时代的假说：如果我们搭建一个超级庞大的神经网络，把通过 Google 搜索索引建立起来的整个互联网的数据全倒进去训练，它会不会学会一切？回望历史，从某种视角来看，Google 在 2005 至 2015 这十年间研发与收购各项产品的过程，实际上在客观上完成了对 AI 时代所需的三大核心要素——海量全网数据、基础设施算力与顶级科研人才的垄断性吸纳。早在 2000 年，拉里·佩奇就曾极具远见地指出：“人工智能将是 Google 的终极形态。”如果拥有一个终极搜索引擎，它能理解网络上的一切，懂得你想要的一切，并给出完美答案，那就是通用的人工智能；Google 当年拥有的 6,000 台电脑以及堆积起来足有 70 英里高的数据索引，正是这项伟大工程的最初起点。 ### [03:44:11](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13451s) 商业模式剖析：Hamilton Helmer 的“七大力量”框架节目由此切入最后的深度分析环节，主播们引入了汉密尔顿·赫尔默（Hamilton Helmer）著名的《7 Powers（七大力量）》战略框架，用于评估 Google 为何能够长期拥有超越竞争对手的持续超额回报与定价优势。分析 Google 的商业“力量”极其反直觉且复杂，因为大众日常所频繁接触到的消费端产品（Gmail、Maps、YouTube、Android）本身并不是真正的商业交易发生地。要想真正看穿 Google 的护城河，战略视角必须紧紧聚焦于：**为什么广告主愿意将边际上的每一块钱广告预算投向 Google，而不是其他任何竞争平台？** 围绕这一点，Google 构建出了横跨双边市场的多维度商业壁垒。 ### [03:44:52](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13492s) 商业分析：用“七大权力”剖析谷歌商业帝国在大卫和本看来，与其枯燥地列举商业框架，不如用汉密尔顿·赫尔默的《7 Powers》（七大权力）来系统审视谷歌在 2004 至 2015 年间的全面爆发。令大卫惊叹的是，大企业通常极难获得“逆向定位”（Counterpositioning，通常属于挑战者颠覆成熟巨头的利器），但谷歌却创造了商业史上最典型的逆向定位范例——Android。面对微软向 OEM 厂商每台设备收取数十美元 Windows Mobile 授权费的商业模式，谷歌直接祭出了比尔·格利（Bill Gurley）所谓的“低于免费”（Less than Free）策略：“我的竞争对手让你付钱；那我反过来给你分钱怎么样？”凭借搜索引擎带来的滚滚广告现金流，谷歌能将搜索收入的一部分分给运营商和手机厂商，而依赖软件授权收入的微软根本无法跟进这一玩法。这种由搜索驱动的庞大体系同样带来了无以匹敌的规模效应（Scale Economies）。无论是搜索、展示广告还是在线视频，谷歌构建起一个对广告主而言“一站式搞定”（One-stop shop）的庞大分发网络。广告主不再需要分别组建团队、投入人力去对接不同平台；而在底层技术上，基础设施的规模摊薄了每一次计算的边际成本。更关键的是飞轮效应：广告主和用户越多，在每一笔瞬时拍卖中寻找最大出价的能力就越强，谷歌获取的利润也就越惊人。在网络效应（Network Economies）与转换成本（Switching Costs）方面，谷歌的触角同样无处不在。YouTube 构成了典型的双边网络：创作者越多，引来的观众就越多；观众越多，创作者获得的广告分成就越高。Android 的应用开发者与终端用户之间亦是如此。至于转换成本，在核心的搜索广告端，广告主并非因为“系统难以迁移”而留下，而是因为全球具有最高购买意向的搜索入口只有两个——谷歌和亚马逊；但对普通用户而言，切换的门槛却高得惊人：谁会轻易放弃自己在 Gmail 里沉淀了整整 20 年、而且还是免费存储的全部个人邮件记录？更不用说 YouTube 算法早已被调教得对个人的兴趣了如指掌。 - **逆向定位**：Android 对决微软，开创“比免费还要便宜”的商业模式，凭借广告变现支撑免费甚至倒贴的生态。 - **规模效应**：全渠道一站式广告投放平台，配合千次搜索拍卖的价格最优化机制与庞大自研硬件基建。 - **网络效应**：YouTube（创作者与观众）、Android（开发者与用户）构筑坚不可摧的双边生态。 - **转换成本**：核心搜索对广告主是意图流量的不可替代性，对 C 端用户则是 Gmail 二十年历史数据与 YouTube 深度推荐算法的强大锁定。 ```mermaid flowchart TD accTitle: Google 7 Powers accDescr: 谷歌七大权力在各核心业务中的典型应用与飞轮结构 subgraph Defense ["防御与增长壁垒 (7 Powers)"] CP["逆向定位 (Counterpositioning) Android '低于免费' 对决微软授权费"] SE["规模效应 (Scale Economies) 一站式广告投放 & 自建服务器基础设施"] NE["网络效应 (Network Economies) YouTube 双边生态 & Android 开发者池"] SC["转换成本 (Switching Costs) Gmail 20年邮件归档 & 算法偏好锁定"] end CP --> Engine["搜索与广告核心飞轮 (AdWords / AdSense)"] SE --> Engine NE --> Engine SC --> Engine ``` --- ### [03:48:07](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13687s) 品牌声誉、特权资源与超越业界的自研基建在探讨品牌（Branding）时，本和大卫回忆起谷歌鼎盛时期的那种“科技光环”。当时只要冠以谷歌之名发布新产品，技术极客和大众就会争先恐后去尝试。大卫回忆起自己当年疯狂求购 Google Wave 邀请码的往事，尽管该产品后来彻底失败，但在当时人们眼中“谷歌”二字就代表着酷炫与未来。讽刺的是，正是这种巨大的品牌声誉资产，反而在后来的 AI 时代让谷歌束手束脚、迟疑不前——因为品牌价值太大，以至于他们在被推下悬崖之前，根本不敢轻易把自己的金字招牌赌在尚未完全可靠的技术上。至于独占资源（Cornered Resource）与流程权力（Process Power），进入 AI 时代的当下，YouTube 拥有的庞大海量视频库与多模态数据，成了任何竞争对手都可望而不可即的训练宝库。而在基础设施层面，许多人将其视为单纯的规模优势，但大卫指出它更是一种无可复制的流程权力。更令人惊叹的是谷歌内部那些领先于外部世界数年的秘密软件工程系统。谷歌向外部世界开源了许多标准工具（如根据内部系统演化而来的 Kubernetes），但其内部真正运转的核心仍是 Borg 集群管理系统。Borg 始终是谷歌高度保密的独家秘方，不仅成本远低于业界公有云方案，其稳定度与工程优雅度也让离开谷歌的工程师们在走向外部世界后由衷怀念。 --- ### [03:50:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13800s) Playbook 之一：影子平台公司与海量并购飞轮进入 Playbook 阶段，本抛出了一个深刻的战略问题：谷歌一直渴望成为一家平台公司，但它到底算不算成功？大卫的定性非常精炼：它不是苹果或微软那种“纯粹的平台公司”（Pure play platform company），而是一家“影子平台公司”或“生态系统公司”。苹果和微软卖的是硬件或软件授权，建立平台是为了直接强化软硬件销售；而谷歌即便控制了全球最大规模的移动操作系统 Android，其终极落脚点依然是为广告主提供精准触达的管道，赚的依然是广告费。当重大的战略关口来临时，这个本质区别决定了所有决策的取向。紧接着，他们探讨了谷歌在 2000 年代到 2010 年代初展现出的疯狂并购手笔。外界往往只记得 YouTube、Android、DoubleClick 和 AdMob 这类改变产业格局的大交易，但事实上谷歌吞并了数以百计的小型创新团队与初创公司： - 支撑起 Google 生产力套件与社区雏形的团队（Google Groups、Spreadsheets、Docs、Blogger）； - 奠定广告变现核心的 Applied Semantics（带来 AdSense 专利与算法）、Admeld 与 Widevine； - 赋予地图与分析能力的 Where2 Technologies（Google Maps 核心架构）与 Urchin（转化为 Google Analytics）； - 移动社交与工具雏形如 Dodgeball、FeedBurner、reCAPTCHA、Slide、Jambool、Like.com、Punchd、Zagat、Sparrow 以及 Wavii 等。在调研中，本发现了一个极其独特的谷歌文化现象：他们打造了一种“回旋镖式”（Boomerang）的人才与生态闭环。优秀工程师被吸引到谷歌，与顶尖极客共事并结识创业合伙人；随后他们离开谷歌去创立 Web 应用；几年后，谷歌再把这些初创团队高价买回来！这种循环上演了数十次甚至上百次。从局外人看来，放任员工离职又溢价买回似乎是“疯了”，但大卫和本一针见血地指出：只要全网的 Web 生态越繁荣，最终受惠于流量变现的永远是拥有 AdWords 印钞机的谷歌。 --- ### [03:52:45](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13965s) Playbook 之二：Google Glass 与 Meta Ray-Ban 的文化隐喻本回顾了自己在黑客松为 Google Glass 开发应用的旧事，并对比了当年的 Google Glass 与如今爆火的 Meta Ray-Ban 智能眼镜。从底层功能上拆解，两者惊人地一致：镜腿触控手势、拍照摄像、环境交互，甚至当年 Google Glass 还能运行简易的文字应用，而 Meta 即将推出的全息眼镜在产品逻辑上如出一辙。然而，产品的社会命运却有着天壤之别。大卫和本总结道，这绝非单纯的“推出时机”早晚问题，而是两家科技巨头在文化 DNA 上的根本分野： - **Google Glass**：把佩戴者变成了一个古怪怪异的“赛博朋克改造人”（Cyborg），在公共场合引发了隐私恐慌与社交排斥； - **Meta Ray-Ban**：将科技彻底隐形于经典的时尚墨镜外表之下，专为“正常人”的社交审美打造。这一对比成了谷歌与 Facebook（Meta）文化差异最绝妙的隐喻：谷歌由一群聪敏绝顶但略显脱离现实的学者极客组成，他们醉心于技术实现，却完全无法理解为什么“让人看起来像个怪胎”会彻底毁掉一款消费级产品；而 Facebook 骨子里流淌着理解人性虚荣、社交接纳与大众审美的基因。 ### [03:53:47](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14027s) 极客工程文化与顶层极简战略：让伟大的产品自己说话在回顾 Google 的产品哲学时，两位主持人首先提到了与 Meta 在穿戴设备理念上的鲜明对比。Meta 从一开始就意识到必须把产品做得“酷”，因而找来了眼镜巨头 Luxottica 展开合作，把技术塞进普通人日常会佩戴的眼镜里；相比之下，Google Glass 当年的演示虽然在技术上与 Meta 今天的 AR 演示极其相似，但缺乏这种对潮流感与外形审美的拿捏。不过，Google 真正成功构筑的文化内核，在于让工程师全心沉浸于“打造伟大产品”：去攻克计算机科学上极其困难的技术难题，把它变成真正有用的东西，并交付给全球用户。在很长一段时间里，Google 的团队在推出新产品时甚至完全不需要考虑商业模式，产品上线多年后也依然如此。这种文化的能够运转，本质上依赖于一种独特的顶层结构：在公司最顶端，存在一层极其精炼、高度紧凑且无懈可击的宏观战略；而在这层极薄的顶层战略之下，其余所有员工的任务就纯粹是“做出最棒的产品”。 ### [03:54:48](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14088s) Playbook 核心复盘：跨越技术周期的 Android 奇迹关于本期节目的 Playbook 总结，主持人强调了在 Android 章节中反复提及的论断：Android 是科技史上“所有胜利之母”（the mother of all wins）。在科技商业史上，几乎没有任何一家公司能够将其核心商业模式跨越不同的技术周期完美延展，而 Google 不仅做到了，还在下一个时代（移动互联网）中确立了绝对的统治地位。Android 之于 Google，其意义完全等同于微软篇章里的 Azure——它足以赦免公司此前犯下的所有过错（尽管除了 Google+ 之外，Google 并没有犯下太多致命错误）。如果说还有什么比这更完美的结果，那就是当年 Google 不仅推出 Android，还顺带把 iPhone 给做了，从而直接拿走苹果硬件生态的高额利润，而不仅仅是通过移动操作系统筑起护城河来保卫自己的核心搜索业务。 ### [03:55:42](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14142s) 十亿用户产品矩阵：Google 难以企及的产品神话从整体复盘（Quintessence）来看，Google 在 2005 至 2011 年间创造了商业史上独一无二的战绩：在步入这一阶段前，Google 甚至只有一项尚未达到十亿用户的搜索业务，而如今这家公司竟然拥有 8 款用户量突破十亿的超级产品。如果将 Google Play 商店单独计算（Google 官方通常如此划分），这一数字甚至可以达到 9 款。对比之下，全球科技巨头的产品规模如下： - **Google（8~9 款）**：Google 搜索、Android、Chrome 浏览器、YouTube、Gmail、Google 地图、Google 云端硬盘（Drive）、Google 相册（Photos），以及单独统计的 Google Play 商店。 - **Meta（4 款）**：Facebook（蓝标应用）、WhatsApp、Instagram 和 Messenger（虽然 Meta 宣称其 Meta AI 跨产品聚合用户已破十亿，但主持人认为这种将嵌入式功能算作独立活跃用户的算法水分很大）。 - **Apple（严格算 1 款，宽松算 3~4 款）**：iPhone 毋庸置疑达到十亿级；如果将随硬件免费附赠的原生应用独立算入，则可加上 iMessage 与 Safari 浏览器；至于 iPad 与 Mac 则均未达到十亿活跃量级。 - **Microsoft（2 款）**：Windows 操作系统与 LinkedIn。 - **Amazon（0 款）**：没有任何一款产品的用户量达到十亿级别。关于这些产品的独立性，两位主持人展开了趣味辩论：有人认为如果苹果自带的 iMessage 不能单算，那 Google 很多靠 Android 预装的应用（如 Google 地图）是否也该扣除？但经过推敲，Google 搜索、Android、Chrome 实际上来自完全不同的分发渠道；即便把相互借势的 Gmail 与 Drive 合并计算，或者把受 Android 提振的地图像素级扣减，Google 仍然实打实坐拥 6 到 8 款独立的十亿级应用。在整个人类商业史上，从未有任何一家公司在短时间内打出过如此豪华的连击。 ```mermaid flowchart TD accTitle: 科技巨头十亿用户级产品数量对比 accDescr: 对比 Google、Meta、Apple、微软和亚马逊旗下突破十亿用户的产品分布 subgraph GoogleGroup["Google / Alphabet (8-9款)"] G1["Search"] G2["Android"] G3["Chrome"] G4["YouTube"] G5["Gmail"] G6["Maps"] G7["Drive"] G8["Photos"] G9["Play Store (官方口径)"] end subgraph MetaGroup["Meta (4款)"] M1["Facebook"] M2["WhatsApp"] M3["Instagram"] M4["Messenger"] end subgraph OtherTech["其他巨头"] A1["Apple: iPhone (若算预装则加 iMessage/Safari)"] MS1["Microsoft: Windows / LinkedIn (2款)"] AMZ["Amazon: 0款"] end ``` ### [03:58:24](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14304s) 成功的试金石：以底层核心技术洞察为唯一驱动节目引出了本期最深刻的核心洞察：**几乎所有在 Google 取得巨大成功的产品，底层都依赖于一项堪比学术期刊论文级别的“核心技术突破”（Core Technology Insight）。** 这项底层技术洞察如此纯粹、优雅而强大，以至于技术突破本身直接变成了产品的核心用户体验： - **Google 搜索**：底层是 PageRank 算法与广告竞价机制，其机械般的精巧与数学美感直接构筑了商业帝国的基石。 - **Gmail**：通过突破性的系统架构提供 1GB 海量存储空间，并开创性地运用 AJAX 技术实现了桌面级的瞬时响应体验。 - **Google 地图与 Google Docs**：在浏览器内实现了地图流畅拖拽与多人实时协作编辑的底层工程突破。 - **YouTube**：攻克了在全球范围内低成本、高并发交付点播视频的分布式工程极限。 - **Chrome**：在早期的漫画发布说明中就详细罗列了 4 到 7 项关于浏览器多进程架构与 V8 引擎的核心技术创新。 - **Google 相册**：多年来在计算机视觉与 AI 领域的技术突破，直接转化为用户惊叹的智能检索与自动相册功能。 Android 略有特殊，它更偏向于战略层面的完美执行与产业生态的合纵连横，但在技术上也堪称将 Linux 成功驯服并塞入低功耗便携电池设备的工程奇迹（类似于苹果将 OS X 裁切演化为 iOS）。 ```mermaid flowchart LR accTitle: Google 核心成功产品的技术驱动模型 accDescr: 展示学术级底层技术突破如何直接跨过产品层转化为极致用户体验 subgraph TechCore["底层学术级核心技术洞察"] T1["PageRank 算法"] T2["AJAX 响应 & 1GB 分布式存储"] T3["多人实时协作架构"] T4["多进程沙箱 & V8 引擎"] T5["计算机视觉与 AI 算法"] end subgraph TechUX["极客产品形态"] P1["Google 搜索"] P2["Gmail"] P3["Google Docs / Maps"] P4["Chrome 浏览器"] P5["Google 相册"] end TechCore -->|"技术突破直接转化为体验"| TechUX ``` 这种模式也反向揭示了 Google 屡次折戟的原因：凡是脱离了底层技术突破、单纯依靠产品经理构想“交互体验”或“社交概念”的项目（如 Google+、Google Wave），在 Google 内部几乎无一例外走向溃败。Google 的基因根本无法凭空孵化出一个像 Instagram 这样的产品（讽刺的是，Instagram 联合创始人 Kevin Systrom 正是从 Google 离职后才创立了它）。正如前 CEO Eric Schmidt 在其书中所写，他当年在评估项目时总会拷问产品经理："支撑这一切运转的核心技术洞察到底是什么？"（What is your core technical insight that makes it all work?）如果 PM 无法给出有力回答，Schmidt 绝不会给该项目批准预算。 ### [04:02:41](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14561s) 节目福利分享：纽约 CAMP 的 Bluey 沉浸式乐园体验在常规的节目结尾“Carve Outs”推荐环节，主持人在揭晓自己最终到底选购了哪款游戏机之前，先分享了全家在纽约无线电城音乐厅（Radio City）录制完现场节目后的家庭度假见闻。他们带孩子们前往位于纽约市的 CAMP 线下概念店，深度体验了《布鲁伊》（Bluey）官方沉浸式互动展（Bluey x CAMP）。该展览在纽约市区的物理空间中一比一逼真复刻了动画片里标志性的布鲁伊大宅。由于这栋房子在动画中本身就如同一个具备灵魂的角色，当父母与孩子们真正走入其中自由探索时，现场的还原度、沉浸感与趣味性完全不负盛名，给全家人留下了深刻印象。 ### [04:03:42](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14622s) 个人好物推荐（Carve Outs）：Steam Deck 与掌机体验在聊完纽约的 Bluey x CAMP 沉浸式体验后，David 揭晓了自己在游戏主机上的最终选择：他买了 Steam Deck。虽然过去一个月因为筹备 Radio City 活动以及本期超长节目的调研录制，他几乎没抽出时间去玩，但他对这款设备赞不绝口。在做决定时，David 曾极度渴望大女儿已经能陪他一起玩《马力欧卡丁车》（Mario Kart），但女儿现在的年龄显然还没准备好。因此，如果单纯买一台给自己享受的机器，Steam Deck 成为了毫无悬念的胜利者。 David 高度评价了 Valve 的产品设计实力。他指出，Valve 真正厉害的地方在于，他们将一台复杂的 PC 游戏设备完全抽象成了傻瓜式的游戏主机体验。David 一直喜欢 PC 类型的游戏，但很多年没有碰过 PC 游戏，因为他既不想花心思组装机器，也没有空间和需求在家里摆一台多余的 PC。他渴望的是传统主机的极简感——“买下它、打开电源、下载游戏、直接开玩”。Valve 成功地将这种体验压缩到了掌机形态中，玩家完全不需要折腾驱动程序或纠结硬件配置。David 补充说，自己未来一年左右大概率还是会买任天堂的 Switch 2，但眼下 Steam Deck 已经完全满足了他的期待。 --- ### [04:05:22](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14722s) 个人好物推荐：Claude、索尼微单与羊绒衫随后轮到 Ben 分享自己的三个好物推荐： 1. **Claude（Anthropic 出品）：** Ben 发誓这绝非因为对方赞助而硬捧，而是它真正重塑了生产力。使用 AI 彻底颠覆了他为《Acquired》做案头研究和准备节目的方式，“我整天都泡在里面，完全无法想象回到过去没有它的日子”。 2. **Sony RX100 VII（黑卡 7 代便携相机）：** Ben 最近刚买了备受网络推崇的 Fujifilm X100VI，虽然富士胶片模拟色彩极佳、35mm 等效焦距充满仪式感，但挂在脖子上依然是个不小的物件。当天早上 Ben 看到投资人 Shomik Ghosh（注：语音误读为 Shil Monot）在 Twitter 上发帖纠结便携机，让他猛然记起自己有多爱老款的 Sony RX100 VII。这台 2019 年发布的卡片机极其小巧，能轻松塞进裤兜，却塞入了一颗大变焦镜头。当他不想在脖子上挂相机、空间受限时，“手机 + 索尼黑卡”就是绝配。虽然它不是全画幅、也没有富士那种复古的“摄影感”，但对于绝大多数日常场景而言，出片效果远胜手机镜头。Ben 唯一吐槽的是它仍使用老接口，急需升级 USB-C。此外，配合 Adobe Lightroom 最新上线的 AI 降噪（Denoise）功能，画质表现令人惊艳。 3. **Carissimi 羊绒长袖衫：** 一位听友创立的高端服装品牌 Carissimi 给 Ben 寄了一件羊绒衫。Ben 表示整场长达七个半小时的录制中自己一直穿着它，面料在微凉或稍热的天气都极其舒适。David 打趣说自己盯着 Ben 看了七个半小时，一直在心里感叹“天呐，Ben 今天看起来真帅”。Ben 坦言自己最喜欢的胶囊衣橱风格就是不用动脑的深色长袖圆领衫，而 Carissimi 这件是他穿过质感最为顶级的版本。 --- ### [04:08:38](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14918s) 赞助商致谢、节目制作贡献者与尾声节目进入尾声，主持人首先感谢了本季的四家赞助伙伴： - **J.P. Morgan Payments**（摩根大通支付基础设施） - **Anthropic**（大语言模型 Claude 开发者） - **Statsig**（产品研发实验平台） - **Vercel**（全栈 Web 开发与部署平台）接着，两位主播对为本期庞大调研提供帮助与受访的众多嘉宾表达了由衷谢意： - **嘉宾与专家名单：** 包括 Hiroshi Lockheimer、Tim Armstrong、Salar Kamangar（注：语音辨识误作 Sam Schillis）、Hunter Walk、Nick Fox、Shona Brown、Clay Bavor、Max Ross、Greg Corrado 以及 Demis Hassabis（部分专家将在下一期深入探讨）。 - **案头与访谈致谢：** 特别致谢 Worldly Partners 的 Arvind Navaratnam 撰写的深度商业研究报告，以及 Paul Buchheit（Gmail 之父）、Bill Coughran、Jonathan Rochelle、Bradley Horowitz、John Hanke、Ben Edelman、Eisar Lipkovitz、Ben Liebman 等前 Google 核心团队成员，还有诸多不便透露姓名的幕后协助者。最后，主播引导听友收听回顾此前发布的《Google Part I》（搜索业务的起步）、微软三部曲以及史蒂夫·鲍尔默（Steve Ballmer）专访和 Meta 特辑；同时推荐了二台播客 **ACQ2** 最新一期对 Bret Taylor 与 Clay Bavor 的 AI 深度对话。节目在标志性的片尾主题曲《Who Got the Truth?》旋律中正式收官。 ## Notable quotes > [03:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=233s) 如果你公布了营收，大家就会问有多少，而且永远都不会觉得够。原本能涨 100 倍、1000 倍的明星公司，突然之间就变成了只涨 2 倍的平庸之辈。但如果你还没有营收，你就可以说自己是‘预营收（Pre-revenue）’阶段，你是一个极具潜力的‘纯粹概念股（Pure Play）’。关键不在于你赚了多少钱，而在于你值多少钱——谁最值钱？那些一直在亏钱的公司！ — David Rosenthal > [03:53](https://www.youtube.com/watch?v=QhAftC_zFr8&t=233s) tossing balls into the air like a drunken juggler — Steven Levy > [06:49](https://www.youtube.com/watch?v=QhAftC_zFr8&t=409s) morally repugnant — Bill Gates > [10:20](https://www.youtube.com/watch?v=QhAftC_zFr8&t=620s) 你有什么想法？我们这里有几个方向，你能否一个人全栈包揽一个完整产品？ — Larry Page & Wayne Rosing > [28:03](https://www.youtube.com/watch?v=QhAftC_zFr8&t=1683s) 我们必须疯狂复制这个模式，把所有能想到的东西全部搬进浏览器！ — Larry Page, Sergey Brin & Eric Schmidt > [34:16](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2056s) 这正是我想要的，但必须跑在 Web 上。在 Google，我们热爱 Web（We like the web at Google）。 — Larry Page > [43:34](https://www.youtube.com/watch?v=QhAftC_zFr8&t=2614s) 天哪，多人实时编辑才是颠覆性的杀手级特性（killer feature）！ — Sam Schillace > [50:35](https://www.youtube.com/watch?v=QhAftC_zFr8&t=3035s) 微软赚走了所有的钱，Google 赢得了所有的用户，两边皆大欢喜。 — Ben Gilbert & David Rosenthal > [01:16:31](https://www.youtube.com/watch?v=QhAftC_zFr8&t=4591s) 在 YouTube 内部，围绕分发机制曾爆发过长达数年的路线之争：第一派主张‘关注订阅为王’，认为最强的互动来自于用户自主挑选关注的博主；第二派则坚信‘算法至上’。不幸的是，最终被证明正确的正是算法派。绝大多数时候，机器推算出的下一条视频，就是比订阅列表里的内容更能让你一直看下去。 > [02:56:05](https://www.youtube.com/watch?v=QhAftC_zFr8&t=10565s) Droid Does — Motorola / Verizon 广告 > [03:22:30](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12150s) 这是下一代 Google，是 Google+1。 — Vic Gundotra > [03:30:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12600s) God damn it, we missed it. We totally missed it — Eric Schmidt > [03:32:00](https://www.youtube.com/watch?v=QhAftC_zFr8&t=12720s) Facebook 存在成为整个互联网入口的巨大风险。 — Neal Mohan > [03:52:30](https://www.youtube.com/watch?v=QhAftC_zFr8&t=13950s) 人工智能将是 Google 的终极形态。 — Larry Page > [04:01:40](https://www.youtube.com/watch?v=QhAftC_zFr8&t=14500s) 支撑这一切运转的核心技术洞察到底是什么？ — Eric Schmidt ## Mentioned Google, Alphabet, Larry Page, Sergey Brin, Eric Schmidt, Sundar Pichai, Paul Buchheit, Bret Taylor, Bill Gates, Steve Jobs, Steve Ballmer, Bill Gurley, Mark Zuckerberg, Sheryl Sandberg, Neal Mohan, Urs Hölzle, Vic Gundotra, Sam Schillace, Jonathan Rochelle, Chad Hurley, Steve Chen, Jawed Karim, Geoffrey Hinton, Ilya Sutskever, Dario Amodei, Demis Hassabis, Noam Shazeer, Ian Goodfellow, Alex Krizhevsky, Gmail, Google Maps, Google Docs, Google Sheets, Google Drive, YouTube, Google Chrome, Android, Google Photos, Google+, Google Wave, Google Buzz, Project Loon, Google Glass, Google Play, Google Earth, AdWords, AdSense, DoubleClick, Microsoft, Internet Explorer, Windows Vista, Microsoft Office, Motorola Droid, Verizon, Apple, Meta, Transformer, Borg, Steam Deck, Sony RX100 VII, Claude, Carissimi ## Video description In its first six years from 1998 to 2004, Google built one of the greatest products of all time (and certainly the greatest business of all time) with Search. Then in its next six years from 2005 to 2011, Google built seven (!) more billion+ user products: Gmail, Maps, Drive and Docs, YouTube, Chrome, Android, and Photos — all either started from scratch internally or acquired as startups that were still in their infancy. This six-year period of wild innovation STILL stands unmatched in technology history… no other tech company counts more than four billion+ user products in its portfolio total. And of course, this “Google 2.0” era culminated in the transformation of the very company itself into Alphabet. So the question we answer today is… how did they do it?? And why? What was the strategy that led a once “pure play” search company into such far flung fields as email, mapping, funny cat videos and operating systems? We unpack the brilliant (and sometimes accidental) strategies behind each product, the simultaneous three-front war Google fought against Microsoft, Apple, and Facebook, and the spectacular failure of Google Plus that nearly destroyed the company's culture — before ultimately setting the stage for both Alphabet and the AI revolution to come. *Links:* - Jeff Dean and Sanjay Ghemawat New Yorker article: https://www.newyorker.com/magazine/2018/12/10/the-friendship-that-made-google-huge - Eric Schmidt on stage at the iPhone keynote: https://youtu.be/OxUDiS3AR0M?si=bMtVK57n1bFlBj9D - Bill Gurley’s classic “Less than Free” Android post: https://abovethecrowd.com/2009/10/29/google-redefines-disruption-the-less-than-free-business-model/ - Our recent ACQ2 episode with Bret Taylor and Clay Bavor: https://www.acquired.fm/episodes/how-is-ai-different-than-other-technology-waves-with-bret-taylor-and-clay-bavor - Worldly Partners’ Multi-Decade Alphabet Study: https://worldlypartners.com/businesshistory - Episode sources: https://www.acquired.fm/episodes/alphabet-inc#sources *Carve Outs:* - Bluey x Camp in NYC: https://camp.com/bluey-x-camp-nyc - Steam Deck vs Switch 2 (Part 2) - https://store.steampowered.com/steamdeck/ - https://www.nintendo.com/us/gaming-systems/switch-2 - Claude: https://bit.ly/claudecarveout - Sony RX100 VII: https://www.amazon.com/Sony-Premium-Compact-1-0-type-DSCRX100M7/dp/B07VPQV7BY?th=1 - Carissimi clothing: https://bit.ly/acqcarissimi *More Acquired:* - Get email updates https://www.acquired.fm/email and vote on future episodes! - Join the Slack http://acquired.fm/slack - Check out the latest swag in the ACQ Merch Store https://www.acquired.fm/store! Chapters: 0:00 Introduction 3:50 Gmail: Revolutionary Web-Based AJAX Email 16:34 The Web: Google's Strategic Weapon Against Microsoft 32:10 Google Maps and Docs: Expanding the Web Platform 51:52 YouTube: The Early Days of User-Generated Video 1:09:40 YouTube: From Risky Acquisition to Giant Business 1:30:13 DoubleClick: Expanding the Ad Business 1:49:29 Chrome: Google's Entry into the Browser Wars 2:08:57 Chrome: Google's Browser Achieves Market Dominance 2:23:12 Android: Origins of the Mobile Operating System 2:39:43 Android: Google's Response to the iPhone Challenge 3:01:48 Android: Becoming a Global Powerhouse 3:13:23 Google Plus: The Failed Social Media Push 3:26:12 Google Plus: Its Fallout and Enduring Lessons 3:35:25 Alphabet: Laying the Foundation for the AI Era 3:44:11 Business Analysis of Google, 2004-2015 3:50:01 Google's Playbook and Personal Carve Outs _Note: Acquired hosts and guests may hold assets discussed in this episode. This podcast is not investment advice, and is intended for informational and entertainment purposes only. You should do your own research and make your own independent decisions when considering any financial transactions._ [Source](https://www.youtube.com/watch?v=QhAftC_zFr8)

Source quotes

[3:53] David Rosenthal: 如果你公布了营收，大家就会问有多少，而且永远都不会觉得够。原本能涨 100 倍、1000 倍的明星公司，突然之间就变成了只涨 2 倍的平庸之辈。但如果你还没有营收，你就可以说自己是‘预营收（Pre-revenue）’阶段，你是一个极具潜力的‘纯粹概念股（Pure Play）’。关键不在于你赚了多少钱，而在于你值多少钱——谁最值钱？那些一直在亏钱的公司！ [3:53] Steven Levy: tossing balls into the air like a drunken juggler [6:49] Bill Gates: morally repugnant [10:20] Larry Page & Wayne Rosing: 你有什么想法？我们这里有几个方向，你能否一个人全栈包揽一个完整产品？ [28:03] Larry Page, Sergey Brin & Eric Schmidt: 我们必须疯狂复制这个模式，把所有能想到的东西全部搬进浏览器！ [34:16] Larry Page: 这正是我想要的，但必须跑在 Web 上。在 Google，我们热爱 Web（We like the web at Google）。 [43:34] Sam Schillace: 天哪，多人实时编辑才是颠覆性的杀手级特性（killer feature）！ [50:35] Ben Gilbert & David Rosenthal: 微软赚走了所有的钱，Google 赢得了所有的用户，两边皆大欢喜。 [76:31] 在 YouTube 内部，围绕分发机制曾爆发过长达数年的路线之争：第一派主张‘关注订阅为王’，认为最强的互动来自于用户自主挑选关注的博主；第二派则坚信‘算法至上’。不幸的是，最终被证明正确的正是算法派。绝大多数时候，机器推算出的下一条视频，就是比订阅列表里的内容更能让你一直看下去。 [176:05] Motorola / Verizon 广告: Droid Does [202:30] Vic Gundotra: 这是下一代 Google，是 Google+1。 [210:00] Eric Schmidt: God damn it, we missed it. We totally missed it [212:00] Neal Mohan: Facebook 存在成为整个互联网入口的巨大风险。 [232:30] Larry Page: 人工智能将是 Google 的终极形态。 [241:40] Eric Schmidt: 支撑这一切运转的核心技术洞察到底是什么？

Original transcript

[0:00] Are you intentionally wearing a black [0:02] turtleneck for this one? [0:03] >> No. It is actually going to be one of my [0:05] carveouts, though. Yeah. [0:06] >> Amazing. [0:07] >> What? [0:07] >> You think I dress up like Steve Jobs for [0:10] a Google episode? [0:11] >> Well, I thought because of the, you [0:13] know, war between Android and [0:15] >> I I walk in and there's this like smirk [0:16] on your face. [0:18] [Applause] [0:19] >> All right, let's do it. [0:21] >> Who got the truth? [0:24] Is it you? Is it you? Is it you? Who got [0:26] the truth? Now, [0:29] is it you? Is it you? Is it you? Sit me [0:31] down. Say it straight. Another story on [0:35] the way. [0:37] >> Welcome to the summer 2025 season of [0:40] Acquired, the podcast about great [0:42] companies and the stories and playbooks [0:43] behind them. I'm Ben Gilbert. [0:45] >> I'm David Rosenthal. [0:46] >> And we are your hosts. In the late [0:48] 1990s, Google built the best search [0:51] engine for the rapidly growing internet. [0:53] With a breakthrough search algorithm, [0:55] lowcost servers based on commodity [0:57] hardware, and the best business model of [0:59] all time, search ads. They turned that [1:01] search engine into a cash gushing [1:03] business and took it public in 2004. But [1:07] then, curiously, they started doing some [1:09] things that weren't related to search. [1:11] They launched a breakthrough email [1:12] service in your browser with Gmail. maps [1:15] that were far superior to the current [1:16] state-of-the-art docs and spreadsheets [1:19] with real-time collaboration for the [1:20] first time. Of course, YouTube, then [1:23] Android, and their own web browser with [1:25] Chrome. Astonishingly, today Google has [1:29] 15 products with over half a billion [1:31] users. Seven of those have over two [1:34] billion users. David, that is over 25% [1:38] of humans use seven of Google's [1:40] products. [1:41] >> Just unreal. Can't wait to tell all of [1:43] these stories today. [1:44] >> Yes. And they've also launched some [1:48] colossal failures. Plus to try to [1:50] compete with Facebook, Google Wave, [1:53] Buzz, and about half a dozen messaging [1:55] apps. I don't know, maybe a dozen [1:56] messaging apps over the years. Hot air [1:58] balloons to provide wireless internet. [2:00] And of course, [2:01] >> oh man, I forgot about the hot air [2:02] balloons. [2:03] >> Google Glass. [2:06] >> Can't forget about that one, [2:07] unfortunately. [2:08] >> So, why did they do all this? And as a [2:10] business, Google was and still is the [2:13] company that makes the vast majority of [2:14] their money from ads on search results [2:17] on the web. So today, we tell the story [2:19] of Google as the innovation factory of [2:22] the 2000s, their reorganization into the [2:25] parent company Alphabet, and how all [2:27] these different products cleverly serve [2:29] different business purposes, and also [2:31] how it feeds into Google's original core [2:34] mission to organize the world's [2:36] information. And we'll end this episode [2:38] story right at the dawn of the AI era. [2:42] >> Oh, you're giving away the end. [2:43] >> Oh, spoilers. Sorry. So, is Google a [2:46] search engine? Is it the platform [2:49] company of the web era? Or is it an [2:51] incubator that just happens to have [2:52] struck gold with search and perhaps AI? [2:56] Today, we dive in. [2:58] >> Listeners, if you want to know every [2:59] time an episode drops or get early hints [3:02] at what the next episode will be, check [3:04] out our email list. That's also where we [3:06] share corrections and updates about [3:08] previous episodes and we are adding a [3:10] new bonus. You get to help us vote on [3:13] future episode topics. So, the first [3:15] poll is going out soon. Sign up now at [3:18] acquired.fm/e. [3:21] Join the Slack if you want to come talk [3:22] about this with us and the whole [3:23] Acquired community. Acquired.fm/slack. [3:27] Before we dive in, we want to briefly [3:28] thank our presenting partner JP Morgan [3:30] Payments. Yes, just like how we say [3:32] every company has a story, every [3:34] company's story is powered by payments [3:35] and JP Morgan Payments is a part of so [3:37] many of their journeys from seed to IPO [3:39] and beyond. So with that, this show is [3:41] not investment advice. David and I may [3:43] have investments in the companies we [3:44] discuss and this show is forformational [3:46] and entertainment purposes only. David, [3:50] where are we starting this alphabet [3:52] story? [3:53] >> Oh, I have a very, very fun beginning [3:56] for you, Ben. I want to start with a [3:59] quote from Russ Hanaman, [4:02] >> the fictional character [4:03] >> Silicon Valley HBO show. Oh yeah. From [4:08] the TV show. [4:09] >> Awesome. [4:09] >> And the quote is if you show revenue, [4:12] people will ask how much and it will [4:15] never be enough. The company that was [4:17] the 100xer, the thousandxer is suddenly [4:19] the 2x dog. But if you have no revenue, [4:22] you can say you're pre-revenue. You're a [4:24] potential pure play. It's not about how [4:26] much you earn. It's about what you're [4:28] worth and who's worth. The most [4:30] companies that lose money. Immortal [4:33] words of wisdom for the technology [4:35] world. God, that show was so good. Why [4:37] do I bring this up? Why do I start here? [4:39] >> Why are you talking about this? Google [4:40] is a cash gushing machine. Revenue is [4:44] obviously not the problem for Google. [4:47] But what was the problem in 2004, 2005, [4:51] 2006 was being viewed as, in Russ's [4:55] terms, pure play. When Google went [4:58] public in fall of 2004, the stock shot [5:00] up, basically doubled in two months. [5:03] Wall Street loved Google. Adwords, the [5:06] search business model, everybody had to [5:09] own shares. Google had cracked the code [5:12] on monetizing the internet. The more [5:13] people use the internet, the more they [5:15] search. The more they search, the more [5:16] money Google makes. Simple, easy, pure [5:20] play, you might say. [5:22] >> Yep. [5:23] >> That is until Google announced fourth [5:25] quarter 2005 earnings. Full year 2005 [5:30] revenue, $6.1 billion. That's almost [5:33] double the 3.1 that it was in 2004, the [5:36] first year it went public. But earnings [5:40] are flat. Profitability is down. [5:44] >> Google's now investing in all these new [5:45] products and services. Gmail, maps, the [5:48] forthcoming Google Docs. Later this year [5:50] in 2006 would buy YouTube for $1.6 [5:53] billion. Wall Street hates this. Hates [5:57] it. [5:57] >> This is a huge amount of their cash [5:59] they're putting back on the table and [6:01] betting for the future. [6:02] >> So this is January 2006. The stock falls [6:05] 27%. [6:07] Wall Street's like, "God, these guys, [6:09] what are they doing? They're messing it [6:10] up. Steven Levy writes in in the Plex [6:13] that the perception of Google's ventures [6:15] beyond search at the time was that the [6:18] company was tossing balls into the air [6:20] like a drunken juggler. They were a pure [6:23] play in investor's eyes and now they're [6:25] messing it up. They're adding all this [6:26] other stuff. They don't want the other [6:27] stuff. [6:28] >> Yeah. So then Ben, as you teed up in the [6:29] intro, the question is why did they do [6:31] all this? And I think the way to answer [6:33] it is to start and just tell the stories [6:35] of all the individual products. [6:37] >> Let's do it. Strap in. I will say, [6:39] David, doing the research took me way [6:41] back to early acquired grading [6:43] acquisitions. This is the cornucopia of [6:46] hits of iconic product launches in tech [6:49] history. [6:49] >> So, the first and probably the most [6:51] important here because it sets the stage [6:53] for everything else. The first major [6:55] non-arch product was on April 1st, April [6:58] Fool's Day, 2004, Gmail. The most [7:02] famous, infamous, non joke April Fool's [7:06] Day announcement of all time. [7:07] >> Yes. But it sure sounded like a joke. [7:10] Here's the announcement. In 2004, [7:13] entirely webbased email in your browser. [7:17] You can log in and access it anywhere on [7:20] any device. Google search is built in. [7:23] You don't need to spend all this time [7:25] sorting your mail into folders anymore. [7:27] And one gigabyte of storage free. No [7:32] need to delete your mail. No need to [7:34] clean up your inbox. No need to do [7:35] anything ever. And the whole thing is [7:40] free. [7:41] >> Yep. [7:42] >> Of course, this sounds like a joke. This [7:44] is too good to be true. [7:45] >> The universe at the time is Microsoft [7:47] sells sort of enterprisegrade mail for a [7:50] lot of money or there's all these free [7:52] web-based services popping up like [7:54] Hotmail that Microsoft would end up [7:55] buying and Yahoo mail and AOL. You get [7:58] like 5 megabytes of storage. [8:01] >> Yeah. Not even. At the time, Hotmail, [8:03] which as you said, Microsoft owns, had 2 [8:05] megabytes of free storage and Yahoo Mail [8:09] had 4 megabytes. There's another great [8:11] story from In the Plex that Steven Levy [8:13] has. He's interviewing Bill Gates at the [8:16] Newsweek headquarters office in New York [8:18] shortly after Gmail comes out and they [8:20] started talking about Gmail and Bill [8:22] can't believe it. He's like offended by [8:25] Gmail because he thinks that giving [8:28] people all this storage is just [8:30] wasteful. You're doing email wrong. It's [8:33] morally repugnant to leave all of this [8:36] email sitting right on the servers. I [8:39] was thinking about it. Until Gmail, the [8:42] paradigm for email, people treated it [8:44] like regular physical mail. Sort it. You [8:47] file away the important stuff. You throw [8:49] out the pieces you don't need anymore. I [8:50] mean, even freaking Bill Gates operates [8:52] this way. [8:53] >> Yes. [8:54] >> So, Gmail, this is radical. This is a [8:57] radical notion of how email should work. [9:00] and was also correct. I mean, if you sat [9:04] and you thought about it in say 2001 or [9:07] so when Gmail starts getting worked on [9:08] within Google and you thought about the [9:10] combination of the growth of the [9:11] internet, which obviously Google has a [9:13] front row seat to, and Moore's law, you [9:17] would logically come to this conclusion [9:19] that the cost of sending and storing and [9:21] searching email would asmatically go to [9:23] zero. And thus, as that happened, a [9:27] whole lot more email was going to be [9:29] sent in the world. [9:31] >> Yep. So, can I tell you my understanding [9:34] of where this story starts in 1996? [9:37] >> Oh, I was going to go back to 99, but [9:38] yeah, go for it. [9:39] >> All right. So, I know you're about to [9:41] bring up the name Paul Bukite. Is that [9:43] right? [9:43] >> Of course. Yeah. [9:45] >> So, Paul was kind enough to speak with [9:47] me before recording this episode. Paul [9:49] famously the inventor of Gmail. In 1996, [9:53] Paul was a student at Case Western [9:55] Reserve University in Cleveland, which [9:58] you may also know this, David, famously [10:00] was one of Ohio strong. [10:01] >> Yes. The first campuses in the nation to [10:04] have broadband internet in the dorms and [10:06] all over campus. [10:07] >> Oh, okay. I knew about the Paul [10:10] fascination with web mail starting in [10:11] college. But I didn't realize that Case [10:13] Western had broadband. So this is why [10:16] when you're living in the universe of [10:17] broadband everywhere, he was living like [10:20] 15 years in the future temporarily for 4 [10:23] years in college. [10:24] >> Yeah. 1996. [10:26] >> Yes. So he realizes email is kind of a [10:30] bummer if it's a thing that you download [10:32] and lives on your computer. The [10:34] information should just exist at my [10:37] fingertips all the time. Bits are [10:39] becoming free to move around. So he kind [10:43] of gets obsessed with this idea in [10:45] college that email should exist on the [10:48] web in a browser without ever having to [10:51] download it. And he builds a prototype [10:53] for web mail when he's in college. [10:56] >> Wow. [10:57] >> In 2001, famously preipo at Google, [11:01] Larry Page feels like Google is moving a [11:04] little bit too slow and gets rid of all [11:06] engineering managers. So Larry and Wayne [11:08] Rosing, who is leading engineering, go [11:10] and meet with each engineer individually [11:12] to talk about ideas that they could work [11:14] on. This tells you so much about [11:16] googliness, but it also tells you a lot [11:18] about the caliber of the engineers they [11:21] were hiring at the time where they would [11:22] just approach them and say, "What ideas [11:24] are you thinking about? Here are some [11:25] ideas we have. Can you just full stack [11:27] own this product entirely yourself?" And [11:30] so in Paul's meeting, they knew about [11:32] his previous interest in email and [11:34] web-based mail and they sort of float [11:35] this amorphous idea to him and that's [11:39] where it comes from. [11:40] >> Ah, so Larry and Wayne suggested it to [11:43] him. Interesting. [11:44] >> Okay. So here's some other stuff that [11:46] Paul said. So part of the motivation was [11:48] that they were looking to make something [11:50] that would make Google stickier. So [11:52] you'd have sort of this ongoing [11:53] relationship for if there was a next [11:56] Google after Google, there was some [11:57] reason why you would still have a [11:59] relationship [12:00] >> which obviously you know Yahoo would [12:02] have for many many years even though [12:04] there was a next Yahoo after Yahoo in [12:06] Google. [12:06] >> We still get emails from people with [12:08] Yahoo mails. [12:09] >> Do you know how Paul found out about [12:10] Google in 1999? [12:12] >> Oh no. [12:13] >> Slash dot. [12:14] >> Really? That's awesome. [12:16] >> And that he sends an email to [12:17] jobs@google.com. [12:20] >> Unbelievable. fitting that you know he [12:21] gets hired with an email. [12:22] >> Heyo. [12:23] >> Okay, so 2001 Paul gets to work with [12:27] encouragement from Larry and Wayne. [12:29] >> Do you know what the original seed of [12:30] the code is? [12:31] >> Oh no, go for it. [12:33] >> Google had just bought a company called [12:34] Deja News, their first acquisition. It [12:37] was the corpus of all the old Usenet [12:40] posts. [12:41] >> Oh yeah, then this becomes Google [12:42] groups, right? [12:43] >> That's exactly right. [12:44] >> Yeah. And Paul's working on that. And [12:45] part of that was a feature to do real [12:47] time indexing of all the posts that [12:50] would allow you to search the whole [12:52] corpus. So Paul just applies that to his [12:55] own personal inbox. The first [12:58] instantiation of this is just a search [13:00] box to search his personal Unix mail [13:04] directory as if it is the old Usenet [13:07] posts that they had just bought. That's [13:09] the first version of Gmail. [13:11] >> Amazing. As he's building on that [13:14] though, obviously like the first thing [13:15] he needs is a web front, an interface. [13:18] Okay, Hotmail's out there, Yahoo Mail's [13:20] out there, webmail's out there. It [13:22] sucks. It sucks for a lot of reasons. [13:25] There's got to be a way to make it [13:26] better, make it more performant and [13:28] better to use as a web page. And so he's [13:30] playing around with JavaScript and what [13:32] he can do with JavaScript to make this [13:34] application, this web application of [13:36] email better. The history of JavaScript [13:38] is fascinating. Brendan Ike created it [13:41] at Netscape back in 1995. We did a whole [13:43] episode with Brendan years ago about [13:45] this. The idea behind JavaScript was to [13:48] include a programming language as part [13:50] of web browsers so that people could [13:51] make dynamic web pages instead of just [13:54] static HTML documents. The problem was [13:58] it was kind of this casualty of the [13:59] browser wars with Microsoft and Internet [14:01] Explorer and everything that killed [14:03] Netscape. So up until this time, kind of [14:06] 2001, JavaScript existed, but like it [14:09] wasn't super popular. [14:10] >> It wasn't very powerful. You could do [14:12] weird stuff like animate something on [14:14] the page, but I would describe it as [14:17] toy- like and not a real programming [14:18] language, for sure. [14:20] >> Yep. And for what the web was up until [14:22] that point in time, you didn't really [14:23] need it. Static web pages are kind of [14:25] fine for most of what's happen. I mean, [14:26] even google.com was static. You type a [14:28] search into the search box, Google [14:30] servers process the query and they send [14:32] you a whole new static web page with the [14:34] results. But you'd imagine for doing [14:36] something like email on the web or any [14:38] application on the web, you don't want [14:39] the site to reload every time you open a [14:41] new email or you create a draft or you [14:43] move something around in folders. [14:45] >> You might want to move from a website to [14:47] a world of web applications. [14:49] >> Yeah. But this is how Hotmail and Yahoo [14:51] Mail worked. Every time you took an [14:53] action, I mean, it reloaded the page and [14:55] so they were super slow. [14:56] >> Yes. And so Paul's like, "Maybe I can [14:58] use JavaScript to make this better." [14:59] He's working on it and he discovers a [15:02] little known feature of JavaScript [15:04] called the XML HTTP request, which lets [15:09] a web page fetch automatically new XML [15:15] data from a server without reloading the [15:19] page. And Paul's like, "Oh my god, this [15:23] is gold." And this is the birth of Ajax, [15:28] asynchronous JavaScript and XML. [15:31] >> So, David, I assumed you were going to [15:33] go here. I thought that you get it all [15:35] laid up. [15:35] >> You've been letting me go. You just been [15:37] feeding me a rope the whole time. [15:38] >> You trying to tell me that Gmail is the [15:40] first Ajax application. [15:42] >> Well, the first widely adopted around [15:44] the world. [15:45] >> That's fair to say. [15:46] >> That sort of set the bar for what [15:48] dynamic web 2.0, you might say, websites [15:50] could be. [15:51] >> Yes. [15:52] The origin of the XML HTTP request is a [15:57] part of Internet Explorer first [15:58] implemented by Microsoft and used in [16:01] this part of Outlook called Outlook web [16:04] access. [16:05] >> I think I did know this [16:07] >> when I worked for my high school. I [16:08] could log in on any computer into my [16:12] Outlook through their web access and [16:14] that thing used Ajax and I think it only [16:16] worked in Internet Explorer. So that is [16:18] the origin of why this API exists in the [16:22] first place. Ironically for another mail [16:24] client. [16:24] >> Oh, not just for another mail client. [16:26] It's so deeply ironic that this [16:28] originated for a Microsoft mail client. [16:31] >> Yes. [16:31] >> We're going to get deep into that in [16:33] just a minute here. [16:34] >> Yes. So I mean when Paul discovers this, [16:37] this is almost like Google search all [16:40] over again when people realize what you [16:43] can do to create something that looks [16:46] and feels and has all the functionality [16:48] of a application that hereto would have [16:51] been a program that you installed on [16:53] your personal computer [16:56] >> or a app on your Mac [16:58] >> that maybe you downloaded from the [17:00] internet but more likely you went to a [17:02] retail shop like Comp USA or [17:04] installed on your computer. You can now [17:06] just do this in a web browser. This is [17:09] incredible. [17:09] >> The web is the platform of the future. [17:11] >> Yep. So Paul builds the prototype, shows [17:14] it to Larry and Sergey. They're super [17:16] jazzed. So supposedly Larry and Sergey [17:19] become the first beta users of Gmail. [17:21] They are the seed Gmail users and they [17:26] start using it exclusively as their mail [17:28] service within Google. And then by the [17:30] time it launches publicly, all of Google [17:32] is on Gmail and using it, addicted to [17:35] it. And it wasn't called this at the [17:38] time, but it's in the cloud. You don't [17:40] have to have your mail stored on your [17:42] machine or a specific server. You can [17:43] log in, access it anywhere on any [17:46] network, any device. [17:47] >> All this stuff sounds so boring, but it [17:49] was completely breakthrough. [17:51] >> So obviously Larry and Sergey are jazzed [17:53] first because of just the incredible [17:55] nature of this product. And Larry [17:57] especially, he is a product person and [18:00] his view is if we can build a better [18:02] product and it's on the web, then it's [18:03] good for Google and we should do it. And [18:06] that is a huge part of the motivation [18:07] underlying Gmail and everything we're [18:09] going to talk about. But there's also [18:11] another reason and that's Microsoft. [18:15] Because Google was doing great printing [18:17] money, Adwords, search, greatest [18:19] product, greatest business of all time. [18:22] But they've got a big risk, which is [18:25] that everything about Google, everything [18:28] about the web right now flows through [18:30] Microsoft, flows through Internet [18:33] Explorer. [18:34] >> Yeah. Google's entire money printing [18:36] machine was built on top of Microsoft's [18:38] and at two layers. So to this point over [18:41] 90% of Google search queries were done [18:45] on Windows PCs and 90% were done in [18:50] Internet Explorer running on those PCs. [18:52] So Google's got the killer app for the [18:54] web in search and the thing under them [18:56] is a browser owned by Microsoft and the [18:58] thing under that is an operating system [19:00] owned by Microsoft. [19:01] >> Yes, [19:02] >> they exist at the pleasure of Microsoft [19:04] at this point in history [19:06] >> and Microsoft has a different business [19:08] model. So Google's business model, the [19:11] greatest of all time, is people use [19:13] Google search. They discover more of the [19:15] web. They spend more time online on [19:17] these new sites and services that [19:18] they're discovering. As they're spending [19:20] more time online, they search more. [19:23] Searching more leads them to discover [19:25] even more new sites and services, the [19:26] cycle repeats itself and Google just [19:28] monetizes the whole thing. [19:29] >> Yes. And web usage isn't bad for [19:32] Microsoft, but if the platform of the [19:35] next generation becomes the web and [19:37] people are writing web applications [19:39] instead of Windows applications, that [19:42] makes Microsoft's platform a lot less [19:44] valuable versus other operating systems [19:47] like say Mac or say a future where we [19:50] change away from desktop computers [19:52] altogether. Yes, at a minimum, Microsoft [19:55] doesn't business modelwise care about [19:58] the web because they don't monetize the [19:59] web. Microsoft makes money by OEM [20:02] selling PCs that have Windows on them [20:04] and then Microsoft sells software that [20:07] goes on those PCs. So, at a minimum, [20:10] they don't care. And at a maximum, like [20:12] you're saying, web apps are at an [20:14] existential risk to Microsoft. Oh my [20:17] god, there's a future application [20:18] platform that just doesn't really [20:20] require our participation other than the [20:23] fact that we control IE. And at least [20:25] for now, that's really important. [20:27] >> And most of Microsoft hasn't realized [20:29] this yet. Thank God for Google. [20:31] Microsoft's distracted with the [20:33] albatross that was Longhorn that would [20:35] become Windows Vista. [20:37] >> Yes, [20:37] >> a few people in Microsoft realize this, [20:39] but Google for sure realizes this [20:42] though. Eric Schmidt for Doubleshore [20:44] realizes it because he was the CEO of [20:46] Noville before coming to Google. And who [20:48] is Noville's competitor? Microsoft. And [20:51] Microsoft crushed them. So why Google's [20:54] so jazzed about Gmail? They need to [20:56] build up leverage with consumers, with [20:59] users that they're going to demand rich [21:03] web applications so that if Microsoft [21:05] ever tries to disadvantage Google or [21:06] disadvantage web apps and things moving [21:08] to the web, really the only defense [21:10] against that is if consumers have [21:12] already adopted this stuff and love it [21:14] and would revolt. And so this is what [21:16] Gmail is. Yes. So Gmail developments [21:19] trucking along through 2001, 2002, 2003. [21:23] This is hard to remember now. took three [21:25] years to develop Gmail. [21:27] >> Long development cycle. Yeah. [21:29] >> To be ready to release publicly and then [21:30] it was in beta for like 10 years. [21:32] >> Yeah. [21:34] >> I think the reason it took so long was [21:36] this was all new. There wasn't a lot of [21:39] depth of knowledge out there about [21:40] JavaScript. Certainly not about Ajax and [21:42] XML dynamic refreshing. [21:44] >> It was really hard to program. Today [21:46] you've got all these nice abstraction [21:49] layers, these frameworks that people [21:50] have built to do web development that [21:52] really didn't exist to make Ajax [21:54] applications. [21:55] >> Yes. Okay. So, Google's finally getting [21:57] ready to launch it. We're in 2004. [22:00] There's a couple questions. One, the [22:03] service for all the reasons we just [22:05] described. Google, Larry, Sergey, Eric, [22:09] they want it to be so compelling that [22:11] consumers demand it. It takes off like [22:13] wildfire. It builds this strategic mode [22:15] against Microsoft, but it will cost [22:18] money. [22:18] >> There's a reason other people don't do [22:19] this. [22:20] >> Yeah, there's a reason that a gigabyte [22:21] of free storage seems a little crazy. [22:24] Even if you assume, and I think this is [22:26] probably directionally correct, that [22:28] because of Google's commodity [22:30] infrastructure advantage, they could [22:33] launch Gmail at like onetenth of the [22:37] cost that anybody else could. Also [22:39] remember there's no public cloud at this [22:41] point in time. [22:41] >> So you'd have to go build your own data [22:43] center to do this. [22:44] >> You can't just launch on AWS. There is [22:45] no AWS. But even assume that Google has [22:48] a 90% cost advantage on the [22:51] infrastructure side. The state-ofthe-art [22:53] is other competitors are offering 4 [22:55] megabytes of free storage. Google's [22:57] going to offer a gig. Sure, knock that [22:59] down by 90% but the effective cost is [23:01] still 100 megabytes. So how do you get [23:04] around being flooded with cost and [23:06] infrastructure demand when you launch [23:08] it? They come up with the invite system. [23:12] >> Yes. [23:12] >> And this is so brilliant. I actually [23:14] don't know if it was designed as this [23:16] sort of prestigious growth strategy [23:19] thing that it became. [23:20] >> Anyone got any Gmail invites? Please, [23:22] I'll do anything. [23:23] >> Yeah. Yeah. Please, please, please. Or [23:25] if it was truly because of the [23:27] infrastructure cost. Either way, it's [23:29] just brilliant. When they launched it on [23:31] April 1st, 2004, they send out 1,000 [23:35] seed invites to Gmail. It's a private [23:38] inviteonly [23:40] internet service. They send them out to [23:42] influencers. You know, the term didn't [23:44] exist back in the day, but influential [23:46] people. Yes. And journalists. And then [23:48] each user has a set number of invites [23:51] that they can give to other users to [23:54] invite their friends. [23:55] >> And it was low. It was like five or [23:57] something. And then it wasn't clear when [23:59] they would top back up. But you'd give [24:01] out your five. And then at some point [24:02] you'd come in and you'd have five more. [24:04] You'd have three more. It was like super [24:06] dynamic and very clearly whatever Google [24:08] felt like they could give away from [24:10] their servers at the moment. [24:11] >> Yep. But it was so brilliant. It made it [24:13] feel like you're in this special world [24:15] of people in the know that with super [24:17] incentivized [24:19] viral word of mouth growth because I'm [24:22] telling you it's a gigabyte of free [24:24] storage. It's this incredible service. [24:27] They're selling on eBay for 150 bucks. [24:29] There was a monetary value to these [24:30] things. Yes. Yes. They were trading on [24:32] eBay for average price of 150 bucks in [24:35] the early days. And so I'm giving you [24:37] this gift. [24:38] >> Incredible. [24:39] >> And look, everybody wants this, but you [24:41] need to have the product quality that [24:44] cashes the check. [24:45] >> Yes. It needs to be a real gift, [24:47] >> right? And it was it was just better. It [24:50] wasn't just something I'd sign up for [24:52] and then churn and be like, "Cool, I [24:54] locked in my username or whatever." It [24:56] was something that you actually used [24:58] every day. or in the words of Larry [25:00] Page, passed the toothbrush test. It was [25:02] a part of your daily habit, something [25:04] you do once or twice a day. [25:07] >> Uh, I wish I could only refresh Gmail [25:10] once or twice a day. [25:11] >> So, David, was this the first software [25:14] that used a weight list like this? [25:15] Because obviously, it's become very [25:16] popular since, [25:17] >> I think. So, so that's how they take [25:20] care of the cost side of the equation is [25:22] not running out of control is the invite [25:24] strategy. [25:24] >> Well, still not making any money though. [25:26] >> That's question number two. How are we [25:28] going to make money from this thing? [25:30] Because yeah, okay, there's all these [25:31] strategic reasons to do it. It'll [25:33] increase more traffic on the web, time [25:35] spent, people search more, we'll make [25:37] more money indirectly, but they still [25:38] don't really know that. So, they think, [25:40] okay, we need a monetization strategy [25:42] baked into the product itself. [25:44] >> Yes. [25:45] >> Well, how do you make money from [25:46] anything at Google? [25:47] >> This actually came up during [25:48] development. So even in the prototyping [25:50] phase, Paul logs into the database of [25:52] ads, which it's just funny that at that [25:55] point in time, Google's got this big [25:57] database of ads, [25:58] >> right? Yeah. I'm just going to access [26:00] the ads database, you know, all of them. [26:02] >> Yes. And these are the ads that would [26:04] run when you searched and landed on a [26:06] search results page. And so he decided [26:09] to do content matching against your [26:11] inbox and just show those ads on the [26:14] page next to your email. And even though [26:16] they weren't meant for that, it actually [26:17] turned out that these search ads were [26:19] pretty relevant. It actually was a [26:20] decent ad to be showing you while you're [26:22] looking at your inbox about similar [26:24] topics. So he just rolls this out. Even [26:27] though all these people in Google are [26:29] actually using it as their mail client [26:31] at the time, people were pissed. I mean, [26:33] people were like, "Are you looking at my [26:34] emails?" You know, all the things that [26:37] would then sort of come later in the [26:38] public actually happened inside Google [26:40] first. But Larry and Sergey loved it. [26:43] They were like, "Oh, this is so [26:44] obviously the answer." Interestingly, [26:47] this experiment predates AdSense. So, [26:50] Google has the display ad offering for [26:52] website publishers that's called [26:54] AdSense. That's different than AdWords, [26:56] which is the keyword advertisements on a [26:59] search results page. AdSense hasn't [27:01] launched yet. And there's sort of [27:03] multiple versions of history here. How [27:05] much credit for AdSense does Gmail get [27:08] in discovering this? But it is safe to [27:09] say that the idea of display ads that [27:13] are content matched against your Gmail [27:16] did contribute to the idea for the first [27:18] version of AdSense, which are [27:19] essentially the same thing, content [27:21] matched ads just on a publisher website [27:23] instead of the content of your inbox. So [27:27] the product launches publicly April [27:30] 2004. As you would expect, people go [27:33] nuts. It is truly a revolutionary [27:36] product. And Gmail grows over the next [27:40] 20 years from that a thousand initial [27:44] public beta user seed base to over 2 [27:47] billion today. And it's still by far the [27:50] best email service. Even if you use [27:52] another front end for your email, for [27:54] your Gmail like superhuman or whatnot [27:56] today, you still want Gmail on the back [27:58] end, at least as a consumer. [27:59] >> Yes. So once Gmail starts to take off, [28:03] Larry and Sergey and Eric see this and [28:06] they're like, "Wow, [28:09] we should do this a lot. [28:13] Let's go." Like, let's build as many web [28:18] applications as we possibly can imagine. [28:22] >> What else can go into the browser that [28:24] we didn't think was possible before? [28:25] This fires on every single cylinder for [28:29] us. Most importantly, grow the web. [28:34] >> Grow usage. [28:34] >> You grow the web, you grow time that [28:36] people spend in web browsers. They will [28:38] search more. We will make more money. [28:41] And beyond that, with some of these [28:42] products like Gmail, we can monetize the [28:44] products themselves. Great. Two, we are [28:48] building our strategic moat against [28:51] Microsoft. The faster that we get the [28:56] internet using public to fall in love [28:59] with and use web applications, the less [29:02] and less leverage Microsoft has over us. [29:06] >> To use sort of Ben Thompson speak, [29:08] Google realizes the web can become the [29:10] point of integration. Maybe the OS isn't [29:13] what the whole universe has to target. [29:15] The hardware makers, the OEMs, the [29:17] application makers, the users. If [29:20] applications start living in the [29:21] browser, then the web can become the [29:23] point of integration. Users just need a [29:26] browser and OEMs just need an operating [29:29] system that can access the browser. [29:32] >> And what's so great for Google because [29:34] of their business model? Sure, it's [29:36] great when they build and own and [29:38] operate and run and monetize web [29:40] applications themselves, like they do [29:42] with Gmail, like they'll do with maps, [29:44] like they'll do with Docs, like they'll [29:45] do with YouTube that we're about to talk [29:46] about. But if they don't, doesn't matter [29:49] as long as anybody does it, [29:50] >> right? They just need to be wind at the [29:52] back of web adoption. [29:54] >> Yes. So that leads to a whole flood of [29:57] Google web products and services to [30:00] come. But before we tell that story, [30:02] >> yes, now is the perfect time to talk [30:04] about our presenting partner, JP Morgan [30:06] Payments. They're investing billions [30:08] every year into technology and product [30:11] development. In fact, Jamie Diamond even [30:13] referenced it on stage in our interview [30:15] last month. This investment has led to [30:17] them becoming a powerful engine for [30:19] marketplaces, fintech companies, and [30:21] platforms to deliver growth, stability, [30:23] and scale for their customers. And these [30:25] days, the best companies have made [30:27] payments infrastructure invisible to [30:28] their users. If you're booking a ride, [30:31] buying something from a marketplace, or [30:33] managing a subscription, the magic [30:34] happens when payments feel seamless and [30:36] integrated. This is what JP Morgan [30:38] payments brings with their embedded [30:40] finance solutions. Rather than bolting [30:42] on payments as an afterthought, you can [30:44] embed the payments infrastructure [30:45] directly into your platform without [30:48] customers having to leave to complete a [30:49] transaction. This makes your products [30:51] more seamless. And as we've seen from [30:53] Google, the best products with the [30:54] stickiest user bases are often the [30:56] simplest for users. So embedded payments [30:58] are all about integrating payments [31:00] directly into your platform rather than [31:02] sending users to a separate payment page [31:04] or third party processor. JP Morgan has [31:06] the technology and capabilities for this [31:08] with powerful APIs. But behind this [31:10] simplicity, there's an incredible [31:12] complexity that JP Morgan takes on so [31:14] you don't have to. You can get sellers [31:15] up and running quickly, offering [31:17] flexible payment options, provide [31:19] efficient funds management, and secure [31:21] money movement at scale. Plus, they have [31:23] advanced analytics. Marketplaces and [31:25] platforms can provide payment [31:26] acceptance, funds management and payouts [31:28] in secure and compliant infrastructure, [31:30] all without users ever leaving their [31:32] ecosystem. And this reflects JP Morgan's [31:35] broader vision. They're not just a bank [31:36] anymore. their infrastructure with the [31:38] scale, resilience, and security of a [31:40] global leader for trust and safety. They [31:42] were even recognized as the best overall [31:44] embedded finance platform in Tear Sheets [31:46] Big Theory Awards 2024. So whether [31:49] you're building the next great [31:50] marketplace or just need reliable [31:52] payment infrastructure that scales with [31:54] your business, no matter how large you [31:56] scale, JP Morgan Payments is there so [31:58] you can focus on what you do best. To [32:00] learn more about their embedded finance [32:02] solutions and how they're powering [32:03] growth for businesses small and large, [32:05] head on over to jpmorggan.com/acquired [32:08] and just tell them that Ben and David [32:09] sent you. All right, David. So, Gmail, [32:13] we've got our existence proof of a [32:16] Ajaxbased web app. It's going viral. [32:19] People love it. We can really build web [32:21] applications now. Let's go nuts. [32:24] >> Yes. So the next big web apps following [32:27] Gmail were maps, docs, and spreadsheets. [32:32] All absolutely incredible. [32:34] >> And it was not clear that these things [32:37] were possible with web technologies. [32:40] >> These required incredible technical and [32:43] product vision. So first maps. We [32:47] actually did a whole acquired episode [32:49] back in the day just about Google Maps. [32:52] >> The three companies they acquired. [32:54] >> Yeah. It starts in 2003, so even before [32:56] the Gmail launch when a young associate [32:59] product manager APM at Google named [33:01] Brett Taylor that Brett Taylor [33:04] >> of course of ACQ2 fame Brett Taylor [33:07] >> recent ACQ2 guest Brett Taylor. Oh yeah. [33:11] Also friend feed founder, Facebook CTO, [33:15] coco of Salesforce, chairman of Open AI, [33:19] >> former chairman of Twitter. [33:20] >> Yeah. Yeah. that Brett Taylor starts his [33:23] career out of Stanford in 2003 as an [33:26] associate product manager at Google. He [33:29] ends up going to Larry and is like, [33:31] "We're missing out here." AOL has Map [33:35] Quest, which they've just bought for a [33:36] billion dollars. And I'm hearing through [33:39] the grapevine that Yahoo is about to [33:42] make a big push and launch Yahoo Maps. [33:46] And so, as you would expect, Larry's [33:48] like, "Oh, yeah. Is this a web product?" [33:50] Yes, of course. Go do this. [33:52] >> For all these things that we're studying [33:54] here, there's a business rationale which [33:56] might be extremely indirect, but it's [33:58] there. This idea of increasing web use [34:00] increases Google search, which increases [34:02] the money printer. But then there's also [34:04] an abstract rationale, which is our [34:06] mission is to organize the world's [34:08] information and make it universally [34:10] accessible and useful. And maps is [34:14] squarely in the middle of that. [34:16] >> Yeah. Now, the thing was, as big as Map [34:19] Quest and Yahoo Maps were about to [34:22] become at the time, and they were big. I [34:25] remember using them. My parents used [34:26] them. Everybody on the internet used [34:28] these services. They weren't what you [34:31] think of as Google Maps today. They were [34:34] static web pages. [34:35] >> Yep. [34:36] >> They didn't use Ajax. And the whole [34:38] point was to get driving directions [34:41] >> that you could print out. [34:42] >> Exactly. And the business model for [34:44] these services was on the printed piece [34:48] of paper that people would print out, [34:49] you would put ads on there. [34:51] >> Yep. [34:52] >> It was like a Trojan horse newspaper [34:54] business, [34:55] >> right? [34:56] >> So Brett and Larry and Marissa are [34:58] looking at this like, I think we can do [35:00] better than this. So they go out and [35:02] they buy a little company in Australia [35:04] called Where to Technologies which was [35:07] started by these two brothers Lars and [35:10] Yens Rasmusen [35:12] >> who were incredible engineers and they [35:16] had built a real time interactive [35:21] maps application except it was a [35:25] installed desktop app. And so they're [35:28] meeting with them and Larry's like, [35:30] "Okay, this is what we want, but we need [35:32] it on the web." I think actually the [35:34] quote was, "We like the web at Google." [35:38] And this is how good of engineers the [35:40] Rasmusson were. They go off and in I [35:43] think three weeks they rewrite and [35:46] rearchitect the entire application to [35:49] run as a web app and they basically [35:52] independently discover and implement a [35:55] lot of the JavaScript and Ajax features [35:57] that Google was working on internally [35:59] for Gmail. Gmail still hadn't launched [36:01] yet. [36:02] >> Amazing. [36:03] >> So Google ends up buying where to that [36:04] becomes the core of Google Maps. Around [36:07] the same time they also acquired two [36:08] other companies. Zip Dash that did [36:10] traffic data and keyhole which would [36:14] become Google Earth. Now Google Earth [36:17] was an installed desktop application. [36:21] Ultimately everything that Google Earth [36:22] was building would get folded back into [36:24] maps later. [36:25] >> It's actually not true. I thought that [36:27] and just last night I realized you can [36:29] still go to earth.google.com and get a [36:32] completely different 3D experience than [36:34] Google Maps. [36:35] >> Oh, no way. [36:36] >> It's all in the web now. It's [36:37] unbelievably powerful. Oh, so it is a [36:39] web app, but it's separate than maps. [36:41] >> Yes. [36:41] >> Oh, I didn't know that. Oh, I got to [36:43] check that out. [36:43] >> It's amazing. [36:44] >> That's awesome. Yes. Keyhole and Google [36:47] Earth, I think, is my favorite part of [36:50] our first Google episode earlier this [36:52] year that the whole thing ended up just [36:55] being a Trojan horse downloader to get [36:58] Google Toolbar installed on Internet [37:00] Explorer on people's systems. was there [37:02] organizing the world's information and [37:04] making it universally accessible and [37:06] useful, but also it comes with Google [37:08] Toolbar. [37:09] >> Yeah. The greatest distribution hack for [37:12] Google search of all time. [37:13] >> Yes. [37:14] >> Anyway, back to Google Maps and where [37:18] too. February 2005, Google Maps [37:21] launches. People go nuts. Live mapping [37:26] dynamic web application. Do you want to [37:28] know my favorite Easter egg for the [37:30] first day launch of Google Maps? I don't [37:32] know if you know this. When you loaded [37:33] up maps.google.com, do you know what [37:36] visually you saw? [37:37] >> I have no recollection. [37:39] >> You saw a great big ocean and North [37:42] America and then floating in the middle [37:44] of the Atlantic Ocean, you see the UK [37:47] and then there is nothing past it. [37:51] >> They hadn't built it yet. [37:52] >> They hadn't built it yet. Europe, Asia, [37:55] Africa, not included. It's not even like [37:57] it's off limits. It looks like there's [37:59] an ocean where Europe should be. [38:03] How do you decide what the MVP is or the [38:06] minimum viable product to ship on the [38:09] map? That's amazing. [38:11] All right, there's one more really [38:13] important piece of maps, which is the [38:15] next year in 2006, they released the [38:18] API. And this is what really kicks off [38:22] the web 2.0 [38:24] era, Gmail and JavaScript and Ajax had [38:27] inspired developers out there for sure [38:29] to make richer web apps and people were [38:31] doing that. When Google releases the [38:34] maps API, [38:36] this thing called mashups starts [38:38] happening. You remember this? [38:40] >> Absolutely. [38:40] >> It's now super easy to grab Google Maps [38:45] and build stuff on top of it. And it's [38:48] really hot and this enables startups. So [38:50] like Zillow, Uber, eventually Door Dash, [38:54] Airbnb. Think about all the companies [38:57] that just couldn't exist without the [39:01] Google Maps API. [39:02] >> There was that whole web of geo related [39:05] companies too. Remember that era of [39:07] mobile, social, local, mo? [39:10] >> Oh yeah, forsquare and gowala and yeah, [39:13] all those [39:14] >> all this existed because Google maps [39:16] existed. So back to Google's overall [39:19] strategy here and adoption of web apps [39:22] and sort of building this moat and [39:24] defense against Microsoft. This is just [39:27] incredible. I mean here's maps itself as [39:30] a first class rich web application that [39:34] tens eventually hundreds today billions [39:37] two billion plus users use and love [39:40] every day. [39:42] And now here's this API that's making it [39:44] really easy to help other startups and [39:47] other companies go build great web apps, [39:50] too. The lock in just keeps getting [39:52] deeper and deeper and deeper for the [39:53] web. [39:54] >> Yep. And at first, the API was [39:56] notoriously free or very inexpensive at [39:59] very high limits for a long time. That's [40:01] different now. But for the longest time, [40:03] it was just this is a part of the [40:05] mission, so we're doing it and we'll [40:07] figure out the business later. It was a [40:08] very sort of founder driven thing. Now, [40:10] it's popular to create maps. I mean, [40:13] Apple at some point flipped into doing [40:15] it and there's these sort of other [40:16] thirdparty companies and there's the [40:18] Open Street Maps and all this stuff. For [40:20] the first five to maybe 8 years, Google [40:24] was kind of the only ones that had a [40:26] passion for this and a willing to spend [40:28] into the giant hole that you need to [40:31] create maps of the whole world. I mean [40:33] is they incredibly hard data and [40:36] engineering problem. And you know they [40:38] had to go draw all their own maps from [40:39] scratch, acquire the data, figure out [40:41] how to get fresh data all the time, [40:43] create a crowdsourced thing among [40:45] Google, was it Google maps explorers or [40:47] something like that? All the people that [40:49] would update these things. This is a [40:52] extremely googly problem and a founder [40:54] bet to be like, "Nope, we're gonna go [40:56] spend hundreds of millions of dollars, [40:58] billions of dollars on this, drive cars [41:00] around, taking pictures of everything. [41:02] figure out how to not overshare personal [41:05] information on this. Do it dynamically [41:06] because you're capturing a huge amount. [41:08] I mean, it's just it's a wacky wacky [41:11] engineering problem that is daunting and [41:13] they took it on. [41:14] >> Yep. And we're not going to talk about [41:16] this today, but put a pin in for the [41:19] next episode is one of the most [41:23] incredibly strategically valuable data [41:27] assets for the AI era and specifically [41:30] for self-driving cars. Yes. But today, [41:33] MAPS has over two billion active users [41:36] this year. They don't break out revenue, [41:38] but estimates are that Maps does well [41:41] over five billion in revenue, maybe even [41:43] 10 billion in revenue. The larger part [41:45] of that is ads. You see recommended [41:48] places to go around you all the time [41:49] whenever you open Google Maps now that [41:51] are sponsored ads just like on Google [41:53] search. And then the smaller part of [41:55] that is from the API licensing, David, [41:57] that you were talking about. But this is [41:58] a real business for Google today. [42:00] >> Yep. All right. Next ones that we got to [42:03] talk about, docs and spreadsheets. So [42:07] these aren't the biggest Google apps out [42:10] there today. I think if you lump them [42:12] all together into workspace and ad [42:14] drive, it is over a billion users. [42:16] >> That whole suite is among their most [42:18] used products. [42:19] >> That whole uh you call it off office [42:22] suite. Is that what you would call it? [42:24] Yeah. [42:24] >> Sounds like a office type suite. It's a [42:27] good idea. Someone should do that. [42:29] >> So docs and spreadsheets hit Microsoft [42:34] right where it hurts. office. So people [42:38] have tried both before Google and after [42:42] Google to compete with Microsoft in [42:45] productivity forever. [42:47] >> Word Perfect, Lotus Notes, Lotus 123. [42:52] >> We talked all about that on our [42:54] Microsoft episode. By the way, Word [42:56] Perfect acquired by and run by Noville, [43:00] who is the CEO of Noville, Eric Schmidt. [43:03] Eric knows all about this. [43:05] >> But here's what I will say, David. If [43:07] you were starting on the foot of [43:09] competing with Microsoft or trying to [43:11] build a word processor or trying to [43:12] build a spreadsheet, you would be doomed [43:15] for failure. What Google was doing was [43:18] saying there is something that is [43:20] uniquely possible with web applications [43:23] and Ajax in this web 2.0 era for the [43:25] first time. And that thing is real time [43:28] collaboration. real time multiuser [43:32] collaboration. [43:34] These were the first [43:37] I've tried to rack my brain. I talked to [43:40] Sam Chillis, the founder of rightly [43:42] which Google acquired which became [43:43] Google Docs. He believes these were the [43:46] first real time multi-user collaborative [43:50] pieces of software in history. It just [43:53] wasn't possible before the web. [43:55] >> Yeah. Jonathan Rashelle, the founder of [43:57] the company that would be acquired that [43:58] would become Google spreadsheets, [44:00] basically said the same thing. His [44:01] comment was, "We actually didn't know if [44:02] it was possible to do this in the web." [44:05] Google said, "Based on the success we're [44:07] seeing with Gmail, I bet we could do [44:10] actual spreadsheets in the browser with [44:13] real-time collaboration." And when the [44:14] Sheets team came in, it was truly an [44:16] open question of can we make it so you [44:18] and another person can party on the same [44:21] very basic spreadsheet at the same time. [44:23] >> Interesting. the Docs team. So Docs was [44:26] an acquisition. It was a company called [44:28] Wrightly that was founded by Sam and his [44:30] two co-founders who were great [44:32] programmers. They've worked together for [44:33] many years. [44:34] >> I used Wrightley before it became a [44:36] Google. [44:37] >> No way. [44:37] >> You were like one of very few people who [44:39] did that. [44:39] >> Yeah. [44:40] >> Cuz it was not an independent company [44:41] for long. [44:42] >> Product launched August 2005. Google [44:45] bought the company in March 2006. So you [44:48] had about a six-month window. [44:49] >> Wow. But yeah, they built real time [44:53] collaborative word processing as a web [44:57] app inspired by Gmail and everything [45:00] that was going on at Google. The whole [45:02] company started as, oh, for our next [45:05] project, let's explore what we can do [45:08] with JavaScript and Ajax and oh, what [45:10] would it be like if we put a word [45:13] processor on the web? They weren't [45:15] actually even thinking about [45:16] collaboration at first, but then as they [45:18] were working on it together, they [45:19] naturally started collaborating and [45:22] thought, "Oh, this is the killer [45:23] feature." [45:24] >> That's funny. That's different than the [45:26] spreadsheets team. Their whole thing at [45:28] first was, "We're not going to make a [45:30] better spreadsheet than Excel. So, if we [45:32] put it in the web, it has to be about [45:34] sharing collaboration." [45:35] >> Yep. And so, to your earlier point, [45:38] nobody can compete with Microsoft in [45:40] productivity software. One, because [45:43] they'd been doing it so long, they had [45:46] this feature wall of so many features [45:48] that people needed [45:49] >> two proprietary file formats. [45:51] >> They had a network effect of the file [45:53] format. [45:54] >> You built your big model in Excel. Good [45:56] luck. [45:57] >> Other people need to be able to run it [45:58] on their installed desktop applications. [46:02] Good luck getting somebody to try [46:04] downloading or buying a new piece of [46:06] software and installing it on their [46:08] machine. But three, I mean the biggest [46:10] by far the enterprise agreement. This is [46:14] Microsoft's whole entire business model, [46:17] >> right? You don't have to be best in [46:19] breed in any specific thing. You just [46:21] have to be a platform with everything. [46:24] >> Yep. And IT departments will buy it. And [46:26] especially for productivity software, [46:28] really all the money is in B2B and work [46:31] applications. And so if IT departments [46:33] are buying the Microsoft enterprise [46:34] agreement, they're getting everything. [46:36] Good luck unseeding Microsoft Office. [46:39] >> And I'm not sure you could do this as an [46:41] independent business cuz think about how [46:43] long Google went with these things [46:45] before they were adopted by bigger [46:47] companies. For the longest time it was [46:49] oh a Google doc that's like a thing for [46:51] either you use that for your personal [46:53] life or maybe like a startup would use [46:54] it. But even a medium-sized company you [46:57] can't be serious. Get out of here with [46:58] that. And Google was basically able to [47:00] subsidize it because they had a giant [47:03] existing business. [47:04] >> You are so right. [47:06] Nobody except Google could do this for a [47:09] whole bunch of reasons. One, you talk [47:11] about subsidizing. Imagine trying to [47:14] build this software as an independent [47:17] company or really even as any other [47:19] company. It would require a lot of [47:21] infrastructure, real time multi-user [47:24] collaboration in a web app. Gosh, that [47:26] seems like really complicated server and [47:29] backend infrastructure. For Google, it's [47:32] what they do. running Docs and Sheets, [47:34] the incremental load to Google's [47:36] infrastructure was trivial compared to [47:39] search. They already had it built out. [47:41] It was super cheap. [47:42] >> Yeah. [47:43] >> Two, they don't need to make money from [47:45] it. This is the big reason why nobody [47:47] else could compete. Microsoft has all [47:49] the dollars completely on lockdown [47:52] because of the enterprise agreement. [47:54] >> Big dollars. These small and medium [47:55] businesses would of course pay for [47:57] something, but those dollars don't add [47:59] up to be nearly as big. [48:01] >> Right. Exactly. Google though that's [48:03] fine. Microsoft can keep all the [48:05] dollars. All we care about is people use [48:09] the web. And in this instance, [48:12] particularly with Office and [48:14] productivity, [48:16] really this is about putting the screws [48:18] to Microsoft a little bit and [48:21] distracting them. And from Google's [48:23] point of view, this is a cheap [48:24] distraction. If this gets Microsoft all [48:27] spun up, Microsoft is now all of a [48:29] sudden getting asked all the time, [48:30] "What's your web strategy for Office? [48:32] When are you going to add collaboration [48:33] to Word and Excel, etc., etc." They [48:35] don't have any answers. I literally [48:37] worked on this. My internship was at [48:39] Microsoft and I worked on adding headers [48:42] and footers to the Microsoft Word web [48:45] app. We were porting the Windows code to [48:49] have perfect document fidelity to the [48:52] web. So when you looked on the web and [48:54] then printed from the web, the document [48:55] would be laid out pixel for pixel, [48:57] character for character, exactly how it [48:59] would look on the printed page. When you [49:01] have that requirement, that is a hard [49:05] hard engineering task and it's still not [49:08] as good as Google Docs, [49:09] >> right? I love it that this launched your [49:12] technology career. [49:14] >> Yes. [49:14] >> Amazing. But yeah, from Google's [49:17] perspective, this is amazing. Microsoft [49:19] is now forced to bring their crown [49:23] jewels to the web, which they don't want [49:25] to do. And Ben, to your point, because [49:27] they have to make it look and feel and [49:29] function exactly like the installed [49:31] desktop apps, this is going to take them [49:33] a long time and be a big investment. [49:36] Fantastic. [49:37] >> And no matter what, it's going to be [49:38] more complicated because with Google, [49:40] it's install free. There's no licensing. [49:42] Someone just shares a Google doc with [49:43] you. If you have permission to view it, [49:44] you view it. With Microsoft, I remember [49:45] at first it was sort of antithetical. It [49:47] was like, "But what if I haven't bought [49:48] Word? Can I just use Word for free in [49:51] the web then?" [49:52] >> Right? [49:53] >> Is Microsoft okay with that? Am I going [49:54] to hit some weird usage tier? Like what? [49:56] So, it's confusing for users. It forces [49:58] the company to think about pricing and [49:59] packaging. It was a master stroke by [50:01] Google. [50:02] >> Yep. So, fast forward to today. It's [50:05] hard to get real actual apples to apples [50:07] data on Google Workspace versus [50:10] Microsoft Office users. But basically [50:13] the way to think about the market is [50:15] that Google has the vast majority of [50:19] users and usage of productivity software [50:22] and Microsoft still has the vast [50:24] majority of dollars and that's fine. [50:27] Google's super happy about that. [50:29] >> Is that true that there's more active [50:30] users of Google Workspace than there is [50:32] of Office? [50:34] >> Yeah. I mean I think if you look at [50:35] users of docs or sheets or slides it's [50:39] in the billionish 500 million billion [50:43] range for each of those. Office I think [50:46] has [50:48] couple hundred million users worldwide. [50:52] >> Whoa. Yeah, that's crazy. I didn't [50:54] realize that. [50:55] >> Pretty wild, right? But to my point [50:58] about the dollars, so Microsoft's [51:00] productivity and business process [51:03] segment, which is mostly office, I think [51:06] LinkedIn is now part of this too. Last [51:08] year generated over $120 billion in [51:12] revenue. Google reports workspace as [51:16] part of the cloud segment. So all of [51:18] cloud inclusive of their infrastructure [51:20] as a service all the AI infrastructure [51:24] all that the whole cloud segment for [51:27] Google last year did about $50 billion [51:31] in revenue less than 50 so Microsoft's [51:33] productivity segment 120 office is the [51:36] big piece [51:37] >> and that's high margin revenue [51:38] >> high margin revenue Google's office [51:40] products are some small portion of a $50 [51:42] billion revenue segment so yeah [51:44] Microsoft's still got all the money [51:46] >> Google's got all the users and [51:48] everybody's happy. [51:49] >> But you're so right. Everyone is happy. [51:50] This is exactly what Google wants. [51:52] >> Yeah. And ultimately today, Microsoft is [51:55] fine with this arrangement too. The [51:58] ultimate fun KOD though is Sam Schillis, [52:02] founder of Wrightly. He would go on to [52:04] manage you all of Docs and Sheets and I [52:07] think he actually managed maps at some [52:09] point too. He is now the deputy CTO of [52:12] Microsoft. [52:13] >> Careers are long. [52:14] >> Amazing. The interesting thing [52:16] reflecting on Google's actual business [52:18] here and comparing it against all the [52:20] things that we're talking about, Google [52:22] essentially won search by the mid late [52:25] 2000s. I mean, I know Bing hasn't even [52:27] launched yet, and we'll get to that, but [52:29] search was going to continue becoming a [52:32] more and more giant market. And so, all [52:35] this stuff they're doing, it's like, oh, [52:37] we've won, and this market is naturally [52:39] going to become large. I guess let's [52:41] just fuel it getting larger and try to [52:43] do a bunch of stuff under the umbrella [52:45] of our mission. But what do we really [52:48] need to do? And the slightly more [52:50] altruistic answer, I suspect if Larry [52:51] Page was sitting next to us, he would [52:53] say, "What is the goal of a company?" [52:55] The goal of a company isn't build the [52:58] largest business necessarily. It's to [53:00] fulfill its mission. And yeah, we got a [53:01] money printing machine from Search and [53:03] we're investing a lot of money still in [53:04] Search and making that better. But all [53:06] these things fulfill our mission, too. [53:08] >> Yep. And I think these things are all [53:10] true. [53:11] >> Yes. [53:12] >> So on the back of the success of maps, [53:16] Docs, spreadsheets [53:19] really starts to inform Google's [53:22] strategy here. And specifically, they've [53:24] seen, hey, we can acquire [53:26] these web app web 2.0 companies, bring [53:31] them into Google, turbocharge them, [53:34] offer these magical experiences to [53:36] consumers. We get all this strategic [53:38] value out of them both on the offensive [53:40] and defensive front. We can operate [53:43] these things at a fraction of the [53:44] expense that it would cost anyone else [53:47] to do so standalone company or part of [53:50] other big companies. [53:51] >> And some of the things we could buy [53:53] actually fit into our core ads business [53:56] quite well. [53:57] >> Yeah. What if we went big with this? [54:01] like really big, [54:02] >> like something super expensive to run [54:05] that requires storage of massive videos [54:08] and bandwidth for streaming these [54:10] massive videos and lawsuit protection. [54:14] >> Yep. Probably also costs a lot to buy it [54:16] cuz it's wellunded from Sequoia. [54:20] That leads us to YouTube. But before we [54:24] do that, now is a great time to thank [54:27] friend of the show, Anthropic, and their [54:29] AI assistant, Claude. [54:31] >> So, as we were researching Google's [54:33] expansion here from just search into [54:35] being a real platform company here in [54:37] the 2000s with Gmail, Chrome, Android, [54:40] Workspace, everything we're going to get [54:42] to later in the episode. The complexity [54:45] just skyrocketed with all these [54:46] interconnected systems that needed to [54:48] scale to billions of users and keep [54:50] information flowing between all the [54:52] various products and services that [54:53] Google was launching. The funny thing is [54:56] how quaint that problem seems today [54:59] compared to the scale, speed, and [55:00] interconnectedness that you need in the [55:02] AI era. If you're an enterprise building [55:04] today in AI, you need to deal with all [55:07] of this times 10. [55:09] >> Yes. So, enter Claude. What makes Claw [55:11] different for the enterprise is [55:13] sustained performance on complex tasks. [55:16] We're talking about the kind of work [55:17] that would typically take your senior [55:18] engineers weeks, like refactoring entire [55:21] code bases or synthesizing thousands of [55:24] regulatory documents. I wouldn't know [55:26] anything about that. Claude can handle [55:28] these multi-hour projects while [55:30] maintaining context and fewer [55:32] hallucinations throughout. Claude is [55:34] actually the most adopted AI within [55:36] enterprises when it comes to their API. [55:39] >> Yep. That's because Claude integrates [55:40] seamlessly with existing workflows [55:42] through their MCP connector system. They [55:45] have pre-built integrations with tools [55:46] like Jira, GitHub, HubSpot, and Square, [55:49] plus custom integrations for any [55:51] internal system, making Claude your [55:53] central knowledge resource. [55:55] >> So companies like GitLab are already [55:56] using Claude for coding and research [55:58] teams use it to process documents that [56:00] would normally take weeks to analyze [56:02] manually. If you're building the next [56:04] generation of intelligent applications, [56:06] check out Anthropics Enterprise [56:08] offerings to see how teams are [56:09] transforming their workflows with [56:10] Claude. And we've got a special offer [56:12] for acquired listeners to try out Claude [56:14] before making the enterprise commitment. [56:16] Halfpric Claude Pro for 3 months. Go to [56:20] claude.ai/acquired [56:22] to get started. [56:23] >> And just tell them that Ben and David [56:24] sent you. [56:25] >> All right, David. The YouTube story. [56:28] >> The big kahuna. [56:29] >> The big kahuna. Ah, [56:32] the most embarrassing thing in acquired [56:35] history was our early episode on [56:38] YouTube. [56:39] >> All right, I have got a proposal for [56:40] you. [56:41] >> Okay, I'm ready for it. You want to take [56:43] it out of the feed? Delete it [56:44] >> today. We're setting the record [56:46] straight. When we finish this section, [56:48] we are regrading YouTube. We are [56:51] updating the acquired cannon. It's [56:53] happening. [56:53] >> Oh, let's do it. We're bringing grading [56:55] back, baby. [56:56] >> Great. I'm glad you're into it. [56:57] >> I love it. I love it. [56:58] >> Awesome. All right, YouTube 2003. So, [57:02] same time frame as everything we're [57:05] talking about here. Gmail hasn't even [57:06] launched yet. Google starts working on [57:09] Google video. And the idea is that [57:13] there's a lot of information in video [57:16] and thus it fits Google's mission, Ben, [57:18] as you were saying earlier. And also, [57:22] well, there's just so much more [57:26] advertising dollars in TV than anywhere [57:30] else in the global economy. [57:32] >> To this point in time, TV was the bulk [57:34] of ad spend. [57:35] >> Yep. If you go look at some of the old [57:37] Mary Maker Internet Trends decks, [57:39] remember from this time period, and you [57:42] look at the share of global ad dollars [57:44] spent on TV versus any other category, [57:48] it's just so much bigger than anything [57:50] else. [57:50] >> David, I am so glad you did this. We are [57:52] brothers. I did the exact same thing to [57:54] try to tee this up. [57:55] >> Amazing. [57:56] >> I have the stats in front of me. For [57:58] listeners, digital advertising, you [58:00] know, Google's universe would not [58:03] eclipse TV [58:05] until 2017, 2018. [58:09] >> Wow. [58:11] So, almost 15 years in the future from [58:14] when we're talking about here in 2003. [58:17] >> Yes. That is the wildest thing that TV [58:19] was bigger than digital for that long. [58:21] Mary Mer famously had this point that [58:24] she made every single year that the [58:26] attention was all in the digital [58:28] economy, but there was this gap and the [58:30] ad monetization hadn't caught up yet. [58:31] And it took all the way till 2018 for [58:33] the flip to finally happen where digital [58:36] overtook television [58:38] >> thanks to YouTube. [58:40] >> Yes. [58:42] >> And Facebook and Meta and Tik Tok and [58:45] etc. [58:46] >> And the rest of Google too. [58:47] >> I know. I know. So this Google video [58:50] project [58:52] actually came out of the ads or it [58:54] didn't come out of engineering and the [58:56] rest of the Google product or yeah it [58:58] was motivated by hey there's a lot of [59:01] money in TV and [59:04] >> of course this fits the mission. There's [59:06] a lot of information in video. We should [59:08] totally do this. [59:09] >> Here's how Larry describes it. Google [59:11] Video was first launched in 2005 as a [59:14] search service for television content. [59:16] Yes. [59:17] >> Because TV closed captioning made search [59:20] possible and userenerated video had yet [59:22] to take off. But it subsequently evolved [59:24] into a site where individuals and [59:26] corporations alike could post their own [59:28] videos. They were digitizing TV because [59:32] the transcription wasn't as good as it [59:33] is today. So they needed the closed [59:35] captioning data to make it searchable. [59:37] And they were almost like meta-arching. [59:38] They were looking for other websites [59:41] that allowed people to upload video and [59:43] including that in the search results [59:45] also. [59:46] >> Yep. Sure. You can see how this [59:50] conceivably could be a product vision [59:52] you could have at the time, but Google [59:55] video was the wrong product. The problem [59:59] was one, you couldn't actually watch the [60:01] video. It was just search that then [60:03] directed you just like Google's main [60:05] search business model off of Google [60:08] video to go consume it somewhere else. [60:10] Yeah. In the beginning it didn't even [60:11] have a player. [60:13] >> Whoa. I didn't realize that. [60:14] >> Yeah. Crazy. And the bigger problem [60:17] though, well that's a pretty big [60:18] problem. Another big problem, shall we [60:21] say, was that the focus was on [60:24] traditionally produced head kind of [60:28] content, not longtail, not usergenerated [60:30] content. It was really tied to TV. There [60:32] was a press release that said that they [60:34] could search the content of TV programs, [60:36] find programs containing the content [60:37] they're looking for, and discover when [60:39] and where the program will next air. [60:41] >> Yeah. So meanwhile obviously here we are [60:46] in 2004, [60:48] five, six, [60:50] consumer generated digital video is [60:54] becoming a thing either via standalone [60:58] new devices like the flip cam that's [61:00] coming out from [61:01] >> Flip was a startup, right? And then [61:02] Cisco bought it. [61:03] >> Yeah, my other internship employer [61:05] bought Flip while I was there. Yeah, [61:07] this is like Ben Gilbert personal [61:09] history. But more commonly, so I mean [61:11] there were dedicated devices like the [61:12] Flip Cam, but digital point andoot [61:14] cameras had gotten so good by this point [61:18] in time. This is going to come back up [61:19] later in the episode. [61:21] >> People thought this was the big consumer [61:25] electronic device vector before [61:28] smartphones. People were really, really [61:30] excited about how good and how [61:33] universally adopted digital cameras [61:36] were. All of a sudden, in the mid-200s, [61:38] for the first time, anybody could make a [61:42] video at any time. [61:44] >> And iMovie was just becoming a thing. [61:47] So, you could shoot it on your point and [61:48] shoot and you can edit on your computer. [61:51] >> That's right. So, YouTube [61:56] in early 2005, [61:59] three PayPal employees, the PayPal [62:02] mafia, actually fairly junior employees [62:04] at PayPal, Chad Hurley, Javeid Kareem, [62:08] and Steve Chen, leave PayPal and create [62:12] YouTube. Okay, Ben, I have two deep cut [62:18] YouTube corporate history trivia items [62:20] for you. H [62:22] >> number one, do you know what YouTube's [62:26] original tagline was? The name of the [62:29] company was YouTube. What was the [62:32] tagline and the value prop? [62:34] >> I have no idea. [62:36] >> Tune in [62:38] hookup. [62:39] >> Really? [62:40] >> It was a video dating service. [62:43] >> I did know that. [62:45] And they actually posted Craigslist ads [62:50] in the Bay Area for attractive women to [62:54] make videos to post as profiles on the [62:57] site. [62:58] >> Unbelievable. [62:59] >> And they got like no responses as you [63:02] would expect. Thank goodness for them [63:04] and Google though because then they [63:05] pivoted into a general purpose video [63:09] uploading site that anybody could upload [63:11] anything that they made YouTube. Okay, [63:13] so that's trivia question number one. [63:14] >> Okay, [63:15] >> trivia question number two. Do you know [63:19] who Chad Hurley, Chad was the CEO, who [63:22] Chad's father-in-law [63:25] at the time was? [63:27] >> Oh, no. I have no idea. [63:29] >> Jim Clark [63:30] >> of SGI and Netscape [63:32] >> of Silicon Graphics and Netscape. Jim [63:34] Clark. Jim Clark. [63:36] >> Didn't know that. [63:37] >> Yeah. So, not only were they part of the [63:40] PayPal team and PayPal mafia, like they [63:42] had the best adviser of all time. Wow. [63:45] >> To navigate the Silicon Valley ecosystem [63:47] and the internet ecosystem in Jim Clark. [63:50] >> So, the brilliance of YouTube and it [63:54] really was absolutely brilliant was [63:58] three-fold. One, it was super easy for [64:02] anyone to upload a video. So they had a [64:05] killer content acquisition model. [64:08] Anybody, anytime, anything. [64:11] >> And as soon as the servers process it, [64:13] we'll put it live. No copyright checks. [64:16] Unlike Google video, which would take [64:18] one to two days, [64:20] >> all about copyright [64:21] >> for humans to pour over it, make sure [64:24] that it was all good, and bless it, and [64:26] then put it live, which of course won't [64:28] scale in the UGC era. YouTube's just [64:31] like, "Whatever, upload it." [64:32] >> Yeah. two, super easy for anyone to [64:36] watch a video. You need a really good [64:39] viewer in the web app to view the [64:42] videos. Google video didn't have it at [64:44] the beginning. So, killer content [64:46] consumption model. Go to youtube.com, [64:49] find something or find a link or number [64:52] three brilliant thing about YouTube, see [64:55] a YouTube video embedded on another [64:58] website. Boom, you're watching the [65:00] video. killer growth in distribution [65:03] model. And also, YouTube pretty much [65:06] from the beginning had great search. You [65:08] can search YouTube and find videos that [65:12] you're looking for. Pretty quickly, [65:15] YouTube became and still is, Google [65:17] talks about this all the time, the [65:19] second largest search engine on Earth [65:22] behind Google. [65:23] >> It's amazing. [65:23] >> Is searches happening on YouTube. [65:25] >> That happened quickly. I always thought [65:27] that was a more recent last 10 years [65:29] phenomenon. I think that happened very [65:31] quickly. YouTube traffic scaled so fast [65:34] and so big. So you can see how YouTube [65:40] here, not only are they the correct [65:43] video platform for the web and just [65:45] doing it much better than Google's doing [65:46] it with Google video, there's actually [65:49] some version of the world where they [65:50] might become a real competitor to [65:52] Google's core business. If all these [65:55] searches are happening, they could add [65:56] search for other things on YouTube, too, [65:58] >> right? I don't think they had any plans [66:00] to do that, but it's the same rationale [66:02] of Mark Zuckerberg saying, "Uh oh, [66:04] everyone's using WhatsApp for messaging, [66:07] whether or not they put in a social [66:08] media feed stream. They always could." [66:11] And so, it's really dangerous to me for [66:14] them to be out there aggregating all the [66:15] users and attention and habits when they [66:17] always could do something like that. [66:19] >> Exactly. Same dynamic. And whereas in [66:23] the previous categories of apps that we [66:26] talked about, Google had the advantage [66:29] of uniquely being able to do it as [66:31] Google in a way that startups couldn't. [66:34] Here it's a little bit the opposite. [66:37] YouTube as a small startup has the [66:39] advantage of oh copyright rules, laws, [66:44] but I don't know. We're just a platform. [66:46] We're just a startup. Anybody upload [66:48] anything. Google by this point in time [66:50] is a public company. No way they could [66:52] behave like this. [66:53] >> Well, it's funny. They could, but they [66:54] wouldn't. They actually could do it and [66:57] stay in business, whereas YouTube can [67:01] say, "Eh, whatever." But then they're [67:03] going to go out of business because [67:04] they're going to get sued out of [67:05] business. So it's this really [67:07] interesting sort of catch22 of this is [67:09] the way to start and get all the users [67:12] because this is the best user experience [67:15] and at the same time it will not work as [67:19] a resource constrained small company [67:21] >> once it is started it needs to be part [67:24] of Google [67:24] >> right [67:25] >> yes obviously we're going to get to that [67:27] but in the beginning though oh my gosh I [67:30] mean the embeds were a beautiful [67:32] distribution growth mechanic for YouTube [67:35] but people were just uploading [67:37] copyrighted video that people could [67:39] watch for free. I mean, it's almost like [67:41] Gmail. It is so unbelievably compelling [67:44] to a consumer when your friend tells you [67:46] about YouTube or just sends you a link [67:48] or you see an embed page of, "Whoa, I [67:50] can go watch Lazy Sunday from Lonely [67:54] Island and Saturday Night Live in my web [67:56] browser anytime I want for free with no [68:00] commercials." [68:01] >> Yes. [68:01] >> Yes, I want that. And in fact, when [68:05] users started uploading Lazy Sunday, the [68:08] Lonely Island skit from Saturday Night [68:10] Live to YouTube, this is in that brief [68:13] phase where YouTube was an Ascendant [68:14] startup and not yet part of Google. That [68:17] one skit increased YouTube traffic by [68:20] 83%. [68:22] >> Wow. [68:24] >> Unbelievable. [68:25] >> And so they very quickly raised money [68:27] from Sequoia. Is that right? [68:29] >> Yep. So it was basically incubated at [68:30] Sequoia when the three founders left [68:33] PayPal. Sequoia invested right away. I [68:36] think it was Ruf both's first investment [68:38] when he joined [68:38] >> cuz RUF knew them from PayPal. He's also [68:41] part of the PayPal mafia. [68:42] >> Exactly. And then Sequoia led another [68:44] round pretty quickly thereafter because [68:47] the infrastructure costs started as you [68:49] would imagine scaling astronomically [68:51] here. [68:51] >> Yeah. So three things that are very [68:53] expensive. Two of which are ongoing. One [68:55] is a onetime cost. It's the onetime cost [68:57] but still expensive is encoding the [68:59] video. This might eventually play on [69:01] multiple types of devices and multiple [69:03] browsers. So there's a lot of encoding [69:05] that has to happen. Two then are just [69:08] big ongoing variable costs. You have to [69:10] store all this video. And the biggest of [69:13] all the networking, the bandwidth [69:15] becomes extremely expensive and costs [69:17] you every single time someone plays the [69:19] video. Your biggest cost driver scales [69:22] with minutes watched. So that is [69:24] eventually going to kill you unless you [69:26] have an aligned business model. [69:27] >> Yep. By the way, it would also be really [69:29] nice if whoever owns and operates this [69:33] had its own really good, really cheap [69:35] infrastructure with all of these things [69:37] built into it. [69:39] >> Y [69:39] >> So yeah, pretty quickly within a little [69:43] over a year of launching, YouTube is in [69:46] way over its head. the content issues, [69:48] the copyright issues, the infrastructure [69:51] scaling issues. [69:52] >> It's all exactly what they wanted. It's [69:54] going as well as they could have hoped [69:56] and it isn't way over its head. [69:57] >> Yes. And if this had happened today, [70:01] you could probably raise enough capital [70:03] from the private markets to address this [70:05] and scale up as a company fast enough, [70:08] especially with public cloud that you [70:10] could probably build this as a [70:12] standalone company. [70:13] >> Yeah. I mean today you can go raise [70:15] billions of dollars as a series A [70:17] startup if you're in the right space [70:19] doing the most interesting things with [70:20] the big market. [70:21] >> 2005 2006 not the same kind of private [70:24] capital available and of course there's [70:27] no way the company could go public with [70:29] all these issues or anything [70:30] >> right in particular there was a giant [70:32] suit from Viacom. [70:34] >> Yes. So because of these things YouTube [70:38] ends up basically putting itself up for [70:40] sale. I mean they have no leverage in [70:43] content negotiations with rights holders [70:46] and infrastructure is killing them. So [70:49] in November 2006 which is less than 18 [70:53] months after the product launched Google [70:56] buys YouTube for $1.65 [71:00] billion in stock. [71:03] >> In stock. I'm glad you caught that too. [71:05] >> Stock. Yes. We heard in the research [71:07] that after this deal, Patrick Bashett, I [71:10] think was the CFO of Google at the time. [71:12] >> He said, "Never again." Right? This was [71:13] our biggest mistake. [71:14] >> Said, "Never again. This is the last [71:17] stock deal that we ever do." [71:19] >> Google's market cap has 20xed since the [71:21] day that this deal closed. If they had [71:23] paid in cash, they would have made an [71:25] extra 20 times multiple on whatever you [71:28] already think the multiple is on their [71:31] purchase of YouTube. [71:32] >> Yep. The thing is though, this is we [71:35] will correct the acquired record at the [71:37] end of this section. Either way, even if [71:40] Google paid 20 times $1.6 billion for [71:42] this, they got a screaming deal. YouTube [71:45] is so valuable. All right. I have some [71:48] of the numbers from the first few years [71:50] that I was able to cobble together and [71:52] then I want to talk about some of the [71:54] product evolution over the years. [71:55] >> Yes. Great. [71:56] >> All right. So, Google buys it for $1.65 [71:58] billion. And interestingly, Shashir [72:02] Miraa this week went on the Grit [72:05] podcast, the Kleiner Perkins podcast, [72:07] and laid out a bunch of data on this. [72:10] And I actually didn't have a chance to [72:11] reach out to Shashir yet cuz it just [72:12] came out, but a lot of this is from that [72:15] conversation. So after the acquisition, [72:17] he said, and Shashir was the head of [72:19] product and basically the CPO CTO at [72:21] YouTube, not right after the [72:23] acquisition, but within a year kind of [72:25] came in for four or five years. So after [72:28] the acquisition, he said it was doing [72:29] about 30 million in revenue. [72:31] >> Okay. [72:32] >> So they did have revenue. I believe just [72:34] to foreshadow our next chapter that was [72:37] in the form of programmatic advertising [72:39] that was on the doubleclick ad exchange [72:41] that they were using to make money. They [72:44] were losing about a billion dollars a [72:45] year run rate on 30 million in revenue. [72:49] The amount of money they lost was almost [72:51] exactly equal to a penny per view. So [72:54] just imagine every time you loaded [72:56] YouTube in those years, Google would [72:58] just flush a penny down the drain. They [73:00] got to figure out something to do about [73:02] this. So for the first couple years, the [73:04] CFO at the time was terrified of it [73:06] scaling. Like, please don't scale in its [73:08] current state. But of course, there's [73:10] nothing they can do. The cat's out of [73:10] the bag. It's scaling. And the CFO was [73:13] exploring, hey, can we sell this to one [73:15] of the other companies who was bidding [73:16] on it? [73:18] >> That's right. Because Yahoo and the [73:20] media companies also wanted to buy [73:22] YouTube. [73:22] >> Yes. So Shashir says, "We were broadly [73:24] known as Google's first mistake." [73:26] >> Well, back to my tea up in the intro [73:28] being a pure play. Investors didn't like [73:31] this for a long time. This was a huge [73:34] knock. I mean, gez, when we did our [73:36] episode 10 years ago about YouTube, we [73:38] said it was a terrible acquisition. [73:40] >> Yes. The thing we haven't talked about, [73:42] music licensing was really expensive. [73:45] They were one of the top revenue sources [73:48] for the music industry for a long time. [73:50] Maybe even still one of the top few to [73:52] the music industry. [73:53] >> Yeah. Right up there with Spotify. [73:54] >> Yep. So, kind of on the product side of [73:57] things early on, as you were saying, the [74:00] way that you found YouTube is you would [74:01] see it embedded on a different site. You [74:04] would click through and then you may [74:05] stick around to watch something after, [74:07] but then you'd leave and your entry [74:09] point to YouTube again was another [74:10] embed. Most sessions did not start on [74:13] YouTube.com. So you weren't going to [74:15] YouTube with the idea of that they'll [74:17] recommend something to me. And then even [74:18] the people who did go to YouTube.com in [74:21] this sort of four year 5year period [74:22] after the acquisition 90% of that [74:25] traffic was there to search and they [74:26] just ignored anything that you [74:28] recommended to them. I mean it takes a [74:29] long time a to build habits and b to [74:33] build out the technology to make any [74:35] sort of recommendation or browse or [74:37] anything good. [74:38] >> Yep. For first with related videos and [74:40] then ultimately the feed. And just for a [74:42] sense of scale, there was a report that [74:46] estimated that YouTube that year in 2007 [74:49] consumed as much bandwidth as the entire [74:53] internet did in the year 2000. So just 7 [74:57] years before. I have an extremely [74:59] similar stat from Shashir, which is it's [75:02] a later period. It's 2014, but it's [75:03] apples to apples rather than comparing [75:05] that '07 to 2000. He said in 2014, [75:08] YouTube was 20% of the bits on the [75:10] internet. Wow. [75:12] >> I mean, this stat, but especially your [75:14] stat, illustrates just how much this [75:16] thing took off and also just how much [75:18] more bandwidth video took up than any [75:21] other media type on the internet. [75:23] >> Yeah. But the long-term play here [75:25] obviously is the Mary Maker slide of [75:28] yes, video the reason that it gets [75:31] consumed so much is this is what humans [75:33] want, [75:34] >> right? [75:34] >> And you can advertise against it. [75:35] >> And Google realized this. So I think [75:37] they they were very smart to rather than [75:39] trying to continue investing in Google [75:41] video to basically say they got the [75:43] lightning in the bottle, they have the [75:44] consumer brand, they have the attention, [75:46] let's just go buy that thing and on a [75:48] expected value basis if you're making a [75:50] bet. Sure, you could build it on your [75:52] own cheaper, but your chance of [75:53] succeeding is so unlikely relative to [75:55] buying that thing that it's actually a [75:56] deal to get it for 1.65 billion plus the [75:58] billion that we'll need to invest every [75:59] year for a few years to run it in the [76:01] red. So 2009 is the year where the [76:05] business really starts working. Google [76:07] actually discloses nothing about [76:08] profitability but that the ad revenue [76:10] tripled in one year in 2009. [76:14] 2010 2011 they turned profitable. There [76:17] was a report that in 2012 they were [76:20] estimated to make about $4 billion in [76:22] revenue but roughly break even. Then the [76:25] 2012 134 I think they were small [76:28] profitable but profitable. Then in 2013 [76:31] to 2015, that time period on the product [76:33] side, that's when things really changed. [76:35] So the north star really became users [76:38] should go to YouTube to be entertained [76:41] for 15 minutes. And it's our job to do [76:44] whatever we need to do to make that [76:46] true. And a few things really helped [76:48] with this. One was the shift to mobile. [76:51] In mobile, and remember they were a [76:53] launch partner on the iPhone. [76:54] >> Oh, we're going to get into it. [76:55] >> Okay. There were a lot more low intent [76:58] sessions. So people who open the app [77:01] rather than clicking through from an [77:03] embed page. [77:04] >> Low intent being low intent of watching [77:06] a specific thing. [77:07] >> Yes. It's a beautiful thing on mobile [77:10] that you can sort of say I've got [77:11] something to recommend to you. And [77:12] obviously short form vertical video like [77:15] Tik Tok and YouTube shorts and all that [77:16] these days is that on steroids. Mobile [77:19] also made it the case that any given [77:21] user was more likely to be logged in. [77:24] That way all the personalization, all [77:26] the algorithm stuff works well. They [77:28] also adjusted their core metric [77:30] internally away from views and to watch [77:33] time. And YouTube was very early to the [77:37] concept of creator monetization. For a [77:39] long time, it was the only place on the [77:41] internet where creators could make [77:43] money, [77:43] >> share revenue with creators. [77:45] >> And in our old episode, we sort of [77:47] knocked them. We said, "Look, this [77:48] business has to give its first 50% off [77:51] the top of any revenue it makes to the [77:53] creator of the video. That's a way worse [77:55] business than say Google search ads or [77:58] Facebook who you know Facebook has [78:00] influencers on their platforms too all [78:02] the meta platforms Instagram and their [78:05] rev share if it's anything sure sure [78:07] isn't 50% it's probably closer to zero [78:09] and YouTube right early on said you're a [78:12] 50-ish% partner which takes you a decade [78:15] longer to get profitable but helps you [78:18] build that base [78:19] >> just creates amazing incentives for [78:21] people to build businesses and careers [78:24] is on this. I mean, YouTube is the [78:27] ultimate instantiation of the internet [78:30] to me and the power that it can provide [78:34] to individuals to make a living. It [78:36] abstracts the need at all to create or [78:40] run a business. It really just [78:42] simplifies it down to make content that [78:45] people watch, you will get money for it. [78:48] You don't have to do anything else in [78:50] between. There's a little slight of hand [78:51] that you did there, David, which is that [78:54] people watch. [78:55] >> Well, yes. [78:56] >> So, YouTube internally went back and [78:58] forth for years on this, and I think [78:59] we're sort of in this no man's land that [79:01] we've landed today. Camp one is, hey, [79:05] the way to make people most engaged is [79:08] by getting them to follow creators and [79:09] they curate the information sources they [79:12] want. Camp two is in algorithms we [79:15] trust. It turns out camp 2 is actually [79:19] correct, which is unfortunate. It's a [79:22] messed up incentive. Most of the time, [79:24] if you show someone something that [79:26] they're subscribed to or you show [79:28] someone something that the very smart [79:30] computers have figured out, you will [79:32] watch and then watch another video after [79:35] that. Usually the algorithmic approach [79:38] is right. And so there is sort of this [79:40] internal conflict there where they say, [79:42] "Yeah, of course you should subscribe, [79:43] but your views are only loosely related [79:46] to how many subscribers you have." [79:48] >> Yep. This is the dark side to the [79:51] YouTube economy. [79:52] >> Yes. But putting that aside, just the [79:55] sheer concept of anybody and everybody [79:58] in the world has a video camera today [80:02] can create something and if it's good [80:04] and people watch it and the definition [80:06] of good being the algorithm likes it, [80:09] you will make money [80:11] >> with no other steps in between that can [80:14] only happen on the internet. [80:15] >> It's pretty interesting because it kind [80:16] of has these two business models in core [80:19] Google land. They have the Adwords [80:21] business model where they're the first [80:23] party media site. Each search result [80:26] page is a form of media and they run ads [80:28] on that and then they keep approximately [80:31] 100% of the revenue generated from that [80:33] ad. The advertiser pays them and they [80:35] share some in the form of traffic [80:37] acquisition costs that we'll talk about [80:38] later, but it's a largely a first party [80:40] ad. And then they have this other form [80:43] AdSense and the Google content network [80:45] when they show display ads on other [80:47] people's websites where they share like [80:50] 70% of the revenue most of the revenue [80:53] out to the [80:55] >> the publisher publisher [80:55] >> the content owner [80:56] >> right who actually is the reason why [80:58] there's an ad there in the first place [80:59] and YouTube was sort of a interesting [81:02] mix between the two they were [81:04] comfortable and familiar with the idea [81:06] that we can manage a platform where we [81:09] actually share a a lot of the revenue [81:12] with those producing the content, which [81:14] is is interesting. Like, if they had [81:16] never gone into the AdSense world and [81:19] they were purely a search engine, I [81:21] think it would have probably been more [81:22] of a fight to try to do this 50% split [81:24] with creators. All right. So, there's a [81:28] thing that I mentioned earlier, this [81:30] notion of people on mobile are more [81:32] likely to be logged in than people who [81:34] just hit a web page on desktop. [81:36] Loggedinness [81:38] is essential for YouTube's success. [81:42] That is actually new to Google. [81:45] Loggedinness is not essential to the [81:48] effectiveness as of a search engine or [81:50] even the monetization of a search [81:52] engine. We've sort of flirted with this [81:54] idea. You can kind of hear it through [81:55] our episode of things like Gmail are [81:58] good because then you're logged into [81:59] Google, but there's not a giant lift. [82:02] >> It's sort of like a nice to have. [82:04] >> Yes. search, especially on the [82:07] advertising side, is already so bottom [82:09] off funnel, [82:10] >> right? The intent is right there. I [82:12] don't need to know what your [82:12] demographics are. I know what your [82:14] intent is, [82:15] >> right? You search for a shovel, I'm [82:16] going to sell you a shovel. [82:18] >> Yeah. [82:20] >> That's a stark contrast to YouTube where [82:22] the whole YouTube flywheel really only [82:24] works with logged in users, [82:26] >> right? Not just for serving videos and [82:31] content for you to watch, but also for [82:33] advertising. This is how the television [82:35] advertising ecosystem works. It's about [82:37] targeting. You know, why does Chevrolet [82:39] advertise on football games? You need [82:42] demographic data, [82:43] >> right? All right. So, David, are you [82:45] ready to regrade the acquisition of [82:48] YouTube by Google? [82:49] >> Yes. We got to set the context of how [82:52] big it is today. [82:53] >> All right. So, back in 2016, we had two [82:55] knocks on YouTube. One is that it wasn't [82:57] a destination site. Narrative number two [82:59] was they only get to keep 50% of their [83:02] total gross revenue and then they've got [83:05] these crazy infrastructure costs and [83:06] they'll never be able to outrun them. [83:08] Well, they sure have solved both of [83:09] these issues. Clearly a destination [83:11] site. Actually, more than anything, a [83:12] destination app. You open YouTube and an [83:14] algorithm we trust. [83:16] >> It's what I do every night before I go [83:18] to bed. [83:18] >> Yes. And you know, it's some mix of [83:20] things you're subscribed to and things [83:21] you're not, but things YouTube believes [83:23] will grip your attention at that moment. [83:26] On the infrastructure costs, let's just [83:28] start by unpacking their financials. So [83:30] last year in 2024, YouTube ads alone did [83:34] $ 36 billion in revenue. So half goes to [83:38] creators. They have 18 billion left to [83:40] play with, and they've had two decades [83:41] to figure out how to get their variable [83:43] cost down for video hosting, bandwidth, [83:46] compute, for encoding, music licensing, [83:49] all that stuff. They now do insane feats [83:51] of engineering including doing their own [83:54] custom silicon to do video encoding. [83:57] They also have a whole bunch of crazy [83:59] things that they do like change the [84:02] video encoding that is used depending on [84:04] how many views the episode has. [84:08] >> Interesting. [84:08] >> So they do like vanilla H.264 264 when [84:11] you first upload it and then when it [84:13] hits some number of views it switches to [84:16] a more computationally expensive to [84:19] encode but then smaller to distribute [84:22] format and then when you hit another one [84:24] when you have like 5 million views or [84:26] something then they do it yet again they [84:28] re-encode the video and make the file [84:30] size even smaller. So they have figured [84:32] out all these little optimizations to [84:34] make any given stream as inexpensive as [84:37] possible on the sort of variable cost [84:40] basis. [84:40] >> Brilliant. On the revenue side, they [84:43] have gotten much better at selling ads [84:45] and most estimates is that YouTube is [84:47] actually quite profitable. Now, on top [84:50] of the 36 billion in advertising [84:52] revenue, Google reported that if you [84:54] include subscription revenue, so this is [84:56] things like YouTube Premium, YouTube [84:58] Music, NFL Sunday Ticket, they're now [85:00] doing over 50 billion in revenue. [85:03] >> Wow. [85:03] >> So, David, this now makes YouTube the [85:06] second largest media company by revenue [85:08] after only Disney. [85:10] >> And Disney has so many other things [85:12] contributing to that revenue. Theme [85:13] parks, cruise ships, merchandise, etc. [85:16] >> Yes. So, YouTube is already bigger than [85:19] Disney's media business, and this year [85:22] will likely become bigger than Disney's [85:24] entire business. [85:26] >> The question is, how does that revenue [85:28] figure compare to Netflix? [85:30] >> I'm glad you asked. Netflix annual [85:32] revenue for 2024 was 39 billion. So, [85:35] they've alreadycliped Netflix. [85:37] >> Wow. So, there you go. Google doesn't [85:39] release usage data for YouTube, but I'm [85:43] pretty sure that YouTube is the biggest [85:47] single property on the internet in terms [85:50] of minutes spent by humans on it. It's [85:54] not [85:54] >> I believe that [85:55] >> the biggest number of users on the [85:57] internet. Both Facebook Blue app and [85:59] WhatsApp are bigger in terms of total [86:01] number of users, but I think YouTube [86:03] probably dwarfs them in terms of time [86:05] spent by users on the app. I think it is [86:07] the biggest. I could see that [86:09] >> human attention time sync known to man. [86:12] >> So then the question becomes how [86:14] profitable is the 50 billion in revenue [86:17] and officially we don't know but there's [86:19] these great things called research firms [86:20] out there that uh make our jobs at [86:22] acquired here much easier. So storied [86:25] firm Moffett Nathansson published a [86:27] report earlier this year that they think [86:29] YouTube does about $8 billion in [86:31] operating income. 8 billion a year. So, [86:35] I want to diff that against their total [86:38] investment into [86:39] >> Oh, yeah. Okay, great. This is the way [86:41] to grade it. Okay, here we go. [86:42] >> Yes. Now, we're getting into grading [86:44] here. We're landing the YouTube plane. [86:46] >> The definitive acquired regrading of [86:50] YouTube. [86:51] >> So, as mentioned earlier, I don't think [86:52] they ever lost much more than a billion [86:54] a year, and I think they got break even [86:56] within conservatively 5 years. So after [87:00] the $1.7 billion purchase price and the [87:03] 5 billion in additional costs, Google [87:05] paid $6.7 billion. I'd bet it's closer [87:09] to 5.56 to own something that spits off [87:12] $8 billion a year in profit today and [87:16] revenues are growing 10 to 15% every [87:19] year. By the way, also since I think the [87:22] theme of this whole episode is sort of [87:25] like the dual bottom line to Google of [87:27] everything they're doing of both revenue [87:30] and profits and strategic insulation [87:33] versus other large tech companies. [87:35] >> Oh, but David, you're forgetting the [87:36] third of organizing the world's [87:37] information. [87:39] Okay, the triple bottom line. There we [87:40] go. The triple bottom line. Google [87:43] itself, [87:44] not including YouTube, pretty much [87:46] whiffed on social. really really [87:49] strategically good for them that they [87:50] own YouTube, isn't it? Now that you know [87:52] Meta and Tik Tok exist. [87:54] >> Well, here's the crazy thing. They [87:55] whiffed on social and then what ended up [87:58] happening [87:59] >> was social became YouTube. [88:00] >> Yes. [88:01] >> Yes. [88:02] >> It's the craziest thing. We don't open [88:04] apps anymore to look at what our friends [88:06] are posting, a place where Google has no [88:09] presence. But you open Meta's most [88:11] important property with Instagram and [88:13] you look at Instagram's mostused thing, [88:15] reels, or you look at Tik Tok, and what [88:18] do you see? You see videos from people [88:22] you don't know. I mean, it's crazy that [88:25] the rest of social media or almost like [88:27] userenerated media pivoted into Google [88:30] space. Yeah, this was the big D Montto [88:33] to our meta episode last fall was, hey, [88:36] social networking such as the conception [88:39] of it existed in the mid200s and 2010s [88:42] is dead. It's gone. It bifrocated into [88:45] private messaging and public media. [88:48] >> Yes. The sort of middle ground of wide [88:51] group of people you kind of know is [88:54] effectively dead. It's close friends and [88:56] it's I don't really care where it came [88:57] from, but it's entertaining. [88:58] >> Yep. So you could do a discounted cash [89:02] flow on this thing that I just gave you [89:04] this call it 6 billion of investment and [89:08] now 8 billion growing at 10 to 15% every [89:10] year but there's additional strategic [89:13] value too in addition to this thing you [89:15] just said this becoming the winner in [89:16] the short form era they have the largest [89:18] corpus of video to train on for the AI [89:21] era. [89:21] >> Yeah let's go. So Moffett Nathansson [89:24] estimated that if this was publicly [89:26] traded, it would be worth about $500 [89:28] billion as a standalone company. And [89:31] even conservatively, if you sort of take [89:33] media company comps and do a revenue [89:34] multiple and you discount all the [89:37] strategic future value, it's still like [89:38] $200 billion. [89:41] So this is officially one of the best [89:43] acquisitions of all time. And I am [89:45] raising my grade from a C to an A+. [89:48] >> I am obviously right there with you. [89:51] This isn't a plus+. [89:53] >> Now, it's not really fair to say that [89:56] it's like turning 6 billion into 500. [90:00] That initial 1.7 was largely Google [90:03] stock that they traded. So, that had [90:05] real opportunity cost, but it's still [90:07] ridiculous. [90:08] >> Like I said earlier, a screaming deal [90:10] either way. [90:11] >> Yes. [90:12] >> All right. There we go. We have revised [90:14] history. Corrected the record. [90:16] >> Yes. All right. Well, for our next [90:18] chapter, I motion that we go back closer [90:20] to Google's core business of advertising [90:23] on the web [90:25] >> and maybe also [90:28] stay closer to acquire its original uh [90:31] raison detra of discussing the greatest [90:34] acquisitions of all time. [90:35] >> We may as well follow up YouTube with [90:37] Doubleclick. But before we do that, it [90:40] is time to talk about one of our [90:42] favorite companies, Statsig. So, on our [90:45] first Google episode earlier this year, [90:47] we talked about how great the search [90:49] business model is and how once a company [90:50] takes a lead, it's just hard for anybody [90:53] else to catch up. But Google did [90:54] something that kept them in the lead, [90:56] using data to relentlessly improve the [90:58] search experience. Yep. Google really [91:00] was a pioneer in the idea of a [91:02] datadriven product culture. They took [91:04] this to the extreme with a famous [91:06] example where they tested 50 different [91:08] shades of blue for their links on Google [91:11] search result pages to find the optimal [91:13] one. They also famously leverage user [91:15] data when people correct their queries [91:17] to bootstrap the did you mean [91:20] autocorrect feature. More recently, they [91:22] even opted people into AI search via an [91:24] AB test. This obsession with testing [91:27] helped Google find a thousand small [91:29] product and business wins. It also [91:30] helped Google scale its unique culture [91:32] where its employees can quickly test and [91:34] ship new products and features because [91:36] they all have access to great tools. But [91:38] for a long time, smaller companies [91:40] didn't have access to the same quality [91:41] of tools that were available at places [91:43] like Google. Now that's changed thanks [91:46] to Statsig. The smartest new companies [91:48] like OpenAI, Figma, Atlassian, Brex, [91:51] Notion, and Anthropic, plus hundreds of [91:53] startups that you see and use every day [91:55] are using Statsig to build a bottoms up [91:57] datadriven product culture. Statig [92:00] provides all the tools you need to make [92:01] datadriven product decisions in one [92:03] place. advanced experimentation, feature [92:05] flags, product analytics, session [92:08] replays, and more. All backed by a [92:10] single set of product data. And using [92:12] Statsig isn't just about saving [92:14] engineering time. It's about bringing [92:16] that Google level continuous improvement [92:18] culture into your company. Rather than [92:20] arguing about metric definitions or [92:22] troubleshooting broken tools, your team [92:24] can focus on shipping improvements. And [92:26] if you already have your own product [92:28] data, Statsig is warehouse native. So [92:30] they can plug directly into your [92:32] existing data in Snowflake or BigQuery, [92:34] whatever. So if you're interested in [92:36] giving your product team the same [92:37] continuous improvement capabilities that [92:39] keep Google search ahead, go to [92:41] statsig.com/acquired, [92:43] that's satsig.com/acquired. [92:47] They've got a generous free tier, a [92:49] $50,000 startup program, and affordable [92:52] enterprise plans. Just tell them that [92:53] Ben and David sent you. [92:55] So, double click. Well, if buying [93:00] YouTube in October 2006 for $1.65 six [93:05] five billion dollars was a lot. Google [93:09] decided to [93:12] basically double that a few months later [93:15] in April of 2007 when they bought [93:18] Doubleclick for $3.1 billion in cash [93:24] this time, not stock. And this is on the [93:28] display ad side of the house. So [93:30] Google's got two advertising businesses [93:32] at this point. There's Adwords when you [93:34] search and you get the blue links that [93:36] show up above the blue links and then [93:39] there's the off property or the Google [93:41] network ads and at this point in time [93:44] Google just is operating something [93:46] called AdSense which is this ad network [93:48] that they've started. [93:49] >> Yeah. [93:49] >> So Doubleclick actually has a [93:52] fascinating company history before [93:54] Google that I did not know. [93:55] >> Yeah. not a hot rising startup like [93:58] YouTube that they bought for, you know, [93:59] a couple billion dollars, [94:01] >> though it once was. [94:02] >> Yes. [94:03] >> All right. So, here's the Double Click [94:05] story. And huge thank you. There's a new [94:07] book that actually just came out by Ari [94:10] Paparro. The book is called Yield. [94:13] Doubleclick was originally founded in [94:14] 1995. [94:16] >> So, before Google, [94:17] >> before Google, the founders were Kevin [94:20] O'Connor and Dwight Marryman and their [94:22] headquarters were in New York City. The [94:24] original idea was twofold. One, build [94:27] software that could let advertisers [94:30] serve ads across websites. This is [94:32] called an ad server. And the network of [94:34] websites and media, the advertisements. [94:36] When people talk about paid media, it's [94:38] the advertisements themselves that would [94:40] run. Over the next 5 years, they end up [94:43] building and acquiring their way to [94:44] being the leading display ad network and [94:47] ad server. [94:48] >> And they went public during this time. [94:50] Right. [94:50] >> Yep. 1998. shining success of the dot [94:54] industry. However, do crash happens. 70% [94:58] of DoubleClick's customers not only [95:00] churn but go out of business. A huge [95:03] amount of DoubleClick's advertisers were [95:06] actually VCbacked startups. Brand [95:08] dollars hadn't really spread to the web [95:10] yet. Like we talked about, the digital [95:12] advertising was so early and so nent. [95:15] >> Yeah. It was pets.com that was [95:18] advertising on other.com properties. [95:21] >> Exactly. They're almost levered on the [95:23] bubble is probably the right way to [95:25] think about it. So, easy come, easy go. [95:28] So, in 2002, after they're sort of [95:30] limping along for a while, they sell [95:32] that ad network division off for under [95:35] $15 million with an M. [95:38] >> Wow. [95:39] >> So, now all they've got left is the [95:40] software, the sort of ad server part of [95:43] the business. So flash forward to 2004, [95:47] they're this kind of sleepy, slow growth [95:49] company with a shrinking market cap. The [95:52] ad server, their software was still [95:53] widely used, but digital marketing on [95:56] the web just wasn't actually having that [95:58] much spend flow through it. They decided [96:00] to put themselves up for sale. Google [96:02] actually took a meeting to look at it to [96:04] see if they wanted to buy it. They [96:05] decided not to and eventually they sold [96:07] it to private equity. Two different [96:09] firms, Helman and Freriedman and JMI [96:11] Management, bought it in 2005 for about [96:14] a billion dollars. IPO day was double [96:18] this final price tag that they would [96:19] sell it to private equity for. And in [96:22] many stories, this is kind of the end of [96:24] the story. This is the start. [96:26] >> Yeah, it's sort of crazy given the fact [96:28] pattern that you just told us that two [96:31] years later, Google's going to buy this [96:32] thing for $3 billion. [96:34] >> Yes. [96:35] So, David Rosenblat becomes CEO and he [96:40] has a very familiar name that all of you [96:42] will probably recognize who becomes Neil [96:44] Moan. [96:45] >> Yes. The head of product and strategy at [96:47] the company, Neil Mohan. Neil, of [96:50] course, is the CEO of YouTube today. [96:52] >> Yep. [96:53] >> So, from DoubleClick originally, [96:55] >> many would argue the best thing that [96:57] Google got in the DoubleClick [96:58] acquisition. [96:59] >> You could argue that. Now, here's the [97:02] amazing thing. What happens under the [97:05] private equity ownership is that they [97:08] launch a completely different product. [97:10] This new thing called an ad exchange [97:12] when the concept of an ad exchange is [97:14] first invented. Remember it was very [97:16] straightforward before this. There was [97:18] just an ad network and some software [97:20] called an ad server. The ad exchange is [97:22] this sort of brilliant idea that we can [97:24] cross route demand between ad networks. [97:28] And at first what this is sort of used [97:29] for is like the remnant or unsold [97:31] inventory. Oh, we've got some page [97:33] loads. We don't currently have a buyer [97:35] in our ad network forum. So, throw them [97:37] up on this exchange and see if [97:39] programmatically some people will bid on [97:40] it and we'll get more dollars this month [97:43] for the same number of page views. But [97:46] technically, what was going on is it was [97:47] really sophisticated and it allowed for [97:49] some crazy stuff to get done. You could [97:51] bid in real time, including against the [97:53] publishers direct sold ads. You know, [97:55] let's say the New York Times has done a [97:57] specific deal with Ford, then in a [97:59] real-time basis, [98:00] >> right? If somebody else is willing to [98:02] top forward, then you can displace them. [98:04] Yep. Yep. [98:04] >> Exactly. [98:05] >> Gosh, this sounds a lot like Google, [98:08] doesn't it? [98:09] >> It really created the modern [98:10] programmatic display advertising for [98:12] better or for worse. That's basically [98:14] what happened here. And as a publisher, [98:16] when you start working with an ad [98:18] exchange, you can incorporate multiple [98:19] different networks, agency trading [98:21] desks, cuz this became a big thing with [98:23] ad agencies. You can stand up these [98:25] complex rules engines. Effectively, what [98:28] happened is you sort of jumped in front [98:29] of the ad networks. you almost [98:30] disintermediated them. You're the lowest [98:32] level building block that everything [98:34] else has to integrate with and [98:36] eventually what started as this ad [98:39] exchange that just became used for [98:41] remnant and unsold ultimately becomes [98:43] the primary way that digital media is [98:46] bought on the biggest advertisers with [98:48] the biggest publishers and all of course [98:50] bought and sold through these big [98:52] agencies. [98:54] Google is running this little thing [98:57] called AdSense. It's kind of for smaller [99:00] publishers and it's very like DIY self-s [99:04] serve. It's almost like a techie [99:06] utopian's version of how do you run ads [99:09] on websites. Whereas this ad exchange is [99:11] let's acknowledge all the complex [99:13] realities that exist in all these [99:15] business relationships, all these [99:16] purchasing decisions. the way Madison [99:18] Avenue has evolved from the Madman era [99:20] to, you know, this moment in time in the [99:22] early 2000s. [99:23] >> And let's essentially construct fat [99:26] pipes for money to flow [99:29] >> through all this. And what I mean by [99:31] that is direct integrations into ad [99:34] agencies financial systems and the ad [99:36] agencies control the budgets for all the [99:39] big brands and all the big dollars that [99:41] are flowing. [99:41] >> That's exactly right. So, if only Google [99:44] had a way of unlocking and now [99:46] participating in these deeply integrated [99:49] money flows, Google had a few other [99:51] problems. The way DoubleClick worked and [99:53] performed a lot of really fancy stuff [99:55] like frequency capping to make sure you [99:56] don't see the same ad 46 times is third [99:59] party cookies. Google was [100:01] philosophically opposed to using third [100:03] party cookies, so they couldn't do stuff [100:04] like that, but DoubleClick could. Google [100:07] didn't have a lot of these big sales [100:08] relationships since at the time again [100:10] they're very obsessed with self-s serve [100:13] web pages. Advertisers just log in and [100:15] upload and transact. So Google ends up [100:18] kind of locked out of the best ad [100:20] inventory. Advertisers on Google could [100:22] really only be placed on the long tale [100:24] of websites which advertisers were [100:26] willing to pay less to appear on those [100:28] websites. [100:29] >> Again, we're all in the AdSense part of [100:30] the world, not search ads. Yes, there's [100:33] all sorts of things that make them not [100:36] enterprisegrade here. So, Google decided [100:38] they'd like to buy DoubleClick. [100:40] >> Yes. Well, that's sort of the story out [100:42] there. The reality is, think back to how [100:46] you started and when I interjected and I [100:48] said, gosh, a lot of what Doubleclick is [100:50] doing, it really sounds a lot like what [100:52] Google is doing, right? Well, we were [100:55] talking to Tim Armstrong for research [100:58] for this episode. Tim of course was head [101:00] of sales at Google for many years and we [101:03] were asking Tim about Double Click and [101:04] he was like well [101:06] I was close with the Double Click guys [101:08] and I wanted to meet with them here in [101:11] kind of late 2006 early 2007 and just so [101:15] happened I was going to be in Seattle [101:17] for some stuff and I was emailing with [101:18] them and they were like uh oh you're in [101:21] Seattle actually we're in Seattle too [101:23] right now we can get together here. Tim [101:27] immediately sounds the alarm inside [101:30] Google to Eric and Larry. [101:33] >> They're in Seattle. Why are they in [101:34] Seattle? [101:34] >> This is a New York based company. There [101:36] is only one reason why the DoubleClick [101:39] guys are going to be in Seattle, and [101:42] that would be if Microsoft is going to [101:44] buy the company. [101:45] >> Yep. Now, back to everything we've been [101:48] talking about all episode. What does [101:50] Google absolutely not want to have [101:52] happen here? Well, one was Microsoft to [101:57] kneecap them by making changes to [101:58] Internet Explorer or Windows or whatnot. [102:01] They basically neutralized that through [102:02] the whole web app web 2.0 strategy. Now [102:06] the threat is, oh, Microsoft is finally [102:09] going to wake up and do what they should [102:10] have done 10 years ago and compete with [102:13] us, build their own search engine, [102:15] >> right? Be willing to be an adbased [102:18] business. Their DNA was, "Yeah, we'll do [102:21] some ad stuff and MSN kind of has to [102:23] because it's a media business, but [102:24] >> we sell software. [102:25] >> We sell software. That's what we do [102:26] primarily. And we would never trade our [102:29] ability to sell software to make money [102:31] on dirty ads." [102:32] >> Yep. Well, [102:33] >> Microsoft is realizing that for some set [102:37] of users, Google's actually making more [102:39] money on any given PC user than [102:42] Microsoft is. And they're not happy with [102:44] this. And say, "Fine, we at least just [102:46] need to be in that game, too." Yep. [102:49] >> So, the negotiations are kind of [102:51] happening with Microsoft and [102:52] DoubleClick. Tim told us this great [102:54] anecdote where he's invited to present [102:57] and he still thinks it's like a early [102:59] stage conversation in the negotiations [103:01] and somehow he gets sent to the wrong [103:04] floor. The person who is escorting him [103:06] into the Doubleclick building sent them [103:08] to a floor and they sort of freak out [103:10] when the door opens and they're like [103:11] trying to close the door like please go [103:12] to the other floor. And Tim is like [103:14] what's going on here? He steps out. He [103:16] runs down the hall and he sees a [103:18] conference room full of all the [103:20] Microsoft people and their accountants [103:22] and their lawyers and he's like, "Oh my [103:24] god, oh my god, you guys, you're you're [103:27] about to sign this deal with Microsoft. [103:28] You got to." So he gets them to hold off [103:31] so he can kick the tires, do his [103:34] diligence, submit a counter bid. This is [103:36] a crazy process that goes back and [103:38] forth. Yahoo gets involved. There's a [103:41] whole presentation series that happens [103:43] where Yahoo, Microsoft, AOL, and Google [103:46] are basically all getting the pitch and [103:48] DoubleClick is now. [103:50] >> I'm imagining they're all like in an [103:51] auditorium and Double Click is [103:52] presenting on stage. [103:53] >> Dude, there was a spreadsheet called [103:56] YMAG.XLS. [103:58] And the reason they've created it is [104:00] they're effectively trying to show in [104:02] each of these presentations, here's how [104:04] much incremental money you'll make if [104:06] you own DoubleClick and you tie it into [104:09] your existing ad system. And they're [104:11] tweaking the numbers slightly for each [104:13] one. So Google then submits their LOI [104:16] for $3.1 billion. [104:19] And it includes a clause where they [104:21] can't shop the deal around during this [104:23] diligence period. And so the whole [104:26] Google team goes to New York. They rent [104:28] out this big room at a hotel near Double [104:30] Click's offices. And I'm going to read [104:32] an excerpt from the book Yield. The [104:34] company's council, this is Double Click, [104:36] checked her Blackberry and held it up [104:37] for David Rosenblat to see. There was an [104:40] incoming message from Microsoft's [104:41] corporate development team. They were [104:43] willing to match the offer for [104:44] DoubleClick and the message included an [104:46] email from Steve Balmer saying that he [104:48] opened the door for a much higher offer. [104:50] Balmer wrote that if the offer match was [104:52] not acceptable, Doubleclick should [104:54] simply mark up the paper to meet its [104:56] needs and then sign it and Microsoft [104:58] would review and rapidly countersign to [105:00] close the deal with minimal negotiation [105:02] required. Without saying so, Balmer was [105:04] communicating, "Here's a blank check. [105:06] Tell me what closes the deal." [105:08] Ultimately, a week goes by and they're [105:12] in this period where they can't really [105:14] respond and they're supposed to just [105:15] kind of proceed with Google. And a day [105:17] before the LOI is set to expire, the [105:20] DoubleClick team gets an updated term [105:22] sheet from Google. The financial terms [105:24] of the deal are unchanged at 3.1 [105:26] billion. But now the deal includes what [105:28] they call a hell or high water clause, [105:30] which means that Google was committed to [105:31] closing the deal without any substantive [105:33] diligence or any other conditions. It's [105:35] just money in the bank. So no more [105:38] diligence. Doubleclick just signs it. [105:41] 3.1 billion. And the private equity firm [105:43] turns that 1 billion, which was levered. [105:46] It was something like 300 million of [105:47] equity and 700 million of debt into a [105:50] $3.1 billion sale to Google. Then it's [105:53] over. [105:54] >> Not bad work if you can get it. [105:56] >> Nope. [105:57] >> So this was huge for Google. Doubleclick [106:02] bringing it into Google did really help [106:04] with those fat money pipes as I was [106:07] saying of dollar flows from ad agencies. [106:10] >> Yep. But the biggest thing was [106:13] DoubleClick was the number one player in [106:15] the space. There was another company, [106:17] public company called Aquaniv that was [106:19] the number two player [106:20] >> which Microsoft then bought for twice as [106:22] much. They were like, "We really wish we [106:23] had gotten Double Click." And then [106:24] within months, [106:26] >> it was the next month. Right away, [106:28] Microsoft turned around and bought a [106:30] Quanivive for $6 billion. So twice the [106:33] price. But Microsoft getting the number [106:36] two player versus the number one player [106:38] slowed them down. We're heading right [106:40] into Microsoft search efforts with Yahoo [106:43] and then ultimately Bing and getting [106:45] into the advertising business worth [106:48] every penny to Google even for the sole [106:51] reason of keeping the premier number one [106:53] player in the display ad space out of [106:56] Microsoft's hands. Yep. Okay. Hey, the [106:58] one thing I will say here, David, is [107:00] unlike all those other Google products, [107:03] maps, Gmail, [107:06] YouTube, the organizing the world's [107:07] information. This is not organizing the [107:11] world's information and making [107:12] universally accessible. This is we're [107:14] running an ad business and we want to [107:17] expand the ad business and so we're [107:20] going to expand and protect our business [107:23] interest by buying this. And it's a [107:25] chess piece on the table that it being [107:27] in our hands versus other people's hands [107:29] is better. [107:30] >> That's exactly right. And there's ways [107:32] that it like systematically advantages [107:34] you to own the exchange when you also [107:35] own the network. And this is only [107:39] checking the box of strengthening our [107:40] business without checking any of the [107:42] other boxes. And you basically never [107:43] heard Google executives get up on a [107:46] stage where they're inspiring people [107:47] about the future of the company and talk [107:49] about basically anything Doubleclick is [107:51] doing. [107:51] >> Yes. Correct. And you know, even fast [107:55] forward today, unlike YouTube, it's not [107:57] like this has become a world dominating [108:00] thing, [108:00] >> right? If you're in the display ads [108:02] world or you're a publisher or this [108:04] feels like a huge deal. If you're [108:06] Google, let's just look at the numbers [108:08] today. Google in total in 2024 [108:13] made 350 billion of revenue. About 200 [108:18] billion of that is from Google search. [108:22] About 30 billion of that is from Google [108:26] network. This falls under Google [108:27] network. [108:28] >> Plenty of which existed before and would [108:31] have existed anyway in AdSense [108:33] regardless. [108:34] >> I don't know about that. I don't know [108:36] that Google would have become the [108:38] dominant player in display ads [108:42] >> absent double click [108:43] >> without buying double click. Yeah. I [108:45] don't think [108:45] >> Yeah. Yeah. But AdSense was probably [108:47] doing a billion plus in revenue at the [108:50] time. would have kept scaling. So all [108:52] that to say like in the context of [108:54] Google, this isn't a YouTube. [108:57] >> It just doesn't matter that much. Yeah. [108:59] 200 billion in revenue they make from [109:01] search where they get to keep 90ish% of [109:04] that, you know, after paying out traffic [109:06] acquisition costs. Google Network, they [109:09] pay out 70%. And they only made 30 [109:11] billion. So if you start thinking about [109:13] like gross profit, it's comparing [109:17] $9 billion to $175ish [109:20] billion. [109:21] >> Yep. [109:22] >> It's just not that consequential to the [109:24] story. [109:25] >> All right. So speaking of search, catch [109:27] us up on how the search business is [109:28] doing during these years and why [109:31] Microsoft finally said like, "Okay, [109:32] enough. We got to enter this business [109:34] ourselves." [109:35] >> Yeah. So, we've been talking about kind [109:37] of the sideshows like trying to add wind [109:39] to the sales of the web and search is [109:43] cranking on improvements to the core [109:44] product and revenue is growing up right [109:46] along with it. So, here's a little [109:47] timeline to catch us up 03 to08. They [109:50] start updating the index more often. So, [109:52] the index starts to feel not quite real [109:54] time, but it used to be that when you [109:56] would search, you would be getting [109:57] results that were indexed 3 months ago. [109:59] Now, the web is feeling a little bit [110:01] more uh [110:02] >> real timey, [110:03] >> recent when you're searching it. They [110:05] launched Google images, Google News, [110:07] Google Books, Google Scholar. They [110:09] launched Google Suggest, which is when [110:11] it starts autocompleting your searches. [110:14] And later they would launch something [110:15] called Google Instant, which was very [110:18] cool at the time. It's actually kind of [110:19] gone away now, where it would run a [110:21] completely new search based on every [110:23] character you typed and show you the [110:25] results page updated in real time with [110:27] each next keystroke, which was pretty [110:30] amazing. [110:30] >> I remember that being so cool when it [110:32] launched. [110:33] >> Yeah. In 2005 they incorporate your [110:37] search history into your results. So [110:39] this is when they start doing some [110:40] personalization stuff with logged in [110:41] users. They go from in ' 04 they had [110:44] three billion in revenue. 05 they have 6 [110:46] billion in revenue. So doubled even at [110:48] that scale. In 2007 they launched [110:52] universal search across web images video [110:55] you know whatever maps. They try to [110:57] deconstruct your query and understand [110:59] which of these things are you looking [111:01] for rather than they used to basically [111:03] build a completely separate search [111:04] engine for each media type and then [111:05] leave it up to you to decide which thing [111:07] to go search. That year when they [111:09] launched universal search they do 16.5 [111:11] billion in revenue. This 2007 year this [111:16] is when they become the largest seller [111:18] of advertisements in the world. Not just [111:21] digital ads ads. And digital ads would [111:25] not overtake traditional media until [111:27] 2018 as we talked about earlier. [111:29] >> Yeah, I was trying to square this. So I [111:32] guess that means that the market share [111:33] that Google has of digital ads is so [111:35] high massive that it's bigger than even [111:37] in the traditional space or the TV space [111:39] what any one player has. [111:41] >> Yes. So every year for the last 18 [111:44] years, Google has been the number one [111:46] seller of advertising of any kind in the [111:48] world. [111:50] >> Wow. [111:51] This I think helps you understand a [111:53] little bit what's at stake in the era of [111:55] AI. I mean this is literally the [111:57] trillion or five or 10 trillion dollar [111:58] question is can Google keep being the [112:02] number one seller of advertising in the [112:03] world even through this sea change. We [112:06] should do a whole episode on that [112:08] probably. Maybe we'll maybe for next [112:10] time. [112:11] >> But actually there's some great [112:12] corlaries with the mobile wave that [112:14] we're about to talk about. And then just [112:16] to pull forward a few more search [112:17] improvements that they would do later in [112:19] 2009, that's when they really do some [112:22] real- time indexing of the web. 2012 [112:24] they launched the knowledge graph. So [112:26] when you search about a basically a [112:28] thing with a Wikipedia page, you always [112:30] get the kind of snapshot view on the [112:32] right hand side of that entity. And so [112:36] all along the way they're tweaking the [112:37] algorithm in an attempt to reduce spam. [112:40] That's effectively the product changes [112:42] on the people side of things. They [112:45] really had solidified themselves as the [112:47] preeminent computer science research [112:50] company at this point. I mean, if you [112:52] were to refer in 2008 to like a really [112:55] smart programmer, you probably said, [112:57] "Oh, they, you know, they're like a [112:59] Google type engineer. They sort of took [113:01] the mantle from Microsoft and had not [113:04] yet relinquished it to Facebook or later [113:07] to the Stripe or OpenAI or Anthropic or [113:09] any of the sort of companies we would [113:10] talk about in the future as this like [113:12] dense concentration of the best [113:13] engineers." and they had pulled in a lot [113:15] of the people from the big research labs [113:18] that had been collapsing. So you had [113:20] Jeff Dean and Sanjay Gemma coming from [113:22] deck. David, we did Sanjay a total [113:24] disservice on the last episode. A lot of [113:26] the stuff that Jeff did and of course he [113:29] became sort of a Google executive. Jeff [113:30] and Sanjay peer programmed together. [113:33] >> Yes, there's an amazing New Yorker [113:35] article that was published long ago [113:37] about their friendship and career [113:38] partnership and everything that they [113:40] accomplished together. [113:41] >> Yeah, we'll link to it in the show [113:42] notes. And basically if you look at any [113:45] big research paper about giant Google [113:48] infrastructure stuff that was launched [113:50] from I don't know 2002ish [113:52] maybe even earlier through the 201s. [113:55] Jeff and Sanjay are either the two [113:57] authors or two of the five authors. I [113:59] mean it's amazing how much stuff these [114:00] two guys invented. They also got Bill [114:03] Corin and Rob Pike from Bell Labs. you [114:06] had Xerox Park and IBM's labs were sort [114:08] of losing prominence and Google's just [114:10] sucking in all this generational [114:13] heavyhitter computer science [114:16] architecture systems programmers from [114:18] all of those. And so I think that's sort [114:20] of how I would describe where a lot of [114:23] the technical breakthroughs are really [114:25] coming from or at least the culture of [114:27] technical breakthroughs. We talked about [114:29] these incredible products, incredible [114:31] innovations, [114:32] development of the whole concept of a [114:34] web application, but that was coming [114:37] from these people that were coming into [114:38] the company who were just like you say, [114:41] generational talents. Speaking of, that [114:44] was very convenient for a couple things [114:49] that they needed to start doing in 2008. [114:52] namely launching their own web browser [114:56] and then shortly thereafter launching [114:58] their own mobile operating system. [115:02] >> It's astonishing that they did both of [115:04] these things [115:05] >> in the same year. [115:06] >> And this isn't like, oh, I'm going to [115:07] start a browser the way that you can [115:09] start a browser today. I mean, the all [115:10] these AI companies are launching [115:11] browsers. [115:12] >> Yeah, they're using Chromium, [115:14] >> right? This is a giant engineering [115:17] undertaking. You need amazing [115:20] architects. I mean this is equivalent to [115:24] Dave Cutler doing Windows NT. I mean it [115:26] was earthshattering when Google launched [115:29] Chrome [115:30] >> or everything Jeff Dean and Sanjay did [115:32] in the early days of Google. [115:33] >> Yes. [115:34] >> Yeah. And when I say people launching [115:36] web browsers today are using Chromium, [115:37] Chromium of course is the open- source [115:39] version of Chrome. [115:40] >> Yeah. [115:41] >> That Google just gives away for free to [115:43] anybody. That didn't exist. They had to [115:44] build it. [115:46] So, in February of 2008, the shoe that [115:51] Google had been fearing would drop for [115:54] many, many years finally does drop, [115:58] which is Microsoft is officially going [116:00] to enter the search business. They make [116:03] a bid, Microsoft does, to buy Yahoo for [116:07] $44 billion. The giant has finally woken [116:11] up. Fortunately for Google, they get a [116:13] little bit of a reprieve because Jerry [116:15] Yang turns it down. [116:17] >> So dumb [116:18] >> in one of the worst corporate decisions [116:21] of all time because just two years [116:23] later, Bing, after Microsoft would [116:26] launch Bing the next year, in June of [116:28] 2009, Bing would take over powering [116:31] Yahoo search for a deal that paid Yahoo [116:35] $1 billion [116:36] versus the 44 that two years earlier [116:39] Microsoft was willing to pay for the [116:40] whole company. And then Yahoo would sell [116:42] itself to Verizon for [116:43] >> like three I think something like that. [116:45] >> Something like that. Yeah. Single digit [116:47] billions. [116:48] >> Yes. So Google though knew this day was [116:51] going to come eventually. And [116:53] fortunately by this time in 2008 Google [116:56] had its competitive response all ready [117:00] ready to go which was Chrome. [117:02] >> And they had actually been working on [117:04] improving the state of browsers for [117:07] years. [117:07] >> Oh yes they had. The story of Chrome [117:10] goes all the way back to 2001. Larry and [117:14] Sergey wanted to build a web browser in [117:16] 2001 for this very reason that we've [117:19] been talking about the whole episode. [117:20] All of Google rested on [117:22] >> I didn't realize that [117:23] >> Internet Explorer and also I mean it was [117:25] Larry and Sergey. Of course they wanted [117:27] to build a web browser. It's the most [117:28] Google thing. Why wouldn't we build our [117:30] own web browser, [117:31] >> right? [117:31] >> But it was Eric who said in 2001, "No, [117:35] we can't do this now. We can't poke the [117:38] bear right now. Like Google is too [117:41] young, too vulnerable. [117:43] >> Calls like this are why Larry and Sergey [117:45] brought in Eric. [117:47] >> Eric. Yes. The actual quote from Eric at [117:51] the time this is in in the Plex is, I [117:53] don't want to moon the giant [117:57] in 2001. [117:57] >> It's a very Eric Schmidt quote. [117:59] >> But that doesn't mean that Google isn't [118:01] preparing for this. Instead, what they [118:04] do is they decide that they are going to [118:07] become the primary major benefactor for [118:12] the new Mosilla Foundation and what [118:16] would become Firefox. Mosilla was the [118:20] nonprofit organization that was founded [118:23] and spun off from Netscape when AOL [118:26] bought Netscape. [118:27] >> Are they a funer? Are they actually just [118:28] like giving money? Well, I think at [118:31] first it probably was like with grants [118:33] like giving money because this is [118:35] strategically important for Google. But [118:36] then once Firefox actually gets released [118:40] by Mozilla and deployed out there, the [118:44] way Google starts supporting Mozilla and [118:46] Firefox is through paying traffic [118:50] acquisition costs to them to be the [118:51] default search in the Firefox browser. [118:55] Spoiler alert, like they do to Apple for [118:57] Safari today to the tune of like$20 [118:58] billion dollars a year. [119:00] >> Yes, [119:01] >> Firefox is where this all starts with [119:03] Mozilla. Actually, traffic acquisition [119:05] costs originated before Mozilla because [119:09] it's effectively the same thing that [119:10] they were doing with software vendors to [119:13] include Google Toolbar, [119:15] >> right? [119:15] >> And so I think the mechanism of payment [119:18] over time shifted to more of a rev [119:20] share. My understanding now is that they [119:22] share some of the revenue they generate [119:24] from queries that originate [119:27] >> searches that happen in the browser. [119:29] >> Exactly. Which is why it ends up being [119:31] kind of variable yeartoear. But yes, [119:34] Google has a long history of paying for [119:38] distribution of their search engine and [119:41] the new form that it is now taking is [119:43] the Mozilla Firefox browser. [119:45] >> Yep. Starting with toolbar. So, this [119:49] goes on for a couple years. [119:50] >> And by the way, I should say Google [119:52] becomes a giant contributor of source [119:54] code to Firefox. [119:56] >> Well, [119:56] >> oh, is that where you're going? [119:57] >> I'm going to get into this. So, couple [119:59] years, Google's paying Mozilla for [120:03] default search in Firefox, basically [120:05] funding Mozilla. After a little while, [120:09] Google decides that they're gonna hire [120:13] some of the key Firefox engineers at [120:16] Mozilla to come and work at Google [120:18] directly. But they position this as like [120:21] essentially this is the same thing. We [120:22] are still funding Mozilla and Firefox [120:25] just like you know you're Mozilla. You [120:27] can't give these employees, these [120:29] engineers stock options like you're a [120:31] nonprofit. How about instead they do the [120:33] same thing that they're doing which is [120:35] working on Firefox. They'll just come [120:37] and work here at Google and we'll pay [120:40] their salaries and they'll get Google [120:42] stock options. Otherwise, they're going [120:43] to get poached, you know, by all these [120:45] tech companies, etc., etc. [120:47] >> Fascinating. [120:47] >> You can see how this makes sense. This [120:50] team that comes over from Mozilla into [120:52] Google becomes the core of a new quote [120:56] product client group within Google. [120:59] Meaning of this being products on [121:01] clients, i.e. installed applications on [121:05] PCs, not the web apps that the rest of [121:08] new Google is doing. [121:10] >> And the leader of this group, Google [121:14] hires from McKenzie in 2004, [121:18] Sundar Fai. [121:19] >> Oh, I did not realize that that's where [121:22] Sundar came from. [121:23] >> Yep. Well, again, all of this is very [121:25] strategic because if you did someday [121:28] want to build your own web browser, [121:30] >> now you've got the bench of talent. [121:32] They're employees. [121:33] >> So, there's a quote from Mer Schmidt in [121:34] in the Plex. This was very clever on [121:37] Larry and Sergey's part because of [121:39] course these people doing Firefox are [121:41] perfectly capable of going and doing [121:43] another great web browser. [121:46] So this group is sitting there within [121:48] Google for a couple years almost like a [121:51] latent sleeper cell within Google. [121:54] They're just ready to activate as soon [121:56] as the Microsoft threat becomes real. [121:58] >> And the way I heard it was a lot of [122:01] people are working on Google Gears, [122:02] which is this browser extension that [122:04] allows for offline functionality. [122:07] They've built the Google web toolkit to [122:09] make web application development even [122:11] more advanced, even more sophisticated. [122:13] And at some point they kind of lost [122:15] faith that Firefox was going to keep the [122:19] pace and that Firefox was going to stay [122:22] as high quality of a browser as they [122:24] needed it. And of course they had some [122:26] divergent technical ideas like different [122:28] architecture ideas for how a browser [122:30] should function that we'll talk about. I [122:32] think all of these things are true. [122:34] However, [122:36] having your own browser when Microsoft [122:39] does launch Bing hugely, hugely hugely [122:43] important. Imagine if 90% of Google [122:48] happened on Internet Explorer and all of [122:51] a sudden Microsoft launches Bing, [122:53] >> right? There's no amount of money [122:54] suddenly that you could pay Microsoft [122:56] where they would keep you as the default [122:58] search engine because they just want all [123:00] the traffic to go to Bing because they [123:01] now have a great way to monetize. [123:02] >> Absolutely not. Bing default search [123:05] engine done, right? [123:06] >> And the thing that of course Microsoft [123:08] would fail to realize with Bing is you [123:10] can't be second place in search, [123:12] >> right? [123:13] >> The most liquid auction will always win [123:18] and Google has already run away with the [123:20] search ads auction liquidity and so [123:22] traffic on Google searches will forever [123:25] be worth more than traffic on the second [123:29] place browser. [123:30] >> Sure. doesn't mean that that battle [123:34] wouldn't be hugely damaging to Google if [123:36] they didn't have their own web browser. [123:39] So 2006 they finally decide okay it's [123:42] time to start work on Chrome. [123:46] >> Yep. [123:47] >> And also it's clear web apps, [123:50] JavaScript, [123:51] Ajax very important thing and Internet [123:55] Explorer isn't keeping up with the [123:57] technology. [123:58] >> Yeah. So, there's two killer features, [124:02] arguably maybe three, that they're going [124:05] to bake into the Google browser. [124:07] >> Oh, I've got six. [124:08] >> Oh. [124:09] >> So, I'm curious which ones you don't [124:10] think are important. [124:11] >> Okay, I'll go through my three and then [124:13] I'll see what else you have to add. [124:14] >> Okay, great. [124:15] >> Number one, most important, it is going [124:18] to have a super fast, super modern, [124:21] super performant JavaScript virtual [124:24] machine called V8. [124:26] >> Yep. that is going to run big web apps [124:30] fast and stably. [124:32] >> We're the Ajax company, baby. We got to [124:34] speed up. [124:34] >> We are the Ajax company. [124:37] That's right. Two, web apps crashed a [124:41] lot back in the day. They don't so much [124:43] anymore, but they used to crash a lot. [124:45] >> And Larry has this quote when they're [124:47] deciding that they should roll out [124:49] Chrome, and he kind of explains, "We [124:50] have found the web-based service [124:52] delivery model to have significant [124:54] advantages." You don't say. [124:57] But it also comes with its own set of [124:58] challenges primarily related to web [125:00] browsers which can be slow, unreliable [125:03] and unable to function offline. [125:05] >> There you go. And so before Chrome, this [125:08] is impossible to remember now. But if [125:11] you had a tab or a window open and [125:14] running a web app and that web app [125:16] crashed, it took down your whole [125:17] browser. [125:18] >> Yep. [125:18] >> Everything that you had open gone. [125:21] >> Tabs were not their own processes. [125:23] >> Nope. So each tab is going to be a [125:25] separate process on your machine. So if [125:28] the web app running in one tab crashes, [125:31] all it takes down is that one tab. And [125:33] it made sense that before this they [125:34] weren't their own process because one, [125:36] tabs were kind of a new thing. But two, [125:40] web applications were websites. The [125:42] notion of web applications was only [125:44] really fourish years old. [125:46] >> So those are my big two. I suspect one [125:48] of yours is WebKit. I'm not including [125:51] WebKit here because that was an Apple [125:53] innovation. Yeah, [125:54] >> that they borrowed. I'll let you talk [125:55] about that in a sec. [125:56] >> I don't have anything more to say on [125:57] that. It was the best rendering engine. [125:59] >> Yes. So, let's say that's three. And [126:01] then my sort of three and a half is the [126:04] design. So, of course, the web browser [126:06] ultimately comes to be called Chrome, [126:08] which is ironic. Chrome is a reference [126:11] to all the stuff in a web browser, the [126:14] toolbarss, the navbar, etc. that take up [126:16] space around the content. The idea with [126:19] Chrome is and the Google web browser is [126:21] going to be minimal Chrome as little as [126:23] possible. It's just about the content. [126:25] Let the web and the web apps shine. [126:28] >> Yes. Okay. When you said UI, I thought [126:31] you were going to say this. My fifth is [126:33] the omni box. [126:34] >> Ah, yes. [126:35] >> Originally, there was just the URL bar [126:37] and then when search became the killer [126:39] app of the web, there's a second little [126:41] input box that is for search on the [126:43] right side. So, we had that awkward [126:45] teenage years where browsers had the URL [126:48] bar on the left and then the search on [126:50] the right. And it's kind of clean to [126:52] think about it on its own because now [126:54] that we understand that that is [126:56] sponsored that I think for the longest [126:58] time that was not in the public psyche [127:00] that whatever search engine appeared in [127:01] that box in the righth hand corner was [127:03] paying for that placement. That was sort [127:05] of nice because you type in the URL bar [127:07] and that's your organic typing. And then [127:09] the other one is your uh I'm willing to [127:11] give a kickback to Google probably, but [127:14] it could also be Bing, could be Yahoo, [127:15] could be whoever. Google [127:18] correctly from a user experience [127:20] perspective, but also just think about [127:21] their core business model is like the [127:24] right design for web browsers is that if [127:29] you don't type in a URL, it should just [127:31] search [127:31] >> just one bar. Why have two bars? We [127:34] imagine that generating a whole bunch [127:36] more page views on search results pages [127:38] and a whole bunch more opportunities for [127:40] our advertisers to reach your eyeballs. [127:42] But I will say they were also correct [127:43] from a user experience perspective. The [127:45] fact that URLs [127:47] ever leaked to the public is a mistake [127:51] that is letting an implementation detail [127:53] of the technology. [127:54] >> It's an accident of history that [127:56] consumers type [127:58] >> https [128:00] slash. Are you kidding me? Consumers [128:02] never should have known the phrase HTTP. [128:05] >> They should just type New York Times. [128:07] >> Yes. Which AOL tried to do. AOL [128:09] keywords. [128:10] >> Yeah. [128:10] >> So, this is effectively leaning into [128:11] that idea. You can use this box for [128:14] typing in URLs, but like really what you [128:16] use this box for is kicking off the [128:18] Google search. So, brilliantly aligned [128:19] with our business model. [128:20] >> All right. So, that's what I got. What [128:21] else do you have that's not on my list? [128:22] >> That's five. And then lastly, [128:24] sandboxing. Each tab is a sandboxed [128:28] environment. This prevented a ton of [128:30] malware. This was like a big [128:32] breakthrough in computer security where [128:34] anything that was operating in that tab [128:37] was in its own sandbox and couldn't be [128:39] accessed maliciously. [128:41] >> Yeah, that's I guess gosh remembering [128:44] back I mean before Chrome and modern web [128:46] browsers [128:47] browsing the web was like a security [128:49] threat to your PC, right? [128:50] >> Yep, that's exactly right. [128:52] >> Great point. Okay, so they start work [128:54] officially on this in 2006. [128:57] They launch it in early September 2008, [129:00] like a week before Lehman goes down. [129:03] This is wild. [129:04] >> I remember that because I remember [129:05] sitting in North Carolina at my Cisco [129:07] office and I remember reading the Chrome [129:10] comic, which I actually just read a [129:12] couple nights ago for this episode. This [129:13] like amazing web comic [129:15] >> at the same desk where I read the news [129:18] about the great financial crisis and the [129:19] world falling apart and Leman Brothers [129:21] collapsing. [129:21] >> Wow. So they launch it early September [129:25] 2008 with the way this is so Google. The [129:28] way they decide to launch it is they [129:30] hire the famous comic artist Scott Mloud [129:34] to illustrate a digital comic book as [129:39] like a introduction to Chrome explaining [129:41] what it is and like a user manual in a [129:43] comic book form. And it's written for [129:45] this weird halfuser who's like kind of [129:49] technical, but you don't need to be a [129:51] programmer necessarily. It's written [129:53] sort of for the tech enthusiast who can [129:56] understand process independence, [129:57] understand sandboxing, understand V8 and [130:00] the JavaScript speed up, but it's not [130:02] written for the general public. [130:03] >> Yep. But that was exactly the right seed [130:06] crystal user base to get Chrome [130:08] >> Yeah. [130:09] >> into. [130:09] >> Yeah. It's written for the slash dot [130:11] reader. The kind of people who are going [130:13] to go home for Thanksgiving in a couple [130:14] months and install it on all their famil [130:17] family's computers and say you need to [130:19] stop using Internet Explorer right now [130:22] for all of these reasons, Ben, that you [130:24] just listed. Probably security being [130:25] number one amongst them. This is [130:27] actually was me back in the day. Like [130:29] I'm going to go home. I'm going to [130:31] install Chrome on my parents' computers [130:33] so that they don't get hacked and lose [130:35] their financial information, etc. [130:37] >> Yep. Within 18 months, they got 40 [130:39] million users. Then let's see they [130:42] launched it in 2008. By 2010 they had 70 [130:45] million users. Then 2012 they had 200 [130:48] million users. Actually what happened is [130:51] it destroyed Firefox's market share. I [130:55] think the launch of Chrome and the peak [130:56] of Firefox are right around the same [130:58] time. And then after that then it really [131:00] started eating away at Internet [131:02] Explorer's market share. And today aside [131:04] from mostly iPhones but Apple devices [131:07] running Safari it is the browser. I mean [131:10] to say it worked is like the [131:13] understatement of the century. I mean it [131:16] totally liberates Google from Internet [131:19] Explorer and Microsoft. When Chrome [131:21] launched in 2008, Internet Explorer had [131:24] almost 70% market share of browsers and [131:27] Firefox had most of the rest. 2 years [131:31] later, like you said, Chrome had passed [131:33] 100 million users. By 2012, so four [131:36] years after launch, Chrome and Internet [131:39] Explorer are now tied for market share [131:42] with about 30% each. So, Internet [131:44] Explorer has gone from 70% down to 30%. [131:48] And this is both Chrome on the desktop [131:50] side, but you're now also well into the [131:52] rise of mobile. And so, Apple's mobile [131:55] Safari is now becoming huge. And [131:57] Google's Android that we're going to [131:59] talk about in a sec is becoming huge. [132:01] Two years after that, in 2014, [132:05] Chrome is now the clear leader with 40% [132:07] market share. Internet Explorer is down [132:10] to 15%. [132:12] So, [132:13] >> it's over. [132:14] >> It's over. And Internet Explorer is [132:16] basically dead at this point. And today, [132:18] it truly is dead. [132:19] >> 2013, 14. [132:21] >> Yeah, 2013, 14. Today, it's not even [132:24] close. Chrome has almost 70% market [132:27] share according to Cloudflare [132:29] >> including iPhones which all run Safari [132:32] default [132:32] >> right Safari in aggregate across mobile [132:36] and desktop of mobile is by far the [132:38] biggest share of Safari market share is [132:41] about 20%. So Chrome has 70%. Safari has [132:45] 20%. There's 10% left. You know I think [132:48] Microsoft I don't know has a couple [132:50] singledigit percentage points. Talk [132:52] about flipping the tables. Chrome was [132:56] massive. [132:57] >> And it was just better. It was so much [132:59] better. And it really kicked off this [133:02] amazing era for the web between Apple [133:05] needing to then sort of play catchup and [133:07] leaprog. In a lot of years, it was [133:09] actually faster than Chrome and they [133:10] would go back and forth. And it really [133:12] spurred Apple, who was already a steward [133:15] of WebKit. And they sort of had their [133:18] own competitive response to Microsoft [133:20] after Steve Jobs hated the fact that he [133:21] had to keep shipping IE as his best [133:24] option on Mac [133:25] >> cuz that was part of the Microsoft Apple [133:27] deal, right? When Microsoft saved Apple [133:29] with the investment was Internet [133:31] Explorer will become default, right, [133:32] >> on the Mac. [133:33] >> So Safari was created for that. But over [133:35] the years, you know, Apple's incentives, [133:38] especially post iPhone, were not to make [133:41] it so web apps could be great. Apple's [133:43] incentives were to make it so native [133:45] mobile and desktop applications could be [133:48] great. And so Google really pushing the [133:50] envelope in the web's capabilities and [133:52] what a modern browser could do forced [133:55] this like good for the world race [133:58] between Apple and Google to both make [134:00] better browsers. I don't think it is an [134:03] exaggeration to say that Chrome kept the [134:05] web alive. [134:07] >> Yeah. [134:08] >> As a viable platform for applications. [134:11] >> Yep. I mean, Microsoft certainly didn't [134:13] have an incentive to do it in the [134:14] business model they were in at the time. [134:15] And Apple doesn't. [134:17] >> They had every incentive not to, [134:19] >> right? [134:20] >> I mean, who in the world least wants the [134:22] web to be a viable application platform? [134:24] Microsoft, [134:25] >> right? At that point in time, Apple now, [134:27] ironically. So there's one more amazing [134:31] delicious part of the Chrome story which [134:35] is do you remember the Google Chrome [134:37] frame? Google Chrome frame was a plugin [134:43] for Internet Explorer that replaced [134:47] IE's JavaScript engine. [134:50] >> Oh wow. and pulled in the Chrome V8 [134:54] JavaScript engine and I think also [134:56] WebKit [134:58] >> for land. [134:59] >> And so for all the corporate users [135:02] >> who couldn't install a new app [135:04] >> of America and the world who were stuck [135:06] with Internet Explorer, this is the only [135:08] reason that IE hung on to market share [135:11] for so long was just [135:12] >> right lockown PCs. For all those poor [135:14] souls, Google was there for you with the [135:18] Google Chrome frame plugin that let you [135:22] run Chrome quality web apps within [135:26] Internet Explorer. Amazing. [135:28] >> All right, so I have a question for you [135:29] on Chrome before we finish the story. [135:31] Why make Chromium open source? The [135:35] answer that I've read to that is [135:40] mostly about the Google culture and [135:44] trying not to be like too evil about it. [135:47] >> Yeah, I would buy that. [135:48] >> Yeah, [135:49] >> they're like an open-source company. [135:51] Like it's in their bones to contribute [135:53] to open source. There's a thin business [135:56] reason I can think of of why they would [135:59] want to make it open source. Well, the [136:01] reason I can think of is like it doesn't [136:02] need to be closed source. They make [136:04] their money from searches, [136:05] >> right? That's the like it can't hurt. [136:09] Here's the way it could help. At first, [136:11] I was thinking, well, wait, Google wants [136:13] to own as many of the browsers that its [136:18] searches originate from as they can, so [136:20] they don't have to pay out distribution [136:22] costs in the form of traffic [136:23] acquisition. So if someone takes [136:26] Chromium and then builds a better [136:27] browser, it's kind of bad because now [136:30] you have to pay that browser maker. But [136:33] in practice, as long as it's not [136:35] Microsoft and as long as it's more [136:37] fragmented, that's probably a trade [136:39] they're perfectly happy with. If they [136:41] need to go and split some revshare and [136:43] pay, you know, someone who makes uh some [136:46] variant of Chrome based on Chromium, and [136:48] that gets really big and it gets 30% of [136:49] the market, great. Google's deated [136:52] because they it's not Microsoft. I mean, [136:54] look, except for traffic acquisition [136:56] cost, Google's incremental gross margin [136:58] on search revenue is like 98% or [137:01] something like that, they're still going [137:03] to be an 87% gross margin business. [137:05] >> Cuz the whole point of this was to [137:06] prevent a existential risk. And so if [137:10] they have to do some light rev sharing, [137:13] even in their worst case scenario where [137:14] someone builds a successful thing based [137:16] on their open source project, [137:18] >> oh, so horrible. We go down to 87% gross [137:21] margin. [137:21] >> Right. Right. It's still perfectly [137:23] acceptable, which actually may be the [137:26] way it plays out if any of these new AI [137:28] browsers work out. [137:29] >> Yeah, [137:30] >> maybe. [137:31] >> Well, not sure any of these AI browsers [137:32] would be willing to let Google to pay [137:34] them to be the default. [137:35] >> Big chess game to consider there. [137:37] >> Yes, that is the one catch is the owner [137:40] of the browser still has to be willing [137:42] to accept the payment from Google. Yes. [137:44] I have one analogy for all of this that [137:48] is a little far a field, but I think is [137:51] actually the right way to think about [137:52] this. [137:52] >> Okay. Lay it on me. [137:54] >> Walt Disney creates Disneyland. Goes [137:58] great. Uh very handcrafted, curated, [138:03] aurorriven thing, but ultimately he has [138:06] to play within the rules of things like [138:08] city government. [138:09] >> Okay. they go big and get a huge plot of [138:13] land in Florida to build Walt Disney [138:15] World. It would be nice if we controlled [138:19] our underlying foundation a little bit [138:20] more. So they build their own government [138:24] district around the park, [138:26] >> right? [138:27] >> And they say we make the rules here. [138:32] >> It's not totally dissimilar. [138:34] >> Yep. I've been reflecting a little bit [138:36] on like why basically from 1998 onward [138:40] Google's biggest threat was Microsoft [138:42] and not because of Bing, not because of [138:44] building advertising, because of this [138:45] kind of destabilizing thing. There's [138:47] sort of a fine point on it which is that [138:50] spiritually Microsoft was the platform [138:53] of the PC era. And with this platform [138:56] shift, it would be like very convenient [138:57] to just be like, oh, Google's the [139:00] platform of the web era. But even though [139:04] Google is the platform company of the [139:07] web era, they aren't necessarily the [139:10] ones building the platform, [139:13] >> right? Yeah. Yeah. They still existed at [139:15] Microsoft's pleasure. [139:16] >> And no one owns the web as a platform. [139:20] So there's this kind of funky thing [139:21] where like Microsoft built and owned [139:23] Windows and then dominated the PC era [139:26] because of that. Google operates a [139:29] search engine that generates advertising [139:31] revenue. They don't charge anyone for [139:33] anything. They benefit from the web's [139:34] growth. And so they're doing this like [139:36] strange indirection. [139:38] >> Yeah. It's an ecosystem building [139:39] exercise. [139:40] >> Exactly. They're trying to build the [139:41] ecosystem. They're trying to be the [139:42] steward of the open web as a platform. [139:45] And they like put their finger on the [139:47] scale where they need to and take a [139:48] little bit more control and ownership [139:50] like with Chrome or like with some of [139:51] these standards bodies to push the web [139:54] forward to make sure that the place [139:56] where they live, their neighborhood, the [139:58] web, is in good shape. But it's not [140:00] their platform in the way that it was [140:02] Microsoft's platform in that era. [140:04] >> All right. Maybe the Disney World [140:05] analogy is uh even better than we gave [140:08] it credit for a minute ago. [140:09] >> That's sort of I don't know. It's a [140:10] little bit loose, but that's what I've [140:11] been kicking around. [140:12] >> I like it. I like it. Well, no doubt [140:16] Chrome was a huge success. Hell, Sundar [140:19] becomes the CEO of the company. No [140:21] better sign of how successful it was [140:23] than that, [140:23] >> right? It shored up their future. We [140:25] could do a classic acquired what would [140:27] have happened otherwise just for a [140:28] minute or two here. What if Google [140:31] didn't launch Chrome? Let's say Bing [140:34] launches in 2009 and there is no Chrome [140:36] and Microsoft still has 70% share and [140:38] will for a while. [140:39] >> Yep. [140:40] >> And they make Bing the default. Let's [140:42] see. Mobile would get big 3 four years [140:44] later in the 2012ish time frame. [140:47] >> Yep. Still small. So they basically [140:49] would have like four years of 70% of [140:53] people using browsers on any devices [140:56] being defaulted to Bing. [140:58] >> Now obviously a lot of people would [141:00] still want Google. They were used to it. [141:01] They'd switch back. But like I don't [141:02] know, man. Defaults are powerful. [141:04] >> Defaults are powerful. I think you're [141:06] right. [141:07] >> This is why Google pays Apple 20 billion [141:09] a year. [141:10] >> Yeah. To your point, maybe without [141:13] Chrome, Bing would have been a serious [141:15] competitor to Google. There is no more [141:17] important distribution point for search [141:20] than the web browser. [141:21] >> It is the way to monetize a browser. [141:23] Basically, the single way to monetize a [141:25] browser. [141:26] >> Yep. [141:27] >> Which, let's make this relevant to today [141:29] and stop dancing. If the DOJ's ruling is [141:32] that Google has to divest Chrome, there [141:34] is one way that Chrome is a business and [141:37] that is getting paid by Google to drive [141:40] traffic to Google as a search engine. [141:43] It's the only way to operate a business [141:45] of a browser. And like maybe in the AI [141:47] era, it's the AI company having it drive [141:50] traffic, but it's the same exact thing. [141:52] And so, one of two things has to be [141:54] true. Google owns Chrome or someone else [141:56] owns Chrome and then Google pays them. [141:58] But the thing definitely can't make [142:00] illegal or I don't really understand [142:02] what the goal is if you make it illegal [142:04] is if you say they both can't own Chrome [142:06] and they can't pay web browsers to drive [142:09] traffic. Chrome has no potential of [142:12] being a business if that's the case. [142:13] >> Yep. So, Chrome, huge win. In fact, it's [142:17] so much of a win that after a couple [142:20] years, Google starts thinking, well, [142:23] gosh, maybe we should build Chrome into [142:25] an operating system in and of itself. [142:27] Let's go attack Windows. Let's take it [142:29] to Microsoft where it really hurts. [142:31] Chrome OS, you know, but it became [142:34] successful in schools and education. [142:35] >> Yeah, Chromebooks. Yeah, Chromebooks [142:37] have major market share there, but it's [142:40] not a major player in the overall PC [142:42] operating system market. It is wild how [142:45] much the PC computer operating system [142:50] market is still dominated by Windows. [142:52] That has never changed. You and I live [142:54] in this world where like everybody uses [142:56] a Mac. Mac has like 15% market share. [142:59] >> Yeah, [142:59] >> Windows has like 70% market share of [143:02] computer operating systems. [143:05] Well, speaking of operating systems, [143:08] >> it's time [143:08] >> I think it's probably time to talk about [143:10] Google's big one. [143:13] >> Yes. [143:14] >> Which is actually the biggest operating [143:16] system in the world. [143:18] >> Over three billion active Android [143:19] devices now. [143:21] >> Totally freaking wild. That they bought [143:24] for $50 million. [143:26] >> Well, that's a red herring. They've [143:27] invested so much more. But just to make [143:29] the point, it is hit after hit after [143:31] hit. These things are not predicated on [143:37] Google's distribution. [143:39] If you're a company that launches a new [143:43] widget and you can just distribute it [143:44] with your old widget, it's not that [143:46] impressive when your new widget gets [143:48] dominance. But Google Chrome, I mean, [143:52] they could do a little thing and they [143:53] did push it on Google search pages, but [143:56] they managed to get a lot of [143:58] distribution just by being a great [144:00] product on the market with viral [144:02] adoption that everyone told their [144:04] friends to use. And it was the David [144:05] Rosenthal's going home to Thanksgiving [144:07] that were sort of like the seat of it. [144:10] And then within three or four years, [144:11] they just ran the table. And it's not [144:13] just Chrome. Gmail was that way. Google [144:16] Docs and spreadsheets were that way. I [144:18] mean that way it's everything everything [144:20] >> all these things are [144:21] >> independent great products that became [144:25] dominant on their own merits just like [144:28] Google search did. [144:29] >> Yes, I 100% agree and also helped by the [144:33] fact that they were all free. [144:35] >> Yes. [144:37] Yes. Fair. And massively subsidized at [144:39] least in the early years before they [144:41] were able to be businesses on their own [144:42] by the uh the old money printing machine [144:44] in the basement of Google. Good old [144:47] Uncle Edwards. [144:48] >> Yes. [144:49] >> But before we tell the [144:50] >> Before the Android story, now is a great [144:53] time to thank one of our favorite [144:55] companies, Verscell. [144:57] >> Yes, we have talked throughout the [144:58] season about Verscell has become the [145:01] infrastructure backbone for modern web [145:03] and AI development. Highly relevant to [145:06] this episode, powering companies like [145:07] PayPal, Ramp, Under Armour, Notion, [145:10] Runway, Cursor, and many more. Today [145:12] though, we want to spotlight V0ero, [145:14] which is Verscell's AI app builder that [145:17] goes one step further and programs, [145:19] designs, iterates, and deploys full [145:21] stack web applications entirely for you. [145:25] Vzero is a chatbot that looks like [145:27] Claude or ChatGpt or Gemini, except that [145:29] when you give it a prompt, it will go [145:31] build an entire fully functional website [145:34] or application entirely for you. No [145:36] engineering or web design skills are [145:38] required. You don't need to look at a [145:39] single line of code or mock up a [145:41] wireframe. So a marketer can stand up a [145:43] product landing page or a small business [145:45] can generate a homepage in a contact [145:47] form or a creator can spin up an [145:49] independent content hub. Anything. [145:51] >> Versel loves to paraphrase the famous [145:53] line from Pixar's Ratatouille that quote [145:56] everybody can cook which is an [145:58] especially fun Easter egg for us since [145:59] Pixar was our very first acquired [146:01] episode. And the numbers behind Vzero [146:03] are wild. It now has 4 million users. So [146:07] PMS, marketers, creators, founders, [146:10] teachers, students, all building real [146:12] productionready applications on this [146:14] thing. [146:14] >> And speaking of production, every single [146:16] VZero app you generate can be deployed [146:18] to production instantly with Versell [146:20] because Vzero itself entirely runs on [146:22] Verscell's platform. So you go from [146:24] prompt to full stack deployment with [146:26] zero setup using the same secure, [146:28] scalable, automatic infrastructure that [146:30] powers sites like Ramp, Notion, and [146:32] Under Armour. It's the perfect example [146:34] of Verscell being customer zero for [146:36] their own products. They're using their [146:38] own AI cloud to power their own AI [146:40] products. So, if you've got an idea that [146:42] you want to launch, whether you're a [146:43] seasoned developer or someone who's [146:45] never written a line of code, go to [146:47] versel.com/acquired. [146:49] That's verel.com/acquired [146:52] and try it out. Build something real and [146:54] just tell them that Ben and David sent [146:56] you. [146:57] >> All right, [146:59] Android. [147:01] So Google's office spaces [147:05] are legendary. The first one of course [147:08] being Susan Wajiski's garage in Mountain [147:11] View company's first office. And then [147:14] today the Google Plex, the old SGI [147:18] Silicon Graphics campus in Mountain [147:20] View. In between [147:23] Google had another office [147:27] >> for a couple years in downtown PaloAlto. [147:31] at 165 University Avenue would also [147:36] later be the office that PayPal was [147:39] started in. [147:40] >> Oh, really? [147:40] >> Very lucky building. Yeah. August 1999 [147:45] when Google moved out of that office. Do [147:48] you know who moved in [147:50] >> based on the direction this is going? Is [147:53] it danger? [147:54] >> Yes, it is. [147:55] >> Yes, [147:55] >> danger. [147:58] the company started by Andy Rubin and [148:03] Andy of course had been an engineer at [148:05] Apple and then left Apple with a group [148:09] of uh rebels I don't know were they [148:11] rebels that went to go start General [148:13] Magic. General Magic of course legendary [148:16] failed startup in Silicon Valley in the [148:19] early 90s basically was trying to create [148:22] the iPhone just 15 years too early. [148:26] >> Yep. after General Magic, after it falls [148:28] apart, he starts Danger. Now, Andy's [148:31] initial idea for Danger was uh he wanted [148:35] to make a wireless version of the QC cat [148:39] scanner. [148:40] >> What is a QCAT scanner? [148:41] >> This was a device that plugged into your [148:43] computer that looked like a cat, but it [148:46] scanned barcodes. And so Andy's idea [148:48] was, okay, well, man, all this general [148:50] magic stuff we were trying to do, that [148:52] was too far ahead. What if we think [148:53] simpler and we just make a wireless [148:55] version of this to scan barcodes? Okay, [148:58] not a big idea. His first employee at [149:01] Danger, a guy named Hiroshi Lockheimer, [149:05] convinces him that hey, actually, you [149:07] know, a couple years have gone by. Maybe [149:09] we should revisit this General Magic [149:11] stuff. [149:11] >> Wait, Hiroshi was with him at Danger. [149:14] >> He was the first employee. [149:16] >> I did not know that. [149:17] >> Yep. Yes, he was. [149:18] >> I mean, he of course is instrumental in [149:20] the Android story later. I did not [149:21] realize the two of them were at danger [149:23] together too. [149:24] >> I spoke to Hiroshi in research. He told [149:26] me these stories. Great great guy. [149:28] Hiroshi led Android and Chrome at Google [149:32] for many many years and would be the [149:34] authority on this. So Hiroshi's like, [149:37] "Hey, hey, I maybe let's revisit this [149:40] General Magic stuff." And that led to [149:42] Danger building the Sidekick and [149:45] launching it in partnership with [149:47] T-Mobile. This thing was amazing. [149:48] >> That thing was so sick. I was jealous of [149:50] all my friends that had one. [149:51] >> It was a messaging focused sort of rich [149:55] application [149:56] cell phone. I think it was along with [149:59] Blackberry's [150:01] the first vision of a cell phone where [150:03] the primary thing you do on it is not [150:05] talk to somebody [150:06] >> is messaging. [150:07] >> These things were freaking awesome. They [150:09] were really big with celebrities. I [150:10] think it was a plot of an entourage [150:12] episode at some point in time. [150:14] >> So they end up selling this company to [150:17] Microsoft, right? [150:18] >> Yes. [150:19] Microsoft does end up acquiring the [150:21] company but not until 2008 which is the [150:24] same year that Android launches. Andy [150:26] actually had left Danger in 2003 and [150:30] started a new company Android [150:33] >> which in the earliest days it was kind [150:35] of like an opensource competitor to [150:38] effectively Blackberry software. [150:41] >> Yes. In its earliest earliest days, the [150:44] first version of Android the company, [150:48] remember I was talking about point [150:49] andoot cameras and digital cameras back [150:51] in the YouTube section. [150:52] >> Yeah. [150:53] >> Was actually to build a cross-platform [150:57] open-source operating system for [150:59] point-and-oot digital cameras. [151:01] >> Oh, wow. [151:02] >> Yeah. That was Andy's vision was like, [151:03] "Oh, hey, these point- and-oot devices [151:05] like hundreds of millions of consumers [151:06] have them now. What if there were a [151:08] powerful operating system? Could that be [151:12] a Trojan horse to get an operating [151:14] system? You could sort of imagine it, [151:17] >> right? If cameras became phones instead [151:20] of phones becoming cameras, then yes. [151:22] >> Yep. Exactly. But pretty quickly it does [151:25] become clear that phones are going to [151:27] become cameras. So, good thing is [151:30] though, the software they're writing [151:32] still works just as well on phones. So [151:34] Andy pivots the company and has the [151:36] delivery vector shift from cameras to [151:40] smartphones. And at the time the [151:42] smartphone market such that it existed [151:44] and it did exist. [151:46] >> Blackberry, Windows Mobile. [151:48] >> Well, yeah. So here were the players. [151:50] Basically phone companies either were [151:54] full stack like Apple and the iPhone is [151:56] today where they made the phone and the [151:58] operating system. So that was Nokia. And [152:01] then the big player in the smartphone [152:02] market at least was Blackberry. Made [152:05] their own software, made their own [152:06] devices, huge in the enterprise market. [152:09] Or you did have OEMs, device makers who [152:13] made devices and then they bought an [152:14] operating system either from Palm which [152:18] made their own devices but then also [152:19] started selling the operating system to [152:22] other vendors or the big player was [152:24] Microsoft with Windows Mobile. And this [152:26] was a license model as we talked about [152:28] in our Microsoft part two. This was you [152:32] pay Microsoft singledigit dollars and [152:35] you get an operating system and then you [152:38] build the phone stuff on top of the [152:40] operating system. [152:41] >> Exactly. And this was a good business [152:45] for Microsoft. Obviously it wasn't as [152:47] big as the desktop market, but you can [152:49] totally understand why this is their [152:50] strategy. We are the main desktop [152:53] operating system provider. This is our [152:55] business model there. Let's just do the [152:57] same thing here. It seems to be working. [152:59] >> Y [153:00] >> and as far as the phone manufacturers, [153:04] the OEMs and the carriers are concerned, [153:08] things are also pretty good. These [153:11] phones that they're making, they can't [153:13] really do that much. But because of [153:15] that, they don't actually cost that much [153:17] to make. And the consumers, meanwhile, [153:20] are paying through the nose for these [153:21] things. cuz you have a smartphone on a [153:24] carrier contract, you're paying like a [153:25] hundred bucks a month. [153:26] >> And they don't consume that much data [153:28] either because they're not that capable. [153:29] >> Everybody is fat and happy. [153:32] >> Yeah. [153:33] >> So into this [153:35] morass that also Steve Jobs is of course [153:37] looking at the same thing and saying [153:39] this sucks. Enter Andy and Android. and [153:42] he goes around and he starts pitching [153:45] the phone manufacturers and the [153:47] carriers, hey, stop buying an operating [153:51] system from Microsoft or from Palm. I'll [153:53] give you a great one for free and oh, by [153:57] the way, it's going to be open source [153:58] and there'll be third party applications [154:00] that can be written to it and these [154:02] devices will be super powerful. The [154:05] ecosystem's like, no, I don't want this [154:08] one. [154:09] There's just no way in hell that [154:11] >> a startup [154:11] >> AT&T or Verizon is going to work with a [154:15] little rinky dink startup that's valued [154:17] at like $10 million and has eight [154:19] employees. There's billions and billions [154:22] of dollars at stake here. But the other [154:25] part of it too, I think the reason that [154:29] the smartphone market had stagnated for [154:31] so long was this. Everybody was happy, [154:33] >> right? It's a non-priority to upset the [154:35] Apple cart, [154:36] >> right? It's almost like a version of um [154:38] enterprise software in it, right? The [154:41] users don't like it, but the users [154:43] aren't actually the customers here as [154:46] it's the carriers who are the customers. [154:48] >> Yep. [154:49] >> So 2005 rolls around. Andy's now two [154:52] years into the company with Android. [154:54] He's managed to convince HTC, the [154:57] Taiwanese manufacturer, to make a [155:00] prototype with him. He's showing it to [155:01] carriers. He's showing it to other OEMs. [155:04] But for all the reasons we just [155:06] discussed, it's tough sledding out [155:08] there. The company's running out of [155:10] money. As Andy is going around trying to [155:15] jin up investment for another round, he [155:17] ends up meeting with Larry Page. Larry [155:21] immediately is like, "Forget raising [155:23] another round. What if I buy you right [155:25] now?" So July 2005, Google buys Android [155:30] for $50 million. 50 million dollars per [155:34] Android. Oh my goodness. [155:36] >> But of course that's a fallacy because [155:38] they would pour billions in in [155:40] development. [155:41] >> Yeah. Yeah. Yeah. Yeah. They put [155:42] billions. Yeah. Yeah. Yeah. Oh, I know. [155:44] I know. I know. But like to your point, [155:46] this episode, Google is the hit factory [155:48] here. This is the hit parade, [155:50] >> right? The correct way to think about [155:51] Android is that Google built it in house [155:54] with a little kick in the pants from [155:57] this startup kind of got far enough [155:59] along with the idea that forced them to [156:02] do it now. But they needed the kick in [156:05] the pants. [156:06] >> Yes. Because why was Larry so excited? [156:09] Why did they buy Android right away? [156:12] Eric and Larry and Sergey, they all knew [156:14] that they were late to mobile. Here we [156:16] are. It's now mid 2005. Hm. [156:19] >> Were 18 months away from the reveal of [156:22] the iPhone. Apple and Google are very [156:24] close. [156:25] >> Why do you think they knew that they [156:26] were late? [156:27] >> I'm sure they were starting to get wind [156:30] from Apple of what was going on. [156:33] >> That's true. Eric's on the board at this [156:34] point, right, of Apple. [156:35] >> He's not yet on the board, but he's [156:37] about to join the board. But the [156:38] companies are very close. [156:40] >> Okay, [156:40] >> there's that. But but even let's say [156:42] they don't know about the iPhone. [156:44] Blackberry is a thing. Yep. [156:47] >> Big adoption. [156:48] >> Yep. [156:49] >> Smartphones, you know, and even Windows [156:51] Mobile, as bad as it was, proved that [156:54] there is demand. There's clearly [156:56] consumer demand for this, [156:57] >> right? They had a version of Google.com [156:59] for these devices to access and they [157:01] could see the traffic [157:02] >> and they really knew it, especially from [157:04] maps. Google Maps on mobile devices, [157:07] smartphones, was a killer application. [157:10] Google is maintaining, I kid you not, [157:13] 350 [157:15] different versions of Google Maps for [157:18] mobile for all the just sea of phones [157:23] out there. And so like they know [157:25] >> we have built our local government [157:27] district on the desktop around our [157:30] Disney World and uhoh, looks like mobile [157:32] needs a district too. [157:34] >> Yeah. So sure you like you're right. [157:37] they would have done this anyway, but [157:41] they were starting to feel already like, [157:42] "Oh, shoot. We should have started this [157:44] two years ago, buying Android kickstarts [157:46] things." And like from the Google [157:48] perspective, thank God they did cuz [157:50] we're 18 months away from the iPhone [157:52] launch. And if they are starting from a [157:55] cold start in January 2007, good luck. [157:59] We're not telling this story right now. [158:01] >> If they don't buy Android and they don't [158:02] get started basically in the month that [158:04] they did, this market belongs to [158:06] Microsoft. Apple. [158:08] >> No, Microsoft. [158:09] >> Oh, why do you say Microsoft? [158:10] >> There's going to be two players in this [158:13] market. I see what you're saying. [158:14] There's going to be a fully integrated [158:16] player, which Apple was going to be, and [158:18] then there's going to be an OEM plus [158:20] licensed operating system, and the model [158:23] would have just been that Microsoft [158:25] sells operating systems, a mobile [158:28] operating system. [158:29] >> Great point. [158:29] >> To the OEMs who were freaking out that [158:31] Apple was going to run away with it. [158:33] >> Great point. So now back to Android and [158:37] why Android was especially so [158:39] attractive. Andy already had the right [158:41] business model for Google. It's just [158:42] that as Android the startup OEMs and [158:45] carriers are like giving it to me for [158:47] free. Like that makes you less [158:48] attractive to me. But now all of a [158:50] sudden within Google they can run the [158:51] playbook that they run everywhere. They [158:52] go the carriers the OEMs. [158:54] >> It's funny how giving it away for free [158:55] as a startup is counter signal. It makes [158:58] you look desperate. But if you're Google [159:00] it's like oh they must have a really [159:01] good plan here. [159:02] >> Yeah. [159:04] Exactly. So they start work on Android [159:08] as part of Google here in summer of [159:10] 2005. And the plan initially is that [159:15] they're going to be two versions of [159:18] Android. There is a prototype and a [159:21] device that'll be more near-term to [159:23] launch called the quote unquote sooner [159:26] sort of the more Blackberry like device, [159:29] not a touchscreen device. And then there [159:32] was a longer term [159:35] advanced research project cenamed the [159:38] dream for a touchscreen smartphone [159:42] device. [159:44] >> Summer of 2006, that next year, Eric [159:46] Schmidt joins the Apple board. [159:51] >> Sees how far along and how good the [159:53] iPhone is. [159:54] >> Uhhuh. And then January of 2007, the [159:57] iPhone is revealed in the greatest [160:01] corporate presentation keynote of all [160:03] time. [160:03] >> Yeah, [160:04] >> Eric is in the freaking keynote. Steve [160:07] Jobs invites Eric Schmidt on stage. [160:10] >> And Android hasn't been announced yet, [160:11] right? [160:11] >> Nope. Nope. Nobody knows about Android. [160:13] >> This is in January of ' 07. And then [160:16] July of 07 is when it shipped. [160:18] >> July of 07, yes, is when the iPhone [160:20] shipped. Now, I believe Eric had [160:22] disclosed to Steve about the Sooner [160:26] project because obviously it was public [160:28] that Google had acquired Android. And I [160:31] believe that Steve Jobs knew that Google [160:33] was working on like a Blackberry style [160:37] phone, but he did not know about the [160:39] Dream prototype. So, Eric comes on stage [160:44] and you go watch this. We'll link to [160:45] this in this clip in the show notes. [160:49] It's crazy. It's about a, you know, [160:51] three minute long total thing. Eric [160:54] comes on stage. He makes a joke about [160:57] merging the companies that like Apple [160:59] and Google are so close they should [161:01] merge. He says the company should be [161:03] called Apple Goo. And then he jokes and [161:05] he says, "Well, but here's the way with [161:07] the iPhone that we can merge the [161:09] companies without actually merging." [161:11] >> He's making these jokes and the camera [161:13] is focused on Steve Jobs and he just has [161:15] the ick. [161:17] >> I mean, that's the best way to describe. [161:19] He's like trying to be a good sport and [161:21] smile and be like, "Yeah." [161:24] But he has the ick. [161:26] >> It is unbelievable to watch this knowing [161:30] everything that would happen over the [161:32] next 10, 15 years. [161:34] >> This incredibly close collaboration. [161:36] There are two apps that launch in the [161:39] very first version of the iPhone. [161:40] Remember, it didn't have an app store. [161:42] It was not open to third party [161:43] developers. There is a YouTube app and a [161:46] maps app. Both of which are Google [161:49] services. Now, the apps are written by [161:51] Apple. The icons are designed by Apple. [161:53] They're basically just consuming [161:55] Google's data as APIs. The only icons [161:58] and apps on the phone are the ones that [162:00] Apple puts there. And two of the Yeah, I [162:03] don't know how many there were, 10, 12, [162:06] 13 apps are Google apps. It's wild. By [162:10] the way, the uh the YouTube icon with [162:12] the wood green TV. So awesome. [162:14] >> So awesome. I heard uh the YouTube team [162:17] absolutely detested it. [162:19] >> They hated it. Yeah, they hated it. [162:21] Well, because it wasn't the YouTube logo [162:23] and they they knew already. I mean, it [162:25] was obvious this was not going to work [162:26] because the YouTube team is like Apple [162:28] didn't put our logo on there. Of course, [162:29] they're going to start bringing in other [162:30] video content over time. It was a little [162:33] bit pre-algorithm, but it was the [162:35] thinking was there of we have to make [162:36] YouTube a destination and then control [162:38] the experience when they're in. And [162:40] making the app icon reminiscent of an [162:43] old school CRT TV was also just like [162:47] deeply antithetical to YouTube inventing [162:50] the video of tomorrow. [162:52] >> Yes. Yes. It still looked great though. [162:54] >> It fit in with that first iPhone for [162:56] sure. [162:56] >> It totally did. Do you know who was the [163:01] leader of the Google mobile teams that [163:04] developed the backends for these apps? [163:07] >> Oh, no. [163:07] >> Vic Gondotra. [163:09] >> Really? [163:10] >> Yes. That was his first job, I think, [163:12] within uh within Google. First or second [163:13] job within Google. Vic is going to come [163:16] back up here in a minute. So, the iPhone [163:20] keynote truly worldchanging historic [163:23] event. The Android team, of course, is [163:25] watching this. And uh yeah, that whole [163:29] Sooner prototype thing uh right in the [163:32] trash [163:32] >> the next day, right in the trash can. [163:34] Directly in the trash can. The dream is [163:39] no longer a dream. It's happening now. [163:42] >> Get in, kid. You're the A team now. [163:44] >> Yep. Clearly, touchscreens [163:48] are the future of mobile devices [163:51] >> and a capacitive touchcreen at that. [163:53] Yeah. So, remember Eric Schmid is on the [163:57] Apple board. Once Steve Jobs finds out [164:00] about what the Android team in Google is [164:02] now doing, he goes ballistic. [164:05] >> Or perhaps, to use his word, [164:06] thermonuclear. [164:07] >> Yes. Yeah. Yeah. Yeah. Fullon classic [164:10] Steve Jobs supposedly at an Apple all [164:14] hands meeting. This is actually a little [164:15] later, but he's overheard and quoted [164:18] leaked to the press as saying, "We did [164:20] not enter the search business. They [164:22] entered the phone business. Make no [164:24] mistake, Google wants to kill the [164:26] iPhone. We won't let them. [164:30] >> Wow. [164:31] >> Which is, I mean, to this day, fair. [164:34] Apple has been happy to just take a [164:36] spiff of all the traffic that they send [164:38] to Google and not compete in Google's [164:40] core business. [164:41] >> Yep. [164:42] >> Now, I will say I believe Apple [164:44] reputation launderers a little bit. They [164:45] get a lot of the value of being in the [164:47] search business without having to do all [164:49] the uh stuff that they demonize from a [164:51] privacy and data sharing and all that [164:53] ickiness perspective. But fine, [164:56] whatever. It's doing business. [164:58] >> That's fair. Apple did not enter the [165:00] search business. [165:01] >> So, in Walter Isacson's book, Steve Jobs [165:04] says, "I'm going to destroy Android [165:07] because it's a stolen product. I'm [165:09] willing to go thermonuclear war on [165:12] this." [165:12] >> Yes. He also says, "I will spend my last [165:16] dying breath if I need to, and I will [165:18] spend every penny of Apple's $40 billion [165:21] in the bank to write this wrong." [165:24] >> He was pissed. [165:25] >> He was really pissed. Now, [165:27] interestingly, he doesn't actually kick [165:29] Eric Schmidt off the board until 2009. [165:31] >> Yeah, it's interesting. [165:32] >> So, I think it took Steve a little while [165:35] to realize what the dream was within [165:40] Google. And there's also a reasonable [165:42] argument back. Look, both companies took [165:45] stuff from each other. A lot of the [165:47] stuff that, you know, Apple touts that [165:49] they were the first company to ever do [165:50] multi-touch and they own there were [165:52] predecessor companies that did [165:53] multi-touch before them too. The iPhone [165:56] debuted a lot of technologies for the [165:58] first time and a lot of them were also [166:00] just at the right time in history and I [166:03] think Android ar arrived at a lot of [166:04] similar conclusions at the same time. [166:06] >> True. It's interesting you said [166:08] multi-touch. the multi-touch actually [166:10] becomes the battleground [166:12] >> cuz that's the patent that they go to [166:13] war over. [166:14] >> That's the patents that Apple has. So [166:16] Steve Jobs threatens to sue Google over [166:19] implementation of multi-touch gestures. [166:21] And so as a result, Android for several [166:24] years, doesn't have things like pinch to [166:27] zoom or the sort of swipe operating [166:30] system UI navigation gestures. And I'm [166:33] pretty sure if you remember early [166:36] Android phones for the first couple [166:37] years, every single one of them had four [166:40] physical buttons at the bottom of the [166:42] phone to navigate the operating system. [166:44] I think this is why. [166:45] >> Huh. But let's take the Google side of [166:47] this argument for a minute. When Android [166:49] does launch, they have the market, which [166:53] today is the Play Store. Apple didn't [166:55] have an app store. Yep. Android had when [166:58] you swipe down a notification center [167:00] with all the notifications from each of [167:02] your individual apps. You could [167:04] >> took Apple years to get that [167:05] >> drag to rearrange apps on the home [167:07] screen. I mean these are things that [167:09] Apple then directly copied as well. So [167:12] >> Right. [167:12] >> Yeah. [167:14] >> All right. So yeah, let's get into the [167:16] >> great artists steal. [167:17] >> Exactly. Let's get into the launch and [167:19] the competition. So November 2007. So [167:22] what's that? 10 months after the iPhone [167:25] reveal and five months after the launch, [167:29] remember Android launches in 2008, [167:32] Google announces the formation of the [167:34] Open Handset Alliance. [167:35] >> That's right. [167:36] >> And this is a partnership with HTC, [167:39] Motorola, Samsung, LG, T-Mobile, Sprint, [167:41] Qualcomm, Intel, Broadcom, and Texas [167:43] Instruments. And this was so confusing [167:46] at the time. I and everybody else was [167:49] like, [167:49] >> "What does it mean? [167:50] >> What is this? Is Google making a phone? [167:52] is Google not making a phone. So then a [167:56] whole year goes by with basically [167:58] nothing. Then in September of 2008, a [168:02] lot of things happened in September of [168:03] 2008. Chrome, Android, Lehman Brothers, [168:07] Google announces the T-Mobile G1 phone, [168:11] the Gphone. [168:14] And the T-Mobile G1 is manufactured by [168:16] HTC, remember Andy and Android's [168:18] original partner in the prototype. And [168:21] the product name, the HTC product name [168:23] for it is the HTC Dream. [168:25] >> Heyo. [168:26] >> This is the Dream. This was what they [168:29] were working on. And in the US, it's [168:30] called the G1. It actually is a super [168:34] interesting little device. I wrote an [168:35] app in it. I had a class in college. It [168:37] was like a capstone class or something [168:39] where I could like pick my own project [168:42] to do. and we had a fourperson team and [168:45] one of the guys had a T-Mobile G1 and we [168:48] wrote I think it was like a Java um [168:50] thing for it but he then founded the [168:53] company Daily Booth after that. [168:55] >> Oh wow. Yeah. [168:57] >> In fact, it may have even been like a [168:58] Daily Booth for Android app. [169:00] >> You had a lot of founders come out of [169:02] your crew at Ohio State. Awesome. So it [169:05] has a touchcreen on the front with the [169:08] physical navigation buttons like I was [169:10] talking about. It has a slide out [169:12] horizontal query keyboard, sort of [169:14] reminiscent of the Sidekick back in the [169:16] days, unlike the iPhone. It has [169:20] multitasking, [169:22] so you can run multiple apps at once, [169:24] and it has third party applications. [169:28] Now, that's a little bit unfair to the [169:30] iPhone because by the time the G1 [169:32] actually launched, Apple had indeed just [169:35] shipped the App Store. The event where [169:37] they launch it is like T-Mobile event in [169:40] New York City in a commercial kitchen. [169:43] It's a haphazard random launch. You [169:46] can't even find video of it today. [169:47] There's like little clips you can find [169:49] and still images what is widely reported [169:51] and you can actually see in photos. [169:54] Larry and Sergey do show up. They [169:56] rollerblade into the building. They [169:59] rollerblade on stage. There all these [170:01] T-Mobile and HTC executives there in [170:03] suits. Here come Larry and Sergey on [170:06] rollerblades on stage. [170:07] >> Listeners, join the acquired email list. [170:09] We'll throw this in the next email that [170:11] goes out. [170:11] >> Yeah, it was halfhazard to say the [170:14] least. But the G1/Dream [170:18] becomes a pretty decent success. It [170:20] sells over a million units in the US and [170:24] just this one device, this one phone [170:26] gets 6% smartphone market share, which [170:29] puts it roughly on par with Palm. Like [170:31] all of Palm, [170:32] >> the G1 kind of matches in [170:35] >> market share, but the smartphone market [170:38] is still very small. [170:39] >> Yeah. It's important to remember mobile [170:41] really wasn't a thing until 2011 [170:46] of very obviously the next wave and the [170:48] next computing paradigm. [170:49] >> Yep. But to be fair to Apple and the [170:52] iPhone, it is starting to run away with [170:55] the market. And this is to the point of [170:56] man, if Google had not bought Android [170:58] when it had, it would have been too [170:59] late. So the whole lifetime of the G1, [171:02] they sell about a million units. The [171:04] iPhone sold 11 million units in 2008 [171:07] alone, 20 million in 2009. So basically [171:11] overnight, Apple and the iPhone goes [171:13] from not in the smartphone market at all [171:15] to over 50% market share of smartphones. [171:19] But as great as the iPhone was, it did [171:24] have a few weaknesses. [171:27] >> No copy paste. No copy paste. Yep. [171:30] >> No multitasking. [171:31] >> As mentioned before, it didn't [171:32] multitask. Not very customizable. I [171:35] think we're still in the era of you [171:37] can't even change your wallpaper on the [171:39] iPhone. Pretty sure we are. [171:41] >> I think it's still just the black [171:43] background. [171:44] >> Yep. You can't put your own apps on it [171:45] from anywhere but the App Store even [171:47] after it launches. [171:48] >> Yep. A big knock at the time. People [171:51] love that it's a touchscreen. People [171:53] really wanted the physical keyboards. [171:55] >> Yep. And the biggest problem with the [171:58] iPhone, at least in the US, you can only [172:00] get it on AT&T. [172:02] >> And you could only get it with the Edge [172:04] network. It was unusable. [172:06] >> That's right. It didn't have 3G. [172:08] >> It was so terri I mean, eventually the [172:10] iPhone 3G came out within a year, but [172:11] even that was really slow. It was the [172:13] network had not caught up to what you [172:15] wish the device could do for a few [172:17] years. [172:18] >> Yep. [172:19] So, that brings us to holiday 2009 [172:23] and the Motorola Droid [172:26] >> changed everything. [172:27] >> It's sort of funny to say now like, oh, [172:28] the Motorola like the Motorola droid. [172:31] This changed everything. Yes, the Motor [172:33] I mean, when we interviewed Steve Balmer [172:34] a couple months ago, he brought it up. [172:36] When the Droid launched, it was holiday [172:38] 2009. And I think you and I were like, [172:40] "Was it really that late? Wasn't it [172:41] early?" And he was like, "Nope, [172:42] Christmas 2009. I will never forget it." [172:45] That is when Android won the market. [172:48] This was the moment [172:50] >> and Google was really willing to put [172:52] their brand second. Now, were they [172:54] really putting their brand second? It's [172:55] Android versus Droid. So, very [172:57] convenient. But like if you were to go [173:00] survey the American public in 2009, 10, [173:03] 11, 12, maybe even 13 and say, "Do you [173:07] know about Android, the mobile operating [173:10] system?" "No." "Do you know about [173:12] Droid?" "Oh, yeah. I have a Droid [173:13] phone." Well, and then there were a [173:15] couple years after that where it was [173:16] like, "Do you know about Google and [173:20] Android?" Yeah, maybe. Do you know about [173:22] Samsung and Galaxy? Oh, yeah. I know [173:24] about that. [173:25] >> Yep. Exactly. [173:26] >> So, we'll get into that in a sec. The [173:28] Droid [173:28] >> Droid does, baby. [173:30] >> Verizon at this point is getting [173:32] pummeled by AT&T. It's been two years [173:35] since the iPhone launch. AT&T isn't just [173:39] stealing a lot of subscribers from [173:42] Verizon because of the iPhone. They're [173:44] stealing the best subscribers, the [173:47] people that are willing to pay the most [173:49] money for the biggest data plans for [173:55] smartphones. Verizon finally decides [173:57] like, we got to change the game here. We [174:00] got to be able to compete with the [174:02] iPhone. We're going to go all in on [174:04] Android. We are going to buy a device [174:08] and make this our flagship smartphone, [174:11] position it against the iPhone, and we [174:13] are going to invest hugely behind this [174:15] thing. So, the device itself, the actual [174:17] Droid made by Motorola, it was a great [174:20] device. It had a big screen, big screen [174:23] for the time, a slide out keyboard, it [174:25] had a 5 megapixel camera, removable [174:28] battery, all of these things the iPhone [174:31] didn't have. Probably the most important [174:35] feature it had though, the killer killer [174:39] app was on the software side. It was the [174:42] first Android device launched with [174:46] Google Maps turnbyturn [174:49] >> navigation. I didn't realize that. [174:51] >> So before the Droid, there was this [174:54] whole consumer electronics product [174:57] category of dedicated GPS devices. [175:01] People old enough to remember my [175:03] remember this. TomToms, Navex, people [175:06] would buyarm [175:07] >> these devices. Garmin. Yep. [175:09] >> They would put them in their cars and [175:13] you also paid a monthly subscription fee [175:16] for the service of the turnbyturn [175:19] navigation. Overnight, this entire [175:22] product category gets obsoleted, [175:25] sherlocked, gone, because Google Maps is [175:29] a better product with better navigation [175:32] and it's free. No more monthly fees, [175:36] just baked in in your phone, in the [175:38] device you already have with you. Why on [175:40] earth would anybody buy, let alone pay [175:44] monthly, for a standalone GPS product [175:46] again? [175:47] >> Yep. [175:47] >> And you know what doesn't have it? the [175:49] iPhone, the Apple version of Google [175:52] Maps, you had to manually advance the [175:56] steps. So, it would pull up the route [175:59] and then you could tap the button to be [176:01] like, I've made this turn. Now, show me [176:03] the next step. [176:04] >> That's right. [176:05] >> Part of it. [176:06] >> That's such a funny You're exactly [176:07] right. I remember that, too. [176:08] >> Not really what you want to do while [176:10] you're driving, [176:10] >> man. [176:11] >> Yeah. It's crazy how not that long ago [176:14] this was. [176:15] >> Totally. [176:15] >> That was the killer feature. But even [176:18] more important than all the features was [176:20] the marketing and the muscle that [176:21] Verizon put behind this. So they [176:24] licensed the droid name from Lucasfilm. [176:27] >> That's right. And I think Lucasfilm was [176:28] mentioned at the bottom in the credits [176:30] of every commercial. [176:31] >> Yes, every commercial. And they did [176:33] these series of commercials that we've [176:35] been referencing. Man, if you lived in [176:37] the US and you are older than like 12 at [176:42] this time, this is burned in your [176:44] memory. It was so great. The first 80% [176:47] of the ad, 90% of the ad was a Apple [176:50] style ad knockoff with the like bright, [176:52] happy, upbeat music and the white [176:54] background and it had the fading Apple [176:57] style text and it said, "I don't [176:59] multitask. [177:01] I don't have a removable battery, you [177:03] know, etc., etc." And then the very [177:06] last, you know, 5 seconds of the ad, [177:08] there was like a hard cut and like [177:10] static noise and it was black and it was [177:12] edgy and then it said Droid does [177:17] It was so good. So, the CMO of Verizon, [177:22] so Verizon did all of this, said that [177:24] the campaign was designed to quote, [177:26] "Wake up the market." [177:29] And boy did it ever. So, that original [177:32] Droid, I think it sold a quarter million [177:34] units the first weekend it was on sale, [177:37] and then it sold a million units faster [177:41] than the original iPhone had. Like there [177:43] was just so much pent-up demand for a [177:45] real smartphone on the Verizon network [177:48] >> plus all the you know yeah this has [177:51] turnbyturn navigation but even like put [177:53] aside whether it was better or not it [177:55] just was a real smartphone on Verizon. [177:58] Time magazine named the Droid its [178:00] product of the year for 2009. [178:03] >> Wow. [178:03] >> The bigger thing though is that like [178:05] Verizon going all in behind it even [178:07] though they would add the iPhone later. [178:09] It creates sort of this seed of what the [178:13] Android user base would become today, at [178:15] least in America, because Verizon went [178:18] all in on Android, all in on Droid. Over [178:21] the next couple years, they followed the [178:24] original Droid up with, let's see, there [178:26] was the HTC Droid Incredible, the Droid [178:28] X, the Droid 2, the Droid Bionic, the [178:30] Droid Max. All of these had major [178:33] marketing campaigns behind it. game over [178:36] for the the segment of the market that [178:38] is not Apple, this different OEM from [178:41] operating system. Google just runs away [178:43] with it. And before this, Microsoft had [178:45] a shot. [178:46] >> They really did. [178:46] >> They were a systemic disadvantage [178:48] because they were going to carriers and [178:51] saying, "Why don't you pay us five [178:52] bucks, 10 bucks?" And Google was going [178:55] to them and saying, "Here you go. This [178:58] is free. You can have the source code [179:01] and you can modify it as you see fit." [179:03] Even today, the I think um Samsung has [179:06] their own OS, Samsung one or something [179:08] like that that looks different. I mean, [179:10] it's Android, but it's the open-source [179:12] version of Android that they've [179:13] customized. That's the thing that's on, [179:15] I don't know, a billion phones. And [179:17] three, we aren't Microsoft. [179:19] >> Yeah, you guys don't want to be compact, [179:21] >> right? Microsoft managed to suck up all [179:23] the profit in that entire value chain. [179:25] And handset makers, you currently make [179:27] money. So, why would you go work with [179:29] Microsoft who did that to the PC makers? [179:31] And then as a little sweetener on top of [179:33] all this, you know how I mentioned it [179:35] was free in open source? [179:36] >> It's actually less than free. [179:38] >> We're actually gonna pay you. [179:39] >> Yeah. [179:40] >> For searches that originate on your [179:44] phone, we will give a little revshare to [179:48] both the carrier and the OEM, the [179:52] handset maker. [179:53] >> Yep. So this was not widely publicized [179:56] at the time as you can imagine but uh [179:59] Bill Gurley wrote a blog post where he [180:03] had heard from friends that Google was [180:05] paying carriers and OEMs to use Android [180:09] even though Android was free. And he [180:10] wrote this incredible blog post about it [180:12] called the less than free business [180:14] model. And he basically predicted that [180:17] like Android's going to run away with [180:19] this year. If you're a carrier or an [180:21] OEM, sure, there's a segment of the [180:24] market that's going to demand Apple. [180:25] That's fine. But Microsoft is dead. Palm [180:28] is dead. Blackberry is dead. There's no [180:31] way you can compete with free, let alone [180:33] less than free, where they are paying [180:34] you to take something of value for free. [180:37] >> Yep. [180:38] >> And from Google side, it's the exact [180:40] same thing as that thing we talked about [180:41] with open sourcing Chromium. They're [180:44] happy to give a few percentage points in [180:47] traffic acquisition costs of their [180:49] search revenue to people who are [180:51] ensuring that the platform underneath [180:54] them doesn't belong to someone else. I [180:56] mean, there were some risks that it was [180:57] all Apple and then that creates two [181:00] problems for Google. One, they pay Apple [181:03] a lot more money than they pay the [181:05] combination of the carrier and the OEM [181:08] maker. Those get a much smaller spiff. [181:11] Two, this means that Google controls [181:14] more of the underlying environment that [181:16] they operate in. Imagine how terrible it [181:18] would be for them if mobile Safari was [181:20] the new Internet Explorer and their [181:22] entire franchise was at risk of Apple [181:25] saying, "And we're going to point [181:26] traffic over here." Now, Google is happy [181:29] to toss a couple of points over to these [181:31] guys. I can't think of another example [181:36] of a dominant technology business and [181:40] business model that has successfully [181:44] survived and transitioned a major [181:47] platform shift. Yes. [181:49] >> And thrived in that next platform as [181:52] well. [181:52] >> Mobile was a platform shift. A huge one. [181:56] I mean going from PC to the web was a [181:59] platform shift. Going from PCM web to [182:01] mobile was an even bigger platform [182:03] shift. [182:03] >> Play it out even back further in history [182:06] than this. IBM was dominant in [182:08] mainframes and then lost their dominance [182:10] in the PC era. Microsoft was dominant in [182:12] PCs and then lost their dominance in the [182:14] web era. Google was dominant on the web [182:17] and stayed dominant in the mobile era. I [182:20] mean, they didn't derive giant profits [182:22] from the mobile like directly off of [182:24] selling phones or selling the OS or, you [182:26] know, they make some money on the Play [182:28] Store, but not giant amounts relative to [182:30] the rest of their money and what other [182:32] players like Apple make, [182:33] >> but they kept search going, [182:34] >> but they managed to stay relevant to [182:37] consumers with these, you know, hundreds [182:39] of millions, billions of devices that [182:41] they shipped and their business was [182:43] doing better than ever. I mean, all of [182:45] these Android phones that are shipping, [182:48] especially in the earlier years, what is [182:50] the most prominent part of the UI on the [182:52] touch screen? Giant freaking Google [182:54] search bar right there at the top, [182:57] >> right? The state of play of being a big [182:59] tech company, and this dates back 80 [183:01] years, is technology moves fast and the [183:05] new paradigms disrupt everyone that came [183:06] before you. So you get one era and you [183:09] got to make the absolute most of the one [183:12] era that you grew up in and after that [183:15] you're probably going to lose relevance. [183:17] You might keep your money machine going [183:19] for a long time. Famously IBM made more [183:22] revenue than Microsoft for a lot longer [183:23] than people think. [183:24] >> Or even take Microsoft and Windows. Like [183:26] Windows is still big today. [183:27] >> Yep. But the importance of that platform [183:29] is going to fade and fade and fade. [183:31] >> But yeah, you won't be able to [183:32] transition your business model into the [183:34] next era. Google did it [183:36] >> and occasionally someone you know misses [183:39] the second era but comes back for the [183:40] third like Apple figured out mobile. [183:43] They never won a previous era. They were [183:45] a player in PCs but they didn't win. [183:48] Almost no one gets two and almost no one [183:50] gets two successive ones. And that is [183:53] the really impressive thing that Google [183:55] figured out how to do here. [183:57] >> Yeah. I mean guys like we said this [183:59] episode is the hit parade. Android [184:02] basically from end of 2009 onward just [184:06] washes over the world like a title wave. [184:09] In holiday 2009 when the Droid comes [184:11] out, total Android market share of the [184:15] smartphone market is still in the, you [184:17] know, the G1 range, like five 6% global [184:20] market share. One year later, 30%. [184:25] >> Wow. [184:26] >> They go from 5% to 30% in one year. They [184:30] announced that over 200,000 Android [184:32] devices are shipping every day around [184:36] the world. The next year in 2011, [184:39] Android's market share is 50%. And two [184:42] years after that, by the end of 2013, it [184:44] is 80% [184:47] market share. In many ways, it's sort of [184:49] the Visa network of networks thing where [184:52] they don't have to make every phone. [184:54] They don't have just one horse in the [184:55] race. They're getting leverage by having [184:57] two, three, four, five major [184:59] manufacturers of these devices that are [185:02] all independently doing their own [185:04] marketing. And there's a very clever [185:06] arrangement where you can just have the [185:09] Android open source project and you can [185:11] build your own mobile phone and you know [185:15] you can launch it and you don't have our [185:17] app store and you don't have to default [185:19] to Google search and you don't get [185:20] Google map. Like you just have the [185:22] operating system and it's great. Anyone [185:23] can do that. But why wouldn't you want [185:26] to have our app store? It's where the [185:28] all all the apps are. And if you do [185:30] that, then you get all the great Google [185:32] services, all the apps. You you get the [185:34] native Gmail and the native maps and all [185:36] this great stuff we've written. And if [185:38] you do that, then Google's the default [185:40] search and we'll pay you for that [185:41] >> and then you make money. [185:42] >> Yeah. [185:43] >> But by the way, if you want all this [185:44] stuff that your consumers are going to [185:45] demand, you are going to default to [185:47] Google search. Like it's a very uh [185:50] >> that's the payment. Yeah, that's the [185:53] that's the offer you can't refuse. [185:55] >> Yes. [185:56] >> Now, here's the actual sort of crazy [185:57] thing. As I said, by 2013, Android [186:00] global market share is 80%. That's [186:03] actually higher than it is today. Today, [186:06] it's down to like, I think, 72%. [186:09] And Apple is, you know, 27.99%. [186:13] Apple's share has really grown. No [186:16] question, Android pushed iPhone to be [186:20] better on many dimensions. Things like [186:23] cheaper iPhones, bigger screens, better [186:27] cameras. I mean, on and on and on and on [186:30] of things that I don't think Apple would [186:33] have done if Android hadn't been pushing [186:36] them. [186:36] >> Probably not big cheap screens, but some [186:38] of the cameras, I think. So, [186:40] >> maybe. I don't know. I mean, for years, [186:42] the iPhones did not have good cameras. [186:44] Uh big part of that Droid marketing push [186:46] was the five megapixel camera. [186:48] >> That's right. [186:48] >> The original few iPhones had a 2 [186:51] megapixel camera, I think. Like it was [186:54] crappy. [186:55] >> Yeah, they've definitely pushed each [186:57] other. [186:57] >> Yep. So then the other quick thing to [186:59] mention on Android, there was one [187:02] interesting moment and tension with [187:04] Samsung in the early to mid2010s. [187:09] Samsung basically said, "Oh, okay. The [187:11] iPhone is the premium device. Android is [187:13] this incredibly flexible platform. What [187:16] if we just take Android and copy the [187:18] iPhone with Android? [187:20] And they got really good at it. The [187:22] Galaxy devices were just shipping in [187:25] huge, huge numbers. And then Samsung [187:28] started stripping out Google services [187:30] and putting their own Samsung services [187:33] >> in on some of their devices. [187:35] >> That was a bridge too far for Google. So [187:37] this is when Google started the Pixel [187:40] program. Google had done the Nexus [187:42] program making their own hardware [187:44] before. The Pixel though was and is a [187:48] sort of reference device that yeah, [187:51] consumers could buy, but more so to show [187:54] the rest of the OEM market, the non [187:57] Samsung market, hey, here are reference [187:59] designs essentially for great premium [188:02] devices, great cameras, all the features [188:05] you want. Here, copy these. It's the [188:08] same thing as the Microsoft Surface [188:10] strategy. why Bomber was so adamant. We [188:12] got to make a service. We got to show [188:13] the OEMs how to do this, [188:16] >> right? It's funny. So, I have been [188:18] trying to think about what is the [188:20] business of Android. Google having [188:23] Android versus Google not having [188:25] Android. And I tried to pull up the most [188:27] credible numbers I possibly could. [188:29] There's basically [188:31] two [188:32] things that you just have to add [188:33] together to create the value. One is how [188:36] much money they make from the Play [188:37] Store, which has become significant. [188:39] didn't used to be but is now. And then [188:41] the second is how much money are they [188:44] saving by not having the searches [188:47] originate from a platform that they [188:48] don't own. I used to think, oh, because [188:51] it's Android, they don't have to pay [188:54] money. They have to pay $20 billion to [188:56] Apple. It's not zero. They do actually [188:59] have to pay, like we talked about, Dave [189:00] and I sort of figured out as we were [189:02] going through the financial disclosures [189:03] and stuff, they do pay the OEMs and they [189:06] do pay the carriers. And the question is [189:08] how much? Cuz once you can kind of [189:10] figure out how much, then you can do a [189:12] little bit of napkin math to figure out, [189:14] okay, well, how much are they still [189:15] saving by it not being Apple? So Google [189:18] paid out last year, and I'm just using [189:20] the current numbers to try to figure out [189:22] what the splits have always been. They [189:25] paid out last year 55 billion in total [189:28] traffic acquisition costs. Now, traffic [189:31] acquisition costs are actually the sum [189:32] of two different numbers from two [189:34] different businesses because they love [189:35] to offiscate things. one, it's what [189:38] we're actually looking for, the [189:39] acquisition of traffic to Google search. [189:41] And the other component is money that we [189:44] paid to publishers where our ads show up [189:49] in the sort of doubleclick AdSense [189:52] world. [189:52] >> Yeah. [189:53] >> Now, we know that that averages about a [189:56] 7030 split and we know that they made [189:58] $30 billion last year gross in the [190:01] Google network. So you could say, okay, [190:03] they probably paid out about $21 billion [190:06] of that 55 billion in the AdSense [190:10] doubleclick Google network world. So [190:12] that backs our 55 billion down to 34 [190:15] billion. Okay, that's 34 billion in [190:18] actual traffic acquisition for Google [190:20] search. [190:21] >> And we know 20 was iPhone, [190:23] >> right, for Safari searches. So that [190:26] means there's 14 billion that gets [190:29] distributed to non-Apple [190:32] traffic acquisition distribution [190:34] partners which in their annual report [190:36] they define as browser providers, mobile [190:40] carriers, original equipment [190:42] manufacturers and software developers. [190:44] It's basically 14 billion to the the [190:48] Android mobile carriers and OEMs plus [190:51] Firefox. [190:52] >> Yep. [190:53] >> What am I missing? I'm going to guess [190:55] Firefox is less than a billion. Call it [190:58] somewhere around half a billionish. [191:00] >> Yeah, there's probably [191:03] some version of the old portal deals [191:06] that still exist. [191:08] >> Properties on the web that have Google [191:10] search baked into it. [191:11] >> Okay. So, let's cut 4 billion off for [191:13] Firefox and the other web properties and [191:17] other [191:17] >> Yep. Okay. So, 10 billion going to the [191:20] carriers and OEMs. It's actually pretty [191:22] significant that 10 billion going to [191:24] carriers and OEMs. It's half of what [191:27] they're paying Apple. [191:28] >> Half of what they're paying Apple, but [191:30] for many, many, many more devices, [191:32] >> right? And so clearly the rev share to [191:35] the carriers and OEMs is a much smaller [191:38] percent than what they have to pay [191:40] Apple. I'd guess a quarter. Either way, [191:43] I actually think after walking all the [191:45] way through it, the bigger component of [191:47] this is just derisking their future. [191:49] It's not how many billions. They don't [191:51] care about giving 10 billion up for [191:52] this. [191:53] >> Yeah. As we've been saying all episode, [191:55] Google is more than happy to pay traffic [191:57] acquisition costs to any and everyone. [192:00] >> Yes. And so then direct value that they [192:03] make from the Play Store, it actually [192:04] came out in a lawsuit. In 2019, Play [192:08] Store revenue was 11.2 billion. Then [192:11] gross profit was 8.5 billion and 7 [192:13] billion in operating income. Now, 7 [192:16] billion, not nothing, but still a far [192:18] cry from Google's core business of ads [192:20] from search, Gmail, and maps. And that [192:23] same year, the core business did almost [192:24] a hundred billion in revenue. So, [192:26] something like 85 billion in gross [192:28] profit is my best estimate, and around [192:31] 30ish billion in operating income. So, [192:34] even though the Play Store made 7 [192:35] billion in 2019, the important thing is [192:37] that Android is still primarily [192:40] protecting the core search ads business [192:42] and making sure that traffic doesn't go [192:44] elsewhere. This levered Google's web [192:47] business into the mobile era. How [192:49] amazing is that? [192:50] >> Yeah, that's true. It probably generated [192:53] several hundred billion dollars profit [192:56] dollars that they may not have had those [192:59] years otherwise. [193:01] >> Yeah. So I guess what I'm saying is [193:03] obviously Android was a giant success [193:06] and the biggest reason even though they [193:10] save I don't know 10 15 billion a year [193:12] from not having to pay it to Apple and [193:14] even though they generate 8 billion I'm [193:17] sure at this point it's bigger I don't [193:18] know 10 15 billion a year really it's [193:20] about just protecting the core not about [193:22] saving costs. [193:23] >> Yeah. And this one they almost missed [193:26] it. They hadn't bought Android when they [193:28] had like that window was closing fast [193:31] >> and Microsoft did miss it [193:32] >> fast fast fast. Yep. [193:34] >> And so at some point Andy Rubin leaves [193:36] and Sundar actually takes over the [193:39] combined teams. So our hero here who is [193:42] starting to gather more [193:44] responsibilities. It was just the [193:47] application clients and then it was [193:49] Chrome and in 2013 it becomes Chrome and [193:51] Android. And whenever you see Sundar on [193:53] stage, he is very proud of Google's two [193:55] open platforms. [193:57] >> Yep. [193:58] >> So today there are more than 3 billion [194:00] active Android devices. I think it's [194:02] even higher than that now. [194:03] >> This just silly. There's like 7 billion [194:05] people in the world. They're [194:07] >> over three billion active Android [194:10] phones. [194:11] >> Yes. So you're probably thinking coming [194:14] into this 2010 2011 era, they're really [194:18] feeling themselves over there at Google. [194:20] You know, we've jumped over some [194:21] failures, but it's been hit after hit [194:24] after hit in a lot of these areas that [194:26] really matter. Just like we talked about [194:27] on the Microsoft episodes, it really [194:29] doesn't matter when you fail and how [194:31] many times you fail, even the size of [194:33] your failures, if your hits are these [194:36] giant worldchanging [194:39] platform type tech businesses that [194:42] endure for decades. And that's what they [194:44] had on their hands. [194:46] >> Yep. And it sure looked at this time [194:49] like there was another big technology [194:53] category out there [194:55] >> of social [194:56] >> that Google should be playing in of a [194:58] similar size called [195:00] social. [195:01] >> Yes. [195:02] >> And this is the plus story. I'd say rest [195:06] in peace, but I don't think anybody [195:07] misses it. [195:08] >> Yeah. All right. Well, I want to start [195:10] this story the way that people expect us [195:13] to start the story. And I have a little [195:15] bit of a different take on it as we get [195:17] partway in. [195:18] >> Great. [195:18] >> So, Google had been interested in social [195:20] for a long time. They weren't blind [195:22] blind to it. In 2007, they tried to do [195:24] open social and they basically failed at [195:28] that because Facebook didn't participate [195:29] and Facebook was social. So, everything [195:31] else combined didn't really matter. [195:32] >> Oh, you didn't start where I thought you [195:34] were going to start. The craziest thing [195:36] is that Google had Facebook before [195:38] Facebook. Orchid. [195:40] >> Yeah, that's true. which I think was [195:41] like a 20% time project that then blew [195:44] up in Brazil. [195:45] >> Totally. No. No. Yeah. Okay. So, there [195:47] was a Turkish engineer who worked for [195:48] Google named Orchet Botton and his [195:52] passion was social networking and Fster [195:56] was a thing at the time. And so January [196:00] 2004 [196:01] before Gmail, before the Google IPO, [196:04] before Facebook launches on the Harvard [196:06] campus, in his 20% time, he uh launches [196:09] a social network within Google called [196:12] Orchet. And it didn't become that big in [196:14] America, but it got at its peak, I [196:17] think, 300 million users. It was the [196:19] biggest social network in Brazil, the [196:20] biggest social network in India. [196:21] >> Wow. [196:22] >> And Google was like, I don't know, it [196:24] doesn't seem that important. [196:27] >> All right. So, Open Social in 2007, [196:29] Google Wave in 2009. By the way, can we [196:32] just pause and say 2009, this is like [196:34] right after Chrome, right after Android. [196:36] Google is a big place and Google is a [196:39] siloed place at this point. I mean, it [196:42] it's kind of crazy that Android is [196:44] happening over in this other building [196:46] and there's this like fight with Apple [196:48] and that's the same time that they're [196:50] doing Google Wave. Like, it's weird that [196:52] this is all sort of concurrent. The [196:53] company was focused in a lot of [196:55] different directions, [196:56] >> but it was so decentralized that it [196:58] actually worked [196:59] >> well. It worked early. [197:00] >> Yeah, [197:00] >> it worked really well to get all this [197:02] stuff off the ground. [197:03] >> It was so interesting doing the research [197:05] for this episode because so many of the [197:06] people we talked to, even people who [197:08] were leaders of a lot of these products [197:11] because Google was so decentralized and [197:13] so siloed, they were focused on their [197:15] thing on Android or Chrome or whatever. [197:18] And so we'd ask, you know, what was the [197:21] overall strategy? what was the through [197:22] line to all of this? And we kept getting [197:24] answers of like, well, it was just [197:26] googly. People worked on what they [197:28] thought was cool and it was good for the [197:29] web. And like that is absolutely true, [197:32] but there was this all overlay of this [197:35] very, very thin layer of strategy that [197:38] held the whole web together. [197:40] >> I think that the strategy was pretty [197:43] tight at the top level and they just [197:45] didn't actually need to communicate it [197:46] down very far. Like most people that I [197:48] talked to said, I don't know, I was just [197:49] trying to build great products that [197:50] people love. I think it was a feature, [197:52] not a bug is what I'm saying that it [197:54] didn't communicate down because it let [197:55] the teams below build really, really [197:58] great products. [197:58] >> Yep. And never really think about how is [198:00] this going to help the ads business and [198:02] that was okay. [198:03] >> Yep. [198:04] >> So, Wave failed because really nobody [198:06] knew what to use it for despite a [198:08] dazzling and wonderful first [198:10] introductory video. Buzz then in 2010 [198:14] that created this big privacy debacle [198:17] right at launch. It was super shortlived [198:18] that shut down. So then in 2010 Ers Hoa [198:22] very senior Google at this point [198:25] probably distinguished engineer senior [198:26] vice president [198:27] >> yeah the guy who created the distributed [198:29] infrastructure [198:30] >> right after the buzz failure he is [198:33] inspired to write this memo kind of like [198:36] the Bill Gates 1995 internet memo [198:40] there's a sea change going on the [198:42] internet is becoming more people [198:43] oriented social media could be a problem [198:45] for us the social media challenge [198:48] requires a decisive been substantial [198:50] response involving a significant [198:52] deployment of personnel right away. [198:55] Essentially, the internet was now [198:56] starting to organize around people in [198:58] this web 2.0 era, not just pages and [199:01] applications, the things that were sort [199:03] of the domain of Google. And so, here's [199:05] where I want to pause, David, and I'm [199:07] going to take it in a little bit [199:08] different direction than I think you're [199:09] probably expecting, which is so [199:11] therefore they went after Facebook. I [199:12] think it's a little bit more related to [199:16] the palace intrigue at Google [199:18] >> and a little bit less on the nail [199:21] strategic. So [199:22] >> if you zoom out and look at the company [199:24] right now, it's pretty fragmented. It's [199:26] got different fftoms with big [199:29] personalities at the top of each of [199:31] these FFTs. Android, Chrome, search, [199:36] YouTube, developer relations, trying to [199:38] sort of will a Google platform into [199:41] existence. Different products with kind [199:44] of competing goals. Ultimately, they all [199:46] help Google's overarching mission, but [199:49] there's a lot of elbows starting to come [199:50] out. Android was its own FFTM, totally [199:53] off on its own island, fighting an [199:55] existential battle. Chrome is starting [199:57] to do the same stuff as Android. They're [199:59] building their own operating system. [200:01] It's not clear what belongs in an [200:02] Android camp versus a Chrome camp and [200:04] Sundar hasn't unified them yet. Search [200:08] very protected, separate team, [200:10] especially the core people doing search [200:13] ranking and monetization. No one touches [200:14] them. YouTube is totally separate. Gmail [200:17] is massive and it really is the only one [200:20] in 200910 at the company that owns [200:23] identity since it's the only Google [200:25] property that you actually have to log [200:27] into. YouTube has its own entirely [200:29] different username and password system. [200:31] It's a mess, right? It's a complete [200:34] mess. So Larry's sensing this. He's not [200:36] CEO at the time, but he's realizing the [200:39] company's all over the place. He decides [200:40] he's just going to come back and get the [200:42] company on track. And so I think [200:46] Plus is kind of just the thing he picked [200:50] as the single thing to try to galvanize [200:52] and unify the company around. And no [200:54] matter what they picked and how they [200:56] executed it, it was going to create a [200:58] lot of carnage. [201:00] >> I can buy that. [201:02] >> There was a big shift that needed to [201:05] happen in one way, shape, or form. And [201:06] Google+ ended up kind of being the ugly [201:08] thing they did. [201:09] >> Yeah. A recentralizing of authority, so [201:12] to speak, within the company. [201:13] >> Right. So in May 2010, they get the top [201:16] 50 people at Google's leadership [201:18] assembled to discuss what to do. [201:19] >> The argument for this is this is more of [201:21] a convenient crisis. [201:23] >> Yeah. model [201:24] >> and it might be a real crisis also, but [201:26] it's also exactly what you're saying, [201:28] David. So, officially then in January [201:31] 2011, Google announces Larry Pageige [201:33] will return as CEO in a few months that [201:35] April. Right away, Larry moves his [201:37] office into what would become the plus [201:40] building. Wow. Yep. So, they had just [201:44] come out of this chapter. They've got [201:45] this amazing business. The whole Chrome [201:48] and Bing thing was defense against [201:49] Microsoft. Android was defense against [201:51] Apple and Microsoft. and Google+ is now [201:55] defense against Facebook. [201:57] >> Yep. [201:57] >> And legitimately, you could imagine a [201:59] world where social ends up becoming way [202:01] more important and the only places to [202:04] put ads and the places where people are [202:06] asking for information and there was [202:08] rumors for a long time Facebook was [202:09] going to build a search engine. You have [202:11] the attention, you can hijack it and do [202:13] other stuff with it. [202:14] >> These were walled gardens. Facebook was [202:16] a walled garden. Google search couldn't [202:19] index what happened inside of Facebook, [202:21] >> right? And so, yeah, you could see how [202:23] this is an existential threat. Like, the [202:25] traffic is growing. Like, oh my gosh, [202:27] what if this becomes AOL all over again, [202:29] >> right? And that's the main thing. One [202:30] tier down from that is Facebook doesn't [202:32] even allow other ad servers. [202:34] >> At least with AOL, we could do a deal [202:35] with them and power their monetization. [202:38] Facebook just hired Cheryl Samberg. [202:39] They're doing this themselves. [202:40] >> They're doing it all inhouse closed loop [202:42] system. [202:43] >> Yep. [202:44] >> So, Google+, what was Plus and how did [202:47] it get built? So, it was a one-year [202:49] sprint following this point, the 50 [202:52] getting together, and it was built in a [202:54] very, very ungooy way. It was not [202:58] organic, David, like these passion [203:00] projects you're talking about. [203:01] >> It was instilled from on high down upon [203:05] all of the products. [203:06] >> It was not based on a core technical [203:08] insight. It was not consensus driven. It [203:10] was top- down command and control style [203:12] led by the person that you mentioned [203:14] earlier, Vic Gondotra. Now, who was [203:17] Vicond Doorra? [203:18] >> Vic was this interesting character. Like [203:19] we said earlier, he had been leading [203:23] Google's developer efforts in the [203:26] pre-Android days [203:27] >> and he was sort of the frontman. He was [203:29] the MC at Google IO, [203:31] >> right? If you were looking for somebody [203:33] to communicate and push down this new [203:36] top down vision across the company, he [203:38] would be a logical choice. [203:39] >> Yes. I don't know if he raised his hand. [203:41] I don't know if Larry said, "Hey, I [203:42] really think you should do this on our [203:44] behalf." But what is definitely true is [203:47] it became Vic's thing and Eric and Larry [203:51] and Sergey stepped back and let Vic run [203:55] with it. And he was given an enormous [203:58] amount of institutional authority. [204:00] >> And we should say too, you alluded to [204:01] this earlier, what was Plus? It wasn't [204:04] just a social product in and of itself. [204:07] It was baked into all Google. It was [204:11] inserted into every other product that [204:14] Google had. There's a quote from Vic to [204:17] the press at the time about this about [204:19] what Google+ is. He says, "This is the [204:21] next generation of Google. It is Google [204:25] plus one." Oh boy. Oh boy. [204:28] >> There's a lot of these even saying that. [204:30] >> Oh, gives me the heebiejibbies. So yeah, [204:32] it was a Facebook style thing, but its [204:34] goal in addition to being a Facebook [204:36] style thing was to leverage all of [204:38] Google's assets and make all Google [204:41] things Google+ things. So they moved big [204:45] headcounts out of each team and onto the [204:47] Google+ team. They reached deep to [204:49] integrate with these other products and [204:50] it's very clear who the boss was in all [204:52] these negotiations. You had a clear [204:54] mandate like your job this half year, [204:56] this year is do these plus integrations. [204:59] >> Yeah. your OKR. Google famously ran on [205:01] OKRs was now all about plus pluses. [205:04] >> And Danny Kitton, who would go on to [205:06] become the managing editor at Techrunch, [205:07] at this point in time was a Google [205:09] intern. And he wrote about it later. And [205:11] he said, "Due to this integration, much [205:13] of it was forced. The culture around the [205:14] company at Google had become deeply [205:16] poisonous by the time I started. I still [205:18] remember talking to one member of the [205:20] Picassa team, who is Google's photo [205:22] repository that they bought, who told me [205:24] to f off when I asked about integrating [205:27] plus into the product. He was hardly the [205:29] only one. Companywide bonuses were based [205:32] on the success of Google+. They even [205:34] went so far as to put little plus one [205:36] buttons on mobile advertisements like [205:38] those little banner ads at the bottom. [205:40] >> Yes. This is the best. Google had bought [205:43] ADM mob. [205:44] >> Yes. [205:44] >> And the mobile display ad units. [205:47] >> You could plus one it. Who the hell [205:49] wants to plus one an ad? I mean, this is [205:51] like Facebook's like button but plus's [205:54] version. And they're like any Google [205:56] thing should be plus oneable. So they [205:59] even reached into YouTube comments and [206:01] YouTube comments became Google+ posts. [206:06] >> I mean they almost killed the golden [206:07] goose, [206:08] >> right? They almost killed all of these [206:10] golden goosees that they had. [206:12] >> Yes. And so plus from a product [206:16] perspective, it wasn't just Facebook. [206:18] They brought a lot of really interesting [206:20] ideas. Google Hangouts came out of this. [206:22] Google Photos came out of this. There [206:24] were these things called sparks. I mean, [206:26] they really rethought a lot of social [206:28] networking. The issue is nobody really [206:31] wanted to rethink social networking. [206:33] That was a Google priority to get people [206:35] to use this, not a user-driven one. And [206:38] they tried to essentially put rocket [206:41] fuel on to scale something that really [206:43] didn't have product market fit. Well, I [206:45] really think the key huge mistake with [206:50] Google Plus, one of the huge mistakes [206:52] with Google+ was [206:53] >> you don't need a Facebook when there's [206:54] already Facebook. [206:56] >> Not even that, Facebook was already [206:58] dying. Mark Zuckerberg had already [207:00] realized that the future of social was [207:03] not what it looked like at this point in [207:05] time. M [207:07] >> as Google is launching Google+ [207:09] >> in 2011 June 2011 [207:11] >> 2011 2012 2013 these were the big years [207:14] for Google+ what is Mark Zuckerberg [207:16] doing he's buying [207:17] >> Instagram [207:17] >> he's buying WhatsApp and he's remaking [207:22] essentially you know Facebook into what [207:24] meta would become of like hey what we [207:27] used to think of as social networking [207:29] has bifrocated into two things public [207:32] media [207:33] i.e you know, YouTube, Instagram, UGC, [207:36] and private messaging. And here's Google [207:40] >> launching, I kid you not, this is the [207:42] craziest thing, desktop first [207:46] >> with a desktop only UI to arrange your [207:50] friends into circles. [207:51] >> Circles. That's right. Circles. [207:53] >> Which is on its own, it's such a [207:56] computer science way of thinking about [207:57] it. Oh, my friends are in sometimes [207:59] overlapping, sometimes not overlapping [208:01] groups that I want to carefully label so [208:04] that I can identify deterministically [208:06] who I want to share what with, [208:08] >> right? Nobody wants to do that. [208:10] >> Yeah. Here's the thing that just leapt [208:12] out to me about plus. This was Google's [208:15] Windows Longhorn/ Windows Vista. In our [208:19] Microsoft saga, we talked about how [208:21] Vista/Longhorn was the most damaging [208:24] thing to the company because of the [208:26] distraction and the siphoning of [208:28] resources and the best talent away from [208:31] working on what really mattered. Now the [208:34] question I was asking myself and others [208:36] in research of okay what were the [208:38] negative consequences of that? So like [208:40] with Microsoft it was clear Google who [208:43] was the negative consequence like the [208:45] whole reason Microsoft let Google fester [208:48] from their perspective for all these [208:50] years and didn't kneecap them was they [208:53] were tied up with all the distraction [208:54] from Vista [208:56] >> and losing relevance with developers [208:58] because they keep selling them a [208:59] platform that kept not shipping and then [209:01] when it eventually did ship it wasn't [209:02] good [209:03] >> right so then I was sort of trying to [209:05] figure out like okay what what are the [209:06] similar consequences for Google of the [209:10] Plus era. And at first, I couldn't [209:12] really think of any. I was like, "Oh, [209:13] well, Android's pretty good. YouTube's [209:15] pretty good. Chrome's pretty good. [209:17] Search is still pretty good. Gemini AI [209:19] comes out later. It's all pretty good." [209:21] But there were two things. [209:23] >> Messaging probably. I bet in a a non [209:25] plus world, WhatsApp, something like [209:27] that could be owned by Google. [209:29] >> Yeah. Two things. One is messaging. [209:31] Totally missed messaging. So when I was [209:33] a business school student at Stanford, [209:35] Eric was now executive chairman and he [209:38] started co-eing a class at GSB and I [209:40] took his class. I was one of his [209:42] students during these years. It was [209:43] awesome. It was one of the best classes [209:44] I ever took. [209:45] >> And the quarter when I was taking the [209:47] class was when Facebook bought WhatsApp. [209:50] And I remember [209:51] >> Eric coming into class right after it [209:53] happened and just being like, "God damn [209:55] it, we missed it. We totally missed it." [209:58] It's because Google was distracted. So [210:00] that was one thing. And then I realized [210:01] the other as bigger or bigger thing is [210:04] cloud. Google should have been massively [210:08] investing in cloud. And there were all [210:11] sorts of reasons that they did. We're [210:12] going to save this for the next episode, [210:14] but I was like, yeah, especially think [210:17] about where the impetus for this came [210:19] from from this memo, the zquake memo as [210:22] it's known. Should have been focused on [210:25] cloud. [210:27] he should not have been focused on [210:28] social and Google had the wrong strategy [210:31] in cloud for many years and as a result [210:35] that's why their cloud business is way [210:37] behind Amazon and Microsoft [210:39] >> and maybe they turned some talent maybe [210:41] there was some good people that got [210:42] burned by the culture sort of souring [210:45] you could argue this destroyed product [210:47] velocity [210:49] >> so like people complain today that [210:51] Google's always working on really [210:52] interesting technology and they just [210:54] never get cool products out the door [210:56] That is by far the biggest complaint you [210:58] hear about Google from [211:00] >> slow and big and bureaucratic and [211:02] >> yep folks on the inside and outside [211:04] these days is just like too slow and [211:06] Yeah. Yeah. That's probably the biggest [211:08] negative consequences. [211:08] >> Maybe you could trace that here. [211:10] >> I bet you can cuz think about it before [211:12] this we just spent this whole episode [211:14] talking about all these amazing things [211:16] they were building and shipping and [211:17] acquiring and transforming [211:19] >> until Gemini and arguably that's a I [211:22] actually don't know where Gemini stacks. [211:24] Is it a third place product? Still a [211:25] question mark. [211:26] >> Until Gemini, what great breakthrough [211:29] consumer service did they launch after [211:31] Google+? [211:33] >> I got nothing. [211:34] >> O. [211:34] >> Yeah, [211:35] >> that's pretty wild. [211:36] >> After having this incredible 10-year [211:38] run, [211:39] >> and there was a lot of stuff they tried. [211:41] I think some things for Android users. [211:43] Think about Google Now that predated [211:45] Google Assistant, maybe Google Home. [211:48] >> These are not worldchanging products, [211:50] >> right? It's funny. The thing that I keep [211:52] thinking about from the Google+ failure [211:53] is this big existential Facebook threat [211:56] they were worried about. I mean there [211:57] was a strategy memo in 2013 where Neil [212:01] Mohan said there is a risk that Facebook [212:03] becomes the starting point of the [212:05] internet. Google knew social was the [212:07] future and tried to win it. But [212:09] interestingly they didn't and they've [212:11] been fine. [212:12] >> Right. Right. Right. It was all totally [212:14] fine. [212:15] >> And Facebook was really freaked out too [212:17] that Google was going to come in and win [212:19] it. I mean, Google was like this giant [212:21] and Facebook was recently public going [212:24] through their own problems. So, even [212:26] though it was kind of like a nothing [212:28] burger and plus was a footnote in [212:29] history, both companies were completely [212:32] allin on this is the big battle and [212:35] ultimately Google wasn't a credible [212:38] threat to Facebook and [212:40] >> Facebook went in a different direction [212:41] anyway. [212:42] >> Facebook went in a different direction. [212:43] Yeah. So, it's almost like the end of [212:45] Burn After Reading. Have you Have you [212:47] ever seen that movie? [212:48] >> No. I won't spoil anything, but the [212:50] feeling you have at the end is you just [212:51] watched all this crazy stuff happen and [212:52] you're like, whoa, wait, did any of that [212:56] matter? That's how Google+ feels to me. [213:00] >> Yeah. [213:01] >> Funny. Plus did have two great surviving [213:04] products. Hangouts, which became Meet, [213:06] and Photos. [213:07] >> Yep. Photos is a billion user product [213:10] today. [213:11] >> Wow. Huge. The biggest thing and I think [213:14] this is like kind of getting back to my [213:17] original postulate of never waste a [213:20] crisis. You know what we have today? [213:22] Google accounts. [213:25] >> Yep. [213:26] >> You know what Google is today? It's one [213:28] company. It's not these little thiefs [213:30] here and there of different people [213:32] amassing power and building things in [213:34] different ways. I mean, I'm sure there's [213:36] still plenty of that. Everything about [213:38] Google got more unified from this era. [213:39] they have a failed product and a smoking [213:41] crater to show for it, but a unified [213:44] look across all their products, a [213:45] unified login that would, I think, be [213:47] pretty important for them going forward. [213:49] Anyway, my snarky finish on all this is [213:52] it's tempting to say Google lost in [213:54] social because this giant smoking [213:57] creator, but actually all of social [214:00] ended up pivoting to either look like [214:02] messaging or like YouTube. Anyway, [214:05] YouTube is kind of the winning paradigm [214:08] in quote unquote social media in UGC [214:10] media. [214:10] >> Yep. [214:11] >> So, they should have just done nothing [214:13] and just watched the money printer go [214:15] burr. [214:16] >> Yep. [214:18] To put a bow on it, uh, Vic ends up [214:20] leaving the company in 2014. In 2019, [214:23] they finally shut Plus down. There's a [214:26] blog post about it. They cite like a big [214:28] security breach as the reason like, "Oh, [214:30] no. We've discovered there's this huge [214:31] security vulnerability. Thus, we need to [214:34] shut down all of Google+. [214:36] >> Dude, it's so bad. There's been 50 [214:38] Google products that all sound kind of [214:39] the same. They launched this one called [214:41] Currents at one point and when they shut [214:46] Plus down, this is horrible. Many people [214:49] wrote like articles as posts on Google+ [214:53] and they're just gone. [214:54] >> Yeah, that's right. Actually, it was [214:56] sort of an impediment to doing some of [214:57] the research for this episode because [214:59] these posts are gone. [215:00] >> So, if you go to plus.google.com/ [215:01] google.com/ [215:03] anything. It just redirects you to [215:05] Google Currents. However, Google [215:07] Currents has now been shut down. So, it [215:09] is a Google Workspace blog post [215:11] announcing the Currents shutdown. Every [215:14] time you click any Google+ link anywhere [215:16] on the web, you go to a blog post that [215:18] tells you about the shutdown of [215:20] currents. That's the most googly thing. [215:23] >> Ah, [215:24] >> they got to do a better job with those. [215:26] Well, by 2015, [215:28] >> it's our last section, the bridge to [215:30] Alphabet. [215:31] >> Yeah, it's clear it's time for a new era [215:34] at Google. The company announces that it [215:38] is reinventing itself, is becoming an [215:42] entirely different company. Google is [215:45] becoming Alphabet in August of 2015. And [215:49] Larry Pageige will be the CEO of this [215:51] new Alphabet holding company. Sundar [215:55] Pachai will be the CEO of Google, which [215:57] will be by far the largest and really [216:01] primary operating company within [216:04] Alphabet. Interestingly, they didn't at [216:06] all decide to split up YouTube or any of [216:09] the products and yeah, that's they just [216:11] spent all these years unifying it all. [216:13] That's all Google. [216:14] >> They broke out Google X. [216:16] >> Yes, Google X. They broke out. Whimo is [216:18] still part of X at this point in time [216:20] that would later spin out is now part of [216:22] Alphabet on its own. But the other bets [216:25] including an Alphabet really quite [216:27] clever the nomenclature here. Yes. [216:29] >> Uh were Nest that they had just [216:31] acquired, Google Fiber, Calico and [216:34] Verily their two health companies Google [216:36] X Lab and then Google Ventures and [216:39] Capital G the two investing entities [216:41] that they had. So you know then really [216:44] the question is like okay well why did [216:46] they do this? But why did Larry become [216:48] CEO of Alphabet? Why did Sundar become [216:50] CEO of Google? [216:51] >> I think this kind of had to happen as [216:53] like a healing after [216:56] >> plus. Sununar was a leader who had real [216:58] cred going back to the early days and [217:01] with Chrome and with Android, like the [217:03] core [217:04] >> great products, two of these core great [217:06] products, platforms that we've talked [217:08] about the whole episode [217:09] >> that have really driven the Google [217:12] flywheel all along. Interestingly, he [217:14] had never worked in search or ads, [217:16] >> right? But these were the platforms that [217:18] had shoehorned Google into the mobile [217:21] era and protected it from its greatest [217:23] existential threat and just also [217:26] Sundar's personality. I think was a way [217:28] to reunify the company and bring [217:30] everybody back together. [217:31] >> Definitely strikes me as a peacemaker [217:34] among big egos. [217:35] >> Yes. [217:36] And that is where we're going to leave [217:40] Alphabet/Google [217:43] for the moment. Ben, give us a sense of [217:45] how big this company had gotten. [217:47] >> So, at the end of 15, it's gotten huge. [217:49] It's 75 billion in revenue. 52 billion [217:52] of that is first party sites, Google [217:55] websites, Adwords, Gmail, Maps. 15 [217:59] billion, the the smaller part is over in [218:01] doubleclick AdSense land. And actually, [218:03] that's pretty low margin revenue. So [218:05] again, the lion share in Google [218:07] websites. YouTube is profitable at this [218:09] point and their bottom line operating [218:13] income. Google did about $23 billion in [218:18] operating income and their other bets at [218:20] this point lost about 3.5 billion. Their [218:22] other bets are extremely interesting and [218:24] will be the focus of our next episode. [218:26] But the big takeaway here, the business [218:28] was still in 2015 and essentially is [218:31] still today search ads. [218:34] >> Yep. And what so strikes me listening to [218:37] you say those numbers in 2015, a they're [218:41] huge, but also Google is so much bigger [218:44] today on these same businesses with this [218:47] same business model, [218:48] >> right? [218:49] >> There was another 5x scaling to go over [218:51] the next 10 years. [218:53] >> Yes, it's crazy. [218:54] >> Google back then was like 20% the size [218:57] of Google now. And nothing has basically [218:59] changed when it comes to the business [219:01] model and products. I mean, nothing's [219:03] changed since 2002, [219:05] >> right? Well, I think this era, what we [219:07] talked about all the episode, all the [219:08] hits were stewarding that business [219:12] through these sea changes, [219:14] >> but nothing has changed about what the [219:16] core business is. It just turned out [219:18] that that seed of an that search ads [219:20] actually scaled to the biggest market in [219:23] the world. [219:23] >> Yep. All right. Just like the last [219:26] episode with Gmail at the end, I've got [219:28] one little koda, one little teaser for [219:30] next time. Great, [219:32] >> Ben. What if I told you that between [219:35] 2015 and 2016, so this next year, this [219:39] next 12 months after the Alphabet [219:42] transition, [219:44] all of the following people were Google [219:49] employees. [219:50] Alex Kresevki of AlexNet, [219:56] the dawn of machine learning, AI. uh his [220:00] PhD adviser Jeff Hinton the godfather of [220:03] AI his collaborator on the AlexNet paper [220:09] Ilia Skekkever [220:11] founding scientist of open AI Dario [220:14] Amade co-founder with his sister of [220:18] anthropic Andre Karpathy [220:22] until recently chief AI scientist at [220:24] Tesla Chris Ola Nome Shazir Ian [220:28] Goodfellow Hello. And of course, the [220:30] co-founders of DeepMind, which Google [220:33] acquired in 2014, Dennis Hassabis, Shane [220:36] Le, and Mustafa Sullean. Mustafa runs AI [220:40] at Microsoft today. Andrew from [220:44] Stanford, Quac Lee, Oral Vignales, and [220:47] oh yeah, in addition to all of those [220:49] people, the authors of the transformer [220:54] paper because [220:56] Google invented the transformer and [220:59] published the paper in June of 2017, [221:02] >> right? Which is the novel mechanism that [221:04] all LLMs today from every big [221:07] foundational model research lab is based [221:09] on. When I was talking to folks in the [221:12] research for this episode and AI came [221:14] up, one of them said, you know, I have [221:16] to remind people when I'm talking to [221:18] partners out in the ecosystem that the T [221:22] in chat GPT [221:24] stands for transformer and that we [221:25] invented that [221:28] because it is also during this time [221:30] while Ilia is working at Google that he [221:33] poses the question to his research [221:35] colleagues and the Google brain team [221:37] that is working on all of [221:40] Gosh, what do you guys think if we just [221:42] built one really, really, really big [221:45] neural network and we set it loose with [221:48] training data on the entire internet, [221:51] which by the way, of course, we can do [221:53] here at Google because uh you know, [221:54] thanks to the combination of the search [221:56] index, we index the entire internet and [221:59] all the products that we just talked [222:00] about on this whole episode, we have all [222:02] this data [222:02] >> and all this content out there. If we [222:05] did that, do you think it would learn [222:08] everything? [222:10] >> Well, David, that feels like quite the [222:12] groundwork for the next episode. [222:13] >> That feels like a story for next time. [222:15] Yeah. [222:17] But through that lens, there is another [222:19] way to view everything that happens at [222:22] Google during this 10-year period we [222:24] just discussed, which is that they're [222:27] just collecting all the access, all the [222:29] information, and all the talent for AI. [222:32] >> It's nuts. There's this whole other [222:34] world of research. Who would be the [222:36] people that would drive the next decade [222:38] or five decades of change? And they [222:41] basically had them all in one place at [222:43] one time. They were all employees of [222:45] Google. So I want to end with one more [222:48] quote. This time from Larry Page all the [222:52] way back in the year 2000. This is Larry [222:56] talking in the year 2000. [222:58] Artificial intelligence would be the [223:00] ultimate version of Google. So if we had [223:03] the ultimate search engine, it would [223:06] understand everything on the web. It [223:09] would understand exactly what you wanted [223:11] and it would give you the right thing. [223:13] That's obviously artificial intelligence [223:15] to be able to answer any question [223:17] basically because almost everything is [223:19] on the web, right? We're nowhere near [223:21] doing that now. However, we can get [223:24] incrementally closer to that and that is [223:26] basically what we work on and that's [223:28] tremendously interesting from an [223:30] intellectual standpoint. We have all [223:32] this data. If you printed out our index, [223:34] it would be 70 miles high now. We have [223:36] all this computation. We have about [223:38] 6,000 computers. This is 25 years ago. [223:42] We have enough disk space to store like [223:44] a 100 copies of the whole web. So, you [223:47] have a really interesting confluence of [223:49] a lot of different things. A lot of [223:50] computation, a lot of data that didn't [223:53] used to be available. And from an [223:54] engineering and scientific standpoint, [223:57] building things that make use of this is [223:59] a really interesting intellectual [224:00] exercise. So I expect we'll be working [224:03] on that for a while. [224:06] Incredible. This is 25 years ago that he [224:09] said this. [224:10] >> Amazing. [224:11] >> All right. Should we do some analysis? [224:13] >> Let's do some analysis. [224:14] >> All right. Let's do power. And for those [224:18] who are new listeners, power is the [224:20] section where we analyze which of the [224:22] seven powers does Google have from [224:24] Hamilton Helmer's framework that enables [224:26] a business to achieve persistent [224:28] differential returns or be more [224:29] profitable than their nearest competitor [224:31] and do so sustainably. Google is very [224:33] very weird to analyze for this because [224:37] most of the way you think about Google [224:38] is actually not where the economic [224:40] transaction is. If you want to analyze [224:42] the business, it is why are advertisers [224:45] spending a marginal dollar with Google [224:47] versus spending it elsewhere. And Google [224:50] has the seven powers that show up in [224:52] numerous instances all over their [224:53] business. But I think the interesting [224:55] way for us to do this analysis, David, [224:57] is let's look at each one. Just assume [224:59] Google has them all and say where is the [225:01] biggest or a very large example in our [225:03] mind of where each of them show up. [225:05] >> Great. I like that. So [225:07] counterpositioning [225:09] typically doesn't show up for incumbents [225:11] for large companies. [225:12] >> This is the exception though with [225:15] Google, [225:15] >> right? Where you just look at their sort [225:17] of new businesses for example in this [225:20] episode talking about Android. They [225:22] massively counterpositioned against [225:24] Microsoft, the less than free business, [225:26] >> less than free business model. I mean [225:27] this is the clearest example of [225:29] counterpositioning I think that has ever [225:31] existed. Oh hey, my competitors require [225:33] you to pay them. How about I pay you [225:35] instead? [225:36] >> Right? And my competitors can't do that [225:38] because they don't have the business [225:41] model of advertising based on search [225:44] such that they can justify doing this [225:46] >> right scale economies especially as [225:49] they're adding all these apps all these [225:51] users across all this surface area. Now, [225:53] if you're an advertiser and you want to [225:55] reach users, you know, across search or [225:57] display or video, you like Google is a [226:00] one-stop shop, [226:01] >> right? You don't have to sort of [226:03] independently spend operational time and [226:05] headcount on all these different [226:06] platforms. You sort of get the one [226:08] >> and that's not even to mention the scale [226:10] economies on the infrastructure side we [226:12] talked about last time or like I mean [226:13] they show up in every business here, but [226:15] like that's just one example. And the [226:16] fact that the more advertisers there are [226:18] and the more users there are, the more [226:21] profit Google makes because each little [226:23] individual auction on every individual [226:26] search finds a maximal price. [226:28] >> Yes. Network economies. YouTube. [226:32] Hello. More creators, more viewers. [226:36] Creators make money from having views of [226:40] their videos. [226:41] >> Application developers on Android and [226:44] users on Android. the two-sided network [226:46] economies there. Yes, everywhere. Not to [226:49] mention, in the core business too, in [226:50] search, more users searching is more [226:53] valuable to me as an advertiser because [226:54] I have a deeper pool of people I can [226:56] advertise to. So, I can just deploy more [226:58] dollars on your channel if it's working. [227:00] >> Yep. Switching cost. How about Gmail? [227:03] Oh, I've got my last uh 20 years of [227:07] email history in Gmail all stored for [227:11] free. Uh yeah, I'm not switching. In the [227:13] core business, there's not as much [227:15] switching costs. I suppose there's a [227:17] little bit of, oh, because I've spent a [227:19] lot of money, the targeting is very good [227:21] at allocating my spend, but the [227:25] switching costs in the core business for [227:27] an advertiser are not as prominent as [227:30] other powers. I don't think I continue [227:33] to spend on Google because it's hard to [227:35] switch. I continue to spend on Google [227:36] because they have all the high intent [227:38] users for products other than, you know, [227:41] Amazon. That's one of the two big search [227:43] boxes in the world where people type in [227:45] when they want to buy a product. So, I'm [227:47] going to advertise there. Has little to [227:48] do with switching costs, I think. [227:50] >> Yeah. But for users of Gmail, [227:52] >> oh, and several of the other products [227:54] like enormous [227:55] >> for users all across the board. Yeah. I [227:58] won't leave YouTube at this point. The [228:00] algorithm's dialed to my interests. [228:02] >> Oh, yeah. That's a great point of [228:03] switching costs of the algorithm on [228:05] YouTube. [228:06] >> Yeah. [228:07] >> Branding. I think in the heyday of [228:10] Google that we're talking about in this [228:12] episode when they're launching all these [228:13] incredible products, yes, these [228:16] products, one because they're incredible [228:17] and because they were free, but also [228:19] like there was such a halo around the [228:21] company, like if there was a new Google [228:23] product, I would be chomping at the bit [228:25] to go try it. [228:26] >> Yeah, that's super true. I remember I [228:28] was desperate for Google Wave invites. [228:29] The product completely failed, but I was [228:32] completely dazzled by it and I was [228:33] desperate to get invite and access. The [228:36] Google name meant something. Yep. [228:37] >> And still does, by the way, which I [228:39] think held them back in AI for a while. [228:41] They know the Google name means [228:42] something, so they are reticent to to [228:45] throw their name on it until they got [228:46] kind of shoved off the cliff. [228:48] >> Yep. Cornered resource. [228:53] Yes. Well, certainly heading into the AI [228:55] era now. YouTube, the YouTube catalog [228:58] you can train on. [229:00] >> Yeah. [229:01] >> All the data they have. [229:03] >> Yeah. It's funny. Yeah. I was about to [229:04] say their infrastructure, but I think [229:06] that's actually a scale economy that [229:07] they've built out the infrastructure [229:09] they have so they can run all their [229:11] products as cheaply as they can. [229:13] >> Yep. I think the infrastructure is also [229:15] a process power. [229:17] >> Yeah. [229:17] >> In that in the era we've been talking [229:19] about, they could launch all these [229:20] products on their infrastructure just [229:22] way cheaper than anyone else. [229:24] >> You know what's a corner resource? They [229:26] have built internal software and systems [229:28] that is better than what is available [229:30] outside of Google. [229:32] >> Great point. A lot of the time they even [229:34] create open- source projects that are [229:36] similar to their internal stuff, but [229:38] they don't actually give away the [229:39] internal stuff. Inside Google, they [229:41] still run Borg. They run far less [229:43] Kubernetes than they run Borg. And Borg [229:46] is part of the secret sauce. [229:48] >> Yeah. When you talk to engineers who've [229:50] left Google, they miss the [229:53] infrastructure. [229:55] >> So Google has it all. [229:57] >> And we can name a lot more examples, but [229:59] we got to go. [230:00] >> All right. Playbook. [230:02] >> All right. I tried to get most of them [230:03] in as we were going in the story. The [230:06] first is that Google really wanted to [230:08] become a platform company and I was sort [230:11] of noodling on did they ever do this [230:13] successfully. And David, we sort of [230:16] touched on this idea that they are [230:19] advancing the platform of the web [230:21] without owning the platform of the web. [230:23] So if they didn't have Android, how [230:26] would you answer the question, is Google [230:28] a platform company? [230:30] >> Yeah. Yeah. Yeah. [230:32] I would say it's sort of like a shadow [230:34] platform company. It's like an ecosystem [230:35] company, [230:36] >> right? And even with Android, okay, [230:38] great. They own the target development [230:40] platform. Their money is still made [230:43] elsewhere. It's not a platform business. [230:46] They may have a platform orientation as [230:48] a company. They build a bunch of stuff [230:50] that for developers to build their [230:51] applications on top of, but where their [230:54] bread is buttered is really as an [230:56] advertising company. It's important when [230:58] push really comes to shove on big [231:01] strategic decisions the company has to [231:03] make. [231:03] >> Yeah. [231:03] >> Like Apple pure play platform company. [231:07] Microsoft pure play platform company. [231:09] They either sell software or hardware [231:11] and then they need the platform around [231:12] it to bolster their sales. Google's very [231:15] indirect. [231:17] >> Yep. [231:17] >> All right. So that was one. The other [231:19] one is they make tons of small [231:21] acquisitions famously. I mean that that [231:24] run in the 2010s. Aside from the big [231:26] ones from YouTube, from Android, from [231:27] double click, from ADM Mob, there was [231:29] also what became Google groups, [231:31] spreadsheets, Docs, Blogger. They bought [231:34] Applied Semantics with the patents and [231:35] some of the tech for AdSense. They [231:37] bought the technology for Google Maps. [231:39] They bought Urchin for Google Analytics, [231:41] Dodgeball, Feed Burner, Recapture, [231:44] Slide, Jambool, Like.com, Widevine, [231:46] Admeld, Punched, Zaget, Sparrow, Wavy. I [231:50] mean like I could just keep going. Oh [231:52] yeah, there are hundreds of companies [231:53] they bought. In talking to folks in the [231:55] research, there was this amazing part of [231:57] Google culture that also fit the [232:00] strategy perfectly of help the web and [232:02] the rich web and web apps bloom. Come [232:06] work at Google with these incredible [232:08] people, meet your co-founders, go start [232:11] a startup, leave Google, [232:13] >> right? [232:13] >> We will then reacquire you back into [232:16] Google in a couple years. [232:18] >> It happened dozens or hundreds of times. [232:20] I remember seeing this happen from the [232:22] outside and thought Google is nuts to [232:24] let this happen. But I realize now, no, [232:27] this was all part of the strategy. [232:28] >> Yeah, [232:28] >> it's all good for the web. [232:30] >> Yep. You can run very indirect, [232:33] generous, long-term strategies like that [232:36] with a money printer like AdWords. [232:38] >> Yes, [232:39] >> I know I keep coming back to that, but [232:41] that is at the core of what drives [232:43] everything. This one's a little bit less [232:45] playbook, but just an observation. I [232:47] watched the Google IO keynote with Glass [232:50] and I watched a bunch of Glass content. [232:52] I even back in the day at a startup [232:54] weekend launched a Google Glass app. [232:56] >> Oh, nice. [232:58] >> So, after watching all this Glass [233:00] content, and it's the butt of every joke [233:02] now, Meta Ray-B bands and Google Glass [233:06] are the same thing feature-wise. [233:09] >> Yep. [233:10] >> The gestures on the side, the fact that [233:13] it could take a photo. I mean, Google [233:15] Glass was a little more advanced. could [233:16] run these like very basic text based [233:18] apps, but I'm sure when Meta launches [233:21] their little hologram version of the [233:24] glasses, that's going to be eerily [233:26] similar. So, like, of course, you could [233:28] say, "Oh, it's just timing." But here's [233:30] the thing. Google's made you look like a [233:31] cyborg. Metas is for normal people. And [233:34] there is no better metaphor for the [233:36] cultural difference between Facebook and [233:38] Google than this. [233:41] Google's a bunch of wacky academics who [233:43] did not really understand why this would [233:46] make the product fail. Facebook is [233:47] founded on the idea that you're trying [233:49] to be cool. [233:50] >> Yeah. And went and did a partnership [233:51] with SLO Exotica to get the tech Yes. [233:55] >> into glasses that normal people wear. [233:59] >> Yes. It was crazy watching these demos [234:02] because I'm like these are the meta AR [234:04] demos. It just happens to have a cool [234:07] factor versus not. [234:09] >> Yeah. And then my last one is this idea [234:12] that they did figure out a way [234:14] culturally to get people amped about [234:17] just build great products. Figure out [234:20] how to do something really hard from an [234:22] engineering perspective that ends up [234:24] being really useful and ship things that [234:26] people love. It's not that you didn't [234:28] have to think about a business model, [234:29] but a lot of the time for many years [234:32] after launching a product, you really [234:33] didn't. [234:34] >> Yep. It's like we talked about earlier, [234:36] there was this thin layer of really, [234:39] really, really tight, really great [234:41] strategy that was just like a few people [234:44] at the top of the company, but below [234:46] that was just make a great product. [234:48] >> Yes. All right. I've got two for [234:52] playbook. One that I'm going to make my [234:54] quintessence. Just want to underscore [234:56] again, we said this in the Android [234:59] chapter, but like Android was the mother [235:01] of all wins. It was so big to win with [235:06] Android. [235:08] Nobody stretches a business model across [235:12] technology eras. Nobody. And Google did [235:15] it [235:16] >> in a dominant way where they are they [235:17] are the dominant company in the next era [235:19] as well. [235:20] >> Yeah. It is the Google version of Azure [235:23] from our Microsoft series. It absolves [235:25] any and all sins. Not that there were [235:27] many at Google. The only one was plus. [235:29] The only way it could have gone better [235:30] is if instead of launching Android, they [235:32] launched the iPhone and they also got [235:34] the iPhone profits rather than just some [235:38] small dollars that protected their core [235:41] business. [235:42] >> Yep. That was my playbook. And then my [235:45] quintessence is it is wild that this one [235:48] company has eight products with over a [235:52] billion users and started this era with [235:56] just one search that didn't even have a [235:58] billion users yet. [235:59] search, Android, Chrome, YouTube, Gmail, [236:02] Maps, Drive, Photos, and then if you [236:05] count the Play Store as separate from [236:07] Android, which Google does, I think [236:08] that's a bit of a stretch, but if you [236:10] do, then they have nine products with [236:13] over a billion users. And just for [236:15] context, Meta is the next highest count [236:19] of products in one company with over a [236:22] billion users. They have four, the Blue [236:24] app, WhatsApp, Instagram, and Messenger. [236:28] Metal likes to claim they have five. [236:30] They like to say that Meta AI in [236:32] aggregate has over a billion users [236:34] embedded across all their products. [236:35] >> Yeah. But this whole super intelligence [236:36] thing is an admission that the active [236:39] users of Meta AI is a little uh [236:41] stretchy. [236:42] >> If the Play Store doesn't really count [236:44] on its own, the Meta AI for sure doesn't [236:46] really count on its own. So Meta has [236:48] four. [236:48] >> Apple I think only has three, maybe [236:52] four. So, the three Apple has for sure [236:55] are iPhone, iMessage, and Safari. [237:00] iPad, maybe? I don't think so. Mac, [237:04] definitely not. [237:05] >> I basically don't count any iPhone app [237:07] because they all come for free when you [237:08] get the phone. [237:09] >> Okay. So, by your definition, Apple has [237:12] one with iPhone. [237:13] >> I think Apple has one. [237:14] >> Okay. All right. [237:16] >> Let's take that same definition. How [237:17] many of these came for free at Google? [237:20] Google search, Android, those are two [237:23] completely different distribution [237:24] channels. Chrome. [237:25] >> Yep. I don't think any of these came for [237:27] free. [237:28] >> I mean, Google helped Chrome. [237:30] >> No. Yeah. These are all independently [237:32] have achieved billion plus users. [237:35] >> Gmail and Google Drive sort of advantage [237:37] each other. So, I think you can sort of [237:39] subtract one of those out. [237:40] >> Yeah. But it's not to the extent that [237:42] iMessage is default with an iPhone, [237:44] >> right? Maps is advantaged by Android. [237:46] They ship a whole lot of maps. So, but [237:49] probably whatever the phone was would [237:50] have a great Google Maps app. [237:52] >> Yep. Okay. All right. I buy it. Apple [237:54] has one. Microsoft has two. Windows and [237:57] LinkedIn. Amazon doesn't have any [238:00] billion user products. Google's got [238:03] eight. Like, that's incredible. [238:05] >> Call it seven or six. I think it's [238:07] reasonable to subtract. [238:08] >> Okay, fine. [238:09] >> But still, that's exactly right. [238:11] >> Whatever. That's my point. This is my [238:13] quintessence. This period at Google is a [238:18] run like nobody's ever had. [238:20] >> Yeah, absolutely right. [238:22] >> All right, what you got? [238:24] >> So, quintessence for me is the thing [238:26] that I sort of can't stop thinking about [238:28] from the episode and I decided this time [238:31] I knew what it was going in and I [238:33] decided to hide it all the way until the [238:35] end. So, we haven't talked about this [238:36] thing yet. [238:36] >> Oh, okay. Almost all of Google's [238:39] successful products are based on a core [238:43] technology insight that is underneath [238:45] the whole thing. [238:47] >> The type of insight that could be in an [238:50] academic journal. [238:51] >> Yep. [238:52] >> Someone told me this and I've been using [238:54] it as a little litmus test for will a [238:57] product work or not. And as you look [239:00] through I mean you look at the original [239:02] search that is by definition the page [239:05] rank algorithm. It's a core technology [239:08] insight. [239:08] >> I mean, they published it as an academic [239:10] paper. [239:11] >> The way that the ad-based auction works [239:14] is a core, it's almost mechanical in its [239:17] elegance and its brilliance and its [239:18] simplicity. It is a technology insight [239:21] that and everything we talked about in [239:23] Google part one, our first episode. Then [239:25] you look at everything that succeeded [239:26] this episode. Gmail, the way that they [239:29] are able to do the gigabyte of storage, [239:32] Ajax, fast responsive web application. [239:36] You look at maps and docs with real-time [239:38] collaboration, breakthrough core [239:40] technology insight. [239:42] >> Yeah, [239:42] >> YouTube. Uh, [239:44] >> yeah, totally. [239:45] >> Y [239:45] >> serving video on demand to the entire [239:47] world. [239:48] >> Absolutely. [239:48] >> Being able to scale that and make it a [239:50] real going concern. Chrome, four core [239:54] technology insights, maybe six, maybe [239:56] seven in the original comic. [239:58] >> Yep. [239:59] >> Android. I can't name one magical core [240:03] insight. This one may kind of be the [240:05] exception cuz like, yeah, it's [240:06] technically hard and all that, but [240:08] there's not like an elegant thing that's [240:11] the reason that Android succeeded. It [240:14] was perfect execution in a lot of ways [240:17] strategically, distribution, marketing, [240:19] partnerships. [240:20] >> Okay. Wait, no, no. I I got what it is. [240:22] It's the same thing as the iPhone. It [240:24] was an incredible achievement to [240:28] wrestle OS 10 into iOS and to get it to [240:33] run on a battery powered mobile device [240:35] that fit in your pocket. And Android did [240:37] the same thing with Linux. They wrestled [240:39] Linux into a batterypowered mobile [240:42] device that fits in your pocket. [240:44] >> Yep. less of an elegant satisfying core [240:47] insight I think and not the reason that [240:49] it worked. I mean not the reason why [240:51] Android unlike these other ones there's [240:53] a clear line between the it's almost [240:55] like the Google products that succeed [240:58] wildly organically [241:00] except for Android are ones where [241:02] there's almost no product the technology [241:05] solution is just so incredible that it [241:08] is directly the user experience and you [241:11] get the technology breakthrough as the [241:13] experience. [241:14] >> Yeah. Yeah. I I sort of see where you're [241:17] going. [241:18] >> But then look at the other ones. [241:19] Google+, Google Wave, these are like [241:23] products. These are like user [241:24] experiences that people come up with [241:26] that don't necessarily have a [241:28] breakthrough technology underneath them. [241:30] Google Photos is actually quite the [241:32] opposite. All of the AI stuff that's [241:34] been happening on Google Photos for a [241:35] very long time, that's why it worked. [241:37] Like people wanted all these incredible [241:40] magic features that come with Google [241:41] Photos. It's funny as someone told me [241:44] this in the research. It was been sort [241:45] of batting around in my head and then [241:47] I'm reading Eric Schmidt's book and Eric [241:49] Schmidt said he would ask PMs, "What is [241:52] your core technical insight that makes [241:54] it all work?" And if there wasn't a good [241:56] answer, he wouldn't fund the project. [241:58] They figured this out at Google, too. [242:00] >> It's a googly thing that this genius [242:04] technology is the product itself. And if [242:07] you try to craft some cool idea that you [242:10] have that is not just directly [242:12] translating tech breakthrough, [242:15] it's not going to be the type of product [242:18] that succeeds at Google. They don't know [242:20] how. Some people can make an Instagram [242:23] and those people are not Google. [242:26] >> Yep. Ironic that Kevin was a earthw [242:29] Google employee who left to start a [242:31] startup. [242:32] >> Yes. Anyway, I think that has made it [242:35] extremely clear to me when Google [242:37] products succeed and when they fail. [242:39] >> Love it. Spot on. [242:41] >> All right, carveouts. [242:43] >> Carveouts. I've got one and then I've [242:45] got my long awaited followup. [242:49] >> Oh my god. We've been awaiting with [242:51] baited breath. Listeners, what game [242:53] console did David buy? [242:55] >> I'm going to make everybody wait for uh [242:57] one more minute. My actual carve out for [243:00] the episode is when we were in New York [243:04] for Radio City. My whole family came, [243:07] the girls came, and we stayed for the [243:10] rest of the week after the show and we [243:12] took the girls to the Bluey experience [243:17] >> at the camp store in New York City. And [243:20] it was awesome. Lived up to [243:22] expectations, lived up to the hype. They [243:26] basically have recreated the Bluey house [243:30] in this physical space in New York City. [243:35] The house is almost a character in the [243:37] show and they have recreated it and they [243:39] just let you and your kids in to roam [243:42] free in the house. Then you have a [243:44] magical moment at the end of the [243:45] experience. [243:46] >> Ah, [243:47] >> it was cool. Highly, highly recommend if [243:50] you are in the Bluey demographic and [243:51] happen to be in New York. [243:53] >> Okay. What game console did you pick? [243:55] >> I bought the Steam Deck. [243:56] >> The Steam Deck? [243:57] >> I bought the Steam Deck and it's great. [243:59] Although I haven't, truth be told, had [244:01] much time to play it this past month [244:02] with everything we've had going on at [244:04] Radio City and then preparing this [244:07] episode. But it's great. [244:09] >> How'd you pick? What was the ultimate? [244:11] >> It ultimately came down to as much as I [244:15] desperately wanted my older daughter to [244:18] be ready to play Mario Kart with me, [244:21] she's just not. And so if you're buying [244:22] a console for just you to enjoy, you [244:25] went with the Steam Deck. [244:26] >> Yeah. I was like, I would probably enjoy [244:28] the Steam Deck more. And [244:29] >> do you endorse it? Do you recommend it? [244:31] >> Yeah. What Valve has done with the Steam [244:33] Deck, I didn't realize until buying it [244:35] and using it is incredible. They have [244:37] abstracted a PC gaming machine into a [244:40] console experience. So, I've always [244:44] liked PC type games, but I haven't been [244:47] a PC gamer in many, many years cuz like [244:49] I'm not going to build a gaming machine [244:51] or even you could just buy one, but like [244:53] I don't need another PC. Where am I [244:54] going to put it? What am I going to do [244:55] with it? I want the console simplicity [244:57] of just buy the damn thing, turn it on, [245:01] buy the games, play them, [245:02] >> right? [245:03] >> Valve has created that in handheld form. [245:06] It's awesome. You don't have to worry [245:08] about any of the drivers or specs. It's [245:11] really, really impressive. [245:12] >> All right, good to know. [245:14] >> So, I will buy a Switch to at some [245:17] point, probably in the next year or so, [245:19] but for now, Steam Deck. All right, what [245:22] are your carabouts? [245:23] >> I've got three. [245:24] >> Oh, great. [245:25] >> My first one, and I swear to God, this [245:27] is unrelated to their sponsorship, is [245:29] Claude. [245:30] >> Amazing. It's so great. [245:31] >> It's so good. Using AI has completely [245:33] changed the way that I prepare for these [245:35] episodes now, and I cannot imagine going [245:36] back. [245:37] >> I hear AI is a thing. I hear AI is a [245:40] thing. So, that's the first one. I just [245:43] find myself in it all day. Now, two is [245:47] the Sony RX100 [245:50] VIII or 7. So, I recently bought a [245:53] different camera, the Fuji X100 VI or [245:57] the six. [245:58] >> Yeah, that's what you had in New York. [246:00] >> Yeah. And it's great. It's like the [246:02] internet's favorite camera. It has these [246:03] amazing film simulation color profiles. [246:06] It's like a camera, though. I carry it [246:08] around my neck because it's like it's a [246:11] camera that you hold and use and it's [246:12] very fun shooting 35 millimeter [246:14] equivalent. It feels like I'm taking [246:16] pictures the way that pictures were [246:17] meant to be taken. [246:18] >> It's not a full DSLR, but it is like a [246:20] big thing. [246:21] >> Exactly. It's a handheld, but I wouldn't [246:23] call it a pocket camera. Now, the funny [246:25] thing is the thing that I'm actually [246:26] talking about as my carveout is the Sony [246:28] camera. Shil Monot tweeted actually this [246:31] morning. I should have been preparing [246:32] for this episode and I was like replying [246:34] to him on Twitter instead that he's been [246:36] considering getting this Sony or another [246:38] point andoot camera and I just kind of [246:41] remembered how much I love this camera. [246:43] The Sony RX1007 fits in my pocket. It's [246:46] very small. It has a giant zoom lens for [246:50] its size and I was just looking at some [246:53] of the pictures that I've taken with it. [246:55] It's like the perfect thing to bring [246:56] with a phone. Whenever I am space [246:58] constrained, which is usually I mean I [247:00] just don't really want a camera around [247:02] my neck. The perfect combo is bring a [247:04] phone and bring the Sony. I am aware [247:07] that it's not a full-frame camera. I'm [247:08] aware that it's not as photographery as [247:11] my Fuji, but it is the most practical [247:13] one for most things that I want to do. [247:15] And for many, many shots, it is far [247:17] superior to shooting on a camera phone. [247:20] So, I don't know. I just love it. It's a [247:22] 2019 camera and they really need to come [247:25] out with one that has USBC because it's [247:26] annoying to charge, but other than that, [247:28] it's just awesome. So, I highly [247:30] recommend it. [247:31] >> There you go. You're bringing it full [247:32] circle on this episode. A point and [247:33] shoot camera. [247:34] >> Point and shoot camera, especially [247:35] paired with Lightroom has this great AI [247:38] feature called D Noiseise that they [247:40] launched and is now rolled out in [247:41] production. It is incredible. [247:43] >> Love it. [247:44] >> Then I have one more last one. A [247:46] listener recently sent me. He started a [247:48] clothing company called Keraseimi and it [247:52] is an incredible garment. It is just [247:54] this like really really nice cashmere [247:57] shirt. I've been wearing it all [247:58] recording. It's great on a cool day. [248:00] It's great on a warm day. It's my [248:02] current favorite shirt. And so I wanted [248:04] to uh thank the listener that sent it to [248:06] me and say you have built a very uh the [248:09] prices are high. So very nice clothing [248:12] company but just excellent products. [248:14] Well, I I've been staring at you for the [248:16] past seven and a half hours here and [248:19] I've been thinking the whole time, God, [248:20] Ben is looking good. [248:21] >> My normal thing, if I could just wear it [248:23] every day, is a long sleeve dark crew [248:25] neck. Just like I don't have to think [248:26] about it. You can look nice in it. It [248:28] just goes with everything. It's, you [248:30] know, the sort of capsule wardrobe idea. [248:32] And this is the sort of finest version [248:34] of that that I've worn. It's really [248:36] great. Kerosimi. [248:38] >> Nice. [248:38] >> All right. With that, listeners, our [248:42] huge thanks to our partners this season. [248:44] JP Morgan Payments, trusted, reliable [248:47] payments infrastructure for your [248:48] business, no matter the scale. That's [248:49] jpmorggan.com/acquired. [248:51] Anthropic, the makers of Claude, [248:53] claude.ai/acquired. [248:55] Statsig, the best way to do [248:57] experimentation and more as a product [248:59] development team. That's [249:00] statsig.com/acquired. [249:02] And Verscell, your complete platform for [249:04] web development. and Vzero that is [249:07] versel.com/acquired. [249:10] Click the links in the show notes to [249:11] learn more. We have a bunch of people to [249:13] thank for contributing to this episode. [249:16] >> Yes. Yes, we do. Hiroshi Lockheimer, Tim [249:19] Armstrong, Sam Schillis, Hunter Walk, [249:22] Nick Fox, Shona Brown, Clay Bour, a [249:26] bunch of folks who helped a little bit [249:28] here, but more are going to be for the [249:30] next episode. Max Ross, Greg Curado, [249:34] Demisabis, [249:35] and then Ben, you had a bunch of folks [249:37] who you spoke to as well. [249:38] >> Yeah, as always, I want to thank Arvin [249:40] Navaratnam from Worldly Partners for his [249:42] excellent writeup, which you can find [249:44] linked in the show notes, and also Paul [249:46] Bkite, creator of Gmail, Bill Corin, [249:49] Jonathan Rashelle, Bradley Horowitz, [249:52] John Hanky, Ben Idolen, Esar Lipkovitz, [249:55] and Ben Leebald. And that is in addition [249:58] to as always the many folks who helped [250:00] us whose names we can't say here but [250:02] know you are appreciated. [250:04] >> Yes. [250:05] >> And thanks to all of you for listening. [250:07] >> Seriously, if you like this episode and [250:09] uh you're like, "Oh, wait. There's a [250:11] Google episode before this. Most of you [250:14] have probably listened, but if you have [250:15] not, go check out our first episode on [250:18] the origin of Google and the creation of [250:20] the search engine and the search [250:23] business. And of course, our Microsoft [250:25] series part one, part two, and an [250:27] interview with Steve Balmer. [250:28] >> If you want the other side of the story [250:30] to everything we talked about here, [250:32] >> in addition, our giant episode on Meta [250:34] is probably pretty relevant and we [250:36] reference it several times. After this [250:38] episode, check out ACQ2, our second show [250:40] where we talked to founders and CEOs [250:42] building businesses and areas that we [250:44] have covered on the show. Our last one [250:46] was with Google legends, really industry [250:49] legends, Brett Taylor and Clay Bavor [250:52] about the current state of AI and where [250:54] we are headed. So, search ACQ2 in any [250:56] podcast player. Come chat with us in the [250:59] Slack. That's acquired.fm/slack. [251:02] and join the email list for all the [251:04] excellent email goodies, including [251:06] voting on episodes for this fall. [251:09] >> Woohoo! [251:10] >> With that, listeners, we'll see you next [251:12] time. [251:12] >> We'll see you next time. [251:14] >> Who got the truth? [251:17] >> Is it you? Is it you? Is it you? Who got [251:19] the truth? Huh? [251:23] [Music]