Apple’s ‘Pro and Tell,’ Google’s Two New AI Models, OpenAI Eyes Google’s Custom Chips | Diet TBPN
TBPN
AI-assisted summary of Apple’s ‘Pro and Tell,’ Google’s Two New AI Models, OpenAI Eyes Google’s Custom Chips | Diet TBPN — original source. Check the original for full context.
Source published:
Key ideas
- Apple announced the M6 MacBook Pro featuring an OLED touchscreen display and brought back the iconic illuminated lid logo, with configurations starting between $2,900 and $3,400 due to memory and component inflation. (0:00)
- Google introduced Genie 4 as an interactive, real-time steerable world model, alongside Gemini 4 Carbon, which topped the Artificial Analysis Intelligence Index V5 under a reverse periodic table naming taxonomy. (7:30)
- OpenAI and Anthropic are aggressively expanding compute infrastructure with massive multi-gigawatt deals, including OpenAI procuring Google Cloud TPUs alongside an 8-gigawatt Ohio expansion backstopped by Nvidia. (17:47)
- Malaysia has emerged as a computational bridge and 'Switzerland of AI,' receiving hardware through Singapore to provide cloud infrastructure and remote training to global and Chinese labs amid trade restrictions. (21:29)
- Personal AI agent startup Instinct is reportedly seeking a $30 billion valuation, igniting debates over whether consumer adoption velocity can justify multiples compared to revenue-heavy predecessors like Cursor. (24:15)
Detailed notes
## Background This episode of Diet TBPN was published on YouTube by TBPN on October 29, 2026. Hosted by John Coogan and Jordi Hays, the 30-minute daily recap show delivers highlights from their live tech broadcast covering major developments across Silicon Valley. According to the page description, this installment covers Apple's in-person hardware event, Google's newest AI models, major compute deals involving OpenAI and Google TPUs, and emerging valuation rumors around AI agent startup Instinct.
## TL;DR John Coogan recaps attending Apple's 'Pro and Tell' event, breaking down the launch of the touchscreen M6 MacBook Pro, its steep $2,900-plus price tag, and the surprising return of the glowing lid logo. The hosts then analyze Google's generative world model Genie 4 alongside the benchmark-topping Gemini 4 Carbon, while joking about DeepMind's inverted periodic table naming scheme. Shifting to infrastructure, they examine massive compute expansion deals—including OpenAI leasing Google Cloud TPUs and Anthropic expanding into Australia—as well as Malaysia's rise as a global data center hub and an enterprise Nvidia supercomputer listed on eBay. Finally, they debate the personal AI agent race, dissecting rumors of Instinct's $30 billion valuation against rivals like Muse, Cursor, and Big Tech incumbents.
## Atmosphere & style The show feels like an energetic, casual morning conversation between two Silicon Valley insiders balancing sharp market analysis with running internet jokes. The dynamic between Coogan and Hays is breezy and unscripted, transitioning naturally from hardware engineering breakdowns to comedic bits about setting up backyard compute clusters or buying full-sized T-Rex skeletons. Backed by soundboard drops and playful banter, the episode carries a lively, fast-paced momentum that makes high-stakes tech infrastructure and multi-billion-dollar venture deals feel immediate, engaging, and entertaining.
## Bulletin - [00:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=0s) Apple announced the M6 MacBook Pro featuring an OLED touchscreen display and brought back the iconic illuminated lid logo, with configurations starting between $2,900 and $3,400 due to memory and component inflation. - [07:30](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=450s) Google introduced Genie 4 as an interactive, real-time steerable world model, alongside Gemini 4 Carbon, which topped the Artificial Analysis Intelligence Index V5 under a reverse periodic table naming taxonomy. - [17:47](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1067s) OpenAI and Anthropic are aggressively expanding compute infrastructure with massive multi-gigawatt deals, including OpenAI procuring Google Cloud TPUs alongside an 8-gigawatt Ohio expansion backstopped by Nvidia. - [21:29](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1289s) Malaysia has emerged as a computational bridge and 'Switzerland of AI,' receiving hardware through Singapore to provide cloud infrastructure and remote training to global and Chinese labs amid trade restrictions. - [24:15](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1455s) Personal AI agent startup Instinct is reportedly seeking a $30 billion valuation, igniting debates over whether consumer adoption velocity can justify multiples compared to revenue-heavy predecessors like Cursor.
## Key takeaways - [00:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=0s) Apple's 'Pro and Tell' keynote debuted the M6 MacBook Pro with a reinforced, stiffened hinge to solve classic touchscreen bounce issues. - [05:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=300s) The return of the backlit Apple logo received more audience applause than touchscreen support, justified economically as component inflation makes the $2 cutout negligible against $3,000+ base retail prices. - [07:30](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=450s) Google's Genie 4 research preview points toward dynamic, prompt-controlled real-time world generation, eliminating traditional post-production composite VFX for content creators. - [10:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=600s) Gemini 4 Carbon achieved top benchmark scores—59 on Artificial Analysis Intelligence Index V5 and 88.2% on SCC-Bench Pro—though hosts noted historic discrepancies between Google benchmarks and user experience. - [12:33](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=753s) DeepMind's new model taxonomy utilizes an inverted periodic table, where lower atomic numbers denote frontier models, leading to speculation around a future flagship 'Hydrogen' or 'Helium' release. - [17:47](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1067s) OpenAI's decision to lease Google Cloud TPUs highlights how business pragmatism overrides corporate rivalries amid an insatiable compute demand. - [17:47](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1067s) Hyperscale AI leasing hit record milestones with OpenAI's 8-gigawatt Ohio commitment and Anthropic's 1.7-gigawatt global lease, including a 360-megawatt deployment in Queensland backstopped by Google. - [20:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1200s) A 300-rack cluster of 9,600 Nvidia A100 GPUs listed on eBay for roughly $60 million underscores the strong residual value of older compute hardware against its original ~$300 million retail cost. - [21:29](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1289s) Malaysia functions as the neutral computational hinterland for Singapore, leveraging low power costs and regulatory flexibility to train models for international and Chinese AI labs. - [24:15](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1455s) Instinct's rumored $30 billion valuation reflects soaring private market premiums for viral consumer AI adoption, despite lacking a traditional App Store presence. - [25:50](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1550s) The personal agent market is fragmenting consumer attention across specialized tools, forcing standalones like Instinct to aggressively market before Google activates its latent personal data graph. - [28:30](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1710s) Unlike Cursor, which reached a $30 billion valuation backed by over $1 billion in top-line revenue, Instinct's valuation rests on consumer momentum and potential M&A by tech giants like Apple or Amazon.
## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=0s) Apple’s “Pro and Tell” Event and the Touchscreen MacBook Pro
The broadcast opens with Jordi Hays asking John Coogan about his morning, which was spent attending Apple’s in-person "Pro and Tell" event while Jordi worked out. Leading up to the showcase, speculation ran rampant online after Apple hardware executive John Ternus teased a video showing a bright Apple logo spotlight projected against a wall—prompting Christian Reber to jokingly predict on social media that Apple was about to drop the long-awaited Apple Car during a brisk Wednesday morning keynote. Instead, the headline hardware announcement was an M6 MacBook Pro equipped with a touchscreen, accompanied by an updated iPad mini.
Reflecting on the stage presence of Apple’s leadership, Coogan notes that this was his first time attending an Apple keynote in person after growing up on Steve Jobs presentations and watching the evolution toward tightly edited, high-production pandemic videos. Coogan highlights Ternus’s standout delivery, remarking, “The star power is incredible. I'm not joking. Like he is so good on stage. He's incredible.” Ternus moved through the script at a blistering pace, spending a mere 15 seconds recapping the iPad mini and 15 seconds on the M6 processor before introducing the touchscreen display alongside a promotional spot styled after *If You Give a Mouse a Cookie*. Coogan’s personal takeaway on touch support was nonchalant: “So, yeah, my reaction to the launch is like, why the heck not? You know? You should be able to touch the screen of your computer.”
Beyond touch input, the laptop features an OLED panel, bringing deeper black levels and higher contrast ratios to parity with high-end consumer televisions. Testing out the hardware, Coogan and Jordi address the classic engineering flaw of touchscreen laptops: hinge bounce. When testing the display by poking and wobbling it, Coogan found the hinge appeared reinforced to eliminate the distracting springiness found in older laptops, joking that users will likely hold the display with both hands like driving a steering wheel just to flick through Instagram and TikTok reels.
- **Event structure**: Fast-paced, compact keynote with brief updates on iPad mini and M6 MacBook Pro before the main touchscreen reveal. - **Display tech**: Upgraded to an OLED display with significantly higher contrast and deeper blacks. - **Hinge mechanics**: Stiffened hinge tension prevents the screen from bouncing back and forth when tapped.
[05:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=300s) MacBook Pricing Reality and the Return of the Glowing Logo
A notable departure from classic Apple keynotes was the complete absence of pricing slides at the end of the reveal. Rather than closing on a dramatic sticker price reveal, Apple kept the numbers off the screen due to rising component expenses and global memory shortages. On Apple's retail website, the 14-inch MacBook Pro starts around $2,900, while the 16-inch model base tier begins at $3,400, climbing up to $6,000 when fully specced out. To offset the high up-front cost, Apple also offers enterprise customers and power users a 36-month lease program.
```mermaid flowchart TD accTitle: MacBook Pro Cost and Margin Shift accDescr: Diagram illustrating how rising chip and memory prices altered Apple's cost ratio, enabling the return of the glowing lid logo.
subgraph Past ["Era of $1,000 MacBooks"] CostOld["$2 Glowing Logo Cost"] --> ShareOld["~0.2% of Top-Line Price / 0.5% Component Cost"] ShareOld --> DecisionOld["Logo Removed to Cut Bill-of-Materials Costs"] end
subgraph Present ["Era of $3,000-$4,000+ M6 MacBooks"] CostNew["$2 Glowing Logo Cost (Unchanged)"] --> ShareNew["Negligible Fraction of High-End Machine"] ShareNew --> DecisionNew["Logo Reinstated as Premium Brand Identity"] end ```
The reveal that ignited the loudest live reaction in the room was not the touchscreen or the display, but the return of the illuminated Apple logo on the laptop lid. Jordi and Coogan observe that Apple previously removed the glowing cutout strictly to save manufacturing pennies, stripping away one of the most recognizable design flourishes of the Mac ecosystem. As Coogan explains, when base MacBooks cost $1,000, spending roughly $2 per unit on an illuminated cutout took up a noticeable fraction of a percent of component costs. With average configurations now climbing toward $4,000 because of chip and memory inflation, that $2 backlight cost is negligible against top-line margins. Coogan praised Ternus for leaning back into brand aesthetics over pure cost-cutting: “Yes, it has no utility other than like sort of brand and just sort of like enjoyment for the user. It looks nice. And that's still worth it.”
- **Pricing tiers**: 14-inch begins around $2,900; 16-inch begins around $3,400; maximum configurations reach $6,000. - **Financing**: 36-month business and personal lease options aimed at workers expensing hardware. - **Crowd reaction**: The return of the backlit Apple logo received greater applause in the auditorium than the announcement of touch capabilities.
[07:30](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=450s) Google Genie 4 and the Simulation Hypothesis
Transitioning from hardware to generative software, the conversation shifts to Google’s announcement of Genie 4. Quoting Google CEO Sundar Pichai, the hosts break down Genie 4's capabilities as a state-of-the-art interactive world model that allows users to steer dynamic, photorealistic environments in real time with prompt-controlled camera angles, physics engines, and continuous actions. Genie 4 dynamically generates native environmental soundscapes and conversational NPCs, while leveraging Gemini reasoning to maintain coherence across complex, multi-step scenarios.
While currently framed as a research preview, the practical near-term application is moving toward real-time virtual sets and dynamic composite rendering for content creators. Instead of relying on manual VFX or clunky post-production replacement apps like Higgsfield—where creators shoot a video next to a Honda Civic or a back porch and composite it into a supercar or private jet—Genie 4 can parse camera tracking coordinates on the fly and continuously generate matching surroundings. Moving the camera through a door onto a beach or boardwalk will cause the environment to generate logically in real time. Beyond practical production, Coogan reflects on the philosophical creep of the technology: “I think that the primary use case for world models for me personally so far is, you know, making me think that it's more likely that we're living in our own simulation.”
- **Genie 4 capabilities**: Real-time steerable world generation, photorealistic physics, contextual spatial audio, and interactive dialogue driven by Gemini reasoning. - **Creator utility**: Seamless real-time virtual set extensions and backdrop substitution that update dynamically as camera perspectives change. - **Release status**: Debuting initially as an exploratory research preview for AI researchers and developers.
[10:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=600s) Simulation Theory and Google's Rapid AI Rollout
The hosts open by joking about simulation theory, proposing that Earth and the entirety of human existence might just be a racing simulator managed by a higher-level organism whose sole purpose is propagating Formula 1. Shifting focus to Google's DeepMind developments, they discuss Genie 4. While one host has had extensive, fun experiences using Genie 3, Genie 4 remains locked behind an early access application program.
The conversation pivots to Google's relentless release cycle under Sundar Pichai, who recently teased that the company would be rolling out models periodically. Hot on the heels of the Argon release two weeks prior, Google introduced Gemini 4 Carbon. The hosts highlight its early benchmark performance: it ranked number one on the recently updated V5 of the Artificial Analysis Intelligence Index at 59, scored 88.2% on SCC-Bench Pro for finding vulnerabilities, and hit 54.8% on ExploitJim. Even Elon Musk commented publicly that the launch was impressive. However, the hosts maintain some skepticism given Google's historical track record. As one host reflects on past launches, “Yeah, just about a... a year straight where every single time it's amazing on benchmarks and then the sort of lived experience that people have and is less impressive.” One host defends Gemini 3, noting it answered concisely within a single page rather than generating overwhelming, multi-page responses like other LLMs.
- **Artificial Analysis Intelligence Index (V5):** Gemini 4 Carbon scored 59, taking the number-one spot. - **SCC-Bench Pro:** 88.2% score on vulnerability detection. - **ExploitJim:** 54.8% score on exploit evaluation. - **Recent Google Releases:** Gemini 4 Carbon follows the release of Argon by just two weeks.
---
[12:33](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=753s) The Elemental Taxonomy and the AI Naming Wars
The hosts debate Google's new elemental naming taxonomy. While OpenAI has used celestial bodies (Luna, Terra, Sol, Astra) to denote scale—where the moon is smaller than the Earth and the Earth is smaller than the Sun—and Anthropic maps models to literary works (Sonnet versus Opus), DeepMind appears to be using the periodic table in reverse. Because Carbon has an atomic number of 6 and Argon is 18, a lower atomic number corresponds to a higher-capability model.
```mermaid flowchart TD accTitle: Google DeepMind Elemental Model Hierarchy accDescr: Visualizing DeepMind's inverted atomic number hierarchy where lower numbers represent larger, more capable models.
subgraph DeepMindHierarchy["DeepMind Periodic Table Taxonomy (Lower Atomic Number = More Capable)"] direction TB Hypothetical["Hypothetical Frontier (Hydrogen / Helium) Atomic No: 1 - 2"] Carbon["Gemini 4 Carbon (Top Frontier) Atomic No: 6"] Argon["Argon (Smaller Model) Atomic No: 18"] Plutonium["Plutonium (Lowest Theoretical Tier) Atomic No: 94"]
Hypothetical --> Carbon Carbon --> Argon Argon --> Plutonium end ```
The hosts jokingly theorize that this inverted hierarchy opens the door to a massive Helium (atomic number 2) or Hydrogen (atomic number 1) model that could shatter the Ramp Index. As one host quips, “Yes, I think the helium model could be the biggest model for DeepMind in years. It could change everything.” Under this taxonomy, a model named Plutonium would sound terrifying but would theoretically be the least capable model in the lineup. Tyler fires the show's slow "truth nuke" sound effect as they mock the chaotic history of AI model naming, recalling how OpenAI previously confused everyone with version strings like 4.0, 4.5, Omni, and rapid-fire increments before settling on human-legible names.
---
[17:47](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1067s) Massive Compute Leases: OpenAI Buys Google TPUs
The hosts react to the landmark news that OpenAI plans to buy Google Cloud TPUs as part of its infrastructure build-out, following reports from the prior week that OpenAI models would run on Google Cloud Platform (GCP). They marvel at the shift from OpenAI’s founding mission—which was originally positioned as a counterweight to Google—to their present pragmatism: “Business above all solves everything.”
They break down reporting from Dylan Patel at SemiAnalysis, which revealed an unprecedented wave of leasing commitments:
- **OpenAI's Ohio Expansion:** An 8-gigawatt, multi-phase commitment in Ohio, with Nvidia backstopping 4.25 gigawatts in the largest single leasing commitment recorded to date. - **TPU Commitments:** OpenAI is in active discussions with neoclouds for GPU-as-a-service and TPU-as-a-service, including an estimated 1-gigawatt TPU commitment with sites currently being scouted. - **Anthropic Down Under:** Anthropic signed a 1.7-gigawatt lease this quarter, featuring 360 megawatts in Queensland, Australia (signed September 16, with operations starting Q1 2028)—the largest overseas lease by an AI lab during the quarter. Google backstopped 1.0 gigawatt of Anthropic's total capacity.
The hosts banter about Australian inference jokes, warning researchers to stick to "inference and Foster's" instead of "throwing shrimp on the barbie." More broadly, they emphasize that labs are "using every part of the animal," running specialized models across different silicon architectures, hyperscalers, and neoclouds. Pointing to Lisa Su's AMD coverage where Anthropic vibe-coded their AMD implementation rapidly, they observe that software portability has softened the threat of architectural lock-in, fueling a rising tide for silicon and compute providers.
---
[21:29](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1289s) The $60M eBay Supercomputer
Turning to consumer oddities in the high-stakes compute market, the hosts reveal that an entire AI supercomputer is currently listed for sale on eBay. For an entrepreneur dreaming of launching a neocloud, buying tokens, or entering the infrastructure race, the setup is packaged into a turnkey listing.
- **System Specs:** 300 racks containing 9,600 Nvidia A100 GPUs. - **Price Tag:** Estimated fair market value / credit line needed is around $60 million. - **Market Note:** eBay stock is up 16% over the preceding six months, perhaps fueled by ultra-high-ticket listings of enterprise data center hardware.
[20:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1200s) A $60M Nvidia A100 Supercomputer on eBay and the Dinosaur Dilemma
John Coogan and Jordi Hays look over a wild listing on eBay: an enterprise-scale Nvidia supercomputer cluster priced at roughly $60 million. While fair market value for the hardware is estimated closer to $50 million or $55 million, the listing shows that two units are available and six different prospective buyers currently have it sitting in their carts. The hardware, built around Nvidia A100 GPUs—chips that are roughly seven years old—originally commanded nearly $300 million when brand new. Despite the massive depreciation, John notes that there is still significant residual value in older compute racks because the market treats them as raw capacity in an insatiable compute marketplace.
The conversation quickly turns into banter when John asks Jordi whether he would rather spend $60 million on the supercomputer or on "Gus," a full-sized fossilized T-Rex dinosaur skeleton. Jordi admits that, to his own dismay, he would pick the server cluster over the dinosaur because it would let him launch a backyard neocloud called "Malibu Inference." John jokes about how affluent founders claim they will drop data centers in luxury beach towns, imagining Jordi stacking server racks in the shape of a T-Rex right in his home foyer. While turning the entryway of a house into a roaring, high-temperature "hot aisle" would require catastrophic cooling solutions, John quips that it would send an unmistakable message to visitors: you run local compute and you do not pay API pricing.
- **Hardware depreciation**: The cluster of Nvidia A100 servers currently listed at $60 million originally retailed for almost $300 million. - **eBay listing oddity**: The screenshot displayed two available supercomputer units with six users having placed the item in their carts. - **Fair market valuation**: The co-hosts estimate true fair market value for the used cluster between $50 million and $55 million.
---
[21:29](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1289s) Malaysia’s Compute Boom as the "Switzerland of AI"
The discussion pivots to global data center distribution and the sudden explosion of server infrastructure in Southeast Asia. John recalls that Malaysia's compute boom was literally the first topic covered on the very first episode of TBPN after a front-page *Wall Street Journal* feature highlighted the country's transformation into an international data center capital. Malaysia has effectively positioned itself as the neutral broker of the artificial intelligence race, with John explaining, “They are the Switzerland of AI. That was sort of the framing, was that they can buy compute from the US and resell it in China.” The country combines cheap, abundant power across its regions with aggressive government policy aimed at facilitating rapid data center buildouts.
Examining a global chart of compute deployment, Jordi admits surprise that the European Union still commands more total compute capacity than Malaysia, given how aggressively Malaysia has scaled. John points out that the EU is an entire continent while Malaysia is a relatively small nation. More importantly, Malaysia acts as the industrial and computational hinterland for Singapore. While Singapore remains the primary financial hub and global shipping port receiving shipments of advanced hardware, real estate and power constraints force companies to move the chips right across the border into Malaysia. Because trade restrictions and export controls on neoclouds have different regulatory loopholes compared to direct imports into mainland China, Chinese AI labs frequently train models on Nvidia GPUs racked inside Malaysian facilities.
```mermaid flowchart LR accTitle: Singapore to Malaysia AI Compute Pipeline accDescr: Hardware imports flow through Singapore before being racked in Malaysian data centers for regional and Chinese lab access.
subgraph Logistics ["Port & Finance Hub"] A["Nvidia GPU Shipments"] --> B["Singapore Port & Financial Center"] end
subgraph Deployment ["Regional Compute Hub"] B --> C["Cross-Border Delivery<br/>to Malaysia"] C --> D["Malaysian Neoclouds & Data Centers<br/>(Abundant Power & Land)"] end
subgraph Customers ["End Compute Demand"] D --> E["Global AI Labs"] D --> F["Chinese AI Labs<br/>(Remote Model Training)"] end ```
---
[24:15](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1455s) Instinct’s Rumored $30B Valuation and the Private Market Rush
The hosts turn to rumors circulating across Silicon Valley that personal AI agent startup Instinct is negotiating a funding round at an eye-popping $30 billion valuation—having scaled from early-stage pricing to decacorn status in under a year. Addressing social media commentary urging tech workers to get into private venture markets before wealth generation shifts permanently away from public equities, Jordi and John acknowledge why people are "crashing out" over the post. While early-stage investing carries steep risks that require venture firms and LPs to diversify, the sheer velocity of Instinct’s ascent is historically aggressive.
A major challenge in evaluating Instinct's explosive trajectory is that, unlike consumer chat apps or productivity tools like Muse, Instinct does not operate as a standard mobile app on the App Store. Jordi checks the live App Store productivity rankings, noting that Muse remains heavily promoted with over 380,000 ratings and an average score of 4.9 stars, trailing only viral sensations like TikTok Pro and Vinted. This prompts an aside about Jordi browsing Vinted for Rick Owens garments and proposing he become a lifestyle influencer rocking Bottega top hats and Supreme gear, which John brushes off as just another comedic bit. Returning to the startup ecosystem, Jordi stresses that public hot takes about Instinct are detached from operational realities, arguing, “Investors are making a decision based on what they’re seeing in the business. Everyone else is giving takes based on how they feel about the number going up really fast.”
---
[25:50](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1550s) The Personal AI Agent Wars: Instinct, Muse, and Frontier Lab Competition
Beyond valuation metrics, John highlights the brutal marketing and distribution battlefield facing specialized AI agents. Frontier labs regularly release state-of-the-art capabilities, but technical superiority in a model does not guarantee consumer mindshare. John points out that users have limited real estate on their home screens and inside existing apps, making aggressive go-to-market execution non-negotiable: “When new capability comes online from a frontier lab, it can be at the frontier of that new capability. But unless there is a commensurate marketing push, it's not guaranteed that it gets adoption.” For instance, Anthropic launched Duet, but a lack of widespread consumer advertising means many developers compartmentalize their stack—using Claude Code inside the terminal, Muse or Grok in Slack, Codex or Cursor for programming, and Instinct strictly for personal lifestyle management.
Instinct’s distinct advantage has been its relentless focus on pushing its consumer value proposition directly into the public consciousness. Investors backing the startup are seeing monthly revenue and usage charts that show vertical growth, even as the company faces attacks from Mark Zuckerberg at Meta, dedicated agent platforms like Dot, and Anthropic's Duet. Furthermore, Google looms as an existential threat because it already possesses the personal data graph—holding users' Gmail threads, calendar reservations, and restaurant assets like Zagat. Whether the personal assistant space evolves into an oligopoly or a winner-take-all market, Instinct is capitalizing on rapid consumer habit formation before Big Tech can fully package similar native features.
- **Developer fragmentation**: Users frequently segment tools across their workflow, balancing Claude Code in the terminal with Codex, Cursor, Slack bots, and standalone personal agents. - **Google’s dormant advantage**: Google natively owns the email inbox, calendar schedules, flight reservations, and local reviews (Zagat), giving it structural advantages whenever it launches an integrated consumer agent. - **App Store benchmark**: Muse holds the number-one spot in the App Store’s productivity category with 380,000 reviews averaging a 4.9-star rating.
---
[28:30](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1710s) Comparing Instinct to Cursor: Revenue, M&A Exit Realities, and Antitrust
Jordi draws a comparison between Instinct raising at a $30 billion valuation and developer tool Cursor when it hit identical valuation milestones. While Cursor faced intense competition from foundation labs and specialized rivals like Windsurf, Cursor had already crossed over $1 billion in top-line revenue at that stage. While critics debated Cursor’s gross margins and questioned how it would survive competing with its own upstream model suppliers, it possessed massive commercial cash flow. Instinct, by contrast, is commanding a $30 billion price tag primarily off explosive consumer adoption rather than an established enterprise revenue base.
```mermaid flowchart TD accTitle: Cursor vs Instinct $30B Valuation Dynamics accDescr: Comparison of Cursor's revenue-backed IDE model against Instinct's consumer-adoption personal agent model.
subgraph CursorModel ["Cursor ($30B Valuation Stage)"] C1["Primary Asset: Over $1B Top-Line Revenue"] C2["Lead Hybrid IDE ('Centaur Model')"] C3["Competition: Windsurf, followed by Terminal Agents (Claude Code, Codex)"] C4["Investor Debate: Software Margins vs Supplier Dependence"] end
subgraph InstinctModel ["Instinct ($30B Rumored Stage)"] I1["Primary Asset: Viral Consumer Adoption & Growth Curves"] I2["Dedicated Personal Lifestyle Agent"] I3["Competition: Meta (Zuck), Dot, Anthropic Duet, Google"] I4["Investor Debate: M&A Potential (Apple/Amazon) vs Pre-Revenue Multiples"] end ```
The hosts speculate on how Instinct’s investors justify the math, noting that venture capitalists may be banking on an acquisition by cash-rich giants like Apple or Amazon looking to buy their way into consumer AI. When Jordi questions whether federal regulators would block an acquisition of that scale, John vehemently argues that antitrust authorities should leave the category alone, calling it the most aggressively contested frontier in tech: “It should 100% be allowed. It's the most competitive industry right now. It's like a knife fight for the personal agents.” John emphasizes that allowing an enterprise player like Amazon to acquire Instinct to counter Meta, OpenAI, and Anthropic promotes competition rather than harming it, though both hosts concede antitrust regulators frequently apply blunt rules against Big Tech acquisitions regardless of category dynamics.
Wrapping up the show, John and Jordi remind listeners to rate the podcast five stars on Apple Podcasts and Spotify, subscribe to the daily newsletter at TBPN.com, and tune in for their upcoming Friday livestream at 11:00 AM Pacific.
## Notable quotes > [25:50](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1550s) When new capability comes online from a frontier lab, it can be at the frontier of that new capability. But unless there is a commensurate marketing push, it's not guaranteed that it gets adoption. — John Coogan
> [24:15](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=1455s) Investors are making a decision based on what they’re seeing in the business. Everyone else is giving takes based on how they feel about the number going up really fast. — Jordi Hays
> [00:00](https://www.youtube.com/watch?v=OF0Yyvtdgcw&t=0s) You should be able to touch the screen of your computer. — John Coogan
## Mentioned Amazon, AMD, Anthropic, Apple, Apple Car, Argon, Artificial Analysis Intelligence Index, Christian Reber, Claude Code, Codex, Cursor, Dot, Duet, Dylan Patel, eBay, Elon Musk, ExploitJim, Gemini, Gemini 3, Gemini 4 Carbon, Genie 3, Genie 4, Google, Google Cloud Platform, Grok, Higgsfield, Instinct, iPad mini, John Coogan, John Ternus, Jordi Hays, Lisa Su, M6 MacBook Pro, Mark Cuban, Mark Zuckerberg, Meta, Muse, Nvidia, Nvidia A100, OpenAI, Ramp Index, Sam Altman, Satya Nadella, SCC-Bench Pro, SemiAnalysis, Steve Jobs, Sundar Pichai, TBPN, TikTok Pro, Vinted, Windsurf, Zagat
## Video description Sign up for TBPN’s daily newsletter at TBPN.com
Diet TBPN delivers the best of today’s TBPN episode in 30 minutes. TBPN is a live tech talk show hosted by John Coogan and Jordi Hays, streaming weekdays 11–2 PT on X and YouTube, with each episode posted to podcast platforms right after.
Described by The New York Times as “Silicon Valley’s newest obsession,” the show has recently featured Mark Zuckerberg, Sam Altman, Mark Cuban, and Satya Nadella.
TBPN is made possible by: Ramp - https://ramp.com Public - https://public.com Cisco - https://www.cisco.com Console - https://www.console.com CrowdStrike - https://www.crowdstrike.com Figma - https://www.figma.com MongoDB - https://www.mongodb.com NYSE - https://www.nyse.com Railway - https://railway.com Shopify - https://www.shopify.com/
Follow TBPN: https://TBPN.com https://x.com/tbpn https://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231 https://podcasts.apple.com/us/podcast/technology-brothers/id1772360235 https://www.youtube.com/@TBPNLive
Chapters: 00:00 Apple’s “Pro And Tell” Event + Touchscreen MacBook Pro Reveal 05:56 Apple’s New MacBook Pricing + The Glowing Logo Returns 09:27 Google Unveils Genie 4 + Are We Living in a Simulation? 12:33 Gemini 4 Carbon Tops AI Benchmarks + The Model Naming Wars 17:47 OpenAI Turns to Google’s TPUs + Massive AI Compute Deals 21:29 A $60M Nvidia Supercomputer on eBay + Malaysia’s AI Compute Boom 25:50 Instinct’s $30B Valuation Rumors + The Personal AI Agent Wars
[Source](https://www.youtube.com/watch?v=OF0Yyvtdgcw)