亚马逊筑墙封杀AI购物代理 760亿美元广告业务直面脱媒风险

浑水摸鱼清源

剥离高利润赞助广告后亚马逊核心电商在实质上陷入亏损，自主AI购物代理对搜索流量入口的绕过将动摇其需求聚合器的盈利根基。

AI-assisted summary of 【The Motley Fool】亚马逊的下一个最大隐忧是什么？ — original source. Check the original for full context.

Source published: 2026-10-12

Detailed notes

## Background 本内容由Bilibili频道“浑水摸鱼清源”于2026年10月12日发布，取材自The Motley Fool对亚马逊未来核心隐忧的深度探讨。简介指出，视频重点剖析了亚马逊零售电商帝国在智能AI购物代理崛起下所面临的结构性挑战，以及云服务AWS能否凭借自研芯片与AI布局继续支撑其2.7万亿美元的庞大估值。

## TL;DR 节目围绕亚马逊当前的“双引擎”展开深度博弈分析：AWS凭借自研Trainium芯片、对Anthropic的战略投资以及作为AI“公用基础设施”的低成本定位正加速扩张，甚至展现出冲击单体万亿美元营收的潜力。然而，核心零售业务却陷入严重隐忧——其盈利结构高度依赖过去一年创造760亿美元、利润率约80%的站内搜索广告，而一旦消费者转向自主AI购物代理（Agentic Shopping），不看广告的比价机器人将直接瓦解其流量聚合变现模型。结合公司目前超2000亿美元的资本支出与承压的自由现金流，主讲人探讨了亚马逊是否会沦为“加强版UPS”或步西尔斯后尘，并认为当前28倍前瞻市盈率缺乏足够安全边际，回调至20倍左右才更具长线吸引力。

## Atmosphere & style 整场讨论展现出专业、辩证且兼具前瞻性的投资研判氛围。主讲人间互动松弛自然、配合默契，既对亚马逊历史上的颠覆力与云基础设施壁垒报以由衷惊叹，又在剖析零售广告困境与历史兴衰规律时流露出冷静清醒的审慎感。整段对话节奏张弛有度，在严谨拆解财务数据与轻松调侃个人AI使用配额之间无缝切换，给观众带来兼具智力深度与实战警示的沉浸式听觉体验。

## Detailed walkthrough ### [00:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=0s) AWS Acceleration vs. Retail Uncertainty in an AI World

The discussion opens with high praise for Amazon as historically one of the greatest businesses ever built, yet both hosts immediately acknowledge that the company's current landscape is far more nuanced than it used to be. The consensus verdict is split clean down the middle between two fundamentally divergent engines: Amazon Web Services (AWS) and the legacy retail division. On the cloud side, AWS is firing on all cylinders. Revenue growth is noticeably re-accelerating, and Amazon has carved out a compelling, practical niche in the artificial intelligence landscape. Rather than exhausting capital purely competing to train the single premier frontier model, Amazon has positioned itself as the indispensable "commodity supplier" of the AI gold rush: producing its own proprietary custom chips, supplying scalable cloud compute, and securing a major strategic equity stake in Anthropic.

The retail business, however, presents a starkly different, far more precarious picture. The hosts question what Amazon's long-term fate looks like in an "agentic world" where autonomous AI shopping assistants begin to dictate consumer purchasing decisions. Amazon has adopted a defensive, almost hostile posture toward external AI-driven shopping interfaces. It is a striking historical reversal: the company that spent two decades dismantling legacy retail as the ultimate disruptor now suddenly finds itself on the defensive, maneuvering like an entrenched incumbent desperate to protect its castle walls from being disrupted in turn.

```mermaid flowchart TD accTitle: Amazon Dual-Engine AI Landscape accDescr: Diagram contrasting AWS commodity cloud strategy against the retail advertising disruption risk.

subgraph AWS["AWS Cloud Engine"] A["Proprietary Silicon (Trainium / Inferentia)"] B["Anthropic Partnership & Hosting"] C["Accelerating Revenue & Margin Expansion"] A --> C B --> C end

subgraph Retail["Retail Division Dilemma"] D["Aggregator of Demand (Search & App Traffic)"] E["$76B Ad Business (~80% Operating Margin)"] F["External AI Agents (Autonomous Shopping / Price Scraping)"] D -->|Funds core retail ops| E F -.->|Bypasses sponsored ads| D end ```

[02:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=135s) The Agentic AI Threat and the High-Margin Advertising Dilemma

The debate digs into why Amazon reacted so aggressively against external AI shopping assistants like Muse. While rivals such as Shopify and Walmart eagerly integrated with Muse, Amazon deliberately erected walls to prevent any interoperability. The conversation highlights the immediate peril of this walling-off strategy: if a consumer's automated agent cannot execute a query on Amazon, it will simply reroute the order to another vendor, causing Amazon to permanently lose the underlying transaction.

``` +-------------------------------------------------------------------------+ | The Amazon Retail Profitability Reality | | | | Total Advertising Revenue (TTM): ~$76 Billion | | Estimated Operating Margin on Ads: ~80% | | | | Combined Retail Operating Income (North America + Int'l): | | Roughly half of total ad revenue. | | | | Conclusion: Stripped of sponsored ad placements, core retail | | e-commerce operates at an outright structural loss. | +-------------------------------------------------------------------------+ ```

The underlying financial incentive explains this behavior entirely. Over the trailing twelve months, Amazon's advertising business ballooned into a staggering $76 billion powerhouse, carrying astronomical estimated operating margins of roughly 80%. When looking at the combined operating income of both the North American and International retail segments, the total profit sums to barely half of advertising revenue alone. Without high-margin advertising, Amazon's massive retail apparatus is structurally unprofitable.

This creates an existential dilemma: - **Bots do not look at ads:** Amazon's entire core retail model relies on functioning as the definitive "aggregator of demand." Prime exists primarily to lock shoppers into typing queries into the Amazon search bar, where most of the screen real estate is monetized. - **Sponsored search dominance:** A casual query on the mobile app or browser yields a top row that is roughly 75% sponsored listings, functioning exactly like Google search ads. - **The Shopify alignment:** Shopify's business is naturally aligned with AI agents and external platforms because its merchants grew through off-platform Meta (Facebook and Instagram) ad funnels. Amazon, by contrast, relies entirely on captive on-platform traffic. - **Merchant fee rebellion:** Third-party sellers currently surrender roughly 50% of their top-line revenue to Amazon across fulfillment, storage, shipping, and advertising fees. If autonomous AI agents allow buyers to purchase directly from sellers' independent web storefronts, merchants will jump at the chance to bypass Amazon's hefty rake.

[05:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=300s) Rethinking Retail: Just a Better UPS or a Logistics Cloud?

Faced with this vulnerability, the discussion turns to how Amazon might defend its margins. If retail search clicks stall, Amazon must locate incremental margin points elsewhere to remain fast, cheap, and competitive. One clear evolution is opening up their sprawling logistics footprint to non-Amazon packages—effectively creating an "AWS for fulfillment, freight, and logistics" to achieve economies of scale no standalone carrier can match.

Alternatively, Amazon could pivot to maximizing Gross Merchandise Volume (GMV) by running retail close to breakeven and relying on adjacent ecosystem monetization. However, the hosts pose a blunt rhetorical question about valuation: if the retail division's advertising moat is degraded by AI agents and you are buying Amazon for its physical warehouse and fulfillment footprint, are you really just paying an exorbitant premium for "a better version of UPS"?

- Third-party seller takes represent around half of product revenue when factoring in advertising and Fulfillment by Amazon (FBA) logistics. - Amazon's retail listings page has become overwhelmingly saturated, with up to three-quarters of prime initial viewing space occupied by paid placement. - Unlocking internal logistics networks to external third-party merchants mirrors the historical playbook that turned internal computing infrastructure into AWS.

[07:06](https://www.bilibili.com/video/BV1sNpg6FEKK?t=426s) AWS at $2.7 Trillion: Cloud Moats, Low-Cost Compute, and Trainium

The conversation shifts to whether Amazon's massive $2.7 trillion market capitalization is supported almost single-handedly by AWS. Over the trailing twelve months, AWS has reached nearly $150 billion in run-rate revenue, pairing accelerated top-line expansion with widening operating margins. Management recently floated an extraordinary projection: the belief that AWS alone could someday become a $1 trillion annual revenue business. While consensus Wall Street models project the entire consolidated Amazon enterprise to cross $1 trillion in revenue around 2028, the prospect of AWS achieving that milestone on its own would redefine enterprise software.

Evaluating the competitive landscape against Microsoft Azure and Google Cloud Platform (GCP), the hosts observe that while GCP has posted faster percentage growth and Azure has captured significant AI headlines, AWS retains formidable structural advantages: - **Data gravity and switching friction:** Enterprise corporate data already resides inside AWS S3 buckets and databases. Migrating petabytes or exabytes between clouds is notoriously complicated and penalized by punishing egress fees. - **Proprietary hardware efficiency:** Amazon continues to invest heavily in its own silicon, such as Trainium, to drive down the cost curve for model training and inference. - **Horizontal neutrality vs. vertical lock-in:** Unlike Alphabet, which operates in a vertically integrated fashion and strongly encourages enterprise clients to adopt Gemini and Google native databases, Amazon functions purely as a neutral, low-cost utility.

AWS has dominated infrastructure for decades by offering the lowest-cost compute at massive scale. As inference and API costs for models like OpenAI's ChatGPT and Anthropic's Claude inevitably compress toward zero, the host notes that the business that thrives through brutal margin compression is the low-cost commodity infrastructure provider: Amazon.

[10:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=600s) Disruption Risks, Amazon's Adaptation Playbook, and the 28x Forward P/E Debate

Amazon finds itself facing an existential crossroads where the core business may need to radically adapt or risk being disrupted. Yet betting against the company runs counter to more than two decades of history. Time and again over the last twenty-plus years, Amazon has consistently expanded its total addressable market every few years by pushing into new verticals. Even in industries where it is not yet the dominant player—such as healthcare and medical services—it steadily gains ground, packaging those offerings directly into Prime memberships to build a highly stable, recurring subscription moat.

At the same time, the underlying financials present a tricky picture. Free cash flow is practically non-existent right now, weighed down by substantial capital obligations and significantly more debt than casual observers realize, even though operating cash flow continues to trend upward. The looming question remains whether "agentic shopping" will actually cross the chasm into the mainstream. Currently, very few everyday consumers are delegating their product purchases to autonomous AI agents, but if that behavior takes hold, it directly threatens the core Amazon shopping interface and its lucrative advertising machinery.

From a valuation standpoint, the hosts draw parallels to recent historical mispricings in mega-cap tech. Both Alphabet and Meta traded down to 15 or 16 times earnings—and Meta into the low teens—when the market feared their business models were dying, only to see multiples expand back toward 30 times. Amazon currently trades around 28 times forward earnings, which the speakers view as fair rather than an obvious bargain. While retail revenue growth has actually accelerated in the near term, one host notes that a multiple compression down to 17 or 18 times forward earnings over the next year or two would make the stock significantly more intriguing, whereas the other host considers anything in the low 20s to be a compelling "no-brainer buy" given the company's insurmountable global logistics and fulfillment scale.

- **Valuation Multiple Target:** Amazon trades at ~28x forward earnings; hosts view low 20s or 17–18x as attractive entry points. - **Cash Dynamics:** Operating cash flow remains healthy and improving, but free cash flow has turned negligible amid heightened debt and aggressive investment cycles. - **Ecosystem Expansion:** Continued integration of new service verticals (e.g., medical care) into Amazon Prime strengthens the sticky, subscription-driven flywheel.

---

[12:35](https://www.bilibili.com/video/BV1sNpg6FEKK?t=755s) The Agentic Shift: Google Parallels and the Cautionary Tale of Sears

The debate shifts to whether the bear case against Amazon mirrors the ongoing panic surrounding Google Search disruption. The hosts agree that the parallels are striking: just as search engines face displacement from conversational chat, Amazon's search-and-sponsored-ad model faces replacement by AI shopping agents. However, incumbents possess agency and can re-architect their platforms. Amazon could adapt by allowing autonomous agents to query its inventory directly and restructuring its monetization—shifting away from pay-per-click sponsored listings toward alternative transaction fee or surfacing models.

Yet technological history introduces a sobering counter-argument: when massive paradigm shifts occur in retail, the champion of the previous era rarely dominates the next. The classic historical precedent is Sears, which conquered catalog shopping and physical department stores but was completely bypassed when internet e-commerce arrived. If commerce is shifting from human browsing to agentic execution, Amazon's legacy advantage as a consumer search destination might not automatically carry over.

```mermaid flowchart TD accTitle: Paradigm Shift in Retail Shopping Interfaces accDescr: Diagram illustrating how shopping interfaces evolved from mail order to e-commerce and the potential shift to AI agents.

subgraph Past ["Historical Precedents"] A["Catalog & Mail Order (Winner: Sears)"] -->|"Internet Revolution"| B["Web & App E-Commerce (Winner: Amazon)"] end

- **Incumbent Adaptation:** Amazon can redesign its platform to surface products to bots and adjust fee models beyond standard ad placements. - **The "Sears Precedent":** Dominant retail leaders historically struggle to survive platform shifts because their business models are optimized for the previous paradigm. - **Operating Scale:** Amazon is currently deploying more than $200 billion annually across its global operating footprint and investments.

---

[14:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=855s) Capital Expenditure Cycles, Negative Free Cash Flow, and Future Operating Leverage

Amazon's aggressive spending—running well north of $200 billion annually in total operating and capital outlays—explains why free cash flow looks so depressed on paper. Suppressed free cash flow has been Amazon's deliberate operating posture for decades, as Jeff Bezos and Andy Jassy have consistently chosen reinvesting every available dollar over showing short-term GAAP profits. Unlike peers such as Meta, Alphabet, and Apple, Amazon has historically refrained from conducting major share repurchases, leaving that capital return lever untouched for the future.

While the price-to-free-cash-flow metric currently looks bleak—mirroring the sharp FCF contractions experienced by Meta and Google during peak infrastructure buildouts—the hosts emphasize that Amazon has demonstrated its ability to turn on the cash spigot at will. Approximately two and a half years ago, management showed what happens when they optimize for profitability: free cash flow surged dramatically across four consecutive quarters. The hosts expect that once this massive AI and cloud infrastructure investment cycle begins to moderate later in the decade, Amazon will see another tremendous surge in operating efficiency and free cash flow.

- **Capital Allocation Stance:** Amazon remains one of the few mega-cap tech giants that does not actively repurchase its own shares. - **The Profitability Switch:** Amazon previously proved it could unlock massive free cash flow over a four-quarter stretch by dialing back expansion and streamlining retail efficiency. - **Macro Spending Phase:** The current negative-to-flat free cash flow profile reflects peak capital outlays across data centers, robotics, and logistics modernization.

---

[16:05](https://www.bilibili.com/video/BV1sNpg6FEKK?t=965s) Hidden Value: Trainium Chips, the Anthropic Stake, and Skyrocketing Token Consumption

Beyond the core retail website, Amazon possesses massive, underappreciated assets that could redefine its valuation. Among them is its proprietary silicon division, specifically the Trainium AI accelerator chips. If Trainium were spun off as an independent entity, it would rank among the largest semiconductor businesses in the world, serving as Amazon's direct answer to Google's Tensor Processing Units (TPUs). Coupled with Amazon's multibillion-dollar strategic stake in AI lab Anthropic—which the hosts jokingly dismiss as merely "a hundred, two hundred billion between friends"—AWS holds significant structural upside in the AI supply chain.

The conversation closes with an exploration of enterprise AI consumption. Recent reports indicate both Amazon and Meta have urged internal teams to manage their cloud and AI expenses, with Meta reportedly spending as much as $100 million per month on Anthropic models. The hosts marvel at reports of individual heavy users running up seven- or eight-figure annual bills on AI tokens and agentic workflows, bantering about their own personal usage limits and how they manage their quotas before the weekly reset on Sunday. Ultimately, they leave viewers with the central investment question: will Amazon's core retail business be disrupted by AI, will it disrupt itself, or is AWS's cloud and silicon engine robust enough on its own to justify buying the stock at 28 times earnings?

- **Trainium Silicon:** Amazon's proprietary AI training chips represent an under-the-radar business with scale rivaling standalone semiconductor leaders. - **Strategic Anthropic Stake:** Amazon's equity and partnership with Anthropic give it a direct stake in top-tier foundational model development. - **Enterprise Token Spend:** Meta's reported expenditure of ~$100M monthly on Anthropic highlights the astronomical capital flowing into frontier AI workloads.

## Notable quotes > [00:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=0s) AWS在AI浪潮中占据了极具优势的商品供应商定位，不仅自研芯片，还持有Anthropic的大笔股份，近期营收增速呈现显著加速态势。

> [00:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=0s) 管理层甚至预测，未来仅AWS单项业务就有望成长为营收达1万亿美元的巨头。

> [02:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=135s) 如果没有贡献约80%高利润率的广告收入支撑，其庞大的核心零售业务在剥离广告后其实并不具备真正的盈利能力。

> [02:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=135s) 如果未来消费者不再直接登录Amazon.com或App进行搜索，而是交由AI代理全网比价代购，不看广告的AI机器人将直接动摇亚马逊的“需求聚合器”商业模式。

> [07:06](https://www.bilibili.com/video/BV1sNpg6FEKK?t=426s) 第三，合理评估当前28倍前瞻市盈率的安全边际，耐心等待更具性价比的估值回调机会再做长线布局。

> [02:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=135s) Stripped of sponsored ad placements, core retail e-commerce operates at an outright structural loss.

> [02:15](https://www.bilibili.com/video/BV1sNpg6FEKK?t=135s) Bots do not look at ads

> [05:00](https://www.bilibili.com/video/BV1sNpg6FEKK?t=300s) a better version of UPS

> [16:05](https://www.bilibili.com/video/BV1sNpg6FEKK?t=965s) a hundred, two hundred billion between friends

## Mentioned Amazon, AWS, Anthropic, Muse, Shopify, Walmart, Meta, Google, Microsoft Azure, Google Cloud Platform (GCP), Trainium, Inferentia, OpenAI, ChatGPT, Claude, Gemini, Sears, UPS, Jeff Bezos, Andy Jassy, Apple

## Video description 【The Motley Fool】亚马逊的下一个最大隐忧是什么？ This Might Be Amazon's Biggest Issue Going Forward

亚马逊作为美股市场最伟大的企业之一，其未来的增长逻辑和潜在风险正在发生微妙的演变。今天我们来深度剖析亚马逊当前的核心隐忧，探讨其引以为傲的零售电商帝国是否正在面临颠覆性挑战，以及云服务AWS能否继续撑起高达2.7万亿美元的庞大估值。

从当前的业务结构来看，亚马逊的增长引擎主要分为AWS云服务与传统零售电商两大部分。AWS在AI浪潮中占据了极具优势的商品供应商定位，不仅自研芯片，还持有Anthropic的大笔股份，近期营收增速呈现显著加速态势。管理层甚至预测，未来仅AWS单项业务就有望成长为营收达1万亿美元的巨头。然而，零售电商业务却面临着严峻的时代拷问。随着智能AI购物代理（Agentic shopping）和各大竞争对手生态的崛起，传统电商正迎来前所未有的结构性威胁。

核心争议在于亚马逊对第三方AI购物工具采取了相对封闭甚至敌对的态度。例如，亚马逊明确阻止了外部AI助手（如Anthropic旗下的Muse等工具）与自身平台打通，而Shopify、沃尔玛等竞争对手则积极拥抱了这一趋势。亚马逊之所以选择筑起高墙，根本原因在于其零售业务的盈利模式高度依赖广告变现。过去12个月中，亚马逊广告业务收入已高达76亿美元。如果没有贡献约80%高利润率的广告收入支撑，其庞大的核心零售业务在剥离广告后其实并不具备真正的盈利能力。如果未来消费者不再直接登录Amazon.com或App进行搜索，而是交由AI代理全网比价代购，不看广告的AI机器人将直接动摇亚马逊的“需求聚合器”商业模式。

除了广告与流量入口的冲击，亚马逊当前28倍左右的前瞻市盈率是否具备吸引力同样值得商榷。虽然管理层在资本开支上持续砸下巨资用于物流网络和机器人升级，但从自由现金流表现来看，目前依然承受着较大的资金消耗与债务压力。

对于普通投资者而言，当前阶段的操作建议：第一，密切关注AI购物代理对传统电商流量入口的侵蚀速度与巨头自身的防御策略；第二，追踪AWS在云计算与AI算力供给中的营收增速及利润扩张弹性；第三，合理评估当前28倍前瞻市盈率的安全边际，耐心等待更具性价比的估值回调机会再做长线布局。

简洁时间轴 [00:00] 亚马逊业务现状与AWS的AI战略布局 [02:15] 零售电商面临的AI代理冲击与广告盈利依赖 [05:30] 物联网与物流基础设施的开放升级与GMV增长 [07:06] AWS作为核心支撑与2.7万亿美元市值估值解析 [11:15] 自由现金流状况及大规模资本开支背后的思考 [13:10] 当前估值水平下的投资决策与未来展望

[Source](https://www.bilibili.com/video/BV1sNpg6FEKK)

中文简报

亚马逊筑墙封杀AI购物代理 760亿美元广告业务直面脱媒风险剥离高利润赞助广告后亚马逊核心电商在实质上陷入亏损，自主AI购物代理对搜索流量入口的绕过将动摇其需求聚合器的盈利根基。节目围绕亚马逊当前的“双引擎”展开深度博弈分析：AWS凭借自研Trainium芯片、对Anthropic的战略投资以及作为AI“公用基础设施”的低成本定位正加速扩张，甚至展现出冲击单体万亿美元营收的潜力。然而，核心零售业务却陷入严重隐忧——其盈利结构高度依赖过去一年创造760亿美元、利润率约80%的站内搜索广告，而一旦消费者转向自主AI购物代理（Agentic Shopping），不看广告的比价机器人将直接瓦解其流量聚合变现模型。结合公司目前超2000亿美元的资本支出与承压的自由现金流，主讲人探讨了亚马逊是否会沦为“加强版UPS”或步西尔斯后尘，并认为当前28倍前瞻市盈率缺乏足够安全边际，回调至20倍左右才更具长线吸引力。 AWS凭借自研Trainium芯片、对Anthropic的战略投资以及低成本定位加速扩张，展现出冲击单体万亿美元营收的潜力。核心零售高度依赖过去一年创造760亿美元、利润率约80%的站内搜索广告，自主AI购物代理的绕过将瓦解其流量变现根基。在超2000亿美元资本支出与现金流承压下，主讲人认为当前28倍前瞻市盈率缺乏安全边际，回调至20倍左右更具长线吸引力。

Reading summary

The Motley Fool contributors argue that while Amazon Web Services is thriving as a foundational AI cloud provider, Amazon's retail business faces an existential threat from autonomous AI shopping agents that could bypass its high-margin advertising engine. Agentic AI Threats to Retail Advertising (2:15) Amazon's retail profitability relies heavily on its digital advertising segment, which generated roughly $76 billion over the trailing twelve months with estimated operating margins near 80%. Because the combined operating income of the North American and International retail units totals only about half of ad revenue, the underlying e-commerce operation is structurally unprofitable without sponsored placements, which currently occupy up to 75% of initial search results. The rise of autonomous shopping agents threatens this model because software bots do not view sponsored listings. While competitors like Shopify and Walmart have embraced agent integrations like Muse, Amazon has blocked external bots to protect traffic and its roughly 50% merchant take rate across fees and fulfillment. The hosts caution that this defensive posture risks permanently rerouting shoppers elsewhere. They cite Sears as a historical precedent, noting that retail leaders built for one era rarely survive major platform shifts. Fulfillment as a Third-Party Logistics Platform (5:00) To offset potential long-term erosion in search advertising margins, Amazon may need to unlock incremental profitability through its physical assets. The hosts outline an evolution where Amazon opens its internal fulfillment and freight network to external merchants and non-Amazon packages, effectively creating an 'AWS for logistics' that captures unmatched economies of scale. However, this shift creates valuation questions for equity investors. If automated agents weaken the retail division's advertising moat, Amazon's consumer business becomes primarily an asset-heavy logistics and warehouse network. The speakers question whether investors evaluating Amazon at current multiples would effectively be overpaying for an upgraded version of UPS. AWS Compute, Proprietary Silicon, and AI Workloads (7:06) In contrast to retail, AWS has reached an annualized run-rate of nearly $150 billion with re-accelerating top-line expansion and widening operating margins. Management has suggested that AWS alone could eventually reach $1 trillion in annual revenue. Rather than spending capital exclusively to build front-line foundation models, Amazon functions as a neutral utility supplying cloud infrastructure, hosting, and custom hardware. AWS maintains strong switching moats due to enterprise data gravity in existing storage systems. The hosts highlight proprietary silicon, particularly Trainium accelerators, as an underappreciated asset that reduces model training and inference costs. Paired with a multibillion-dollar equity stake in Anthropic, AWS is positioned to capture surging enterprise token demand, highlighted by reports of Meta spending as much as $100 million monthly on Anthropic models. Valuation Multiples and Heavy Capital Cycles (10:00) Amazon currently trades at approximately 28 times forward earnings, which the hosts view as fair rather than deeply discounted. While one host suggests the low 20s would represent an obvious entry point given Amazon's fulfillment advantages, another considers a multiple compression to 17 or 18 times forward earnings necessary to make the risk-reward compelling, comparing the situation to prior dips in Meta and Alphabet. Free cash flow is currently flat to negligible because Amazon deploys more than $200 billion annually across operating footprints, data centers, robotics, and logistics modernization, alongside holding elevated debt. Unlike many mega-cap peers, Amazon does not conduct major share buybacks. However, the hosts emphasize that management proved two and a half years ago it can generate substantial free cash flow across consecutive quarters whenever it decides to curb expansion spending. The hosts leave open whether agentic shopping will achieve widespread consumer adoption quickly enough to cripple Amazon's advertising flywheel, or whether the company can re-architect its monetization and rely on AWS's high-margin compute utility to sustain its $2.7 trillion market valuation.

Source quotes

[0:00] AWS在AI浪潮中占据了极具优势的商品供应商定位，不仅自研芯片，还持有Anthropic的大笔股份，近期营收增速呈现显著加速态势。 [0:00] 管理层甚至预测，未来仅AWS单项业务就有望成长为营收达1万亿美元的巨头。 [2:15] 如果没有贡献约80%高利润率的广告收入支撑，其庞大的核心零售业务在剥离广告后其实并不具备真正的盈利能力。 [2:15] 如果未来消费者不再直接登录Amazon.com或App进行搜索，而是交由AI代理全网比价代购，不看广告的AI机器人将直接动摇亚马逊的“需求聚合器”商业模式。 [7:06] 第三，合理评估当前28倍前瞻市盈率的安全边际，耐心等待更具性价比的估值回调机会再做长线布局。 [2:15] Stripped of sponsored ad placements, core retail e-commerce operates at an outright structural loss. [2:15] Bots do not look at ads [5:00] a better version of UPS [16:05] a hundred, two hundred billion between friends

Original transcript

[0:00] Hey everyone and welcome back to another episode for you today. We are going to talk about the one and the only Amazon in my eyes, still one of the best businesses out there. Although in today's episode, we are going to talk about Amazon. the good and maybe the bad as well I'm joined here with Travis Amazon. What's the verdict nowadays on this tremendous business? such an interesting place for Amazon because it's just not what it once was on the retail side of things. I think we should probably talk about two businesses. There's AWS and then there's the retail business. I actually like a lot of what's going on with AWS today. You look at accelerating revenue growth at this point. They, I think, have a really compelling strategy in the world of AI where They're not building models. They happen to own a big chunk of Anthropic, which is good news today. They're building their own chips. They're the commodity supplier. I like that positioning in what I think is... is a pretty commodity market in AI. for retail I've got a lot of questions you know what happens to Amazon in an agentic world if we go in that direction Not a super easy answer, and they have been the most hostile to some of the AI shopping, which is an interesting position for a company that used to be the disruptor, and now they're potentially the one getting disrupted. And it is interesting because I do feel like they have been pushing AI search for shopping, but... only on their own platform, whereas when Muse came out, they made sure that Muse and Amazon will not be able to work with each other, but then again, we've had Muse and Shopify, we've had Muse and Walmart, I think we had Muse and many other competitors out there that did partner up. So it's-- It is interesting to see how Amazon works around this agentic let's say, shopping assistant feature. Because on the one hand, if you don't do anything, you're going to be left behind. Because at the end of the day, my AI agent is just going to go and find that same product just somewhere else. So you just lost the sale. Is losing a sale worth it purely because you need to protect the who wrote in the advertising business has been, of course, tremendous. for Amazon. Over the past couple of years, it's now a business generating $76 billion over the last 12 months. And I've included here... Because Travis asked... international operating income and North America operating income, which You can see makeup almost. half of it. half of the whole advertising revenue yes so let's let's talk about the incentives there so what that is telling you is that without advertising revenue which is going to be somewhere around 80 percent margin. Without that advertising revenue, the retail business is not profitable. So if you want to know why they blocked Muse, that's your answer right there, is that if bots aren't watching ads which they're not They don't want bots on the site. Amazon's entire business model is built on the idea of being the aggregator of demand. Amazon is the place that you go to shop for whatever it is that you want. Not only, you know, Prime is like a good business, but Prime is a reason to keep coming back to Amazon. That's the real reason that Prime is a magical subscription. If you go to a world where, you know, somebody else, an agent or whatever is doing your shopping for you, maybe keeping an eye on your fridge. I mean, these are all kind of hypothetical things, but we're talking about, you know, where's the growth going to come from five or 10 years from now? that's a real change to the business. And you can just you can try this yourself, you can go and you can search for something on Amazon, whether you're on a mobile phone, or whether you're doing it on a What you're going to see is ads. Most of what you see is ads. And Amazon gets paid for that placement, just like Google gets paid for that placement. So how does the business change? How do they adapt to that in the future? Are they going to have to raise their prices? You mentioned Shopify. plugging into Muse right away. Shopify's business model is perfect for Muse. I mean, Shopify really was built on the back of... of meta Facebook and Instagram ads because they were able to go in and say Okay. Merchant. you do your advertising on let's say Instagram that points to a Shopify shop It's sort of a, those companies are married people. to each other, grow together in a lot of different ways. Amazon's completely different because it's predicated on you going to Amazon.com or the Amazon app. If you're not doing that anymore... That's a real threat to its business. So it brings in a lot of questions about what that future looks like. And remember, most of their business is now third-party sellers. So it's the same people that have a Shopify shop. Have a shop on Amazon. If they go, hey, we would love it if you would shop directly on our site through an AI agent instead of going through Amazon. So we have to pay somewhere around 50% of our revenue. [5:00] revenue to Amazon for advertising, for logistics, for all that kind of stuff. That's a real question for them long term. So the question now becomes, how do they... fix it or how do they work around it? How do they find a couple of percent in margin here and there to still be extremely Competitive. cheap enough, fast enough. On the one hand, we have the other parts of the business that might subsidize it. So, you know, you have a lot of people who are interested in, or the logistics network. could probably run much better than anybody else. I think that's also one of the main reasons why they've opened up their logistics network to outsiders, sort of AWS, but for logistics, shipping, et cetera. So that's one way. Then the other way is I guess opening up and saying, you know what, we are going to focus on GMV growth, and making sure that we still do have quite a lot of sales that go through our platform and try and make some extra money here and there. Still through advertising, I'm pretty sure that they could still grow. the advertising business in various ways, but yeah, if you go on Amazon.com right now, you search for something, the top row is likely to be 75% sponsored listings. And like you said, AI agents, well, they don't really. They don't really care that much about ads. Yeah, so I think what you're getting at here is if you're buying Amazon for the retail business, are you really buying just a better version of UPS? And is that worth the valuation that you're paying? On the other hand, we could say, are we buying... Amazon purely because of AWS, which AWS right now is generating close to $150 billion over the last 12 months, has seen growth accelerate while margin expand as well. is Amazon at $2.7 trillion. purely a AWS story or... something else because management did say more recently, I think that they believe that AWS alone could be a trillion dollar business. In the future, revenue-wise, I mean, at this rate, first of all, the overall Amazon business, I think in 2028, it's expected to generate a trillion dollars in revenue, but it's the whole business, not... not AWS alone. But yeah, at these prices, Does this justify... buying Amazon purely because of AWS. Of course, you still have all the rest at the moment, but let's say that all the rest does not continue to grow as much as as it is today. I think that's the most compelling part of the business is in what's been so impressive. And it showed in that chart there that. the growth has accelerated in a way that I think is really meaningful. I think you look back, you know, even just a handful of quarters and you could see, you Azure... maybe taking share... GCP in Alphabet doing extremely well. GCP is still growing faster. But AWS is where a lot of companies already have their data. So if it's just a matter of, okay, my stuff's already there. It's really hard to move clouds. They make it really expensive to leave too. Maybe, you know, as they get better at AI stuff, they have more models that are available. You know, Tranium gets better. Maybe we'll just stick with Amazon. And I think that's a pretty compelling business model. The other thing is they can think about this completely differently because they're not a vertically integrated company. So if you think about like Alphabet. They're vertically integrated, so they would love for you to use Gemini. Maybe you don't have to use Gemini, but they would love for you to use Gemini. They would love for you to use their databases as a customer. They want you to use all their systems. Amazon can. Just say, hey, we are going to provide the lowest cost compute. And that's the way that we've been competing for decades now in AWS. You know, we don't often think about that, but they are the most efficient at The cloud today, now that we're now we're just moving to a world where you're providing GPUs instead of instead of CPUs and storage, I think they're gonna be able to do that pretty effectively in the future. And so as you see some of these costs come down for you know, ChatGPT or for Claude. Who is the winner there? If you're... If your revenue starts to go down, who survives and ultimately thrives? It's a company like Amazon. So we are here in a difficult... situation because on the one hand we can see Amazon being extremely successful because of the position that they're in right now, but on the other hand, di lezzelle [10:00] I guess the Amazon business might need to change. Or else... it might be, the one that gets disrupted. So it's a very difficult situation, I guess. Then again, I always go back and look at the size of this business, look at their track record, look at... everything that they do. And I feel like somehow they'll figure it out. Because over the last 20 plus years or so, they've always managed to figure it out. that they've managed to grow First of all, their total dressable market seems to expand every single X couple of years. They managed to go into another industry. They might not be the biggest player, medical for example, They might not be the biggest player there, but slowly but surely they're entering that space. They're including this with your Prime membership, which again makes In the case that the prime subscription business continues to grow and that could be a stable business, very profitable for them. as well So, yeah. Plus all the investment that they're making as well, but it's a very interesting case here for Amazon. It's one of course that's As of right now, you look at free cash flow, it's non-existent. And there's a lot more debt than you might think too, yeah. Yeah, but cash from operating activities is still improving and doing quite well. So interesting situation with Amazon, especially when you look at their core legacy business that, yes, it seems like with... And that only if. agentic shopping and all of that does become mainstream because as of right now, not that many people are asking their ai agent to go shop and and buy some some stuff online so It's a if... it gets... much bigger, what does that do to the core Amazon business? Yeah. And the other thing, you know, from an investment perspective to think about, too, is just the valuation. This is something that we've seen with a lot of these big tech companies. You know, I think we were kind of on the same boat with Alphabet when shares were trading for 15, 16 times earnings going, hey, this is this is a little bit crazy. The business is not going to be dying anytime soon. Same thing with Meta. You can go back to when they were trading in the teens. What often happens with these bigger companies is the valuation goes up to 30 times. Right now, Amazon's trading for about 28 times more. forward earnings Maybe not a great buying opportunity, but sometime in the next year or two, I wouldn't be shocked to see the price earnings multiple drop to 17 or 18 times forward earnings. That's probably when I would be more intrigued in picking up a company like this, where maybe you do see a little bit of a slowdown in the retail business long term. It's actually accelerated a little bit recently. So so things are actually going pretty well there short term. But, you know, long term, you're able to answer some of those questions with the valuation starts to come down. That's where it becomes a little bit more compelling because I. You're right. Ten years from now, Amazon is still going to be around. They're going to figure this out. They're a behemoth in the world. There's scale that so many companies are just not going to be able to compete with. I think the bigger question is just what's the appropriate price to pay? What do you think? What would be a no-brainer buy price or price to earnings multiple for you for Amazon? And if we could go to the low 20s, I think that that would be good enough for me. Of course, with... And by the way... I was just thinking about this. Is the bear thesis about the core Amazon business being disrupted similar to what we've heard? for Google and Google Search. It could. It absolutely could be. Yeah, it absolutely could be because they could figure it out. They could make a surf, allow people to surface their stuff to agents. You know, maybe your pricing is just a little bit different. So instead of, you know, taking your ad revenue, you yeah, there is a way to figure this out. And that's. Companies have agency. So I wouldn't be shocked if they do. I just I think it's notable that when you have these paradigm shifts in the way that we shop, typically the company that wins the last paradigm is not the one that wins the next paradigm. So if we are shopping with agents, like you said, big, big, big if it was not Sears that won in. you know, the on the end shopping and the Internet and e-commerce. Because they were a huge winner in the past. That would be my one caution is always what does this look like in the future? Is it the same as the last 20 years or is there something different coming? Yeah, fair enough, I think. Also with Amazon. It's a company that's spending over $200 billion right now a year. Yeah. And... Free cash flow, although free cash flow with Amazon has never really been their strong suit because it's always a business that's just... reinvest to make sure that the future of this business looks better and better. But I do think that at one point we are going to see just free cash flow chump because the core business just becomes so much better, so much efficacere. [15:00] so much more efficient. and It's one of the only ones that doesn't buy any shares as of ever. So that could, of course, help with valuation in the future. But the price to free cash flow is negative, just like it's negative with a Meta, for example. It's negative for Google last quarter. It's not negative for Microsoft as of right now. but with Amazon it's You know that a core business is very profitable. It's just on paper right now. If you look at the price of free cash flow, if you look at free cash flow, It doesn't exist. But at one point, I do think that they'll just, with a flip of the button, I think we've seen that two and a half years ago when they decided to say, you know, let us show you how profitable we are. And there was like four quarters where suddenly the free cash flow started to go up an insane amount. Where did that come from? I do think that we might see it later in this decade after the huge spending cycle does slow down. Plus, like you said, terrarium, Trinium could also be a huge business. I think they said if it's a standalone business, it could be one of the biggest businesses. ships businesses out there today. Just like TPUs for Google, I do think that that's another... business that's flying under the radar for for Amazon. Yeah, don't forget the stake in Anthropic 2. Just, you know, $100, $200 billion between friends. Eh, no big deal. Yeah, but I did see, I think, a report... earlier where Amazon and Meta are telling their employees maybe to use cloud a bit less. For Meta, I guess that's a good sign that Muse is working for them and it can save them billions of dollars. For Amazon, I wonder what they will be using because So like Amazon has a cloud or muse equivalent. So And the numbers behind those are kind of crazy. I think it was $100 million a month that I saw that Meta is spending on Anthropic. Yeah. crazy, crazy amount of spend that you have. I even saw that there are several individual people spending into seven maybe even eight figures a year on uh on ai agents that's i'm definitely not using that many tokens maybe you are No, I think Jose might be the one we should talk about, the token usage. But no, I'm nowhere near. I never actually reach. the limit. And whenever I... I'm close to it. Just say, you know what, let's wait for Sunday and we'll get a little refresh. But yeah, Amazon, let us know what you think down in the comment section below. core business, legacy business, could that be disrupted? Do they have to disrupt themselves to make sure that they continue to survive? Or is AWS just... good enough to buy Amazon at these prices. Let us know down in the comment section below, like, subscribe, do all of that, and we shall see each other in the next one. Bye-bye. . Thank you.