Amazon

Amazon delivered a Q1 2026 print that broke three records simultaneously and crashed one ratio that may matter more than all of them. Net sales of $181.5 billion (+17% YoY) beat the Street by over $4 billion, AWS revenue grew 28% to $37.6 billion (its fastest pace in three years, putting AWS on a $150 billion annualized run-rate), and net income nearly doubled to $30.3 billion. But $16.8 billion of that net income came from a mark-to-market gain on Amazon’s Anthropic equity position — not operations — and free cash flow collapsed 95% to just $1.2 billion as capex hit $44.2 billion in a single quarter.

What makes Amazon uniquely positioned in 2026 is the AWS-Anthropic vertical integration. The new Anthropic deal disclosed this quarter commits the AI lab to up to 5 gigawatts of current and future-generation Trainium chips for model training and inference, against a backlog now sitting at $364 billion (excluding the $100B+ Anthropic commitment which is reported separately). Project Rainier — Amazon’s purpose-built AI compute cluster for Anthropic, with hundreds of thousands of Trainium2 chips already deployed — is the most concrete Trainium-vs-NVIDIA comparison data the market has. CEO Andy Jassy committed to roughly $200 billion in calendar-year-2026 capex, the largest single-year capital commitment in Amazon’s history.

Amazon’s smart-money position is unusual: Berkshire Hathaway holds AMZN as one of its top-25 positions (originally initiated under Todd Combs in 2019), and the position has not been trimmed despite Buffett’s broader cash-hoarding posture. Bill Ackman’s Pershing Square does not currently hold AMZN. Stanley Druckenmiller cycled out of Amazon in 2024 but rebuilt a position in Q4 2025 according to his most recent 13F.

Insider activity has been muted but coherent: Andy Jassy has made no open-market sales in the past 24 months and his last 10b5-1 plan execution was in October 2025 at $182. CFO Brian Olsavsky has not transacted since 2023. The most-watched insider event remains Jeff Bezos’ 10b5-1 plan disclosed in Q4 2024, which authorized up to 25 million shares of sales through May 2026; Bezos has executed approximately 18 million of that authorization, with the remaining roughly 7 million expected before the program expires. The market has absorbed this distribution without sustained downward pressure on the share price — a signal of demand depth at the megacap-AI-exposure tier.

Amazon trades at $215 per share, roughly 38× consensus FY2026 EPS of $5.65 (after stripping non-operating Anthropic mark-to-market). On an EV/AWS-revenue basis, Amazon trades at approximately 8.5× — versus Microsoft (Azure) at 12× and Alphabet (GCP) at 9×. Wall Street consensus across 56 covering firms averages $260 (Morgan Stanley $275, JPMorgan $270, Goldman $265, Bernstein the bear at $230), implying ~21% upside. The forward multiple looks expensive on headline GAAP earnings but cheap on AWS-specific metrics — a divergence that gets resolved when Amazon eventually breaks out AI-specific revenue rather than reporting it embedded in the AWS line. The valuation case becomes asymmetric to the upside if AWS sustains 28%+ growth into Q3 with margin expansion, since the inference-layer take rate is materially higher than legacy IaaS.

1. July 31, 2026 (estimated): Q2 2026 earnings — first full quarter to test whether AWS can sustain growth above 25% with margins expanding, and first opportunity to provide AI-specific revenue disclosure separate from total AWS
2. Mid-2026: Project Rainier full deployment — completion of the Anthropic-dedicated Trainium2 cluster announced at re:Invent 2024; first quarter of full-scale revenue contribution from the largest single AWS-AI customer commitment
3. October 2026 (estimated): Bezos 10b5-1 plan expiration — the remaining ~7 million authorized shares should clear before May 2026 if the cadence continues; subsequent quarters remove a structural overhang on share supply

Amazon is the cleanest AWS-AI compounder I track if you’re willing to look through the quarterly free-cash-flow chaos. AWS at 28% growth with $150 billion run-rate, sub-40% operating margins on the AI tier, and the Anthropic-Trainium vertical lock-in are arguably the strongest combined operating signals across hyperscalers right now. My add-trigger is any quarter where AWS grows above 25% AND retail North America margin expands sequentially — that combination would signal that the $200 billion capex is genuinely funding revenue rather than running into depreciation drag. I would not chase AMZN above $230; I am building the position on every print where AWS holds above 25% growth and capex stays inside the $200 billion calendar-year band. The thesis breaks if AWS deceleration drops below 22% before fiscal 2027 — at which point the free-cash-flow compression becomes structural rather than cyclical.