Amazon's Best Quarter in Years Was Half-Built on a $16.8B Anthropic Paper Gain

Amazon reported first‑quarter 2026 net sales of $181.5 billion on Thursday — a 17% increase year over year that beat analyst estimates by more than $4 billion. Net income reached $30.3 billion, nearly doubling from $17.1 billion a year earlier. Earnings per share came in at $2.78, crushing the $1.64 consensus. On the surface, it was Amazon's best quarter in years.

But the composition of those profits tells a more complicated story. According to TNW, roughly $16.8 billion of Amazon's pre‑tax income came from a revaluation of its investment in Anthropic — the AI company behind the Claude family of models — not from selling products, cloud services, or advertising. That's more than half of the net income behind the earnings‑per‑share beat.

Strip out the Anthropic paper gain, and Amazon's operating profit was still a healthy $23.9 billion. But the distinction matters: the strongest earnings report in recent memory was substantially powered by owning a piece of an AI startup that has yet to turn a profit.

Amazon has committed up to $25 billion to Anthropic, and a portion of that investment converted from convertible notes into preferred stock during Q1 — triggered by Anthropic's most recent funding round. Under current accounting rules, significant financing events require marking the investment to fair value. Anthropic's shares have been trading at an implied $1 trillion valuation on secondary markets, and the company is reportedly in early discussions for an IPO as soon as October 2026, per TNW.

The accounting gain pushed Amazon's $8 billion cumulative investment in Anthropic to a mark‑to‑market value exceeding $70 billion. Each upward revaluation flows directly into net income, inflating earnings per share without any corresponding cash inflow. This isn't unique to Amazon. Motley Fool reports that Alphabet also holds a substantial Anthropic stake, and a Fortune analysis found that roughly half of both Amazon's and Alphabet's Q1 2026 AI profits came from their Anthropic holdings rather than their actual operating businesses.

The dynamic creates a strange situation: the AI companies spending the most on infrastructure — and generating the least free cash flow — are simultaneously reporting record earnings because the private companies they've invested in are being marked up by other investors.

The operational story underneath the Anthropic gain is genuinely strong. AWS revenue grew 28% to $37.6 billion, its fastest growth rate in more than three years and an acceleration from 24% in the previous quarter. Operating income from AWS reached $14.2 billion, well above the $12.8 billion consensus. CEO Andy Jassy told analysts that AWS demand continues to outpace supply, with the backlog growing as enterprise customers commit to multi‑year cloud and AI contracts, TNW reports.

Advertising revenue jumped 24% to $17.2 billion. Amazon's custom chip business — covering Trainium, Graviton, and Nitro — now generates more than $20 billion in annualized revenue, growing at triple‑digit rates. Jassy has described the chip division as potentially a $50 billion annual business if sold on the open market.

But the growth has a price. Capital expenditure hit $44.2 billion in the quarter, up from $25 billion a year ago. Amazon's trailing twelve‑month free cash flow compressed to $1.2 billion, a 95% decline year over year. The company has committed approximately $200 billion in total capex for 2026, making it the largest infrastructure spender among the hyperscalers in absolute terms.

Trainium2, Amazon's AI training and inference accelerator, has largely sold out, and the next generation Trainium3 — which began shipping in early 2026 — is nearly fully subscribed. Uber joined Amazon's Trainium customer roster during the quarter, and Meta signed a multibillion‑dollar deal for Graviton5 processors as its own AI compute demand outstripped infrastructure capacity.

These chip wins signal a strategic shift, TNW notes: the custom chip strategy is turning AWS from a cloud platform into a vertically integrated compute provider that competes not just with Microsoft Azure and Google Cloud but, increasingly, with Nvidia itself.

For builders, the implication is clear. If Trainium and Graviton continue gaining traction, the AI infrastructure market could diversify beyond Nvidia's near‑monopoly — potentially lowering costs and creating new deployment targets for AI workloads. Amazon's chip business is no longer an experiment; it's a $20 billion revenue stream with triple‑digit growth.

Amazon's relationship with Anthropic is deeper than a passive investment. Anthropic has committed to using AWS as its primary cloud provider, and the computing commitments that come with Amazon's $25 billion investment effectively lock Anthropic into Amazon's infrastructure for years. As Anthropic's models — Claude Code, Claude Cowork, and the newly released Claude Design — gain enterprise traction, AWS captures the compute revenue.

Motley Fool reports that Anthropic has reached 30.6% of U.S. companies paying for its services, rapidly closing the gap with OpenAI's 35.2% — a level that has been stable for the past year. Both Amazon and Alphabet added to their Anthropic stakes recently, with Amazon putting in $5 billion more and Alphabet adding $10 billion. Those investments come with computing commitments that will flow directly into cloud revenue.

With OpenAI reportedly missing its own revenue and user targets, per Motley Fool, Amazon's Anthropic bet looks increasingly well‑timed — both as a financial investment and as a strategic hedge against the OpenAI-Microsoft axis.

The Anthropic gain raises a question that applies to every major tech company with large stakes in private AI firms. Each time Anthropic raises money at a higher valuation, Amazon (and Alphabet) must mark their stakes to the new price. The gains flow directly into net income, inflating earnings per share — but produce no cash. Meanwhile, the massive capex required to serve AI demand destroys free cash flow.

The divergence between Amazon's reported earnings (rising) and its actual cash generation (collapsing) illustrates the degree to which Big Tech earnings in 2026 are being shaped by accounting conventions as much as by operational performance. The TNW analysis notes that if Anthropic's next valuation event is flat rather than up, the earnings narrative could shift dramatically.

Amazon guided Q2 revenue of $194 billion to $199 billion and operating income of $20 billion to $24 billion, assuming current tariff conditions hold. The guidance also incorporates approximately $1 billion in incremental costs from Project Kuiper, Amazon's satellite internet constellation now ramping manufacturing. For builders watching the cloud market, the signal is clear: AWS is accelerating, AI demand is real, but the financial engineering behind Big Tech's AI profits deserves a closer look.