Amazon’s AI division is generating $15 billion annually within AWS, chip demand is outpacing supply, and CEO Andy Jassy says the company’s massive capital commitment is anything but speculative. Amazon’s free cash flow fell sharply last year as capital spending surged by over $50 billion, a deliberate trade-off Jassy argues will pay off significantly in the years ahead. Two major customers attempted to purchase Amazon’s entire available Graviton chip supply for 2026, a move Amazon declined but which Jassy cites as a clear signal of where enterprise demand is headed.
In his annual letter to shareholders, Amazon CEO Andy Jassy made the most detailed public case yet for why the company is committing approximately $200 billion in capital expenditures in 2026. Far from a defensive posture, the letter reads as a confident, data-forward argument that Amazon is not chasing an AI trend — it is actively shaping one. “We’re not investing approximately $200 billion in capex in 2026 on a hunch,” Jassy wrote, framing the spending as a calculated response to demand signals that most companies would envy.
The headline figure drawing immediate attention is the $15 billion annualized revenue run rate for AI services within Amazon Web Services, a number reflecting performance through the first quarter of 2026. Amazon has not previously broken out this type of AI-specific revenue figure, making the disclosure a meaningful benchmark for the industry. For context, AWS as a whole carried a $142 billion annual revenue run rate as of the fourth quarter of 2025, meaning AI is now accounting for a notable and rapidly growing slice of the cloud division’s total business.
Jassy’s letter also sheds light on Amazon’s internal chip ambitions, revealing that its custom silicon business — anchored by the Graviton and Trainium processor lines — is now generating more than $20 billion in annual revenue. The disclosure about the two enterprise customers attempting to secure all available Graviton capacity for 2026 underscores a supply-demand imbalance that Amazon sees as a long-term tailwind rather than a short-term headache. The company turned down the requests, but Jassy made clear the episode illustrates just how tight the market for high-performance AI infrastructure has become.
The financial trade-offs embedded in this strategy were addressed openly. Amazon’s free cash flow dropped sharply from $38 billion to $11 billion over the past year, a direct consequence of capital spending rising by more than $50 billion — nearly all of it directed toward AI infrastructure, data centers, and custom hardware. That decline came even as the company grew total revenue by 12 percent, from $638 billion to $717 billion. Jassy framed the FCF compression not as a warning sign but as a deliberate investment decision, one the company expects to reverse as AI infrastructure scales and monetization accelerates.
Jassy’s language throughout the letter is notably assertive about Amazon’s competitive positioning. He described AI as “a once-in-a-lifetime opportunity” with growth that is both unprecedented today and potentially even larger in the future, and stated plainly that Amazon intends to be a dominant force in that landscape rather than a cautious participant. “We have never seen a technology more quickly adopted than AI,” he wrote, adding that enterprises are actively choosing AWS as their platform of record for artificial intelligence workloads. The shareholder letter effectively functions as an early preview of Amazon’s upcoming quarterly earnings report, which has yet to be released, and sets a high bar for the narrative the company is building around its AI returns.
Beyond AI, the letter touched on Amazon’s broader portfolio of long-horizon bets, including its Kuiper satellite internet initiative and its continued push toward faster last-mile delivery, with 20-minute fulfillment windows now part of the conversation. But the unmistakable center of gravity in Jassy’s message was artificial intelligence — and the conviction that Amazon’s willingness to absorb short-term financial pressure will translate into a structurally larger and more profitable business on the other side of this infrastructure buildout.
