Amazon CEO Jassy Discloses AWS AI Revenue for the First Time: $15 Billion Run Rate and Growing 260 Times Faster Than Early AWS

Amazon CEO Andy Jassy made two disclosures in his annual shareholder letter on April 9 that reshape how investors and the industry understand AI economics. First: AWS's AI services have reached a $15 billion annualized revenue run rate — the first time Amazon has ever put a dollar figure on its AI business. Second: Amazon plans roughly $200 billion in capital expenditure this year, almost entirely for AI infrastructure. "We're not investing approximately $200 billion in capex in 2026 on a hunch," Jassy wrote.

Amazon stock rose approximately 5% on the day.

How $15 billion in AI revenue compares

The $15 billion figure represents about 10% of AWS's $142 billion total revenue run rate. That share is growing at a pace Jassy called 260 times faster than AWS experienced at a comparable stage of its development — a comparison that carries weight given AWS itself grew from an internal experiment into a $100 billion business over roughly two decades.

The number also arrives capacity-constrained. Amazon Bedrock, the company's managed AI model service, processed more tokens in Q1 2026 than in all prior periods combined. Inference volumes were "nearly doubling month-over-month" through March, suggesting the $15 billion figure understates actual demand.

For context: Microsoft's AI business crossed a $13 billion annual run rate in late 2024. Anthropic reported $30 billion in annualized revenue this month. Amazon's disclosure turns the AI revenue conversation from a two-player comparison into a three-way race — with Amazon holding the infrastructure advantage of running the cloud platform that many AI companies, including Anthropic and OpenAI, build on.

The $200 billion bet and the chips behind it

Jassy framed the spending as a repeat of Amazon's early AWS investment cycle: concentrated capital spending now, durable cash generation later. "Of the AWS capex we expect to spend in 2026, much of which will be monetized in 2027-2028, we already have customer commitments for a substantial portion of it," he wrote. The OpenAI deal with Amazon alone is reportedly worth more than $100 billion.

Amazon's custom silicon business adds a second dimension to the infrastructure story. The company's chip lines — Graviton for general computing, Trainium for AI training, Nitro for virtualization — are generating over $20 billion in annualized revenue at triple-digit year-over-year growth. Demand is outstripping supply: two large customers asked to purchase all of Amazon's available Graviton capacity for 2026, and Amazon turned them down.

Jassy hinted the chip business could expand beyond AWS. "It's quite possible we'll sell racks of them to third parties in the future," he said — a statement that would put Amazon in direct competition with Nvidia as a chip supplier, not just a cloud provider.

Amazon's overall 2025 financials show the trade-off behind this strategy. Total revenue hit $717 billion, up 12% year over year. Operating income reached $80 billion, up 17%. But free cash flow dropped from $38 billion to $11 billion as AI capital expenditure consumed the difference.

What accelerating AI infrastructure means for consumer products

AWS powers the backend of thousands of consumer-facing AI products — shopping assistants, travel planners, healthcare tools, financial advisors. When Bedrock's token volumes double monthly, it means the number of companies embedding AI into consumer experiences is expanding at a rate the market has not seen since the smartphone app boom.

For Amazon's own consumer business, the implications are direct. AI optimization already runs through Alexa, Amazon's product recommendation engine, logistics routing, and warehouse automation. Amazon Pharmacy's same-day delivery of GLP-1 weight-loss medication — announced earlier this year — depends on AI-optimized supply chain management to make one-day pharmaceutical delivery economically viable at scale.

Mubboo's Take

Jassy's letter contains one line that matters more than the $15 billion figure: AWS AI is growing 260 times faster than AWS did at a comparable stage. AWS took nearly two decades to reshape how the internet works. If AI infrastructure is on a trajectory 260 times steeper, the transformation of consumer services — shopping, travel, healthcare, local services — will happen in years, not decades. For every platform building on AWS, Jassy's $200 billion bet on AI infrastructure is a bet that the tools used to serve consumers will get dramatically more powerful and more affordable. That trajectory is why building now, rather than waiting, is the right call.