Amazon discloses $15B AI run‑rate
Amazon told investors that AWS’s AI services are now generating more than $15 billion in annualised revenue and CEO Andy Jassy defended a roughly $200 billion AI and data‑centre spending plan as necessary for the market’s importance. The disclosure gave markets firmer numbers to judge hyperscaler capex and signals that spending will be evaluated on contracted revenue, not just ambition. (manilatimes.net) (spokesman.com)
Amazon just gave investors a number they had been guessing at for more than a year: Amazon Web Services says its artificial intelligence services are now running at more than $15 billion a year in revenue. Andy Jassy put that figure in his April 9, 2026 shareholder letter, and Reuters said it was the first time Amazon had broken out the business this way. (aboutamazon.sg) (finance.yahoo.com) That number is not a full-year booked total. It is a run rate, which means Amazon took recent quarterly performance and annualized it, like looking at one fast lap and projecting the full race. (finance.yahoo.com) Jassy paired that disclosure with an even bigger claim: Amazon expects roughly $200 billion of capital spending in 2026, and he said the company is not making that bet “on a hunch.” He wrote that a substantial portion of the Amazon Web Services spending already has customer commitments attached and will be monetized in 2027 and 2028. (geekwire.com) (finance.yahoo.com) The reason investors wanted a hard number is simple: the giant cloud companies have been spending on data centers, networking gear, and chips at a pace that makes even strong profits look smaller. Amazon said free cash flow fell to $11.2 billion after a $50.7 billion year-over-year increase in property and equipment purchases, even as 2025 revenue rose 12% to $717 billion. (geekwire.com) (stocktitan.net) (s2.q4cdn.com) Amazon Web Services is the part of Amazon that rents computing power and software to other companies. Its overall revenue was running at about $142 billion a year, so Reuters said the new artificial intelligence figure is roughly one-tenth of the cloud unit’s current scale. (finance.yahoo.com) What Amazon is selling here is not one single chatbot. Amazon Web Services has been pushing Amazon Bedrock for access to outside and Amazon-built models, Amazon SageMaker for building and tuning systems, and Trainium-powered infrastructure for customers that want cheaper computing than top-end Nvidia gear. (aws.amazon.com 1) (aws.amazon.com 2) Jassy’s letter also tried to show that Amazon is not only renting the factory floor but also building some of the machinery. GeekWire reported that Amazon’s internal chips business, including Graviton processors, Trainium artificial intelligence chips, and Nitro networking cards, is now running above $20 billion a year. (geekwire.com) (finance.yahoo.com) That matters because the economics of artificial intelligence are still dominated by the cost of chips and electricity. Jassy told shareholders that better chip price-performance, changes in model design, and more efficient infrastructure should push the cost of artificial intelligence down over time, the same way cloud computing got cheaper and then spread everywhere. (cnbc.com) Amazon is not the first hyperscaler to use this framing. Microsoft said on January 29, 2025 that its artificial intelligence business had passed a $13 billion annual revenue run rate, so Amazon’s new figure gives Wall Street a second benchmark for how quickly the biggest cloud firms are turning demand into contracted sales. (microsoft.com) (finance.yahoo.com) The thread running through all of this is that the market is starting to judge artificial intelligence spending less like a science project and more like a utility build-out. Amazon’s message was that if customers are already reserving future capacity, then the right way to read a $200 billion spending plan is not “too much,” but “already spoken for.” (finance.yahoo.com) (geekwire.com)
