Amazon’s AI Business Scale

Amazon’s AI services are reported to have reached about $15 billion in annualised revenue, signalling that AI is moving from a cost centre to a material business line for the company. The figure highlights Amazon’s advantage in combining cloud infrastructure with distribution channels that can monetise AI across logistics, workplace software and healthcare. (Economic Times)

Amazon just put a number on a business investors had mostly been guessing at: its artificial intelligence services inside Amazon Web Services are now running at more than $15 billion a year, based on first-quarter 2026 revenue. Reuters said it was the first time Amazon disclosed a direct revenue figure for that business. (reuters.com) That number came from chief executive Andy Jassy’s shareholder letter, where he said Amazon Web Services reached an artificial intelligence revenue run rate of over $15 billion in the first quarter of 2026 and was “ascending rapidly.” He compared that with early Amazon Web Services history, saying the cloud unit was about 260 times smaller at a similar point in its buildout. (aboutamazon.com) A revenue run rate is not booked annual sales. It is the current quarter’s pace multiplied out over a year, like taking one lap time and estimating the full race if the speed holds. (reuters.com) The backdrop is Amazon’s spending bill. Jassy wrote that Amazon is not planning roughly $200 billion in capital spending in 2026 “on a hunch,” and he tied that spending to strong demand for artificial intelligence computing. (cnbc.com) Amazon has been building the plumbing as well as renting it out. In February, Amazon said its Trainium and Graviton chip businesses together had already passed a $10 billion annual revenue run rate, and this week Jassy said the broader chips business is now over $20 billion. (ir.aboutamazon.com) (usnews.com) That matters because Amazon is trying to lower one of the biggest costs in artificial intelligence: the chips. Jassy said custom chips such as Trainium can offer better price-performance than some general-purpose options, which lets Amazon sell computing more cheaply and keep more of the economics for itself. (aboutamazon.com) The software layer is moving too. On Amazon’s February earnings call, the company said Amazon Bedrock, its service for using foundation models from Amazon and outside providers, had reached a multibillion-dollar annualized run rate and was being used by more than 100,000 customers. (fool.com) Amazon also has more places to sell artificial intelligence than a pure cloud company does. Jassy’s letter pointed to uses across shopping, fulfillment, advertising, devices, healthcare, and business software, which means Amazon can make money from the same technology inside its own operations and by selling tools to other companies. (aboutamazon.com) The scale of Amazon Web Services makes the new figure easier to place. Reuters said the $15 billion artificial intelligence run rate is roughly one-tenth of Amazon Web Services’ overall $142 billion revenue run rate, so artificial intelligence is still a slice of the cloud business, but no longer a rounding error. (reuters.com) For the last two years, the big question around artificial intelligence was whether the spending would ever show up as real revenue. Amazon’s answer on April 9, 2026 was that the meter is already running fast enough to clear $15 billion a year, and it says demand is still above available capacity. (reuters.com) (qz.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.