Amazon’s cloud business is posting its strongest growth in years, and the story underneath the numbers is that a small number of AI model companies are becoming some of the largest customers in the history of cloud computing.
Amazon is reporting Q1 2026 earnings today, and the headline number for AWS is expected to be the most important figure in the report. Wall Street has been forecasting revenue of around $36.8 billion for the quarter, which would represent 26 percent year-over-year growth, faster than Q4 2025’s 24 percent and well above the 17 percent posted in Q1 last year. If the number lands near those estimates, it will mark the strongest cloud growth Amazon has posted since the 2022 boom, and it will arrive at a moment when the entire tech sector is under pressure to show that AI capex is producing real returns rather than deferred promises. AWS appears to have a credible answer.
The structure of the story matters as much as the revenue figure itself. The growth is not coming primarily from incremental enterprise adoption of traditional cloud services. It is coming from AI model companies that are consuming compute at a scale that was not legible in cloud pricing models just two years ago. Anthropic is the most visible example. The company’s annualized revenue hit $9 billion at the end of 2025, then climbed to $30 billion by early April 2026, a rise that is almost hard to process even when the numbers are written out. Analysts at KeyBanc estimate that AWS accounts for roughly 60 percent of Anthropic’s total compute spending, which at $30 billion of annualized revenue implies Anthropic alone could be contributing billions in quarterly AWS revenue. That is a material number for any customer, and extraordinary for one that did not exist a decade ago.
The OpenAI relationship adds another dimension. Amazon has entered a separate cloud partnership with OpenAI, whose annualized revenue has also been growing fast. Taken together, Anthropic and OpenAI represent a new class of cloud buyer: AI-native model companies that need compute not in the way an enterprise software firm needs compute, but the way a power grid needs fuel. Continuous, massive, non-negotiable and expanding. That is exactly the kind of customer a cloud provider wants at scale, as long as the infrastructure can keep up.
AWS has been racing to build that infrastructure. CEO Andy Jassy said on the Q3 2025 call that the company had added 3.8 gigawatts of power capacity in the prior twelve months, doubling its 2022 footprint, with plans to double again by 2027. Amazon has also guided toward roughly $200 billion in total capex for 2025 and 2026 combined, with the bulk going toward data centers. That is a commitment large enough to be audacious in any other context, and the market has been watching closely to see whether those costs produce a revenue acceleration that justifies the spending. The Q1 numbers, if they hold, are the best evidence yet that the answer is yes.
There is also a backlog signal that matters here. AWS reported a contract backlog of over $200 billion at the end of Q3 2025, and that figure has been growing. A large and expanding backlog means customers are not just spending now, they are committing future revenue. For investors trying to figure out whether the cloud infrastructure buildout has a real demand foundation, a $200 billion-plus backlog is a more durable signal than any single quarter’s revenue result. It says that the buying is not opportunistic. It is locked in.
The Customer Concentration Risk
There is a sharper angle here that is worth taking seriously. If AI model companies are becoming some of the most significant cloud customers in the market, then cloud growth is increasingly driven by a very small number of buyers. Anthropic and OpenAI are not like having thousands of enterprise customers each spending a few million dollars. They are two companies each spending at a scale that would previously have been the domain of global financial institutions or major government contracts. That concentration creates real upside for AWS. It also creates real risk.
If the model companies slow down, hit a funding gap, or shift their compute to competing infrastructure, the effect on AWS revenue would be visible and fast. Amazon’s relationship with Anthropic includes an equity stake estimated at around 17 percent, revenue-sharing agreements tied to Claude model sales through AWS, and the Project Rainier cluster of roughly 500,000 Trainium2 chips built specifically for Anthropic’s training workloads. That level of entanglement makes the relationship sticky, but it also makes it more complex than a standard vendor agreement. Amazon is no longer just selling compute to Anthropic. It is a co-investor with financial exposure to Anthropic’s success and a platform partner with structural incentives to keep it growing.
That is a different relationship than AWS has had with most of its customers, and it reflects a broader shift in how cloud providers are thinking about the AI era. The biggest upside in cloud revenue is no longer coming from convincing enterprises to migrate off on-premises hardware. It is coming from the frontier model labs that are running the most compute-intensive workloads in civilian technology history. Winning those customers and keeping them is now a central part of cloud strategy, not a side effect of it.
What The Quarter Tells The Market
AWS growing at 26 percent or better tells the market several things at once. It tells them that AI capex is converting to cloud revenue faster than the skeptics expected. It tells them that Amazon’s decision to accelerate data center spending was the right call, not a miscalculation. And it tells them that Anthropic’s growth trajectory, which has accelerated faster than almost anyone predicted, is showing up directly in one of the most visible financial metrics in American business.
The deeper implication is that cloud growth in the AI era may be stickier and more concentrated than anything the industry has seen before. When a frontier AI company triples its revenue in less than four months and sixty percent of its compute bill flows through one provider, that provider is no longer just a utility. It is a strategic partner, a financial stakeholder and an infrastructure dependency all at once. AWS has positioned itself at the center of that relationship with Anthropic, and the Q1 results are the first major proof point for what that position is actually worth.
Also read: AWS is growing fastest when it needs to most and Anthropic is why • Meta’s Louisiana AI campus is pulling utilities into the AI boom • Meta’s Louisiana AI campus is turning utility spending into an AI cost center