OpenAI signs $38 billion compute deal with Amazon (AWS), breaking Microsoft’s AI cloud grip

OpenAI is cutting a massive new path through the AI infrastructure race. According to a report from CNBC, the maker of the popular ChatGPT has signed a $38 billion agreement with Amazon Web Services, marking its first deal with the world’s largest cloud provider—and a major shift away from its longtime dependence on Microsoft.

The partnership, announced Monday, allows OpenAI to immediately start running workloads on AWS using hundreds of thousands of Nvidia GPUs across U.S. data centers. Over time, Amazon plans to expand that capacity with new infrastructure built specifically for OpenAI’s operations. Amazon’s stock jumped 5% following the announcement, reflecting investor optimism about the deal’s scale and strategic importance.

“It’s completely separate capacity that we’re putting down,” said Dave Brown, AWS’s vice president of compute and machine learning services. “Some of that capacity is already available, and OpenAI is making use of that.”

Inside OpenAI’s $38 Billion Amazon Partnership And What It Means for Microsoft

For OpenAI, the move is part of a broader effort to secure more compute as demand for its large language models continues to soar. In recent months, the company has struck buildout agreements worth roughly $1.4 trillion with Nvidia, Broadcom, Oracle, and Google—sparking both excitement and skepticism about whether the U.S. has the infrastructure to deliver on such ambitious plans.

Until recently, OpenAI relied almost exclusively on Microsoft’s Azure cloud. The two companies have been tightly linked since 2019, when Microsoft made its first investment in OpenAI and eventually poured in a total of $13 billion. But in January, Microsoft said it would end its exclusivity arrangement, keeping only the right of first refusal for future compute needs. That status expired last week, freeing OpenAI to engage other major cloud providers—AWS, Google Cloud, and Oracle among them, CNBC reported.

“Scaling frontier AI requires massive, reliable compute,” said OpenAI CEO Sam Altman. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”

Even with the new alliance, OpenAI isn’t walking away from Microsoft. The company confirmed it plans to spend an additional $250 billion on Azure services. Still, the AWS deal underscores its desire to diversify partnerships and reduce single-vendor dependence—especially as it eyes an eventual IPO. CFO Sarah Friar has described OpenAI’s recent restructuring as laying the groundwork for going public.

For Amazon, the contract is a coup. It deepens its role in the AI supply chain and highlights AWS’s ability to handle workloads of OpenAI’s size and complexity. The move also positions Amazon closer to the center of the generative AI race, even as it continues to invest billions in OpenAI rival Anthropic. AWS is currently building an $11 billion data center campus in Indiana dedicated to Anthropic’s workloads.

“The breadth and immediate availability of optimized compute demonstrates why AWS is uniquely positioned to support OpenAI’s vast AI workloads,” AWS CEO Matt Garman said in the announcement.

The current deal centers on Nvidia’s Blackwell GPUs, though future expansions may include other chips. Amazon’s Trainium processor, already in use by Anthropic, could eventually become part of OpenAI’s compute mix.

The infrastructure will serve both inference—powering real-time applications like ChatGPT—and the training of new frontier models. OpenAI’s agreement runs through at least 2026, with room to expand as its compute needs grow.

By locking in massive long-term capacity with multiple cloud giants, OpenAI is signaling a new phase: one where it’s no longer just a Microsoft-backed startup but an independent force shaping how AI is built, scaled, and distributed globally.