OpenAI has entered into a $38 billion, seven-year cloud infrastructure agreement with Amazon Web Services (AWS), ending its exclusive reliance on Microsoft Azure and moving toward a multi-cloud strategy. The partnership gives OpenAI access to large-scale NVIDIA GPU clusters and extensive CPU capacity to support its growing artificial intelligence workloads.
The agreement, announced on Monday, grants OpenAI immediate access to hundreds of thousands of NVIDIA GPUs through AWS’s high-performance infrastructure, with plans to expand to tens of millions of CPUs by the end of 2026.
“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.”
Market Impact and Strategic Context
Amazon shares rose about 5% following the announcement, reaching a record high and adding nearly $140 billion in market value. The deal is AWS’s largest AI infrastructure engagement to date and comes amid growing competition among major cloud providers to support large AI model development.
The partnership will utilize AWS’s NVIDIA GB200 and GB300 GPU clusters via Amazon EC2 UltraServers, which are optimized for low-latency, high-bandwidth AI workloads, including training and inference for products like ChatGPT. AWS CEO Matt Garman said the collaboration demonstrates why AWS’s infrastructure scale and flexibility are suited to OpenAI’s evolving compute requirements.
Microsoft Ties Continue Despite Lost Exclusivity
OpenAI’s agreement with AWS does not end its collaboration with Microsoft, which retains a 27% stake in OpenAI Group, valued at around $135 billion. Microsoft has committed to purchase $250 billion in additional Azure services, though its exclusive cloud rights expired earlier this year.
The shift follows OpenAI’s restructuring into a public benefit corporation, which allows greater operational flexibility and enables partnerships beyond Microsoft’s previous “first right of refusal” clause.
The move also aligns with OpenAI’s broader effort to diversify infrastructure providers. In 2025, the company is expected to commit approximately $1.4 trillion in total compute contracts across multiple partners, including Oracle, Broadcom, and AMD, to support global AI scaling needs.
