Following the conclusion of its corporate restructuring last week, which included waiving Microsoft’s right of first refusal to be its exclusive compute provider, OpenAI has moved quickly to diversify its cloud infrastructure. The AI giant has just finalized a new, massive multi-year partnership with Amazon Web Services (AWS), reportedly valued at $38 billion.
Read: Nvidia becomes first $5-trillion company amid AI boom
The deal, announced Monday, immediately provides OpenAI with access to a vast supply of cutting-edge hardware. AWS will supply the company with “thousands” of NVIDIA GB200 and GB300 GPUs for both the inference and training of its next-generation models. Amazon expects to deploy the entire capacity that OpenAI has agreed to purchase by the end of 2026, with provisions included for optional capacity expansion in 2027 and beyond.
AWS stated that the partnership is expected to help “millions of users continue to get value from ChatGPT,” positioning the cloud provider as a critical partner in supporting the scaling of OpenAI’s services.
See also
While the deal provides OpenAI with crucial compute power from a Microsoft competitor, it raises significant questions about the company’s financial model and capital needs. The $38 billion commitment to AWS comes on top of an existing, long-term agreement with Microsoft. As part of its recent restructuring, OpenAI agreed to spend a staggering $250 billion on Azure services.
For context, The Information recently reported that OpenAI is generating an estimated $12 billion in annualized revenue. The combined cloud commitments, totalling nearly $290 billion across both Microsoft and Amazon, underscore the massive, unprecedented capital expenditure required to finance CEO Sam Altman’s vision of pouring trillions of dollars into global AI infrastructure. Despite securing multiple cloud partners, the path to funding this colossal need remains a key challenge for the company.
