OpenAI, the inventor of ChatGPT, announced a $38 billion, seven-year strategic partnership with Amazon Web Services (AWS). This deal is one of the largest cloud infrastructure agreements, which secures the massive computing power OpenAI needs to train and operate its next-generation AI models.
This partnership will help OpenAI with a strategic shift, as before it relied only on Microsoft Azure, its longtime partner and primary investor.
AWS will provide OpenAI with access to an extensive network of high-performance infrastructure, including hundreds and thousands of NVIDIA's most advanced GPUs, such as the GB200 and GB300 accelerators. OpenAI will also tap into tens of millions of CPUs via Amazon's EC2 UltraServers to power its agentic AI workloads: the complex, real-time reasoning tasks behind its applications. The full capacity of the deal is expected to be deployed before the end of 2026.
According to The Hindu, OpenAI co-founder and CEO Sam Altman, in a joint statement, said that, “scaling frontier AI requires massive, reliable compute. Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”
The move is the latest in OpenAI's aggressive campaign to diversify its infrastructure and secure an astonishing amount of computing resources. The $38 billion AWS pact follows many significant deals cracked by OpenAI.
OpenAI has signed approximately $1 trillion worth of infrastructure deals in 2025, which includes a $300 billion Oracle deal and a $500 billion Stargate project with Oracle and SoftBank, as cited in The Hindu reports.
The deal is a monumental win for AWS: It strengthens Amazon’s position as an essential infrastructure provider for frontier AI, even while it hosts OpenAI’s main rival, Anthropic. Amazon shares surged 4% to an all-time high after the announcement.
As per CNBC reports, AWS CEO Matt Garman said that, “The breadth and immediate availability of optimized compute demonstrates why AWS is uniquely positioned to support OpenAI's vast AI workloads.”
