OpenAI struck a $38 billion, multi‑year cloud deal with Amazon Web Services, sending Amazon’s stock up about 5% as investors cheered a marquee win for AWS in the race to power next‑generation AI. The pact underscores how surging demand for compute is reshaping alliances across Big Tech.
Announced on November 3, 2025, the seven‑year agreement gives OpenAI immediate access to AWS infrastructure—including hundreds of thousands of Nvidia GPUs—to train and run models like ChatGPT. The companies say initial capacity will come online right away, with capacity targeted by end of 2026 and room to expand into 2027 and beyond. The deal follows OpenAI’s recent restructuring that ended Microsoft’s exclusive role in supplying cloud resources, opening the door to multiple providers.
For AWS, the contract is a high‑profile validation at a moment when investors have questioned its AI momentum relative to Microsoft and Google. For OpenAI, diversifying compute is strategic: securing reliable, large‑scale infrastructure is essential to push frontier models and keep popular services running. The $38 billion commitment highlights the capital intensity of AI and the industry’s pivot to multi‑cloud strategies to ensure scale, price, and resiliency.
Amazon shares climbed about 5% on the news, hitting an all‑time high during Monday’s session as the market priced in fresh confidence in AWS’s competitiveness. The move came days after a strong quarterly update and amid broader debate about whether massive AI infrastructure bets—often spanning chips, data centers, and power—are sustainable for model makers that are still building toward profitability.
The OpenAI–AWS pact arrives as cloud providers jockey for a central role in AI’s build‑out. Amazon backs Anthropic, another leading model developer, even as it courts OpenAI’s workloads. Meanwhile, OpenAI continues to line up long‑dated infrastructure commitments across the ecosystem, a sign of both escalating technical ambitions and the sheer scarcity of top‑tier compute.
Watch for AWS to roll out dedicated clusters—featuring Nvidia GB200 and GB300 accelerators—to serve both training and inference at scale. OpenAI, newly freed to shop across clouds, is expected to keep balancing capacity, performance, and cost as it develops its next generation of models and agentic systems. Investors will be gauging whether rising AI demand translates into sustained cloud revenue growth—and whether spending ramps can be matched by cash flow over time.