AWS and OpenAI sign a USD 38 billion seven-year partnership

2025-11-03 · Publisher : AWS · Domain : Cloud & DevOps

OpenAI commits USD 38 billion to AWS: hundreds of thousands of NVIDIA GB200 and GB300 GPUs in EC2 UltraServers, with all targeted capacity before end 2026 and possible extensions in 2027.

### Background On 3 November 2025, AWS and OpenAI announce a **multi-year strategic partnership worth USD 38 billion**, with continued growth over seven years. OpenAI accesses AWS compute comprising hundreds of thousands of NVIDIA GPUs, with the ability to expand to tens of millions of CPUs for agentic workloads. OpenAI starts using AWS compute immediately, with all capacity targeted for deployment before the end of 2026 and the ability to expand into 2027 and beyond. ### Key facts - Amount: USD 38 billion, with continued growth over seven years. - Hardware: NVIDIA GB200 and GB300 GPUs clustered in Amazon EC2 UltraServers on the same network. - Expansion: up to tens of millions of CPUs for agentic workloads. - Schedule: immediate use; targeted capacity fully deployed before end 2026; extension possible in 2027. - Use: from ChatGPT inference to training next-generation models. ### Why it matters After the Oracle agreement and the revision of its Microsoft contract, OpenAI adds AWS to its suppliers. For customers, it confirms that large AI labs spread their workloads across several clouds, which makes hyperscaler capacities interdependent. The amount is a contractual consumption commitment, which says nothing about the real pace of use. ### Takeaway AWS joins OpenAI's supplier circle, with a USD 38 billion commitment over seven years.

Tags : AWS, OpenAI, Hyperscaler, Datacenter, NVIDIA