Amazon Web Services (AWS) and OpenAI have announced a multi-year, strategic partnership under which AWS will provide its infrastructure to support and scale OpenAI’s core artificial intelligence (AI) workloads.
This agreement, valued at $38 billion with planned growth over the next seven years, grants OpenAI immediate access to AWS compute resources, initially comprising hundreds of thousands of state-of-the-art NVIDIA GPUs. This capacity includes the ability to expand to tens of millions of CPUs to rapidly scale agentic workloads.
AWS is providing large-scale AI infrastructure, leveraging its experience in securely and reliably operating clusters of up to 500,000 chips. This collaboration aims to combine AWS’s leadership in cloud infrastructure with OpenAI’s advancements in generative AI, supporting the continued value delivered to millions of users through products like ChatGPT.
Addressing Demand for AI Compute
The rapid evolution of AI technology has created significant demand for powerful computing resources. Frontier model providers, in their pursuit of more advanced models, are increasingly utilizing AWS for its performance, scale, and security.
OpenAI will begin using AWS compute capacity immediately as part of this partnership. The goal is to fully deploy this capacity before the end of 2026, with flexibility for further expansion into 2027 and beyond.
Infrastructure Deployment and Technology
The infrastructure AWS is building for OpenAI features a sophisticated architectural design focused on maximizing AI processing efficiency. This design includes clustering NVIDIA GPUs (both GB200s and GB300s) via Amazon EC2 UltraServers on a unified network. This setup is intended to ensure low-latency performance across interconnected systems, enabling OpenAI to run diverse workloads with optimal efficiency. The clusters are designed to support various operations, from serving inference for ChatGPT to training next-generation models, offering the flexibility required for OpenAI’s evolving needs.
Continued Collaboration
This announcement builds upon existing collaboration between the two companies aimed at delivering cutting-edge AI technology to global organizations. Earlier this year, OpenAI’s open-weight foundation models were made available on Amazon Bedrock, offering additional model options to AWS customers.