OpenAI and Amazon Web Services (AWS) have entered into a strategic partnership, effective immediately, to leverage AWS infrastructure for OpenAI’s advanced artificial intelligence workloads.
This collaboration will see OpenAI utilizing Amazon EC2 UltraServers, a platform featuring substantial computational power. These UltraServers are equipped with hundreds of thousands of NVIDIA GPUs and the capability to scale to tens of millions of CPUs, specifically designed to handle the intensive demands of advanced generative AI applications.
The seven-year agreement entails a financial commitment of $38 billion. This partnership is designed to facilitate the rapid expansion of OpenAI’s computational capacity while benefiting from AWS’s established infrastructure, including its pricing model, performance capabilities, scalability, and robust security features. An official press release highlighted AWS’s extensive experience in managing large-scale AI infrastructure with a strong emphasis on security and reliability, supported by clusters of over 500,000 chips. The collaboration aims to provide continued value to millions of users of ChatGPT.
The AWS capacity provided under this agreement is scheduled for full deployment by the end of 2026, with provisions for further expansion beginning in 2027. The architectural design of this deployment involves clustering NVIDIA GPUs, including both GB200 and GB300 models, on a unified network to ensure low-latency performance across interconnected systems. This configuration is optimized to allow OpenAI to run its AI workloads with peak efficiency.
