Generative AI giant OpenAI announced a new collaboration with cloud services provider AWS to help it run and scale its core AI workloads. The seven-year partnership comes with a $38 billion commitment from OpenAI to expand its compute capacity through AWS’s infrastructure.
AWS will grant OpenAI access to its Amazon EC2 UltraServers, which feature hundreds of thousands of NVIDIA GPUs. Clustering GB200 and GB300 chips on the same network is expected to enable low-latency performance across interconnected systems, helping OpenAI run its workloads with maximum efficiency.
This infrastructure deployment can expand to tens of millions of CPUs, allowing OpenAI to rapidly scale its advanced agentic workflows. The AI company will immediately start utilizing AWS compute under this agreement, aiming to deploy all capacity before the end of 2026, with the ability to scale further into 2027 and beyond.
