Overview
- OpenAI will begin running workloads on AWS immediately under a $38 billion, seven‑year agreement, with full capacity targeted before the end of 2026 and options to expand beyond 2027.
- AWS will provision EC2 UltraServers that cluster hundreds of thousands of Nvidia GB200 and GB300 GPUs on a single network with the ability to scale to tens of millions of CPUs for low‑latency performance.
- The arrangement follows a corporate restructuring and ends Microsoft’s prior exclusive status as OpenAI’s cloud provider.
- Amazon shares rose about 4% to record highs after the announcement, adding roughly $9.8–$10 billion to Jeff Bezos’s net worth that day.
- Reporting indicates the deployment uses Nvidia chips rather than AWS Trainium and fits into OpenAI’s wider multi‑supplier infrastructure buildout.