Overview
- OpenAI will begin using AWS capacity immediately, with full rollout targeted by the end of 2026 and potential extensions into 2027.
- The agreement gives OpenAI access to large-scale Nvidia GPU infrastructure on AWS to train advanced models and operate ChatGPT.
- The move follows disclosures of about $1.4 trillion in prior compute and chip commitments that analysts say could require roughly 30 gigawatts of electricity.
- Amazon shares rose roughly 4–5% to a record and Nvidia also gained after the announcement.
- Microsoft remains a major OpenAI shareholder at around 27% but has reported limited available cloud capacity, and it is leasing additional chips and servers to Iren in a $9.7 billion deal.