OpenAI and NVIDIA announce strategic partnership to deploy 10 gigawatts of NVIDIA systems

Source: OpenAI Date: 2025-09-22 URL: https://openai.com/index/openai-nvidia-systems-partnership

Summary

Summary

OpenAI and NVIDIA announced a strategic partnership in September 2025 to deploy 10 gigawatts of NVIDIA compute systems — a supply commitment at a scale that dwarfs prior enterprise GPU contracts. The announcement came as OpenAI was building out Stargate infrastructure and needed guaranteed hardware access to support its frontier model training and inference roadmap.

Implications

Infrastructure/platform thread. Ten gigawatts is a headline number chosen for its scale signal — it represents both a hardware supply guarantee and a public statement about the pace of AI infrastructure buildout. For NVIDIA, OpenAI as an anchor customer at this scale validates the continued dominance of GPU-based AI compute and keeps NVIDIA’s roadmap tightly coupled to frontier model training requirements. For OpenAI, the partnership is infrastructure diversification and supply chain security during a period when access to frontier compute is a primary competitive constraint. The announcement reinforced that AI infrastructure is now a geopolitical and macroeconomic variable, not just an engineering cost.