Pranav Hotkar.jpg)
San Francisco, Santa Clara, September 23, 2025- NVIDIA and OpenAI have unveiled a landmark partnership to deploy up to 10 gigawatts of AI data center capacity, in what the companies describe as the largest expansion of its kind. The deal, yesterday, involves NVIDIA progressively investing as much as USD 100 billion into OpenAI to support the build-out, which will run on NVIDIA’s forthcoming Vera Rubin platform.
The infrastructure will roll out in phases beginning in the second half of 2026, with the first GW equating to millions of GPUs. Under the structure, OpenAI will purchase NVIDIA hardware with cash, while NVIDIA will receive non-controlling equity in the AI firm, most recently valued at USD 500 billion.
NVIDIA founder and CEO Jensen Huang said, “The scale of the agreement marks the next leap forward deploying 10 gigawatts to power the next era of intelligence.”
OpenAI CEO Sam Altman emphasized the centrality of compute, noting, “Everything starts with compute infrastructure … we will utilize what we’re building with NVIDIA to both create new AI breakthroughs and empower people and businesses with them at scale.”
Analysts say the partnership reduces OpenAI’s reliance on Microsoft while easing ongoing compute shortages, and it positions NVIDIA to secure one of the industry’s most important customers. Market reaction was immediate, with NVIDIA shares rising about 4% following the news.
The collaboration comes against the backdrop of escalating global AI infrastructure races, including OpenAI’s own USD 500 billion “Stargate” initiative. With one GW of capacity estimated to cost USD 50-60 billion, the 10 GW program underscores both the scale of AI investment and the growing scrutiny over NVIDIA’s dominance in advanced chips.
.jpg)
