NVIDIA Is Investing $100 Billion In OpenAI - Here's Why

As demand for frontier-model compute skyrockets, NVIDIA is scaling up its role in the AI chip market with plans for a $100 billion investment in OpenAI. The deal is tied to a massive infrastructure project that will scale OpenAI's compute through 10 gigawatts of data center capacity. These systems will use NVIDIA's next-generation Vera Rubin GPUs, with the initial one gigawatt installation scheduled to come online in late 2026.

The partnership builds on a relationship that began in 2016, when NVIDIA delivered its first DGX system to OpenAI. Now, the two titans of the AI industry are scaling that collaboration into what NVIDIA CEO Jensen Huang calls "the biggest AI infrastructure project in history." The project is designed to power OpenAI's next wave of models — those built for reasoning, autonomy, and longer context — across both training and real-time inference.

Instead of a standard funding round, the deal ties NVIDIA's investment directly to infrastructure. OpenAI will use the capital to buy NVIDIA's GPUs, networking hardware, and systems software, while NVIDIA takes a non-voting stake in return. The two companies also plan to co-develop future platforms, with NVIDIA now acting as OpenAI's preferred compute partner. Once fully deployed, the 10-gigawatt AI data center network is expected to involve millions of NVIDIA graphics cards and use as much electricity as several million U.S. homes.

OpenAI's user base has grown to over 700 million weekly active users since ChatGPT's launch, with newer workloads focused on advanced reasoning, deep research, agent-like tools, and real-time responses. Model complexity is growing, demand is exploding, and capacity hasn't caught up. The NVIDIA partnership gives OpenAI a guaranteed path to that capacity while providing NVIDIA with a long-term pipeline of hardware demand anchored to a fixed roadmap. The $100 billion will roll out in stages, with each round tied to OpenAI bringing a new gigawatt of capacity online.

For NVIDIA, it locks in a steady revenue stream for its most advanced GPUs and data center platforms. For OpenAI, it secures access to power capacity at a time when both access to electricity supply and AI accelerators have become a limiting factor for model development. This investment also folds into OpenAI's broader "Stargate" program, a $500 billion push to expand global AI infrastructure. 

Other suppliers are part of that push. On September 25, 2025, GPU cloud provider CoreWeave announced a $6.5 billion contract with OpenAI, bringing its 2025 total to $22.4 billion across three separate deals this year. Oracle is also delivering 4.5 gigawatts under an annual leasing agreement, while SoftBank is developing two additional U.S. sites under the same Stargate program. What sets NVIDIA's deal apart is the sheer amount of capital involved and the fact that it mirrors NVIDIA's own expansion strategy, which includes plans to produce $500 billion worth of AI servers over the next four years — matching OpenAI's infrastructure needs.