In a bold move that signals the intensifying race in artificial intelligence infrastructure, Nvidia has announced a massive $100 billion investment in OpenAI to build a next-generation AI compute network. The agreement, revealed on September 22, 2025, outlines a vision to deploy at least 10 gigawatts (GW) of Nvidia-powered infrastructure that will support OpenAI’s growing demand for AI training and deployment at scale.
Nvidia–OpenAI Deal: What’s in It?
- The investment begins with a letter of intent signed between Nvidia and OpenAI, establishing a strategic partnership to create an AI factory of the future. At its core is the commitment to deploy 10 GW of Nvidia GPUs and networking technologies to build dedicated AI infrastructure for OpenAI. This will enable OpenAI to scale its large language models, like GPT and Codex, far beyond their current capacity.
- The deal also involves co-optimisation of software and hardware, meaning Nvidia and OpenAI will jointly plan their development paths—improving efficiency and performance for future AI models.
Strategic Importance of the Deal
For Nvidia
- This agreement ensures Nvidia continues to dominate the AI chip market. As the preferred compute and networking partner for OpenAI, Nvidia not only sells more of its high-end chips but also gains deeper integration into OpenAI’s long-term roadmap. This move also comes just days after Nvidia pledged $5 billion to support Intel, reinforcing its expansive reach across the semiconductor ecosystem.
For OpenAI
- The partnership provides OpenAI with long-term, secure access to the compute power it needs to train next-generation models. By investing in its own infrastructure (instead of relying entirely on cloud services like Microsoft Azure), OpenAI strengthens its autonomy and ability to scale faster.
For the AI Ecosystem
- A 10 GW infrastructure implies millions of GPUs, massive energy use, and the emergence of AI-specific data centres as national-level infrastructure. This level of investment will push other players—like Google, Meta, Amazon, and startups—to rethink their own AI infrastructure strategies.
Static Facts
- Announcement Date: September 22, 2025
- Investment Amount: Up to $100 billion
- Target: Deployment of 10 GW Nvidia chips for AI compute
- First Deployment: Expected in H2 2026
- Partnership Model: Strategic, with hardware-software co-optimization
- Nvidia Role: Preferred compute & networking partner