In a strategic alliance of unprecedented scale, Nvidia has pledged up to $100 billion to OpenAI, a move designed to construct the world’s most powerful AI infrastructure. This partnership effectively merges the industry’s top hardware and software players, creating a unified front to accelerate the journey toward super-intelligence and solidify their dominance in the rapidly evolving tech landscape.
This isn’t just a financial transaction; it’s a deep integration of their roadmaps. In exchange for its massive investment, Nvidia will acquire a significant equity position in OpenAI. The deal is structured in stages, beginning with a $10 billion payment upon the activation of the first gigawatt of computing capacity. This milestone-driven approach underscores the practical, results-oriented nature of the collaboration.
The long-standing relationship between the two companies provides a strong foundation for this venture. Nvidia’s CEO, Jensen Huang, reflected on their shared history of innovation, stating, “Nvidia and OpenAI have pushed each other for a decade.” He framed the 10-gigawatt infrastructure project as the next chapter in their joint mission to pioneer the future of intelligence.
OpenAI’s Sam Altman highlighted the foundational importance of this project, stating that “Everything starts with compute.” He envisions the new infrastructure as the bedrock of a future economy driven by AI. This collaboration will empower OpenAI to overcome previous growth limitations and harness unprecedented computational power to unlock new frontiers in artificial intelligence research and application.
The project complements existing partnerships with other tech giants like Microsoft and Oracle, placing this new Nvidia-OpenAI “AI factory” at the heart of a growing ecosystem. Set to launch its first phase in the latter half of 2026 with Nvidia’s Vera Rubin platform, this initiative is poised to meet the immense computational demands of OpenAI’s 700 million weekly users and fuel the creation of its next generation of world-changing models.
