Nvidia’s Direct Partnership With OpenAI Marks a New Era for AI

​A First-of-Its-Kind Alliance
In an exclusive interview, Nvidia CEO Jensen Huang revealed that OpenAI is entering its first direct purchasing partnership with the chipmaker — a major shift in how AI companies access computing power.
“Usually, cloud providers like Microsoft or Amazon buy our chips and rent them out,” Huang said. “This time, OpenAI is buying directly from us.”

The deal cements a closer relationship between two of the most powerful players in artificial intelligence — and underscores how the infrastructure behind AI models is evolving toward massive, independent systems.

Building the Next Generation of AI
Under the partnership, Nvidia plans to invest up to $100 billion in OpenAI to help it expand its data center capacity. The new systems will demand around 10 gigawatts of power, equivalent to roughly 4–5 million GPUs, according to Huang.

The goal: create the kind of computing muscle capable of running and training the next generation of AI models. “We’re helping them build infrastructure they’ll ultimately operate themselves,” Huang explained. “We’re setting them up for what comes five years down the road — when OpenAI will effectively run its own global cloud.”

The AI Boom’s Power Couple
The partnership brings full circle the relationship that helped spark the AI explosion. Nvidia’s chips became the engine of the generative AI boom after OpenAI’s ChatGPT launch three years ago. That success sent Nvidia’s market value soaring past $4 trillion, making it the most valuable company in the world.

Together, the two companies are now positioned at the heart of a global race to dominate AI infrastructure — from model training to hardware design.

Expanding the Ecosystem
Huang emphasized that this collaboration doesn’t replace Nvidia’s existing partnerships with Oracle, CoreWeave, AMD, and Intel. Instead, it adds another layer. “This is incremental,” he said. “Each relationship helps push AI computing further.”

He also described how Nvidia’s hardware and software ecosystem is becoming “one big computer,” spanning entire data centers built specifically for AI workloads.

The Bigger Picture
For Nvidia, this direct deal with OpenAI marks a strategic pivot — from selling chips through cloud intermediaries to forging end-to-end partnerships with the world’s top AI developers. For OpenAI, it’s a chance to control its own infrastructure and scale independently.

​In the words of Huang: “AI now needs entire data centers acting as one unified computer. This partnership is how we make that happen.”