Nvidia commits $100B to OpenAI for massive GPU-powered data centers

Nvidia will pour $100 billion into OpenAI to help the artificial intelligence lab build massive data centers centered on Nvidia’s processors.

OpenAI plans to deploy Nvidia systems that together will need 10 gigawatts of power, the companies said. The gigawatt measure, used more often now to better describe large AI chip clusters, sets the scale of the project.

Nvidia CEO Jensen Huang told CNBC’s Jon Fortt in San Jose, California, that 10 gigawatts equates to around 4 – 5 million GPUs. That matches what Nvidia expects to ship in total this year and is “twice as much as last year.” “This is a giant project,” Huang said in the interview, seated with OpenAI CEO Sam Altman and the company’s president, Greg Brockman.

Nvidia shares surged to a new record high of $184.55 after the announcement, adding $200 billion in its market cap in a single hour.

Huang called the partnership “monumental in size,” underscoring how closely OpenAI and Nvidia remain linked after kicking off the recent AI boom.

“Nvidia invests $100 billion in OpenAI, which then OpenAI turns back and gives it back to Nvidia,” Bryn Talkington, managing partner at Requisite Capital Management, told CNBC after the announcement. “I feel like this is going to be very virtuous for Jensen.”

The commitment signals how much Nvidia technology OpenAI expects to require for the next wave of AI that goes beyond its current systems. OpenAI’s chip needs were already rising to serve users at scale. The company said it has 700 million active weekly users.

“You should expect a lot from us in the coming months,” Altman said in the interview. “There are three things that OpenAI has to do well: we have to do great AI research, we have to make these products people want to use, and we have to figure out how to do this unprecedented infrastructure challenge.”

The funding will roll out progressively

The companies said the funding will roll out “progressively” as new capacity is built, and that Nvidia will be a “preferred” supplier of chips and networking gear. Nvidia leads the market for AI accelerators, but it faces rising competition from Advanced Micro Devices and from cloud providers that are building their own chips and tightly integrated systems.

Huang told investors in August that it takes between $50 – $60 billion to construct a data center with one gigawatt of capacity, and almost $35 billion is linked to Nvidia chips and systems.

The first phase of the new buildout is scheduled to come online in Q2 2026, using Nvidia’s next-generation Vera Rubin systems.

Nvidia’s move follows a recent secondary sale that valued OpenAI at $500 billion, backed by a roster of investors. Microsoft was one of OpenAI’s early backers and has a strategic pact to bring OpenAI models to its Azure cloud and Microsoft Office. Thrive Capital and Softbank are also major investors in OpenAI.

The Nvidia-OpenAI effort will complement infrastructure work underway with Oracle, Microsoft, the Stargate project, and SoftBank. In the CNBC interview, Altman described Microsoft and Nvidia as “passive” investors and as two of OpenAI’s “most critical partners.”

Huang said Nvidia’s investment is “additive to everything that’s been announced and contracted.” He told CNBC that the plan sits on top of what Nvidia has already told Wall Street to expect.

While this commitment far exceeds Nvidia’s earlier outlays, the company has been spending widely across the AI landscape. According to an earlier Cryptopolitan report, Nvidia’s AI spending is expected to reach $1.5 trillion in 2025.

Last week, Nvidia announced that it took a $5 billion Intel stake and that they would collaborate on AI chips. Nvidia has also invested $700 million in Nscale, a UK-based data center startup.

According to a CNBC report from Thursday, Nvidia spent more than $900 million to bring the CEO of Enfabrica, Rochan Sankar, on board, along with other employees from the firm, and license Enfabrica’s technology.

If you're reading this, you’re already ahead. Stay there with our newsletter.