Nvidia May Guarantee $250 Billion So OpenAI Can Build a Data Center the Size of a Small Country's Power Grid
The proposed backstop would let OpenAI borrow against Nvidia's credit to fund a 10-gigawatt Ohio campus that could ultimately cost more than $500 billion.

OpenAI and Nvidia are discussing a financial arrangement that sounds almost abstract until you look at the number attached to it: up to $250 billion. That's the size of the backstop the two companies are reportedly negotiating, according to CNBC, which would let OpenAI borrow heavily to build one of the largest data center projects ever conceived.
The mechanics matter here. This isn't Nvidia writing OpenAI a check. It's Nvidia lending its credit strength so OpenAI can go raise debt on better terms than it could on its own. A source familiar with the discussions told CNBC the guarantee would apply specifically to the lease and construction costs of the project, not to the Nvidia chips that will eventually fill it. The chip financing is reportedly being worked out as a separate matter entirely.
The project in question is a 10-gigawatt data center campus planned for Pike County, Ohio, on a site that once housed a uranium-enrichment plant. To put 10 gigawatts in perspective, CNBC's analysis of Energy Information Administration data puts that roughly on par with the annual electricity consumption of 8 million American households. The source told CNBC the full campus could ultimately run north of $500 billion once everything is built out.
Why OpenAI needs a guarantor
OpenAI has been on an infrastructure spending spree that outpaces almost anything in the company's history, chasing the compute it says it needs to keep its models ahead of a fast-closing field. Anthropic, Google, Amazon and Meta are all pouring huge sums into their own AI infrastructure, and OpenAI is trying to make sure it doesn't fall behind on the one resource that actually constrains how good these models can get: raw computing power.
The problem is that data centers on this scale require financing on a scale most companies, even ones as flush with investor cash as OpenAI, can't access on their own signature. That's where a backstop from Nvidia comes in. Nvidia's balance sheet and market position give it a level of creditworthiness that can make lenders far more comfortable extending massive sums, even when the borrower is a company that isn't yet profitable in any traditional sense.
This wouldn't be the first time Nvidia has put real money behind OpenAI's ambitions. In September, Nvidia said it planned to invest up to $100 billion in OpenAI as part of a partnership under which OpenAI would deploy at least 10 gigawatts of Nvidia's systems. That particular investment never actually happened. Nvidia did, however, put $30 billion into the record-setting funding round OpenAI closed in March.
Nvidia CEO Jensen Huang has suggested the company's direct investing in OpenAI may be winding down as OpenAI approaches the public markets, saying it "might be the last time" Nvidia invests in the company before an IPO. OpenAI confidentially filed for an initial public offering with the SEC in June, though it hasn't said when it actually plans to go public.
The stakes for OpenAI's valuation
Private investors currently value OpenAI at close to $1 trillion, a figure that assumes the company can hold its lead in AI and eventually land on a business model that actually works at scale. That bet is getting harder to make with a straight face. A wave of open-weight AI models, many coming out of China, are improving fast and threaten to squeeze the pricing power OpenAI has relied on to justify its spending.
That tension is part of what makes a deal like this notable. OpenAI is trying to lock in enormous amounts of physical infrastructure years before it's clear whether the revenue to support it will materialize at the pace investors are pricing in. A quarter-trillion-dollar backstop from Nvidia would be a strong vote of confidence, and also a sign of how tightly the fortunes of the two companies have become intertwined. Nvidia sells the chips that make OpenAI's models run; OpenAI's continued growth, in turn, is one of the biggest demand drivers for Nvidia's chips. Propping up OpenAI's ability to build more data centers is, in a roundabout way, propping up Nvidia's own sales pipeline.
The Ohio campus itself isn't an OpenAI-only production. SoftBank and its energy arm, SB Energy, are developing the site in partnership with the U.S. Department of Energy. SoftBank is both a major OpenAI investor and, along with OpenAI, announced plans earlier this year to put $1 billion into SB Energy. Nvidia declined to comment on the talks when asked by CNBC, and the discussions remain fluid. A second source told CNBC that details of the site and its financing are still in progress and subject to change. The Wall Street Journal first reported on the negotiations.
Nothing here is finalized, and the structure could still shift substantially before any papers get signed. But the scale of the number under discussion says something on its own: the AI industry's infrastructure buildout has reached a point where even the companies leading it need backstops measured in the hundreds of billions just to keep pace with their own ambitions.


