Nvidia may guarantee $250 billion for OpenAI’s Ohio AI campus

The Wall Street Journal reported on July 26 that Nvidia is discussing a roughly $250 billion financing guarantee for OpenAI’s planned lease of a huge AI data centre in southern Ohio. SoftBank’s energy subsidiary is developing the 10-gigawatt campus; Reuters and Bloomberg also reported the details overnight. Talks remain early, so there is no final agreement and the figure is not cash already committed to OpenAI.

The complete project could cost more than $500 billion including chips. An initial phase of about 800 megawatts is expected in 2028. Nvidia’s possible guarantee would cover OpenAI’s lease obligations and debt financing for the project, which could reassure lenders and reduce borrowing costs. It would not cover the Nvidia hardware itself. Separate discussions could involve financing as much as $350 billion of OpenAI’s future chip purchases. Both amounts are possible ceilings, not signed orders.

The structure matters because it puts Nvidia in several roles at once. It supplies the accelerators, invests in OpenAI and could use its balance sheet to support infrastructure from which it expects future chip sales. OpenAI is growing rapidly but remains a private, loss-making company without an investment-grade credit rating. A guarantee from Nvidia would transfer part of that credit risk to one of the world’s most valuable public companies.

For the AI market, this is another sign that competition is moving beyond models and into industrial finance. Training and serving frontier systems now require power plants, transmission lines, buildings, cooling, networking and enormous quantities of advanced chips. Companies with the strongest balance sheets can influence not only which hardware is used, but also which projects lenders are willing to fund. That concentration may accelerate construction while increasing financial dependence between AI labs, chipmakers and infrastructure owners.

Users and developers will not see an immediate product change. If the campus is built, more computing capacity could eventually support faster models, fewer usage constraints and broader enterprise deployment. The uncertainties are just as important: the parties have not announced a deal, the first phase is years away, and financing, power, permits, construction and demand must all align. For now, the report is best read as a measure of the capital required to keep scaling AI—and of how closely Nvidia’s fortunes are becoming tied to OpenAI’s growth.