Nvidia may be preparing to support one of the most ambitious artificial intelligence infrastructure projects ever proposed. The chipmaker is reportedly discussing a financial guarantee worth roughly $250 billion to help OpenAI lease a sprawling data centre campus planned for southern Ohio. The entire development could eventually cost more than $500 billion.
That number is difficult to absorb. It is larger than the annual economic output of many countries, all tied to a single infrastructure project built around artificial intelligence. Nothing has been finalised yet. Nvidia, OpenAI and the companies connected to the development have not publicly confirmed that a binding agreement has been signed.
Nvidia May Provide a $250 Billion Financing Backstop
The proposed arrangement would not necessarily involve Nvidia handing OpenAI $250 billion in cash. Instead, Nvidia could provide what is often described as a financial backstop or credit guarantee. That support would make it easier for OpenAI to secure long-term financing and commit to leasing capacity from the planned data centre.
OpenAI remains a privately held company without an investment-grade credit rating. Financing a project of this size would therefore be difficult without backing from a company with Nvidia’s financial strength and market position.
The guarantee could reassure banks and other lenders that the lease obligations would still be covered if OpenAI ran into financial trouble.
For Nvidia, the logic is fairly direct. More OpenAI infrastructure means more demand for Nvidia GPUs, networking equipment and data centre systems. Still, the arrangement creates an unusual relationship. Nvidia would potentially help finance a customer that could then spend a significant portion of that money buying Nvidia technology.
The Ohio Campus Could Require 10 Gigawatts of Power
The proposed site is expected to be developed in southern Ohio by SB Energy, a renewable energy company owned by SoftBank Group. Plans reportedly call for a campus capable of consuming as much as 10 gigawatts of electricity when fully developed. That is a staggering amount of power for one location, even by hyperscale data centre standards.
Construction would take years. The first phase could deliver around 800 megawatts of capacity by 2028, while later phases would gradually expand the project toward its full target.
OpenAI would reportedly serve as the principal tenant, using the site to train and operate future generations of artificial intelligence models. The campus could also become part of a wider network of computing facilities designed to reduce OpenAI’s dependence on traditional cloud providers.
OpenAI Wants More Control Over Its Computing Infrastructure
OpenAI has historically relied heavily on Microsoft Azure to provide the computing power behind ChatGPT and its other AI services. That model is changing.
As AI workloads grow, OpenAI appears increasingly interested in controlling more of the infrastructure itself. The company has signed major computing agreements with Microsoft, Oracle, Amazon Web Services and other providers while becoming more directly involved in data centre development. Its planned cloud and infrastructure spending could reach approximately $750 billion through 2030, according to recent reporting.
OpenAI is also pursuing a separate 3.2-gigawatt data centre project in Georgia. That facility, known as Project Camellia, is expected to receive power in phases between 2028 and 2032. The Ohio proposal would go much further. A fully developed 10-gigawatt campus would give OpenAI access to computing capacity on a scale rarely attempted by a single AI company.
Nvidia Could Also Finance OpenAI’s Chip Purchases
The discussions may extend beyond the data centre lease. Nvidia is reportedly considering financing as much as $350 billion in chip purchases for OpenAI. Those chips would likely form the computing foundation of the Ohio campus and other OpenAI infrastructure projects. This would deepen an already close relationship between the two companies.
In September 2025, Nvidia and OpenAI announced plans to deploy at least 10 gigawatts of Nvidia systems across OpenAI’s next-generation AI infrastructure. Nvidia said it could invest up to $100 billion as each gigawatt of capacity entered service.
The first phase was scheduled to use Nvidia’s Vera Rubin platform and begin coming online during the second half of 2026. The newly reported Ohio discussions appear even more aggressive. Nvidia would not simply supply the hardware. It could help make the financing behind the entire facility possible.
The Deal Raises Questions About Circular AI Financing
The structure will attract scrutiny. Nvidia sells the processors needed to build large AI systems. OpenAI needs those processors but also needs enormous amounts of money to purchase them. Nvidia could help OpenAI secure that money, allowing OpenAI to spend it on infrastructure filled with Nvidia hardware. It is a powerful growth loop. It can also look circular.
Similar concerns have emerged around other AI infrastructure agreements involving chipmakers, cloud providers, model developers and data centre operators. Each party benefits from the expansion, but the economics depend on AI revenue eventually catching up with the spending.
OpenAI continues to grow quickly, although operating frontier models remains expensive. The company would need strong and sustained demand from consumers, businesses and governments to justify hundreds of billions of dollars in long-term infrastructure commitments. A slowdown in AI adoption would make those obligations much harder to carry.
Energy Could Become the Bigger Problem
Money is only one part of the equation. A 10-gigawatt data centre campus would require new power generation, transmission lines, substations, cooling systems and water infrastructure. Local communities would also want clear answers about electricity prices, environmental impact, land use and long-term employment.
Data centre resistance is already increasing in several parts of the United States. Residents and lawmakers have questioned whether enormous AI facilities could raise household electricity costs or place too much pressure on regional power grids.
OpenAI has tried to address similar concerns around its Georgia project. The company said residential customers would not carry the project’s electricity costs and that its facilities could reduce power use during periods of peak demand. A project five times larger would face even greater scrutiny.
Nvidia and OpenAI Are Building Beyond the Cloud
The reported talks show how quickly AI competition has moved beyond software. Model quality still matters. So do coding ability, reasoning performance and inference speed. None of those improvements happen without physical infrastructure.
AI companies now compete for land, electricity, chips, cooling equipment, construction workers and access to financing. The winning model may not simply be the one with the smartest architecture. It may be the one backed by the largest and most reliable computing network.
Nvidia understands that shift better than most companies. Supporting OpenAI’s Ohio data centre would create another enormous customer for its hardware while strengthening its position at the centre of the AI infrastructure economy.
The deal remains under discussion, and its final structure could change significantly. Even so, the proposed numbers reveal where the industry believes the next battle will take place. Not inside a chatbot interface. Inside data centres measured in gigawatts.
