Nvidia to Provide Up to $105 Billion in Financing for New OpenAI Data Center in Ohio
Key Takeaways
- •Nvidia disclosed up to $105 billion in financing for the OpenAI-linked Ohio data center project in an August 17 SEC filing.
- •The company is also investing $1.5 billion in SB Energy, which is helping develop the site with OpenAI.
- •The PORTS-Pike Technology Campus in Pike County is planned to provide 8 gigawatts of compute capacity, with initial service expected in 2028.
- •OpenAI said the project could generate 35,000 construction jobs and 2,500 long-term operating jobs, alongside $4.2 billion in regional grid investment.
- •The project has drawn scrutiny because critics cite land use, turbine pollution, and concerns about circular financing.

Nvidia plans to provide up to $105 billion in financing for a new OpenAI data center in Ohio, a commitment disclosed in an August 17 filing with the U.S. Securities and Exchange Commission (SEC). The financing and infrastructure role places the AI chip vendor at the center of yet another multibillion-dollar data center project.
Nvidia also disclosed that it is investing $1.5 billion in SB Energy, a U.S. data center company owned by Japan's SoftBank Group that is heavily involved in the Ohio project. The equity stake deepens Nvidia's financial ties to the developer behind the site, extending its role beyond supplying and financing the compute itself.
The facility will be located at the PORTS-Pike Technology Campus in Pike County, Ohio. Nvidia will be the exclusive provider of the compute, while SB Energy will build, own and operate the facility under a 20-year lease to OpenAI. The campus name is a nod to the site's history: it occupies land tied to the former Portsmouth Gaseous Diffusion Plant (PORTS), a Cold War-era uranium enrichment complex near Piketon that the U.S. Department of Energy has been decommissioning for decades.
Ultimately, the campus is slated to deliver 8 gigawatts of compute capacity to OpenAI. For scale, one gigawatt is roughly the output of a single large nuclear reactor, so a fully built 8-gigawatt campus implies electricity demand comparable to a fleet of major power plants. Nvidia's initial credit will support 4.25 gigawatts of that capacity, with the company holding an option to take on the remaining 3.75 gigawatts. Whether that option is exercised is among the project's biggest open variables. The planned compute is expected to start coming online in 2028.
According to Nvidia, the facility will use its infrastructure, including GPUs, CPUs and networking. In a statement, the company said the setup will deliver "resiliency across the full stack -- facilities, hardware, and software together -- reducing infrastructure overhead and accelerating time to tokens."
The scale of the Ohio project is vast. OpenAI says it will create 35,000 construction jobs during the six-year buildout, plus another 2,500 long-term operating positions. In tandem with SB Energy, OpenAI has also pledged to build power-generating facilities locally capable of delivering 10 gigawatts of energy and to invest $4.2 billion in regional grid infrastructure. Pairing campuses with dedicated generation has become common practice for hyperscale AI projects as developers confront limits on available grid power nationwide.
The project has local support as an economic revitalization initiative at an old uranium enrichment site, but critics warn of over-intensive land use, pollution from the gas turbines that will power the data center, and a volatile funding structure.
Nvidia continues to take a more prominent role in data center financing. Earlier this month, it formed partnerships with Wall Street firms Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to establish compute financing platforms that could generate more than $500 billion to build out AI infrastructure.
That ever-increasing involvement has led to suggestions that such financial arrangements, including the Ohio project, amount to circular financing. The label describes a vendor-financing loop, in which a supplier extends credit that customers use to purchase the supplier's own products -- a structure with precedents in the late-1990s telecom boom, when equipment vendors financed carrier network purchases. Nvidia CEO Jensen Huang countered this in a blog post. Huang acknowledged the site will be lucrative for Nvidia, noting that each generation of AI factory system deployed in Ohio represents about 1.5 million Nvidia GPUs, worth up to $200 billion.