Nvidia Backs OpenAI’s Ohio AI Campus With Up to $105 Billion

Key Takeaways

• Nvidia plans to provide financing and supply the chips for OpenAI’s planned Pike County campus.
• The facility is projected to support 8 gigawatts of computing capacity and require 10 gigawatts when supporting systems are included.
• Grid costs, water consumption and the durability of promised construction jobs are becoming central issues in Ohio’s 2026 elections.

Nvidia is backing one of the most ambitious infrastructure projects of the artificial intelligence boom: a massive computing campus planned for Piketon in Pike County, Ohio. The chipmaker is expected to provide financing for the $33 billion project and serve as the sole supplier of processors for the facility, which SB Energy will build, own and lease to OpenAI.

The facility, called the PORTS Technology Campus and also referred to as the PORTS-Pike Technology Campus, is scheduled to begin operating in phases. Business leaders expect the first phase to come online in 2028, followed by a full rollout in 2032. The project was initially announced in March by the Department of Energy and business leaders, while reporting on the complete buildout has placed total expenses at $33 billion.

Its defining number is 8 gigawatts. That represents the reported computing capacity, while cooling, power conversion and other supporting systems could add another 2 gigawatts, producing a total site requirement of 10 gigawatts. The Cincinnati Enquirer reported that the 8-gigawatt figure is equivalent to the electricity used by roughly 6.4 million homes, based on U.S. EIA consumption data.

That comparison is illustrative rather than a forecast that the campus will literally divert power from millions of households. Still, it conveys the scale. A congressional definition classifies a hyperscale data center as a facility covering at least 10,000 square feet and consuming more than 100 megawatts. At 8,000 megawatts of compute capacity, the OpenAI campus would sit in a different category from conventional hyperscale developments.

Ohio’s power system is already confronting a queue of data-center requests that dwarfs its existing customer demand. AEP Ohio said contracted data-center projects reached 17,861 megawatts as of Feb. 12, 2026. By comparison, peak demand across all AEP Ohio customers typically ranges from about 8,000 to 10,500 megawatts. How quickly can generation and transmission capacity expand without shifting financial risk to households and smaller businesses?

OpenAI says SB Energy will cover the full cost of grid upgrades and new transmission lines required for the campus. According to OpenAI, those expenses will not be shifted to Ohio ratepayers, and the resulting investment could strengthen regional reliability. The commercial details behind that commitment will matter, including interconnection terms, minimum-payment provisions and how costs are handled if later phases are delayed or canceled.

Efficiency will matter too. Power Usage Effectiveness, or PUE, compares a data center’s total energy consumption with the energy delivered to computing equipment. Even a relatively small efficiency difference becomes significant at this scale. EPRI projects U.S. data-center electricity consumption could rise from 177 to 192 terawatt-hours in 2024 to between 383 and 793 terawatt-hours by 2030. Aggregate peak load could reach 45 to 94 gigawatts by then, up from 21 to 22 gigawatts in 2024.

Water is another pressure point. OpenAI says the Pike County facility will use closed-loop, air-cooled systems that recirculate water instead of relying on cooling towers that consume it continuously. The company expects the campus to use significantly less water than the former Portsmouth gaseous diffusion plant. That said, local stakeholders will likely seek operating data once the first phase starts, not just design projections.

For southern Ohio, the economic case rests on jobs and a broader supplier ecosystem. OpenAI projects a surge in construction jobs over six years and says the campus could encourage business investment while giving younger residents more reasons to remain in the region. Long-term employment levels after construction, however, will be watched closely.

The debate is now political as well as industrial. Republican gubernatorial candidate Vivek Ramaswamy and incumbent Sen. Jon Husted have supported data-center development while calling for safeguards against ratepayer burdens. Democratic gubernatorial candidate Amy Acton and former Sen. Sherrod Brown have been more skeptical, with Acton proposing a conditional moratorium on new projects. For Nvidia, OpenAI, SB Energy and Ohio policymakers, the campus has become a test of whether AI infrastructure can scale without leaving the public to absorb its largest costs.