NVIDIA and OpenAI Announce $105 Billion AI Supercomputer Campus in Ohio at Former Uranium Site

NVIDIA and OpenAI have finalized an agreement with SoftBank's SB Energy to build what is projected to be the world's largest artificial intelligence data center in Pike County, Ohio. Located on the grounds of the decommissioned Portsmouth Gaseous Diffusion Plant—a Cold War-era facility that enriched uranium for the U.S. government for nearly half a century—the new PORTS-Pike Technology Campus will exclusively host NVIDIA computing hardware. OpenAI has signed on as the anchor tenant under a 20-year lease, securing a site that is designed to eventually support up to 10 gigawatts of power capacity.[1][3][4][5]

The massive undertaking marks a turning point in the technology sector, signaling that the primary bottleneck for artificial intelligence development has shifted from the availability of advanced microchips to the physical constraints of land and electricity. As frontier models require exponentially more compute to train and operate, tech giants are realizing that traditional data centers are no longer sufficient. Securing long-lived land, power, and shell capacity allows companies like OpenAI to repeatedly upgrade their physical sites as new generations of computing technology become available.[1][6]

The financial structure of the deal places NVIDIA in an unprecedented role as a guarantor for its largest customer. To secure the massive footprint required for OpenAI's future models, NVIDIA has agreed to provide a financial backstop of up to $105 billion, which covers OpenAI's lease and power payment obligations. By using its own balance sheet to underwrite the physical infrastructure, NVIDIA is ensuring that its exclusive computing platform will be deployed at a site capable of supporting the next wave of artificial intelligence development.[1][2][5][6]

In addition to the credit guarantee, the semiconductor giant is investing $1.5 billion directly into the project's developer, SB Energy, cementing its position at the center of the physical AI buildout. The initial deployment is designed to support 4.25 gigawatts of compute capacity, with an option to expand to 8 gigawatts. The total cost of the campus buildout, factoring in chips, labor, and power generation, is expected to approach $500 billion by the time it is fully completed in 2032.[2][4][6]

Powering a facility of this magnitude requires infrastructure upgrades that rival the output of multiple nuclear reactors. To meet the campus's immense energy demands without destabilizing the local grid, SB Energy and SoftBank have committed to building at least 10 gigawatts of new energy generation. The developers are also investing $4.2 billion into regional grid infrastructure, including high-voltage transmission lines, structured specifically to ensure that Ohio ratepayers do not absorb the costs of the expansion.[1][3]

Beyond electricity, the developers are implementing advanced cooling techniques to mitigate the environmental footprint of the gigawatt-scale campus. The facility will utilize closed-loop, air-cooled systems to recirculate water, avoiding the continuous water consumption typical of traditional cooling towers. By building on remediated Department of Energy land, the project is also able to leverage the site's existing heavy-industrial water infrastructure, repurposing a federal nuclear site for the digital age.[3][5]

For rural southern Ohio, the project represents a massive economic revitalization effort. The construction phase is expected to generate 35,000 jobs over the next six years, followed by 2,500 permanent operating positions once the facility is fully online. To secure local support, OpenAI is adding $40 million to an existing $40 million community benefits fund established by SB Energy, creating an $80 million pool to support local priorities such as workforce development, affordable energy, and public safety.[1][3]

Additionally, OpenAI is offering $84 million in Codex credits to over 800,000 Ohio college and technical school students during the 2026-2027 academic year, aiming to build a local pipeline of AI talent. The first 800 megawatts of computing capacity are scheduled to come online in 2028, utilizing existing infrastructure from AEP Ohio while the broader power generation projects are completed. As tech giants race to build out the physical foundation for next-generation AI, the Ohio campus serves as a blueprint for how the industry plans to deploy gigawatt-scale AI factories.[3][4][5][6]