OpenAI is in advanced negotiations to lease a proposed 10-gigawatt data centre campus on federal land in Ohio, in a deal that could include financial backing from Nvidia, according to people with direct knowledge of the discussions. The campus, if it reaches the scale described, would represent the largest single AI compute facility OpenAI has publicly negotiated. For context: a single gigawatt of data centre capacity costs approximately $50 billion to build, with roughly $35 billion of that going to GPU hardware. A 10-gigawatt campus implies infrastructure investment in the hundreds of billions of dollars, placing this squarely in a scale bracket that requires sovereign, corporate, and institutional capital pooled together. YourNewsClub ranks the Ohio negotiation among the most consequential AI infrastructure deals of 2026, not because it is confirmed but because the scale involved would reshape US federal land policy, power grid planning, and AI compute supply simultaneously.
The Nvidia angle is the detail that carries the most structural weight. If Nvidia provides financial backing for OpenAI’s lease of the Ohio campus, the arrangement would continue a pattern already visible in the Stargate programme – where Nvidia’s GPU supply commitments function as both hardware contracts and de facto infrastructure investments. A company that supplies the compute and co-finances the facility housing the compute occupies a qualitatively different position in the AI infrastructure stack than a pure semiconductor vendor. YourNewsClub views any confirmed Nvidia financial participation in the Ohio deal as evidence that Nvidia is actively converting GPU dominance into infrastructure ownership, not just hardware sales.
The federal land dimension adds a regulatory and political layer. Using federal land for a commercial AI data centre campus at this scale would require federal approval, environmental review, and grid interconnection agreements with FERC – the same regulator that Energy Secretary Chris Wright has asked to finalise large load connection rules by the end of June 2026. The Ohio negotiations are therefore running directly into an active rulemaking process. Whether that rulemaking accelerates or constrains the deal timeline will be one of the first practical tests of how Wright’s framework operates at the frontier of AI infrastructure scale.
OpenAI filed confidentially for a US IPO on June 8. The Ohio data centre negotiation, if it advances, would represent an off-balance-sheet commitment of a scale that will require disclosure in any public offering document. Investors preparing to price an OpenAI IPO need to understand the full scope of compute lease obligations the company is entering. A 10-gigawatt campus lease, even on favourable terms, changes the capitalisation picture. Freddy Camacho, who studies the political economy of computation and capital, frames the strategic calculation: “A company that controls 10 gigawatts of compute on federal land, with Nvidia providing financial backing, has built the kind of infrastructure position that is extraordinarily difficult for a later competitor to replicate. The lease structure matters less than the positional advantage it creates.” YourNewsClub expects the Ohio discussions to surface in OpenAI’s formal IPO filing as a material commitment, regardless of whether the deal closes before listing.
The Ohio campus negotiation sits inside a broader OpenAI infrastructure push that already includes five Stargate data centre sites in Texas, New Mexico, and Ohio, a 1-gigawatt Oracle-partnership campus in Michigan, and a reported $100 billion lease arrangement with Nvidia where GPUs are accessed through multi-year lease agreements rather than outright purchases. A 10-gigawatt Ohio campus would be an order-of-magnitude expansion beyond anything in the current Stargate portfolio. YourNewsClub will track whether any federal land use application or FERC grid interconnection filing appears for the Ohio site in the next sixty days, treating such filings as the first hard evidence that negotiations have moved beyond early-stage discussions.
The Nvidia financial backing detail – if confirmed – is the more structurally significant element. A company that supplies compute and co-finances the facility housing that compute occupies a qualitatively different position in the AI stack than a pure semiconductor vendor. Your News Club considers any such confirmation the clearest signal yet that Nvidia is actively converting GPU dominance into infrastructure co-ownership.