The Information : OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Bac

OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Backing From Nvidia

The Takeaway
  • OpenAI is in advanced talks to become the tenant of a planned 10 gigawatt AI campus in Ohio.
  • OpenAI is negotiating a 20-year lease that would represent its largest infrastructure commitment to date.
  • Nvidia has discussed providing credit support for the project.

OpenAI is in advanced negotiations to lease a proposed 10 gigawatt data center campus on federal land in Ohio as part of a deal that could include financial backing from Nvidia, according to two people with direct knowledge of the discussions.

The campus under discussion would be among the biggest of its kind, with a total cost of at least $500 billion if fully built out based on today’s prices for chips, labor, power and other materials, the people said. OpenAI would control the equipment in the facility under a long-term lease and would be on the hook for payments once the project starts operations, with the first phase expected in 2028.

The discussions haven’t been finalized and plans could change. If completed, the Ohio project also would represent OpenAI’s largest infrastructure commitment to date—and would show its deep interest in controlling its own data centers despite having committed to large deals to rent servers from cloud providers such as Microsoft, Amazon and Oracle over the next decade. It is reminiscent of OpenAI’s Stargate plan, announced at the White House in January 2025, to build $500 billion dollars worth of data center capacity through a joint venture with Oracle and SoftBank—a plan that ultimately fizzled out.

The new project involves a campus on land in southern Ohio owned by the Department of Energy that is already under development by SB Energy, which is majority-owned by OpenAI shareholder SoftBank Group. Under the plan being discussed, OpenAI would use Nvidia hardware in the facility, and Nvidia would provide a financial guarantee for OpenAI’s lease and SB Energy’s future project financing, using its enormous balance sheet for what is known as a backstop.

Such a deal would also be new terrain for Nvidia, which has signaled its interest in helping finance its customers’ data center projects but has yet to commit to a project at this scale.

Nvidia chip rival Google already has agreed to backstop debt and data center lease payments associated with data centers where Anthropic plans to use Google’s tensor processing units.

Breaking Ground

A groundbreaking ceremony for SB Energy’s work on the project was held in March, with company and government officials including SoftBank founder Masayoshi Son and U.S. Energy Secretary Chris Wright. But the possible roles of OpenAI and Nvidia haven’t been reported.

Numerous gigawatt-scale data centers are being developed across the U.S., including one in Abilene, Texas, run by Oracle Cloud for OpenAI. But 10 GW, which is 4.5 times the generating capacity of the Hoover Dam, would dwarf even those sites. It’s not clear when the entire Ohio facility would be completed, but the first 800 megawatts is expected to be ready in 2028.

Not all of OpenAI’s massive infrastructure bets have materialized.

Last year, OpenAI and Nvidia announced a tentative deal in which Nvidia agreed to provide OpenAI with up to $100 billion in funding to build up to 10 gigawatts of its own data centers. Nvidia also discussed leasing OpenAI its GPUs as part of that arrangement. The companies didn’t follow through on completing the deal, and earlier this year Nvidia made a $30 billion equity investment in OpenAI.

Earlier this year, during the demise of the Stargate plan, OpenAI shelved its near term plan to build its own data centers. The potential deal with SB Energy would get it much closer to having more control over its cloud capacity and potentially lower its compute costs. OpenAI’s Sam Altman has long sought to build multiple gigawatts of data center capacity at the same campus, with conversations for a $100 billion campus dating back to early 2024 conversations with Microsoft.

Though having more control over its compute capacity could in the long run be less expensive, it’s still a costly endeavor.

OpenAI would need to make at least tens of billions of dollars in payments for the Ohio lease over a 20 year duration, the people said. That doesn’t count the costs of equipment such as chips and servers, which typically account for around 70% of data center costs, meaning OpenAI would need around $350 billion to obtain Nvidia AI chips.

OpenAI has discussed arranging financing for those chip purchases, according to the two people. It isn’t clear whether Nvidia or other parties would lease out or finance the eventual chip purchases.

OpenAI is racing to obtain the power and chips needed for future AI systems, while also trying to ensure it does so in a way that is financially sustainable. It has signed cloud rental agreements with providers including Oracle, Microsoft and Amazon Web Services totaling at least $665 billion over the next five years. Its deals for computing power are sure to be a central subject of scrutiny around its planned initial public offering, which it filed confidentially for on Monday.

Already one of the biggest customers of Nvidia, OpenAI also has announced agreements to use chips from Advanced Micro Devices, as well as Trainium chips from Amazon. OpenAI is also developing its own custom AI chip with Broadcom.

Repurposing an Old Nuclear Weapons Site

Founded in 2019, SB Energy is majority owned by SoftBank and counts as investors Ares Management and OpenAI, which said in January it had invested $1 billion in the energy company and hired it to build and operate a 1.2 GW data center site in Milam County, Texas. SB Energy in May announced that it intended to file for an initial public offering in the U.S. Reuters reported SB Energy could seek a valuation north of $50 billion.

The Ohio project originated in an agreement between SB Energy and the Trump administration for the company to redevelop a former uranium enrichment site in Pike County, Ohio, which sits just over 50 miles south of Columbus. That grew out of the work of a team within the Energy Department led by employees Tim Walsh and Ankur Bansal, who at the direction of Energy Sec. Wright began last year exploring how to use DOE-owned land for developing power plants and data centers for AI.

Conversations accelerated in February, when the U.S. Commerce Department announced that as part of Trump’s new trade deal with Japan, SB Energy would invest $33 billion to build a new, 9.2 GW natural gas power plant at the Ohio site.

The U.S. government will own the power plant, which SB Energy will operate. SB Energy has said it will fund the cost of power grid upgrades so that costs are not passed onto the public.

At the March ceremony, also attended by Commerce Secretary Howard Lutnick, officials touted the plans to use the power plant for a massive AI data center, which officials described as one of the largest planned AI infrastructure projects in the world. Officials also emphasized the value of revitalizing a site for commercial purposes that was once used to make material for nuclear weapons.

The Energy Department’s ownership of the site is part of what makes it attractive. That could help the project avoid some of the public backlash to large data center projects around the country, most of which are located in areas more prone to community resistance and local zoning boards.