OpenAI signs twenty-year lease for Ohio data centre with Nvidia as guarantor

OpenAI signed a twenty-year lease agreement on 17 August 2026 for a data centre on the PORTS-Pike Technology Campus in Pike County, Ohio. Nvidia is guaranteeing up to $105 billion in residual value of the facilities and will become the exclusive supplier of computing systems for the complex.

The site, located in part on a former uranium enrichment facility belonging to the US Department of Energy, is being developed by SB Energy, a subsidiary of SoftBank Group. The campus will have a total capacity of 8 gigawatts, making it one of the largest data centre complexes in the world. The first construction phase covers 4.25 gigawatts of IT capacity, and the Nvidia guarantee applies to that phase.

The deal brings together several large financial commitments that largely sit outside conventional corporate balance sheets, an arrangement that is becoming increasingly common in the AI sector.

Where the money comes from

Nvidia plays a central role in the financing structure. The company is guaranteeing up to $105 billion for the residual value of the facilities built in the first phase, and is also investing $1.5 billion directly in SB Energy. In return, Nvidia becomes the sole supplier of chips and computing systems for the campus. Nvidia itself estimates it could generate between $150 billion and $200 billion in revenue from this site.

SB Energy and SoftBank are investing at least $4.2 billion in regional grid infrastructure. OpenAI and SoftBank had previously, in January 2026, jointly invested $1 billion in SB Energy. OpenAI is also contributing $40 million to SB Energy's community fund, bringing that fund to $80 million.

AEP Ohio, the US Department of Energy and the Department of Commerce are involved in the development of the campus, although the precise role of the government agencies is not fully elaborated in available sources.

If OpenAI expands the project further, total spending on Nvidia chips could reach $600 billion by 2030, according to available reporting. OpenAI reportedly has annual revenues of around $40 billion, which puts the ratio between income and the commitments entered into in perspective.

Commitments that do not appear on the balance sheet

The structure whereby Nvidia guarantees the residual value rather than OpenAI or SoftBank carrying the capital on their own balance sheets fits a broader trend. According to the Wall Street Journal, nine major technology companies together have approximately $3 trillion in AI-related commitments outstanding that do not appear in their regular financial statements.

Guarantee obligations and long-term infrastructure lease contracts are formally accounted for differently from direct capital investments. This gives companies more financial headroom on paper, but also means that the true scale of their AI investments is less visible to shareholders and regulators. Nvidia's $105 billion guarantee is a notable example of this approach: it is a substantial guarantee obligation for a single project, carried by the chip supplier rather than the end user.

The site and its scale

The PORTS-Pike Technology Campus in Pike County is located on the grounds of the former Portsmouth Gaseous Diffusion Plant, a uranium enrichment facility that was in use for decades as part of the US nuclear weapons programme. The Department of Energy has remediated the site in recent years and made it available for new industrial development.

The complex will have a total capacity of 8 gigawatts, which is exceptionally large for a single location. By way of comparison, large hyperscale data centres operated by companies such as Microsoft or Amazon typically reach hundreds of megawatts per campus. One gigawatt is equal to one thousand megawatts. The energy demand of the complex therefore places significant requirements on the regional electricity grid, which explains AEP Ohio's involvement and the investments in grid infrastructure.

What this means for the European AI scene

The scale and financing structure of this project stand in stark contrast to the AI infrastructure currently being built in Europe. European initiatives, including Dutch and European AI factories being encouraged under the EU AI Act, are operating in the range of hundreds of millions to a few billion euros. The gap in investment volume is therefore tangible.

For European investors and policymakers, the structure also raises questions about how comparable off-balance-sheet commitments should be valued and regulated over time. If AI infrastructure is increasingly financed through guarantees and lease obligations rather than direct capital investment, this has implications for how financial risks are visible in the sector, including outside the United States.