Nvidia has agreed to provide up to $105 billion in residual-value support for OpenAI’s 20-year lease at a planned Ohio AI data-center complex, placing the chipmaker directly behind one of the largest infrastructure commitments yet announced for frontier artificial intelligence.
The project, developed and operated by SoftBank-owned SB Energy at the PORTS-Pike site in Ohio, is designed to reach roughly 8 gigawatts of “AI factory” computing capacity and add at least 10 gigawatts of renewable generation once fully built. OpenAI will use the completed capacity as a tenant, while Nvidia systems will supply the site’s computing infrastructure.
Nvidia said the arrangement responds to a growing constraint in AI deployment: securing power, land, transmission access and finished data-center buildings has become as consequential as procuring advanced chips. The Ohio project shows how suppliers are moving beyond hardware sales and using financial guarantees to help customers lock in decades of physical infrastructure.
Nvidia backstop is tied to completed assets
The $105 billion commitment does not function as a conventional promise to pay OpenAI’s rent. Under the structure described by Nvidia, it is a capped residual-value guarantee tied primarily to the land, energy and data-center infrastructure developed by SB Energy.
If OpenAI were to stop leasing completed capacity, SB Energy would first seek a replacement tenant or sell the relevant assets. Nvidia would make a payment only if those measures did not recover a contractually defined minimum value. The guarantee is expected to take effect gradually as facilities are completed and activated, with exposure broadly spanning 2028 through 2030.
OpenAI’s rent payments begin only when individual portions of the complex are finished and available for use. That structure limits the company’s obligation to capacity that has been delivered, while Nvidia’s potential liability falls over time as lease payments are made and the assets mature.
SB Energy and SoftBank plan to invest at least $4.2 billion in regional grid infrastructure. The first phase is expected to provide 4.25 gigawatts of capacity, including an initial 800 megawatts slated to enter service in 2028 through existing AEP Ohio infrastructure.
Reaching the full target will require additional generation plants, transmission lines and grid facilities. Nvidia may make another 3.75 gigawatts of capacity support available depending on demand and infrastructure progress, though it has no obligation to take the entire additional amount.
Ohio buildout depends on power and transmission
The project’s scale puts electricity development at the center of the construction timetable. An 800-megawatt initial deployment would already rank among the larger individual data-center power loads planned in the United States, while the longer-term 8-gigawatt target would require a regional buildout measured in years rather than conventional server-farm construction cycles.
Jensen Huang, founder and chief executive officer of Nvidia, said AI infrastructure planning now extends across chips, advanced packaging, memory, networking, land, electricity and data-center shells. Large cloud providers and companies with strong credit profiles can often commit to long-term infrastructure contracts themselves, he said, while frontier AI laboratories may lack the balance-sheet capacity to do so at comparable scale.
Huang said Nvidia would apply this type of infrastructure support only to a limited number of sites where the company sees durable computing demand. His comments frame the Ohio transaction as an effort to secure future deployment capacity rather than simply finance a single order of current-generation GPUs.
Nvidia estimates that each generation of systems deployed at PORTS-Pike could correspond to around 1.5 million of its GPUs, representing roughly $150 billion to $200 billion in potential Nvidia revenue. That figure is an estimate of systems that the site could host over successive hardware cycles, rather than a firm purchase order for a particular GPU generation.
Huang also said OpenAI’s existing and planned commitments represent about 12 gigawatts of Nvidia computing capacity, a figure that could rise if the Ohio site expands. Nvidia estimated the associated deployment opportunity at about $600 billion by 2030.
AI spending is moving off balance sheet
The Ohio agreement arrives as technology companies increasingly use leases, joint ventures, long-term purchase commitments and guarantees to fund capital-intensive AI infrastructure without owning every asset directly.
A review of corporate filings by the Wall Street Journal found that nine technology companies had disclosed nearly $3 trillion in AI-related off-balance-sheet commitments. The total was roughly three times their combined current lease liabilities and long-term debt, according to the newspaper, and exceeded the approximately $600 billion they had spent on capital expenditures over the preceding year.
Uncommenced lease payments — obligations associated with facilities that have not yet begun operating — accounted for about $1.2 trillion across the companies reviewed, the Journal reported. Purchase commitments and other contractual obligations totaled another $1.9 trillion.
Meta offers one example of the financing model. Its Hyperion data-center project in Louisiana is being built through a joint venture involving a fund managed by Blue Owl Capital. Meta is a minority partner and tenant, with lease payments intended to support cash flows to bondholders. Meta disclosed $347 billion in lease obligations that had not yet commenced as of June, including the Hyperion project.
Alphabet reported $811 billion in purchase commitments and contractual obligations as of June 30, up from $332 billion three months earlier. The company said the obligations mainly covered technical infrastructure, inventory and agreements to secure data-center energy, some of which extend to 2054.
Morgan Stanley warned in an April accounting analysis that the expanding scale and complexity of such arrangements can make corporate leverage harder to assess. Nvidia’s role at PORTS-Pike illustrates that concern: a chip supplier is now assuming potential long-term liability connected to infrastructure built for one of its largest AI customers.
Power competition reaches cryptocurrency mining
For cryptocurrency miners, the immediate effect is unlikely to be a uniform rise in electricity prices. Power markets are regional, and large industrial users often negotiate specialized contracts, build dedicated generation or pay for transmission upgrades. Yet a surge in data-center construction could make access to low-cost, dependable power more contested in regions with limited grid capacity.
The International Energy Agency has projected that global data-center electricity consumption could approach 1,000 terawatt-hours by 2030 as AI workloads expand. That trajectory places increasing value on sites with available generation, transmission connections and long-term power agreements — the same assets sought by large-scale Bitcoin miners.
The Cambridge Centre for Alternative Finance has reported rising use of sustainable energy in Bitcoin mining, though miners generally retain a more flexible operating model than hyperscale data centers. That flexibility can help miners absorb excess generation or curtail during grid stress, but it also means they may struggle to match 20-year commitments from companies able to support multibillion-dollar infrastructure guarantees.
The Ohio arrangement therefore gives a clearer view of the competitive landscape facing energy-intensive digital-asset operations: access to cheap electricity is increasingly being secured years in advance through leases, grid investments and corporate guarantees, rather than through short-term power procurement alone.
Want deeper insight into AI-driven trading? Explore our advanced AI copy trading tools tailored for fast-moving tech megatrends.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
