Nvidia disclosed a $105 billion financing commitment for a new OpenAI artificial intelligence data center in rural Ohio, a deal that dwarfs most infrastructure projects in American history and raises hard questions about who bears the risk when the AI boom demands its next upgrade.
The chipmaker filed a securities disclosure with the SEC on Monday revealing it will extend credit to support an initial 4.25 gigawatts of computing capacity at the PORTS-Pike Technology Campus in Pike County, Ohio, with an option for an additional 3.75 gigawatts. SB Energy will build and manage the facility. OpenAI has signed a 20-year lease, and capacity is expected to begin coming online in phases in 2028, CNBC reported.
The numbers are staggering even by Silicon Valley standards. Nvidia is also investing $1.5 billion directly in SB Energy, the development arm tied to SoftBank. SB Energy and SoftBank, in turn, will build power sources supporting 10 gigawatts of energy and invest at least $4.2 billion into regional grid infrastructure. OpenAI CEO Sam Altman was an early investor in SB Energy, a relationship the filing does not explain in detail.
The location itself tells a story about shifting American priorities. The PORTS-Pike campus sits on the grounds of a former uranium enrichment facility in Pike County, a rural stretch of southern Ohio far removed from the coastal tech corridors. The Washington Examiner reported that the planned facility will surpass the largest operating U.S. data center, SpaceX AI's Colossus 2 at 1.563 gigawatts, and even the nation's largest nuclear plant, Vogtle, at 4.658 gigawatts.
OpenAI issued a statement calling the project a chance to give "a site that once supported American industry and national security a new role building infrastructure for the Intelligence Era." That framing is convenient. It also sidesteps the question of what happens to Pike County if the AI boom cools, the lease terms shift, or the technology leapfrogs the hardware before the concrete dries.
OpenAI claims the project will support 35,000 new construction jobs through 2032 and 2,500 long-term positions. Those figures come from OpenAI itself, not from an independent labor analysis, a distinction worth noting when a company is selling the public on a project of this scale.
The broader AI infrastructure race is pulling skilled tradespeople away from other sectors of the economy. Electricians and construction crews are already in short supply nationwide, and projects of this magnitude only intensify the squeeze on an already strained labor market.
The $105 billion figure did not arrive out of thin air. CNBC previously reported in late July that Nvidia was in discussions to provide a backstop of up to $250 billion for a 10-gigawatt facility at the same Ohio location. The Wall Street Journal then reported last week that Nvidia was set to cut the guarantee to less than $120 billion. The final number landed at $105 billion, lower than the original discussion, but still enormous by any measure.
Nvidia CEO Jensen Huang framed the commitment as a long-term infrastructure play. In a company release, he said:
"We are securing long-lived infrastructure for NVIDIA compute so OpenAI can deploy the most productive AI factories that can be upgraded repeatedly with each new generation delivering more intelligence and better economics."
That language, "AI factories" that can be "upgraded repeatedly", points to the real business model. Nvidia is not just financing a building. It is locking in a customer for decades of chip purchases, each new generation more expensive than the last. The $105 billion credit essentially guarantees demand for Nvidia's own products.
OpenAI President Greg Brockman appeared on CNBC's "Squawk Box" on Monday and put the stakes in blunt terms:
"Compute is really becoming the new oil, the new limited resource of the AI age."
If compute is the new oil, then Nvidia just positioned itself as both the drilling company and the bank that finances the wells. That kind of vertical entanglement deserves scrutiny, not just applause.
Altman offered his own pitch for the project's scale. The New York Post reported his statement:
"This is going to be a huge site, with enough computing power to help millions of people use AI to do things we can only start to imagine today, from finding new medicines to starting businesses and solving hard problems."
Grand promises about medicine and problem-solving are easy to make when someone else is writing a $105 billion check. The harder question, who absorbs the loss if the "things we can only start to imagine" don't materialize at a pace that justifies this investment, remains unanswered.
Altman's early investment in SB Energy, the entity that will build and manage the campus, adds another layer. The SEC filing does not disclose the size of that stake or whether it creates any formal conflict of interest. For a deal of this magnitude involving public filings and taxpayer-adjacent infrastructure commitments, that opacity is notable.
The Ohio project does not exist in isolation. Last week, Nvidia teamed up with six large asset managers, whose names have not been publicly disclosed, to build financing platforms designed to deploy $500 billion in third-party capital for data center projects. That initiative suggests the $105 billion commitment is a single piece of a much larger strategy to make Nvidia the financial backbone of the AI buildout.
Separately, a consortium called the Artificial Intelligence Infrastructure Partnership, which includes BlackRock, Nvidia, and Microsoft, is acquiring Aligned Data Centers in an approximately $40 billion deal. Aligned operates 50 campuses with more than 5 gigawatts of operational and planned capacity across the U.S. and Latin America. BlackRock Chairman Larry Fink said the partnership "is positioned to meet the growing demand for the infrastructure required as AI continues to reshape the global economy," AP News reported.
The capital flowing into AI infrastructure is extraordinary. Jeff Bezos recently unveiled his own AI venture, Prometheus, after a $12 billion funding round. Anthropic has filed confidential IPO paperwork in a race to beat OpenAI to Wall Street. The sheer volume of money chasing AI capacity raises a basic question: is this investment driven by proven demand, or by the fear of being left out?
SB Energy and SoftBank have pledged to cover all grid upgrade costs so local ratepayers are not burdened, a promise worth watching closely over the life of a 20-year lease. Pike County residents did not sign up to subsidize Silicon Valley's next bet, and the commitment to shield them from costs should be enforceable, not aspirational.
The recent $15 billion loss at Jane Street after an AI hedge fund's meltdown is a reminder that enormous bets on artificial intelligence do not always pay off. The technology is real. The productivity gains may be real. But $105 billion in credit, a 20-year lease, and a former uranium site in rural Ohio add up to a wager that assumes the AI boom has no ceiling, and no bust cycle.
Several critical details remain unclear. The exact nature of Nvidia's $105 billion commitment, whether it functions as a loan, a guarantee, a credit facility, or some hybrid instrument, is not specified beyond the words "financing" and "credit." The terms of Nvidia's $1.5 billion investment in SB Energy, including whether it represents equity or debt, are also undisclosed.
Concerns about circular financing in the AI trade have surfaced in industry commentary. Nvidia finances the infrastructure; Nvidia supplies the chips that fill the infrastructure; the customer uses the chips to generate demand for the next generation of Nvidia chips. Each dollar Nvidia commits comes back as revenue for Nvidia products. That loop is not inherently corrupt, but it deserves the kind of independent scrutiny that a $105 billion SEC filing ought to invite.
When the companies building the future are also financing it, guaranteeing it, and supplying it, the taxpayers and workers left holding the bag if it unravels deserve more than press releases and morning-show sound bites.