Our Blog

Blog Index 

Nvidia to Back OpenAI's Ohio Data Center With a $105 Billion Guarantee — What It Means for the AI Race

Posted on 18th Aug 2026 06:04:36 in Artificial Intelligence, Machine Learning

Tagged as: nvidia, openai, data centers, ai infrastructure, sb energy

One of the largest infrastructure financing commitments in the history of the chip industry was announced on August 17, 2026, and it did not involve a new chip at all. Nvidia has agreed to guarantee up to $105 billion in lease payments so that OpenAI can secure a sprawling data center campus in Pike County, Ohio — a site with a total planned capacity of up to 8 gigawatts of AI computing, built on the grounds of a former Cold War-era uranium enrichment plant. The deal, reported by Reuters and Bloomberg, ties together the two most dominant forces in the AI boom in a way that will shape how artificial intelligence infrastructure gets financed for the rest of the decade.

The structure is unusual because Nvidia is not simply selling chips. The company will act as a financial backstop, an exclusive hardware supplier, and an investor all at once. It plans to put $1.5 billion into SB Energy, the SoftBank Group-backed developer that will build, own and operate the facility, and OpenAI will lease the campus for 20 years. The first 800 megawatts of capacity are expected to come online by 2028, with the project growing in phases after that.

Inside the Deal: What Each Party Brings

The Ohio campus is enormous even by the inflated standards of the AI buildout. The site will begin with an initial capacity of 4.25 gigawatts, with an option to expand by a further 3.75 gigawatts to reach 8 gigawatts in total. For scale, a single gigawatt is enough electricity to power roughly 750,000 U.S. homes at any given moment, and the entire city of San Francisco consumed just over 5 gigawatts in 2024. This one facility, at full build-out, would draw more power than San Francisco.

The key terms of the arrangement include:

  • The guarantee: Nvidia will cover defined portions of OpenAI's lease and power payments, up to $105 billion, plus a commitment that the site retains a minimum value. If OpenAI defaults, Nvidia covers the gap between that guaranteed minimum and whatever the owner recovers by re-leasing or selling the site.
  • Exclusivity: The campus will run exclusively on Nvidia chips. Nvidia has already secured land and power for the facility and is providing credit support for the construction tied to the initial capacity.
  • Equity: Nvidia is investing $1.5 billion in SB Energy. This follows a $1 billion investment in January from OpenAI and SoftBank to expand the data center and power infrastructure behind their Stargate AI initiative.
  • Power build-out: SoftBank and SB Energy plan to build at least 10 gigawatts of new power generation and will invest $4.2 billion in regional grid infrastructure through a partnership with AEP Ohio.

Nvidia CEO Jensen Huang framed the deal as supply-chain discipline applied to infrastructure. "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," Huang said. OpenAI CEO Sam Altman described the site in simpler terms: "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."

The economics are staggering at every level of the stack. Huang has said the Ohio site alone could contribute as much as $200 billion to Nvidia's revenue, and that OpenAI's total spending on Nvidia compute could reach roughly $600 billion through 2030 — equivalent to about 16 gigawatts of computing capacity when the expansion is included.

The Circular Financing Question

The announcement arrives at a delicate moment for AI infrastructure finance. For months, analysts and investors have questioned whether the AI boom rests partly on circular deals — arrangements in which chipmakers and other suppliers effectively fund their own customers, who then spend the money back on their chips. Nvidia has provided financial backstops before, notably to cloud provider CoreWeave, but the Ohio guarantee is by far the largest arrangement of its type for the company.

The scrutiny is not hypothetical. When The Wall Street Journal and Bloomberg reported in July that Nvidia was weighing a backstop of as much as $250 billion for the same Ohio project, Nvidia shares fell about 5%. The final figure — $105 billion — is less than half that, and the stock has since recovered, still up more than 19% for the year.

Nvidia addressed the criticism head-on. "OpenAI will pay the lease," the company said, adding that the deal was made with "the same discipline we apply to supply-chain management." The financing structure is not yet fully defined, but people familiar with the matter told Reuters it will include equity — potentially from a future SB Energy IPO and direct SoftBank investment — followed by a debt layer that could include project finance loans and public bonds.

Market commentators remain split. "Investors are right to be worried about what seems to be a never-ending loop of AI deals, but realistically the field of players isn't all that vast and there was always going to be a degree of circular financing," said Danni Hewson, head of financial analysis at AJ Bell. "The biggest test is whether these investments ultimately generate decent returns for all those laying out cash, and that's something that can only be figured out further down the line."

The Ohio guarantee also lands one week after Nvidia partnered with six major financial institutions, including BlackRock, to launch financing platforms targeting more than $500 billion in third-party funding for AI infrastructure. The pattern is clear: Nvidia is moving deliberately from being a chip vendor to being an architect of the entire AI infrastructure capital stack.

Why Power Is the Real Bottleneck

The most revealing part of the announcement may not be the money but the megawatts. Land and electricity have become the primary constraints on AI expansion, as aging and strained power grids collide with surging data center demand. Community opposition to new construction is also rising, driven by concerns about higher electricity prices and water use. A Gallup poll from March 2026 found that more than 70% of Americans oppose data centers being built near their homes.

The Ohio project is designed to sidestep these frictions. The state has been supportive because of the jobs — OpenAI expects about 35,000 construction jobs through 2032 and roughly 2,500 long-term operating jobs — and OpenAI and SoftBank have committed $80 million toward community projects. The site involves federal land and has the involvement of the U.S. Departments of Commerce and Energy, giving it a degree of institutional weight few data center projects enjoy.

Power, however, must be generated first. The 10 gigawatts of new generation that SoftBank and SB Energy plan to build will not appear overnight, which is why the campus comes online in phases beginning in 2028 rather than next year. The guarantee structure itself reflects this reality: OpenAI will only begin paying as capacity becomes available for lease, which spreads the financial risk across the construction timeline.

What It Means for the AI Race

Scale is now a strategic weapon in AI. The Ohio campus is a centerpiece of the Stargate initiative — the multi-year, $500 billion data center program announced by OpenAI, Nvidia, SoftBank, Oracle and others — and it would rank among the largest AI computing facilities in the world. For comparison, SpaceXAI's Colossus data centers in Tennessee and Mississippi currently hold about 1 gigawatt of capacity, and Meta's Louisiana project is planned at 5 gigawatts. The Ohio site, at 8 gigawatts, would eclipse both.

For businesses, the takeaway is that AI compute is being treated less like a utility and more like strategic national infrastructure, with financial structures to match. The next generation of AI models will require orders of magnitude more compute than today's, and the companies that lock in power, land and hardware first will define the economics of the AI industry for years. As Huang put it, the goal is infrastructure that "can be upgraded repeatedly with each new generation."

What remains to be seen is whether the returns justify the capital. With $105 billion in guarantees, $1.5 billion in equity, 10 gigawatts of new power generation and a 20-year lease, the Ohio project is a bet measured in decades — and every major player in AI is watching to see whether it pays off.

Sources

whatsapp me