The Circular Empire: Nvidia's $250 Billion Bet to Finance Its Own Customer

On July 26, 2026, the Wall Street Journal reported that Nvidia is in talks to provide a roughly $250 billion financial backstop for OpenAI. The money would let the ChatGPT maker lease a 10-gigawatt data center campus in Piketon, Ohio — a former uranium enrichment plant being redeveloped by SB Energy, a subsidiary of SoftBank. A separate tranche of up to $350 billion in chip financing is also under discussion. If the full deal comes together, the total commitment would push past $600 billion, anchored to a single campus that could cost at least $500 billion to build.

Reuters could not independently verify the Journal's account. Neither Nvidia nor OpenAI has confirmed the talks. But the structural signal is clear regardless of whether this specific deal closes on these exact terms: the chip supplier is now prepared to underwrite the existence of its largest customer. That is not a sale. It is a dependency.

The Arrangement

The $250 billion would not involve Nvidia handing over cash. It is a guarantee — a promise to cover OpenAI's lease and construction debt obligations if the project falls short. That promise is what would make lenders comfortable extending credit to a company that does not carry an investment-grade rating and whose spending commitments already stretch well beyond what its revenue supports. Without Nvidia standing behind the debt, the borrowing would be expensive, limited, or impossible. With it, OpenAI gets access to compute capacity it could not finance on its own.

The Ohio campus is staggering in scale. Ten gigawatts is roughly the output of ten large nuclear reactors dedicated to a single facility. It is approximately double the entire Northern Virginia data center market as of 2025. The full Stargate footprint — seven active or planned sites including the flagship campus in Abilene, Texas — totals nearly 7 gigawatts of planned capacity. This single site in Piketon would exceed that combined total.

The power infrastructure is being built separately. In March 2026, the US Department of Energy announced a partnership with SoftBank and SB Energy to redevelop the Piketon site. SB Energy committed to building 10 gigawatts of new power generation, including at least 9.2 gigawatts of natural gas capacity, backed by $33.3 billion in SoftBank funding. AEP Ohio is co-developing the grid and transmission infrastructure under a separate $4.2 billion investment. The power allocation is government-controlled, with Commerce Secretary Howard Lutnick involved in deciding who receives access. The first phase is expected to bring only around 800 megawatts online in 2028.

Why Nvidia Would Do This

The strategic logic is straightforward. Nvidia sells the chips. Guaranteeing OpenAI's ability to buy and deploy them protects Nvidia's largest revenue stream. In September 2025, the two companies signed a letter of intent for a partnership to deploy at least 10 gigawatts of Nvidia systems, with Nvidia committing to invest up to $100 billion in OpenAI progressively as each gigawatt comes online. The first gigawatt, using Nvidia's Vera Rubin platform, is targeted for the second half of 2026. The guarantee now under discussion would dwarf that equity commitment, turning a headline investment into a far larger and more entangled bet on OpenAI's ability to pay.

For OpenAI, the arrangement fits a wider strategic shift. The company is trying to own more of its infrastructure rather than rent it from Microsoft, Amazon, and Oracle. Owning the compute means controlling costs, latency, and capacity allocation — but it also means bearing the capital burden, which is where Nvidia's creditworthiness enters the picture. OpenAI cannot borrow $250 billion on its own. Nvidia can make the market believe it can.

The Circular Question

The reception this weekend was not universally enthusiastic. Michael Burry, the investor known for The Big Short, posted on social media: around and around we go. His observation cuts to the structural concern. If Nvidia guarantees the financing that lets OpenAI lease a data center, and OpenAI uses that data center to buy Nvidia chips, then Nvidia is effectively underwriting the demand that drives its own revenue. The capital circulates between a small number of players, and revenue at one node depends on financing provided by another.

The tech critic Ed Zitron called the structure such an insane thing to do on so many levels. The unease is partly structural. A guarantee of this size behaves like a debt obligation on Nvidia's balance sheet even if OpenAI never misses a payment. It could weigh on the company's financial flexibility and invite questions from credit rating agencies about whether Nvidia's growth is being driven by organic demand or by self-financed demand it is creating.

This pattern is not unique to the Ohio deal. Nvidia has taken equity stakes in numerous AI companies that are also its customers. Cloud providers borrow to buy Nvidia chips against contracts with AI labs that are themselves burning venture capital. Nvidia has spent much of the cycle offering AI startups compute now, payment later. Each link is individually rational. The aggregate is a system where the same capital circulates between a shrinking number of participants, and the appearance of organic growth becomes harder to distinguish from demand stimulated by the supplier itself.

The Counterargument

Vendor financing is not new, and it is not inherently pathological. Capital-intensive industries from aircraft manufacturing to telecommunications have long relied on supplier-backed financing to bridge the gap between what customers need and what they can afford upfront. Boeing and Airbus routinely help airlines finance aircraft purchases. Telecom equipment vendors have underwritten network buildouts for decades. The model works when the underlying demand is real and growing, and when the financed assets generate sufficient returns to service the debt.

The AI compute demand is, by every available metric, real and growing. Nvidia's data center revenue reached $115.2 billion in fiscal year 2025, up 142 percent year over year. OpenAI's CFO Sarah Friar addressed similar concerns at the Abilene campus launch in September 2025, telling CNBC that bold infrastructure bets are a historical pattern in technology booms. That defense has logic. The telephone network, the internet backbone, and the cellular buildout all required capital commitments that seemed outsized at the time and looked obvious in hindsight.

But both things can be true simultaneously. The demand can be real, and the financing structure can concentrate risk in ways that amplify a downturn if one arrives. The critical distinction is whether Nvidia's capital commitments translate into durable third-party demand — enterprises, governments, and developers buying Nvidia chips because they need them — or whether they primarily function as internal demand stimulation, where Nvidia is creating the appearance of growth by funding the purchases that generate its own revenue. The Ohio deal makes that question impossible to defer any longer.

What Changes

If the guarantee goes ahead, three things change.

First, Nvidia ceases to be primarily a chip company and becomes an infrastructure-financing company that also makes chips. A $250 billion contingent liability on its balance sheet is not a marginal commitment. It reshapes the company's risk profile, its capital allocation strategy, and the questions analysts must ask about its revenue quality. Nvidia's shares closed Friday down 0.92 percent at $206.84 and slipped further after hours, suggesting the market is already pricing in unease.

Second, the competitive landscape shifts. If Nvidia is willing to finance OpenAI's infrastructure but not Anthropic's, or Google's, or Meta's, then access to Nvidia capital becomes a competitive moat in itself. Labs that can secure Nvidia's financial backing gain a capacity advantage that labs paying cash cannot match. That dynamic could accelerate consolidation in an industry that is already concentrating around a small number of frontier players.

Third, the regulatory surface area expands. A deal of this magnitude, involving government-controlled power allocation, a former federal uranium site, and a single supplier underwriting a single customer's entire infrastructure build, will attract scrutiny from antitrust regulators, financial watchdogs, and national security reviewers. The Commerce Department is already involved in power allocation. The Treasury and the FTC will not be far behind.

The Honest Caveat

The deal is not confirmed. The Journal's report rests on sources described as having direct knowledge of the discussions. Reuters could not verify it. No agreement has been signed. The terms — including the fee Nvidia would charge for the guarantee, the timeline, and the conditions under which the backstop would be triggered — have not been made public. The largest number in the AI buildout so far rests on a single weekend report.

But the direction is visible regardless of whether this specific deal closes. Nvidia is not waiting for customers to build capacity. It is helping finance the capacity itself, then supplying the chips to fill it. The question is no longer whether the AI buildout requires capital beyond what any single company can provide from its own balance sheet. The Ohio deal answers that. The question is whether the structure being built to finance it is durable — or whether it is a circle that holds only as long as everyone keeps spinning.