$250 billion. That is the size of the financial guarantee Nvidia is reportedly weighing so that OpenAI can build a single data center in southern Ohio. The Wall Street Journal broke the story on July 27, and the figure is so large it reads like a typo. It is not. If the deal comes together, it would be one of the biggest corporate financing arrangements ever attached to a piece of computing infrastructure.
The biggest building project in tech
The site sits about 50 miles south of Columbus, on the grounds of a decommissioned uranium-enrichment plant. SoftBank’s energy arm is developing the project, and OpenAI would lease it once the concrete and cabling are in place. The plan calls for 10 gigawatts of capacity. To put that in perspective, a single large nuclear reactor produces roughly one gigawatt, so this one campus would draw the power of about ten reactors running flat out.
The total price tag, once you add the Nvidia chips that would fill the halls, could climb past $500 billion. Reporting from Bloomberg and CNBC puts it on track to become the largest data center project anyone has announced. For a sense of scale, that number rivals the annual budgets of entire government departments. And it buys one thing: a place to run and train the next generation of OpenAI models.
Why OpenAI needs a co-signer
The awkward part is that OpenAI, one of the most valuable private companies on earth, still cannot borrow half a trillion dollars on its own terms. The company has not turned a profit, which means it cannot secure an investment-grade credit rating. Lenders looking at a lease of this size want someone solid standing behind it. Nvidia, sitting on enormous cash reserves and the highest market value of any chipmaker in history, is that someone.
The proposed $250 billion guarantee would cover the data center lease and the debt raised to build it. Notably, it would not cover the chips themselves. According to the Journal, Nvidia is separately discussing financing for OpenAI’s chip purchases, a commitment that could reach $350 billion on its own. Stack those together and Nvidia is potentially underwriting well over half a trillion dollars of its largest customer’s spending.
Around and around the money goes
That structure has set off alarm bells among people who watch balance sheets for a living. Nvidia would be guaranteeing the financing for a customer whose main purchase is Nvidia hardware. Money leaves Nvidia, travels through OpenAI and a data center lease, and comes back to Nvidia as chip orders. Michael Burry, the investor who called the 2008 housing crash, summed up the mood with a short jab about going around and around.
Defenders see it differently. Nvidia has every reason to make sure its biggest buyer can keep buying, and locking in a decade of guaranteed chip demand is exactly the kind of moat a chipmaker dreams of. Whether that is shrewd vertical strategy or a warning sign about how much of the AI boom is being financed by the same handful of players depends on who you ask. Both readings can be true at once, which is what makes the deal so hard to categorize.
OpenAI’s bid to own its own house
Strip away the eye-watering numbers and there is a clear strategic logic here. For years, OpenAI has rented its computing power from Microsoft, Amazon, and Oracle. Renting is fast, but it leaves you dependent on a landlord who is often also a competitor. Controlling a purpose-built facility of this size would let OpenAI set its own priorities, its own timelines, and its own hardware roadmap. This Ohio campus would be the first serious step toward that independence.
The catch is that everything remains early. The talks could fall apart, the terms could shift, and the eventual guarantee could land well below or above the reported $250 billion. Deals of this magnitude rarely close on the first draft, and the parties involved have said little in public. What we have is a detailed picture of an ambition, not a signed contract.
Still, the direction of travel is unmistakable. The companies building frontier AI have decided that raw electrical and computational capacity, not talent or clever algorithms, is what now caps how far they can push, and they are prepared to spend at a scale once reserved for national infrastructure to get it. Watch for whether other chipmakers and cloud providers respond with backstops of their own, because if Nvidia sets this precedent, it will not stay a one-off for long. For more coverage of AI infrastructure and the companies racing to build it, visit Mylistingo.
Source: Nvidia And OpenAI Discussing $500 Billion Data Center: Here’s What We Know (Forbes, YouTube).
The Netherlands, for internationals
One email a week: the week's news for internationals in the Netherlands, new practical guides and what changed in the cities. No spam, unsubscribe any time.
