Nvidia in Talks to Guarantee $250 Billion for OpenAI's Giant Ohio Data Center
Nvidia is negotiating to guarantee roughly $250 billion in financing so OpenAI can lease a 10 gigawatt data center SoftBank is building in Ohio, on a site that once housed a uranium enrichment plant. The project's total cost, including chips, could exceed $500 billion, according to the Wall Street Journal, making it the largest data center ever announced.
Nvidia in talks to guarantee $250 billion
Nvidia is in talks to guarantee roughly $250 billion in financing that would let OpenAI lease a 10 gigawatt data center project that SoftBank's energy subsidiary is building in Pike County, Ohio, according to a Wall Street Journal report echoed by Qz, Yahoo Finance and Proactive Investors. The project's total cost, including chips, could exceed $500 billion, which would make it the largest data center project ever announced.
A site with a nuclear past
The project is being built on a site that once housed a now decommissioned uranium enrichment facility, roughly 50 miles south of Columbus. The location was chosen, according to the WSJ's reporting, for the electrical infrastructure and grid connections an industrial site of that scale already has, which cuts down on the construction time normally needed to make a data center of this size viable.
Two separate financing deals: lease and chips
Under the terms being discussed, Nvidia's $250 billion commitment would cover the data center's lease and construction, leaving out the cost of the AI chips that will equip the site. In parallel, Nvidia is negotiating a second financing arrangement, this time to help OpenAI buy the chips, in a figure the Wall Street Journal estimates could reach $350 billion. Combined, the two deals push Nvidia well beyond the role of hardware supplier into that of a direct financier of demand for its own products.
Why OpenAI needs a guarantor
The reason for this financial engineering is straightforward: OpenAI is not yet profitable and therefore cannot secure an investment grade credit rating on its own that would make financing at this scale viable on reasonable terms. Nvidia's guarantees work as borrowed credit standing, letting OpenAI access capital that would otherwise have to come entirely from investors willing to take on the direct risk of a still unprofitable company.
The US government and Japan in the middle of the deal
Power for the project is controlled by the US government and funded separately by Japan, under a recent trade deal between the two countries, according to the Wall Street Journal. US Commerce Secretary Howard Lutnick is directly involved in deciding which companies get access to that power, turning the project into an instrument of industrial policy and energy diplomacy as well, not just a private deal between Nvidia, OpenAI and SoftBank.
Market skepticism over circular financing
The structure drew attention from investors critical of the financing model dominating the AI infrastructure race: investor Michael Burry, known for having predicted the 2008 financial crisis, publicly commented on the deal with the phrase around and around we go, according to Benzinga and Yahoo Finance, a direct reference to the fear that suppliers like Nvidia are, in practice, financing demand for their own products, a circular arrangement that inflates the market without necessarily reflecting real revenue on OpenAI's side.
One more piece in the AI infrastructure investment escalation
The announcement adds to a string of billion dollar moves this outlet has been tracking: OpenAI recently signed a 25 year energy deal for another data center, Project Camellia, in Georgia, valued at $30 billion; Nvidia also invests up to $5 billion in Anthropic and signed a deal worth more than $500 billion with SK Hynix to lock in memory supply. The Ohio project stands out for its sheer size, though: $250 billion just for the lease guarantee is a figure larger than the annual GDP of entire countries.
Why it matters to Brazilian agencies and SMBs
No Brazilian agency or SMB will ever negotiate directly with this kind of infrastructure, but the sheer volume of these investments is the direct reason the cost of accessing frontier AI models keeps falling year after year: the more computing capacity comes online, the greater the supply available to run models like GPT or Claude through an API. The point of caution is the same one Michael Burry raises: a significant share of this growth is financed by debt and cross guarantees among a handful of companies, meaning any hiccup in that ecosystem, a risk repricing, a construction delay, a shift in energy policy, could ripple quickly up the supply chain that Brazilian automation companies also depend on indirectly.