Nvidia Backs OpenAI’s $105 Billion Ohio AI Campus — But the Number Needs Unpacking

Nvidia has disclosed a financial commitment to OpenAI’s planned Ohio data centre campus that carries a headline figure of $105 billion — but the structure of the deal is more nuanced than that number suggests, and understanding the difference matters as much as the announcement itself.

Nvidia disclosed the arrangement in an August 17 filing with the US Securities and Exchange Commission. The $105 billion figure represents the maximum aggregate payment obligation under residual-value guarantees Nvidia is providing for SB Energy’s PORTS-Pike Technology Campus in Pike County, Ohio, where OpenAI will lease AI computing infrastructure under a 20-year agreement. SB Energy builds, owns, and operates the campus. OpenAI leases it. Nvidia has guaranteed a minimum residual value — meaning if specified default or insolvency events occur and SB Energy is forced to re-lease or sell the facilities at a loss, Nvidia could be liable for the shortfall between the guaranteed minimum value and what SB Energy actually recovers. OpenAI has agreed to reimburse and indemnify Nvidia for any amounts it actually pays.

Nvidia’s direct investment in the project is a separate and considerably smaller number: $1.5 billion into SB Energy itself. The $105 billion guarantee is a contingent liability, not a cash commitment.

The campus’s initial build is targeting 4.25 gigawatts of IT capacity, with an option for Nvidia to extend its guarantee to cover an additional 3.75 gigawatts — bringing the eventual campus to 8 gigawatts total, with SoftBank and SB Energy separately planning at least 10 gigawatts of new energy generation to power it. Nvidia describes itself as the exclusive AI compute infrastructure provider for the deployment.

What makes the structure genuinely interesting is what Nvidia’s CEO Jensen Huang is signalling about where the company sees its role in the AI economy going. Rather than simply supplying chips to whoever can pay for them, Nvidia is positioning itself as a long-term infrastructure partner — essentially treating its GPUs the way Ford’s financing arm treats vehicles: as assets with residual value that can underpin financial structures, unlock capital, and create a new investable asset class in AI hardware. Analysts have noted that high-end AI chips can have useful lifespans of six to seven years or longer, creating room for exactly this kind of financing model.

The practical scale of what’s being built is genuinely unprecedented. Nvidia said it has secured land, power, and shell capacity at the Ohio site and will be the exclusive AI compute infrastructure provider — a structure that locks three pieces of the AI infrastructure stack together: Nvidia’s hardware, OpenAI’s long-term compute demand, and SB Energy’s physical ownership of the data centre. The forthcoming detailed guarantee agreements, expected in Nvidia’s next quarterly report, will clarify how much financing capacity the residual-value structure actually creates.

For Africa and Uganda, the Ohio campus matters indirectly but significantly. Every gigawatt of AI compute built in the United States is compute that US-aligned AI companies will use to train and run the models that increasingly power the world’s digital services — including the platforms, financial tools, and infrastructure that East African governments, businesses, and consumers rely on. The concentration of AI infrastructure in a small number of giant campuses in the US and China, backed by financial structures that the world’s largest chipmakers are now directly underwriting, is part of the same global AI alignment dynamic that is narrowing the window for countries that have not yet built their own AI policy capacity to engage on their own terms.