The AI infrastructure race has a new landmark, and it sits on the grounds of a Cold War-era uranium enrichment plant in Pike County, Ohio. On August 17, 2026, Nvidia and OpenAI announced plans to build what they are calling an AI factory at the PORTS-Pike Technology Campus, a redevelopment of the former Portsmouth Gaseous Diffusion Plant, a decommissioned Department of Energy facility.
The scale is difficult to overstate. OpenAI will lease approximately 8 gigawatts of computing capacity from the campus over 20 years, with the initial phase covering 4.25 gigawatts supported by Nvidia credit guarantees capped at $105 billion.
What is actually being built here
The campus is being developed by SB Energy, a SoftBank subsidiary, which will build out a planned 10-gigawatt generation capacity at the site. Nvidia is putting $1.5 billion directly into SB Energy, and the campus will run exclusively on Nvidia’s DSX AI factory platform.
Phased construction is set to begin in 2028. The project is expected to generate roughly 35,000 construction jobs over the build-out period and create 2,500 permanent positions once operational. OpenAI and SB Energy have also committed a combined $80 million in community benefits funds, with each contributing $40 million. On top of that, OpenAI is providing $84 million in AI tool credits for students in Ohio.
The $600 billion question
OpenAI’s revised compute spending target through 2030 sits at approximately $600 billion. That figure represents a significant downward revision from earlier estimates of $1.4 trillion and puts the Ohio project in context: 4.25 gigawatts is the opening act, not the finale.
Nvidia’s credit guarantees on the initial Ohio phase, capped at $105 billion, effectively de-risk a project that would otherwise require OpenAI to carry the full capital burden of building unprecedented compute infrastructure from scratch. OpenAI gets compute capacity without front-loading the capital expenditure. Nvidia gets a locked-in, decade-long customer with guaranteed hardware demand and a showcase deployment for the DSX platform.
Why the power-first model changes the calculus
The AI factory model treats power capacity as the primary design constraint and builds the compute layer around it. The Portsmouth site, with its existing federal infrastructure and planned 10-gigawatt generation capacity, fits that model precisely.
The $1.5 billion Nvidia is putting into SB Energy is an investment in power generation and delivery capacity, which then becomes the substrate for compute. OpenAI and SB Energy have committed a combined $80 million in community benefits funds alongside $84 million in AI tool credits for students in Ohio.