Why it matters
The AI race is no longer constrained only by model architecture or GPU supply. Large training and inference campuses increasingly depend on power generation, transmission, cooling, networking, construction and financing moving together.
The Ohio project makes that stack unusually visible because the reported commitments connect a major AI customer, a major compute supplier and a very large energy buildout in one campus plan.
The reported campus is measured in gigawatts
Axios reported that OpenAI signed a 20-year lease for a southern Ohio AI campus planned for as much as 8 gigawatts of IT compute. The first 800 megawatts were targeted for 2028 in contemporaneous reporting.
The U.S. Department of Energy and the PORTS Technology Campus describe a broader energy plan totaling 10 gigawatts of new generation, including 9.2 gigawatts of natural-gas generation, at the former Portsmouth Gaseous Diffusion Plant site.
NVIDIA is exposed to more than chip demand
The Wall Street Journal reported credit support from NVIDIA of up to $105 billion tied to the first portion of completed data-center capacity. The figure is a contingent support structure, not equivalent to NVIDIA simply writing a $105 billion check today.
That distinction matters, but so does the strategic direction: the compute vendor is participating in the financing architecture needed to make future GPU demand physically buildable.
The bottleneck expands into the industrial stack
At campus scale, incremental AI capacity can pull demand through turbines and generators, transformers, switchgear, substations, fiber, liquid cooling, construction labor and advanced packaging. Any one constrained layer can become a schedule bottleneck.
RFDELTA therefore treats the project as an infrastructure signal rather than a single-company product story.
The RFDELTA takeaway
The important shift is that frontier compute is behaving more like heavy infrastructure. Capacity is increasingly gated by permitting, power, equipment lead times and finance as much as by model design.
All capacities, jobs, financing commitments and target dates remain planned or reported terms and can change with final agreements, construction and grid execution.
Share this Signal
Use the memorable RFDELTA path below when sharing this sourced Signal. Original social-video links will appear here once their exact post IDs are wired.
Memorable path: https://rfdelta.com/001
Video transcript
Stop thinking the AI race is about chatbots. It just became an energy war. OpenAI reportedly signed a 20-year lease for a giant AI campus in southern Ohio. NVIDIA could provide up to $105 billion in credit support. The scale: 8 gigawatts of AI compute, backed by 10 gigawatts of new power — 9.2 gigawatts of natural gas — on a former uranium-enrichment site. The first 800 megawatts are targeted for 2028. The campus projects 35,000 construction jobs and 2,500 permanent roles. The real signal: NVIDIA is moving from selling chips to financing the customers who buy them. So the next AI winners may be power, grid equipment, cooling, networking and advanced packaging — not just model labs. RFDELTA takeaway: compute is infrastructure. Follow the power. Follow the capital. See the trade before the headline.
Frequently asked questions
Is NVIDIA spending $105 billion immediately?
No. Reporting describes up to $105 billion of credit support tied to specified data-center capacity. It should be treated as contingent exposure rather than a simple cash purchase.
How large is the planned Ohio campus?
Contemporaneous reporting described up to 8 gigawatts of IT compute, with an initial 800 megawatts targeted for 2028, alongside a larger 10-gigawatt generation plan.
Why does RFDELTA call this an energy story?
At this scale, AI capacity depends on generation, grid equipment, cooling, networking, construction and financing. Those physical systems can become binding constraints on compute deployment.
Primary sources
Continue exploring RFDELTA
RFDELTA Signals map the hidden systems, technology transitions and operational dependencies underneath fast-moving headlines.