Nvidia Guarantees Ohio Campus for Exclusive AI Compute
Nvidia announced on August 17 that it is guaranteeing SB Energy's PORTS-Pike Technology Campus in Portsmouth, Ohio, will exclusively host Nvidia AI compute. India Today reports that OpenAI is the intended tenant under a 20-year lease, with an initial 4.25-gigawatt AI factory using Nvidia's DSX full-stack platform. The arrangement extends Nvidia's involvement from chips and systems into data center capacity and power infrastructure.
Nvidia announced on August 17 that it is guaranteeing SB Energy's PORTS-Pike Technology Campus in Portsmouth, Ohio, will exclusively host Nvidia AI compute. The campus is a proposed large-scale AI data center and power project whose development, capacity and operations remain forward-looking, according to Nvidia's release.
India Today reports that OpenAI is intended to occupy the site under a 20-year lease and to build and operate an AI factory using Nvidia's DSX AI factory platform. The publication describes the initial facility as having 4.25 gigawatts of capacity and reports that Nvidia could secure an additional 3.75 gigawatts at a later stage.
From accelerators to site capacity
In a post cited by India Today, Nvidia CEO Jensen Huang wrote that AI factories require chips, packaging, memory and networking alongside "land, power and shell," referring to the physical data center building. Huang also wrote that AI-company growth is increasingly constrained by compute availability rather than algorithms or customer demand.
Nvidia's release describes SB Energy as an integrated data center and power-infrastructure company that develops and operates gigawatt-scale campuses and utility-scale generation. It also identifies Nvidia's investment in SB Energy and credit support among the matters subject to forward-looking-statement disclosures, without providing investment terms in the material provided.
The transaction therefore combines several layers normally procured separately: a data center site, power infrastructure, Nvidia systems, and a prospective large AI-compute customer. Nvidia's guarantee that the campus will host its compute exclusively is a notable commercial constraint for a site at this scale.
Why power availability matters
AI clusters require not only accelerators but also grid interconnection, substations, cooling, networking and construction capacity. Industry reporting has increasingly identified electricity and time to power as gating factors for new data center deployments, particularly for multigigawatt campuses. Yahoo Finance, citing an August 7 Information report, separately described a proposed Nvidia investment of up to $3 billion in power-infrastructure developer Lancium, whose assets include the Abilene, Texas site associated with the Stargate project.
That Lancium transaction is distinct from PORTS-Pike, but it illustrates the broader shift in AI infrastructure procurement toward power and land access. For ML infrastructure teams, the practical consequence of this pattern is that capacity planning increasingly extends beyond GPU specifications and cluster topology to long-lead physical constraints, including utility commitments and facility readiness.
Nvidia's release cautions that the PORTS-Pike campus's timing, scale, capacity, operation, potential expansion and OpenAI's expected role are forward-looking and subject to risks and uncertainties.
Key Points
- 1Nvidia guaranteed exclusive Nvidia-compute hosting at the PORTS-Pike campus, linking accelerator supply to a multigigawatt physical infrastructure project.
- 2India Today reports an initial 4.25-gigawatt facility and a prospective OpenAI tenancy, making the project significant for frontier-model compute capacity.
- 3Comparable AI infrastructure buildouts increasingly depend on grid access, cooling and construction timelines, not solely GPU availability or cluster design.
Scoring Rationale
The announced Ohio campus is a large proposed AI infrastructure deployment, with 4.25 gigawatts of initial capacity reported by India Today. It matters to ML practitioners because power and site availability are becoming material constraints on training and inference capacity, although major project details remain forward-looking.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
