NVIDIA backs Ohio AI campus for OpenAI at PORTS-Pike

NVIDIA has partnered with SB Energy to secure land, power and shell capacity at the PORTS-Pike Technology Campus in Ohio for AI infrastructure, with OpenAI as the tenant.

The arrangement covers an initial 4.25 gigawatts at the Portsmouth campus, where OpenAI plans to build and operate an AI facility using NVIDIA GPUs, CPUs, networking and infrastructure software. NVIDIA may extend the deal to cover the remaining 3.75 gigawatts at the site.

The move highlights how access to land, electricity and physical data centre space has become a constraint alongside chips, packaging, memory and networking as demand for large-scale AI computing grows. NVIDIA is applying the same supply-chain approach it has used for semiconductors to these infrastructure inputs.

Under the structure it outlined, NVIDIA is supporting parts of the lease and power commitments for about 4 gigawatts over a 20-year term, with the site reserved for NVIDIA compute. The support does not cover the full cost of the campus or all of the tenant's obligations, and the guarantee is due to take effect in phases as data centres enter service between 2028 and 2030.

OpenAI will pay the lease, while NVIDIA is using its scale and visibility into customer demand to help secure a location that can host multiple generations of its systems. As capacity comes online and lease payments are made, NVIDIA's remaining exposure would fall.

Site economics

NVIDIA described the Ohio site as a long-life asset that can be upgraded repeatedly as new generations of compute equipment are installed. Each generation of systems deployed at PORTS-Pike could amount to about 1.5 million NVIDIA GPUs and generate roughly USD $150 billion to USD $200 billion in revenue for NVIDIA.

Over two decades, the campus could support several upgrade cycles. The emphasis is on securing a physical site with long-term access to power and real estate, while replacing the computing equipment inside it as newer systems become available.

NVIDIA also set out the broader scale of its business with OpenAI, saying the company's existing and planned commitments represent about 12 gigawatts of NVIDIA compute through 2030, with scope to rise to about 16 gigawatts if the Ohio arrangement expands beyond the initial 4.25 gigawatts.

At that level, the opportunity would represent around USD $600 billion of compute through 2030. The figures underline the scale of spending now being attached to a handful of large AI customers as the sector races to secure data centre capacity and electricity supply.

Selective approach

Most NVIDIA customers will continue to arrange their own land, power and shell capacity. That includes cloud service providers, enterprises and state-backed AI projects with the balance sheets and financing structures to sign long-term infrastructure contracts directly.

Its intervention is aimed at frontier AI labs whose demand for training and inference computing is rising faster than their ability to secure long-dated infrastructure on their own. In NVIDIA's view, these groups may have revenue growth and customer demand but still lack the financing profile needed for large site commitments.

NVIDIA argued that this is not a form of circular financing because OpenAI remains responsible for lease payments. Instead, it presented the model as a way to lock in scarce infrastructure where it has confidence in long-term demand for its computing systems.

A key part of that argument is that the site would retain value even if one tenant stepped away. Capacity at PORTS-Pike could be reassigned to another customer from NVIDIA's base of cloud providers, enterprises, AI labs and start-ups because the underlying systems are widely used and can be redeployed.

NVIDIA linked that claim to the breadth of its software and developer ecosystem, saying standardisation around its platform makes installed compute assets more reusable across customers over time. That position supports its case for treating long-term site commitments as productive infrastructure rather than a one-customer wager.

The Ohio project marks a further shift in NVIDIA's role from chip supplier to a company involved in financing and securing the physical foundations needed for AI deployment. It will remain selective, focusing on what it described as exceptional sites where sustained demand can support multiple generations of NVIDIA compute.