NVIDIA provided credit support to secure exclusive compute rights at OpenAI’s Ohio data center, tying financing directly to hardware allocation.
NVIDIA will provide credit support for the land, power, and shell buildout at SB Energy’s PORTS-Pike Technology Campus in Ohio, where OpenAI will lease 8 IT-GW of AI factory capacity for 20 years. NVIDIA will also invest $1.5 billion in SB Energy and deploy its full-stack DSX platform at the site. Futurum sizes the unstated backstop from NVIDIA’s own per-gigawatt revenue math and examines what exclusivity bought.
NVIDIA announced on August 17 that it has secured land, power, and shell (LPS) capacity through a partnership with SB Energy at the PORTS-Pike Technology Campus in Pike County, Ohio, which will exclusively host NVIDIA AI compute. OpenAI will be the customer for 8 IT-GW of AI factory capacity, and SB Energy will build, own, and operate the data center under a 20-year lease to OpenAI. NVIDIA will provide credit support on the land, power, and shell buildout to secure the initial 4.25 IT-GW, with an option to take the remaining 3.75 IT-GW, and will invest $1.5 billion in SB Energy, joining existing investors SoftBank Group and OpenAI. The campus, which is reindustrializing the decommissioned Portsmouth Gaseous Diffusion Plant in collaboration with AEP Ohio, the US Department of Energy, and the US Department of Commerce, is expected to come online in phases beginning in 2028. The site will deploy NVIDIA’s full-stack DSX AI factory platform, including GPUs, CPUs, and networking.
“We are securing long-lived infrastructure for NVIDIA compute so OpenAI can deploy the most productive AI factories that can be upgraded repeatedly with each new generation delivering more intelligence and better economics,” said Jensen Huang, Founder and CEO of NVIDIA. “This is going to be a huge site, with enough computing power to help millions of people use AI to do things we can only start to imagine today, from finding new medicines to starting businesses and solving hard problems,” said Sam Altman, CEO of OpenAI. Goldman Sachs and JP Morgan advised SB Energy; Morgan Stanley advised NVIDIA.
Futurum ended our analysis of NVIDIA’s $500 billion compute financing platforms with a specific watch item: confirmation of the reported backstop for an OpenAI facility in Ohio, a question Jensen Huang had declined to answer on August 10. Confirmation arrived in 7 days. NVIDIA will guarantee the land, power, and shell buildout for OpenAI’s Ohio data center at PORTS-Pike, backing the initial 4.25 IT-GW with an option on the remaining 3.75. Futurum views the deal as a purchase of exclusivity. OpenAI spent the past year diversifying its silicon suppliers across AMD, Broadcom, and its own designs, and NVIDIA has now closed the largest announced campus in that pipeline to every one of them. The financing platforms announced on August 10 turned NVIDIA’s cost of capital into a product. PORTS-Pike shows what that product buys when NVIDIA is the customer for its own credit.
The exclusivity clause becomes the headline when considering OpenAI’s compute commitments. OpenAI has agreed to deploy 6 GW of AMD Instinct GPUs under a partnership that includes warrants for up to 160 million AMD shares, has commissioned roughly 10 GW of co-designed custom accelerators with Broadcom, and continues to build Stargate sites with Oracle and SoftBank. Against that field, NVIDIA secured a contractual guarantee that all 8 IT-GW at PORTS-Pike will run NVIDIA silicon for the life of a 20-year lease. AMD’s Helios racks compete on memory capacity and on equity-linked deal terms. Broadcom competes on cost per token for inference at scale. Neither structure blocks NVIDIA from the biggest deal. Credit support on real estate and power creates an instrument that no competing chipmaker has offered at this scale. AMD lacks a similar third-party financing platform, and Broadcom leaves facility financing to its customers. OpenAI’s silicon allocation is now decided partly in the credit market, where NVIDIA’s over $100 billion in annual free cash flow prices everyone else out. The counter-move to watch is whether AMD converts its OpenAI warrant relationship into a comparable guarantee vehicle, because the warrant aligns equity upside without solving OpenAI’s financing problem, and financing is the constraint that this deal just proved matters most.
Futurum argued last week that NVIDIA’s financing platforms exist to solve a collateral problem, since GPU depreciation follows a roadmap only NVIDIA controls. PORTS-Pike extends the logic to the other side of the balance sheet. Land, shell buildings, and grid interconnection are assets lenders can underwrite over 30 years with actuarial confidence. The tenant risk sitting on top of them is the hard part, and that is precisely what NVIDIA absorbed. Jensen Huang detailed the structure in a companion essay. The commitment covers defined portions of lease and power payments along with a specified residual value commitment, not the full cost of the site, phases in as data centers deploy from 2028 through 2030, and declines as OpenAI makes its payments. His fungibility argument solves the collateral problem, since CUDA compatibility means another tenant could take over the systems if circumstances change, which is the remarketing story every lender asks for. This is demand insurance for a tenant whose obligations dwarf its income. A 20-year lease on 8 IT-GW rests on OpenAI revenue growth that remains a projection. The DSX platform closes the loop: PORTS-Pike becomes the flagship reference-design build, the standardized, upgradeable AI factory Futurum described as underwriting infrastructure, where the 20-year site outlives 5 or more generations of silicon inside it.
Huang’s post supplies enough math to size the exposure that neither announcement quantifies. He writes that each GPU generation at the initial 4.25 IT-GW deployment means roughly 1.5 million NVIDIA GPUs and $150 billion to $200 billion in NVIDIA revenue, which works out to $35 billion to $47 billion of NVIDIA content per gigawatt. His separate figure of roughly $600 billion of NVIDIA compute across OpenAI’s 12 GW to 16 GW pipeline through 2030 lands in the same band, at $37.5 billion to $50 billion per gigawatt. Set those against the $50 billion to