Strategy5 min

Nvidia Secures $105 Billion for OpenAI’s AI Campus in Ohio, with Exclusive GPUs for 20 Years

Canteiro de obras do PORTS-Pike Technology Campus em Pike County, Ohio, com guindastes ao entardecer e subestação elétrica iluminada em primeiro plano

Nvidia commits to up to $105 billion in credit for the PORTS-Pike Technology Campus in Pike County, Ohio. SoftBank's SB Energy builds and operates; OpenAI leases for 20 years with exclusively Nvidia hardware.

Nvidia will finance up to $105 billion in credit for the construction of the PORTS-Pike Technology Campus, an OpenAI data center complex in Pike County, Ohio, announced on Monday, August 17. The agreement involves SB Energy, a SoftBank subsidiary, as the builder and owner of the site, and includes a 20-year lease under which OpenAI commits to using exclusively Nvidia hardware.


The initial planned capacity is 4.25 gigawatts of computing power, with a contractual option to expand by an additional 3.75 gigawatts, totaling 8 GW. The site can accommodate up to 1.5 million Nvidia GPUs, which could represent between $150 billion and $200 billion in potential revenue for the manufacturer per hardware generation deployed by 2030. Nvidia also invests $1.5 billion directly into SB Energy as a partner in the venture. The first wave of computing capacity is expected to come online in phases starting in 2028.


The Exclusivity Clause and the Circularity Debate


The most talked-about aspect of the agreement is not the value, but the structure: Nvidia finances the campus where its own chips will be installed, and OpenAI pays the rent with revenue generated from those same chips. Jensen Huang, Nvidia's CEO, responded to criticisms regarding "circular financing" in a post on platform X, arguing that the company is leveraging "its scale and long-term visibility" to enable a project that would otherwise take additional years to capitalize. Sam Altman, CEO of OpenAI, described the complex as "a massive site, with enough computing power to assist millions in using AI to do things we can barely imagine today, from discovering new drugs to starting new businesses."


The exclusivity clause locks the campus out from competing chips, including Google Cloud’s TPU accelerators and AWS’s Trainium. For OpenAI, which has operated its inference capacity entirely in third-party clouds until now, the campus represents the first dedicated frontier-scale facility on its own terms. In exchange for $105 billion in guaranteed credit, Nvidia ties the world’s largest GPU buyer to its portfolio for a period that will cover at least three hardware cycles.


SoftBank, Ohio, and the Geopolitics of AI Infrastructure


The choice of Pike County is not operationally coincidental. The county is home to the old Portsmouth uranium enrichment plant, a symbol of the industrial decline of the American Midwest. Masayoshi Son, chairman of SoftBank, had committed $100 billion to AI investments in the United States by 2024. The Ohio campus materializes the largest individual share of this commitment, transforming SB Energy into a strategic global computing asset operator, far beyond the renewable energy business the subsidiary currently conducts.


The agreement is also a calibration data point for governments outside the United States. In Europe, state-funded digital sovereignty programs are planning inference capacity that operates on another order of magnitude compared to a single block of 4.25 GW. In Japan, SoftBank is transitioning from a financial investor in AI to a physical operator of one of the largest computing assets on the planet, a positioning that alters the geopolitics of GPU access throughout the Asian supply chain.


What Changes for the Enterprise Market


For consulting firms and IT companies delivering products with embedded OpenAI inference, such as Accenture, Deloitte, and SAP, the assurance of 8 GW starting in 2028 resolves a central roadmap uncertainty: the availability of frontier capacity for the next two years. API contracts closed today for 2027 and 2028 will be priced before this infrastructure goes live, creating a negotiating window for large-scale corporate buyers who understand the timing.


The debate around financial circularity will continue in the coming weeks. The relevant question for the CIO is not whether Nvidia finances itself to some degree, but whether the $105 billion commitment signals that the race for AI infrastructure has shifted from the experimental phase to an irreversible capital bet. The implicit answer in Monday's agreement is yes.

The week's analysis, by email

One weekly edition with what matters to people who decide. No ads, no sponsorship.

One-click cancellation, at any time.

Strategy