NVIDIA Backs $105 Billion Ohio AI Factory Campus With OpenAI as the Sole Tenant

NVIDIA is providing up to $105 billion in credit and a $1.5 billion equity stake to secure a 4.25-gigawatt AI factory campus in Portsmouth, Ohio — guaranteeing the site exclusively for NVIDIA compute with OpenAI as the sole tenant, in what amounts to the largest single AI infrastructure commitment in history and a fundamental shift in NVIDIA's role from chip supplier to infrastructure financier.

LPS — The New Strategic Resource in AI Infrastructure

Land, power and shell — collectively abbreviated LPS in the data center industry — have quietly become the binding constraint on AI expansion. Advanced chips can be manufactured at scale; interconnect and memory supply chains can be expanded; but securing a large-scale site with long-term power agreements and physical infrastructure takes years and requires the kind of balance-sheet certainty that most frontier AI companies do not yet have. In a post on the NVIDIA blog, CEO Jensen Huang framed the PORTS-Pike commitment as applying the same supply-chain discipline NVIDIA has used to secure critical semiconductor resources — now turned toward the infrastructure layer beneath the chips.

The world's largest cloud providers — AWS, Google Cloud, Microsoft Azure — have the balance sheets and long-term credit profiles to secure LPS independently. They build and operate AI factories using NVIDIA accelerated computing under multi-decade infrastructure contracts. That model continues to represent most of NVIDIA's business. Frontier AI labs are structurally different: extraordinary compute demand, rapidly growing revenue, but growth rates that outpace the balance-sheet depth needed to independently finance infrastructure at this scale. NVIDIA is stepping in to provide the financing capacity those labs lack. The Goldman Sachs–NVIDIA $500 billion financing framework and the Apollo–BlackRock AI infrastructure vehicles established the financial architecture; PORTS-Pike is the first operating instantiation at this scale.

PORTS-Pike — 20 Years of NVIDIA Compute Upgrade Cycles

The PORTS-Pike Technology Campus, developed by SB Energy — a SoftBank-linked data center developer in which NVIDIA is now investing $1.5 billion — will host an initial 4.25 gigawatts of AI factory capacity, with NVIDIA retaining the option to extend to the full 3.75 gigawatts remaining, bringing total campus capacity to 8 gigawatts. According to the NVIDIA Newsroom announcement, each generation of NVIDIA AI factory systems deployed at PORTS-Pike could represent approximately 1.5 million NVIDIA GPUs — equivalent to roughly $150 billion to $200 billion in NVIDIA revenue per generation. Over a 20-year site commitment, PORTS-Pike is designed to support multiple hardware upgrade cycles, meaning the LPS infrastructure is effectively a permanent NVIDIA revenue platform that gets refreshed with each new GPU generation.

The compute platform specified for PORTS-Pike is NVIDIA's full-stack DSX AI factory system: GPUs, CPUs, networking and infrastructure software. The PORTS-Pike campus project is expected to create tens of thousands of Ohio jobs, pay for its own power infrastructure and invest hundreds of millions in the surrounding community, anchored by an initial $80 million community benefits fund — the kind of economic development commitment that has become standard political table stakes for large US data center projects in the post-CHIPS Act environment. The AMD $5 billion bond sale for AI data center expansion illustrates just how aggressively the entire semiconductor industry is now competing for infrastructure capacity at this layer.

OpenAI's $600 Billion Compute Commitment Through 2030

OpenAI will be the building operator and sole tenant at PORTS-Pike, running the AI factory using NVIDIA's DSX platform. More broadly, OpenAI has committed to approximately 12 gigawatts of NVIDIA AI infrastructure through 2030 across existing and planned deployments — with an opportunity to expand to approximately 16 gigawatts if NVIDIA extends the PORTS-Pike arrangement to full campus capacity. At those levels, the total opportunity represents roughly $600 billion of NVIDIA compute through 2030, a figure that makes OpenAI the single largest customer relationship in NVIDIA's history and arguably in the history of enterprise technology procurement. The scale reflects the OpenAI product expansion across enterprise and consumer AI, where compute availability — not algorithms or customer demand — has become the primary growth constraint.

NVIDIA Becomes an AI Infrastructure Financier

The PORTS-Pike arrangement marks a categorical expansion of NVIDIA's business model. For the past decade, NVIDIA has been a hardware and software company: it designs and sells chips, systems and developer platforms. PORTS-Pike makes NVIDIA a principal in AI infrastructure finance — providing credit support, taking equity stakes in site developers and guaranteeing exclusive compute deployment on long-lived campuses. The $105 billion credit facility is not a product sale; it is a structured infrastructure financing commitment of a scale typically associated with sovereign wealth funds and major infrastructure banks. As TechCrunch noted, NVIDIA's $1.5 billion equity stake in SB Energy ensures it is not merely a creditor but a co-owner of the infrastructure platform.

The strategic logic is straightforward: by securing LPS capacity exclusively for NVIDIA compute, NVIDIA locks out AMD, Intel, and any future competitors from the site for 20 years — turning infrastructure financing into a competitive moat. Each upgrade cycle refreshes the compute inside the same physical shell, generating another $150–200 billion revenue opportunity without requiring new site acquisition. It is the AI infrastructure equivalent of a long-term mining rights agreement: the land secures the resource; the resource generates compounding value over decades. As the Oracle–AWS AI infrastructure partnership and Ryanair's multi-year cloud commitments demonstrate, the enterprise AI investment wave has moved decisively from evaluation to long-term structural commitment — and NVIDIA is positioning itself as the permanent infrastructure layer beneath all of it.