Nvidia guarantees up to $105B for OpenAI’s Ohio data center

Nvidia is putting $1.5 billion into SB Energy to serve as the exclusive compute infrastructure provider for OpenAI’s PORTS-Pike campus in Ohio, a facility designed to eventually reach 8 IT-gigawatts of capacity. To put that number in perspective, most large data centers operate in the range of tens to hundreds of megawatts. This one would be measured in thousands.

The initial commitment covers 4.25 GW of AI infrastructure, with an option to scale to the full 8 GW. SB Energy, SoftBank’s US renewable energy arm, will build and operate the site under a 20-year lease with OpenAI, with capacity rolling online in phases starting in 2028.

The money behind the megawatts

The total financing architecture for the Ohio initiative is projected at roughly $500 billion. That figure includes a $250 billion financial guarantee backing OpenAI’s lease obligations and $350 billion earmarked for accelerator purchases, the GPUs and related hardware that will fill the facility.

Nvidia’s $1.5 billion investment in SB Energy is the anchor, but it’s hardly the only capital flowing into Appalachian Ohio. SB Energy and SoftBank are planning at least 10 GW of new power generation, sourced from natural gas and potentially nuclear, along with $4.2 billion in regional grid upgrades.

Japan is contributing too. The country has committed $33.3 billion in natural gas funding to support 9.2 GW of generation capacity for the project. When a single data center campus requires its own international energy diplomacy, you know the scale has shifted.

SB Energy and OpenAI have also pledged an $80 million community benefit fund for the surrounding region. The project is expected to generate approximately 35,000 construction jobs and 2,500 permanent positions, transforming a stretch of southeastern Ohio into something resembling an AI boomtown.

Why Nvidia wants to be the landlord’s landlord

This deal extends Nvidia’s influence well beyond selling chips. By investing directly in the energy and infrastructure layer, Nvidia is effectively guaranteeing demand for its own products while locking in a massive, long-duration customer relationship with OpenAI.

It’s a vertical integration play disguised as a partnership. Nvidia provides the compute hardware. SB Energy builds the physical campus. OpenAI signs a two-decade lease. And the whole thing runs on power generation that wouldn’t exist without the project itself. Each layer depends on the others, which creates sticky economics but also concentrated risk.

The $250 billion guarantee for OpenAI’s lease obligations is a particularly striking detail. OpenAI, still a company navigating its own complex transition from nonprofit to for-profit structure, is essentially getting its long-term real estate underwritten at a scale that rivals sovereign debt issuances. For Nvidia, the bet is straightforward: if AI workloads keep growing, this campus will be printing money for decades. If they plateau, the guarantees become a very expensive lesson in optimism.

The $350 billion accelerator purchase commitment, meanwhile, represents a pipeline of hardware demand that would be material even by Nvidia’s standards. The company’s data center revenue has been growing at triple-digit percentages in recent quarters, and a guaranteed multi-year procurement channel of this magnitude helps smooth out the boom-bust cycles that have historically plagued semiconductor companies.

What this means for the broader landscape

The PORTS-Pike project crystallizes a trend that’s been building for years: AI infrastructure is becoming indistinguishable from energy infrastructure. The 10 GW of new power generation alone would be enough to supply a mid-sized US city. The $4.2 billion grid investment signals that existing electrical infrastructure in the region simply cannot handle what’s coming without fundamental upgrades.

For Nvidia’s competitors, the deal raises the barrier to entry even higher. AMD, Intel, and a growing roster of custom chip startups are all competing for data center GPU share, but none of them are writing billion-dollar checks to energy companies or guaranteeing the lease payments of their largest customers. Nvidia is playing a different game entirely, one where chip performance is almost secondary to ecosystem control.

The phased 2028 timeline also matters. Three years is a long runway in AI, a field where the dominant architecture can shift in months. Nvidia is betting that its hardware will still be the platform of choice when the first racks power on. Given its current market position, that’s a reasonable assumption, but the history of technology is littered with companies that looked invincible three years before they weren’t.

For investors watching Nvidia’s stock, the deal adds both upside potential and tail risk. The guaranteed demand pipeline supports the kind of forward revenue visibility that Wall Street loves. But $500 billion in total project financing, spread across multiple entities with interlocking dependencies, creates the sort of systemic exposure that doesn’t show up in quarterly earnings until something breaks.

The energy commitments alone introduce variables that Nvidia has never had to manage at this scale, from natural gas price volatility to nuclear permitting timelines to grid interconnection delays. Any of those could push the project’s economics sideways without a single GPU failing to perform.

AI-linked crypto tokens like FET and RENDER have historically shown sensitivity to Nvidia’s trajectory, and a commitment of this magnitude could amplify that correlation in both directions. When the world’s most valuable chipmaker essentially merges its fortunes with OpenAI’s operational footprint and Appalachian Ohio’s electrical grid, the ripple effects will be felt well beyond semiconductor equities.

Disclosure: This article was edited by Estefano Gomez. For more information on how we create and review content, see our Editorial Policy.