Nvidia May Guarantee $250B to Back OpenAI’s Ohio Data Centre

Nvidia is in talks to provide around $250 billion in financing guarantees for OpenAI as part of a large data centre project in southern Ohio, according to a Wall Street Journal report published on Sunday, July 27, 2026. The site, developed by SoftBank's energy subsidiary, is a 10-gigawatt project whose power supply is controlled by the US government and separately funded by Japan under a recent trade deal. On the same weekend, Nvidia and South Korea's SK Group announced a distinct $500 billion AI initiative covering data centres and next-generation memory chips.

What happened

Detail	Fact
Nvidia financing guarantee (reported)	~$250 billion
Ohio site capacity	10 gigawatts
Site developer	SoftBank’s energy subsidiary
Power oversight	US government; US Commerce Secretary Howard Lutnick
Separate Nvidia-SK Group initiative	$500 billion+
SK Telecom planned data centre	2 gigawatts, powered by Nvidia Vera Rubin chips and SK Hynix HBM4
First SK Telecom facility due	2027

The Wall Street Journal reported on Sunday that Nvidia is in advanced discussions to act as a financial guarantor for OpenAI’s lease of a 10-gigawatt data centre in southern Ohio. Nvidia, OpenAI, and the US Commerce Department had not responded to comment requests by the time of publication, and Reuters noted it could not independently verify the report.

OpenAI is said to be among the most interested parties in the project. Other major players, including Anthropic, Microsoft, and Google, have also reportedly held conversations with Commerce Secretary Howard Lutnick, who is involved in deciding which companies gain access to the government-controlled power supply. The electricity for the Ohio site is funded separately by Japan as part of a recent US-Japan trade deal, adding a geopolitical layer to what is ostensibly a commercial real estate transaction.

The Nvidia-SK Group deal is a separate, bigger story

On Friday, Nvidia and South Korea’s SK Group announced a joint initiative valued at more than $500 billion. The initiative covers large-scale AI data centres and next-generation memory supply chains. Key components include:

• A long-term partnership between Nvidia and SK Hynix to secure next-generation memory supply for Nvidia and jointly develop high-bandwidth memory (HBM) for AI training, AI agents, and physical AI applications.
• SK Telecom’s plan to build a 2-gigawatt AI data centre running on Nvidia’s Vera Rubin chips and SK Hynix’s HBM4 memory.
• A separate plan involving Nvidia, Naver, and Brookfield to expand Naver’s existing AI data centre in South Korea.

The first SK Telecom facility is scheduled to come online in 2027, according to Nvidia.

Why it matters

These numbers represent a step-change in AI infrastructure spending. A single 10-gigawatt data centre is roughly equivalent to the entire generating capacity of a mid-sized country’s grid. If Nvidia acts as guarantor on a $250 billion lease, it shifts from being a chip supplier to a financial stakeholder in the AI build-out, something that affects every company buying Nvidia hardware.

The involvement of US government power allocation and Japanese trade-deal funding also signals that AI infrastructure is now a matter of national industrial policy, not just corporate capital expenditure. Who gets access to compute, and at what scale, is being decided partly in government offices, not just boardrooms.

For businesses evaluating AI integration today, the practical implication is that the largest cloud and AI providers are locking in computing capacity years in advance. That creates both concentration risk (fewer, larger players controlling resources) and long-term pricing pressure on inference and API services.

The SK Group deal reinforces a supply-chain story worth watching: HBM4 memory is the limiting factor in training frontier AI models, and a locked-in supply agreement between Nvidia and SK Hynix tightens that market further. Coverage of related AI infrastructure moves is collected in our AI news section.

Our take

The Ohio project, if it materialises at this scale, would be one of the largest single capital commitments in the history of corporate America. But the source is a single WSJ report citing unnamed people, and none of the named companies confirmed it. Treat the specific $250 billion figure as unverified for now.

What is confirmed is the SK Group deal, which is already announced and carries concrete specs: 2 gigawatts, Vera Rubin chips, HBM4 memory, online in 2027. That timeline matters for anyone planning AI infrastructure purchases in the next 18 months, because it shows where Nvidia’s highest-priority chip allocations are going.

For most businesses, the practical takeaway is simpler: the gap between frontier AI labs and everyone else is being measured in gigawatts now. Your AI strategy does not need its own data centre, but it does need a clear plan for which APIs and platforms you depend on, and what happens if pricing shifts as these infrastructure deals reshape the market.

What to do about it

1. Audit which AI services your business relies on and whether they are tied to OpenAI, Google, or Microsoft infrastructure, given that all three are competing for the same Ohio capacity.
2. Watch HBM4 memory availability news. SK Hynix supply constraints directly affect when Nvidia can ship next-generation GPUs, which affects cloud inference pricing.
3. If you are evaluating a larger AI or automation build, get quotes and commitments now rather than waiting for 2027 hardware cycles to stabilise.
4. Consider whether your current AI tooling could be diversified across providers to reduce concentration risk as infrastructure ownership consolidates.

The compute arms race is real, but for most operators the action item is vendor diversification, not panic.

Source: Bing News · OpenAI