Nvidia guarantees up to $105 bln for OpenAI Ohio data center lease

Author: Maham Fatima Read Time: 4 minutes

NVDA -2.34%

On August 17, Nvidia (NASDAQ: NVDA) agreed to guarantee up to $105 billion to help OpenAI lease a massive data center in Pike County, Ohio, being developed by SoftBank-owned SB Energy. The site could reach 8 gigawatts of capacity, with the first 800 megawatts due online in 2028. OpenAI has signed on for a 20-year lease.

Nvidia will also invest $1.5 billion directly into SB Energy and will serve as the facility's exclusive chip supplier.

This marks the latest example of Nvidia financing the very infrastructure that runs its own chips—a practice that continues to expand even as it draws increasing scrutiny.

Bull Case: An Ecosystem Built to Last

The guarantee reflects Nvidia's strategy of using its balance sheet to secure long-lived sites where its chips can operate across multiple hardware generations. The numbers behind that bet keep climbing:

• Revenue growth is expected to accelerate toward 97% this quarter, up from 85% the previous quarter, as the company rolls out its new Rubin platform.
• CEO Jensen Huang has projected $1 trillion in combined revenue across 2026 and 2027, implying roughly $600 billion in fiscal 2028 alone.
• Demand is not limited to hyperscalers:
• SpaceX has announced it will build its AI infrastructure exclusively on Nvidia chips, targeting 10 gigawatts of compute by the end of 2027, with an estimated build-out cost of $150 billion to $250 billion.
• The four largest hyperscalers are expected to spend roughly $700 billion on AI chips this year.
• A potential OpenAI or Anthropic IPO could free up even more capital.

The Ohio guarantee, paired with the $500 billion financing consortium Nvidia formed with six major financial institutions the week prior, follows the same playbook: widen the pool of customers locked into its hardware so that no single client's slowdown can shake the business.

Bear Case: Money Flowing in a Circle

The primary criticism is circularity. Nvidia is effectively helping fund a customer's ability to buy Nvidia chips:

• First, through the $500 billion consortium with BlackRock, Blackstone, KKR, Apollo, Brookfield, and Goldman Sachs.
• Now, through a direct guarantee tied to OpenAI's Ohio lease.

Huang insists this is not circular financing, but the plan to securitize AI infrastructure into a tradable asset class invites comparisons to how mortgage pools were packaged before 2008.

Nvidia's exposure is not unlimited. It covers:

• A portion of lease and power payments
• A minimum-value commitment on the site

It does not cover the full $105 billion or all of OpenAI's obligations. However, if OpenAI defaults, Nvidia absorbs the difference between that floor value and whatever the site fetches if re-leased or sold.

Concentration Risk

• Data center sales made up 92% of Nvidia's revenue last quarter, meaning the business relies heavily on a small number of AI labs and hyperscalers continuing to spend at current levels.
• Nvidia's balance sheet carries its own volatility, including:
• A $21 billion stake in SpaceX
• Roughly $30 billion in Intel
• These positions added $15.9 billion in other income last quarter but could just as easily cut into earnings if those stocks decline.

Wall Street's Read on It

| Metric | Value | |---|---| | Hedge fund ownership | 275 funds (up from 264 last quarter) | | Short interest | 1.26% of float | | Forward P/E (as of Aug 18) | 25.64x |

Hedge fund ownership has increased, signaling growing institutional conviction. Short interest remains minimal, indicating almost no organized skepticism despite the circular-financing debate. The forward P/E multiple assumes continued fast growth without appearing stretched against current growth estimates.

Where This Leaves Investors

The $105 billion guarantee does not cost Nvidia much today, but it ties the company's fortunes more tightly to whether OpenAI and its peers can actually pay for the compute they are committing to years in advance.

For the growth case to hold, hyperscaler and AI-lab spending must continue to compound.