A petition signed by 197,000 people over the weekend, including Elon Musk, demanded the reinstatement of Nvidia's ousted CEO Jensen Huang, highlighting how deeply tied the AI boom has been to the chipmaker's corporate balance sheet. Huang did not merely champion hardware adoption; he used Nvidia's surging earnings to help underwrite the expansion of his buyers. The company reported net income of $64.9 billion for the third quarter on Tuesday, climbing from $59.7 billion in the prior quarter. Yet this profit engine now faces a decisive shift in financial discipline under interim leadership.

Chief Financial Officer and interim CEO Colette Kress has moved to rein in open-ended infrastructure commitments, establishing strict conditions on loan backstops that previously masked the capital risks of external AI deployments. Circular dealmaking—where the vendor finances compute buyers who immediately reinvest those funds into hardware orders—is giving way to a rigid credit posture. Market participants must now confront whether cloud builders can fund multi-billion-dollar facilities independently when vendor safety nets are contingent on third-party capital.

The Anatomy of the Ohio Backstop

Under Jensen Huang, Nvidia aggressively directed capital into AI startups and neocloud operators while increasingly acting as a debt guarantor for massive compute facilities. The primary showcase of this mechanism was an August agreement covering OpenAI's PORTS-Pike data center campus in Ohio, where Nvidia pledged a $105 billion loan backstop. In exchange for the credit support, OpenAI committed to using Nvidia's equipment exclusively across the site. However, internal friction over balance-sheet exposure had already taken hold before the leadership shakeup.

Reports indicate that Kress pushed back against Huang's expansive deal structures, scaling down the backstop guarantee for OpenAI's Ohio campus from an initial $250 billion to $105 billion. In its quarterly CFO commentary, Nvidia clarified that offering a guarantee does not mean the company supplies direct project financing.

"What we are indicating in our statement there is that is when our guarantee starts. So if all of that begins and unfortunately they are delayed or they have a challenge in terms of building that, our guarantee doesn’t exist anymore,"

Kress stated to analysts on Tuesday during the quarterly conference call, emphasizing that the guarantee activates only after the client secures the debt or equity needed to complete the structure and install Nvidia's hardware. If construction encounters delays or execution challenges, the financial umbrella vanishes entirely.

Physical Assets and Future Underwriting

When questioned by analysts regarding this aggressive posture, Kress maintained that the disclosure represented an explanation of existing terms rather than a formal policy shift. Even so, the specific operational triggers governing Nvidia's guarantees had never been disclosed publicly prior to this call. By making the limits explicit, Nvidia is effectively forcing cloud builders and debt syndicates to shoulder primary project execution risk.

Nvidia does not plan to exit infrastructure financing altogether, but Kress stated that the vendor will be far more selective with future projects. Rather than subsidizing speculative hardware buildouts, the company will prioritize physical fundamentals over speculative expansion schedules.

"The availability of land, power, shell is such a critical part of the whole structure in terms of building out those data centers. So they may need the assistance. That’s a great asset for anyone to hold. So we will look at that as probably one of our top priorities,"

Kress noted, adding that Nvidia is scrutinizing the operational profiles of prospective borrowers. Securing viable grid access, land rights, and real estate now dictates whether Nvidia considers providing credit support at all.

Vendor-subsidized compute demand has reached its structural limit. By conditioning billions of dollars in backstops on secured capital, physical shells, and flawless execution timelines, Nvidia is removing the artificial floor under debt-heavy data center expansions.

Side by side

Can the artificial intelligence boom sustain its pace without cheap balance-sheet insurance from the chipmaker? Colette Kress is dismantling the practice where the supplier absorbed the construction and debt exposures of its own buyers to drive sales. While in mid-August the planned infrastructure guarantee had already dropped from $250 billion to just under $120 billion, the current Ohio campus deal shrinks that support to $105 billion and conditions it on external funding. Unconditional vendor financing is over: Nvidia is shifting execution risk directly back onto infrastructure developers and private credit syndicates.

Our stories: Nvidia Cuts OpenAI Guarantee to $120B Amid CapEx Pushback (August 15).

AI InvestmentNVIDIAAI InfrastructureOpenAIAI Economics