- Nvidia CFO Colette Kress pumped the brakes on a landmark deal that would have seen Nvidia financially backstop a data center project for OpenAI expected to cost more than $500 billion — after a WSJ report on the negotiations triggered negative investor reaction; Kress pushed for a scaled-back arrangement, and Nvidia ultimately agreed to backstop less than half the original amount discussed, revealing the internal tension between Jensen Huang’s expansive “bank of Nvidia” vision and the CFO’s obligation to manage investor perception of financial risk.
- Nvidia publicly maintains that its sales “flywheel” — the self-reinforcing cycle in which AI companies buy Nvidia chips, generate revenue, and reinvest in more compute — is more than robust enough to sustain its financial commitments to frontier AI labs and data center developers; on its Q2 earnings call, Kress characterized the investment deals as primarily supply commitments rather than circular financing, saying “the majority of what we have been helping folks with… is one very important thing called supply.”
- The circular financing concern remains the most legitimate structural risk in Nvidia’s investment strategy: if Nvidia financially backstops AI companies that then use those funds to buy Nvidia chips, the transactions inflate reported revenue without reflecting genuinely independent demand — a dynamic critics argue makes the AI buildout appear more economically robust than it is, and one that Kress’s intervention suggests the company is managing more carefully than Huang’s public statements imply.
- The scaled-back OpenAI deal lands alongside Nvidia’s blowout Q2 results ($96.2 billion revenue, 70% growth forecast, supply-constrained demand) — a combination that illustrates the paradox of Nvidia’s position: its earnings justify enormous financial confidence, yet its CFO is simultaneously working to limit the financial commitments that would be the most natural expression of that confidence, because investors are scrutinizing those commitments for signs of circular financing risk.
What Happened?
The WSJ reported that Nvidia was approaching a landmark deal to help OpenAI secure financing for a data center project expected to cost more than $500 billion. After the WSJ published an earlier report on the negotiations, CFO Colette Kress pushed to scale back Nvidia’s commitment, concerned about investor reaction to the scope of the potential financial guarantee. In the end, Nvidia agreed to backstop less than half the original amount discussed. The development comes the same week as Nvidia’s Q2 earnings, which showed $96.2 billion in quarterly revenue, a 70% sales growth forecast for fiscal 2028, and supply-constrained demand — the strongest set of results Nvidia has ever reported.
Why It Matters?
The Kress intervention is the clearest signal yet that there are internal limits to how far Nvidia will extend its “bank of Nvidia” strategy — the approach of providing financial backing to AI companies that are simultaneously Nvidia’s largest customers. Those limits are set not by Nvidia’s financial capacity (which is enormous given its cash generation) but by investor scrutiny of circular financing dynamics. The Q2 earnings call’s “supply commitment” framing is important: by characterizing investment deals as supply agreements rather than loans, Nvidia attempts to position the financial relationships as demand acceleration rather than demand creation. But the scaled-back OpenAI deal suggests the company knows those distinctions have limits in terms of investor credibility, and that Kress will continue to intervene when commitments become large enough to generate material market concern.
What’s Next?
Watch for details on the final structure of the OpenAI financing arrangement — the amount Nvidia actually backstops, the terms, and whether third-party lenders fill the gap left by Nvidia’s reduced commitment. More broadly, Nvidia’s total financial commitment exposure is the key balance sheet risk to monitor: the Q2 report described the commitments as primarily long-term purchase agreements with vendors, but investors will press for more specificity about the liability profile. If the AI buildout slows and some commitments come due, the structure of those obligations will matter enormously to Nvidia’s financial position.
Source: The Wall Street Journal
