Nvidia clarifies $500 billion data-center plan: no open-ended risk

Nvidia says, at least from CEO Jensen Huang’s public comments, that it isn’t taking on unlimited risk to bankroll customer purchases. The issue came into focus after the chipmaker’s five-year credit-default swap spread widened to a record 82 basis points in mid-2026, according to market pricing. That debt-market scare was driven by investor worries about “circular financing,” especially once reports tied Nvidia to a possible $250 billion guarantee for an OpenAI data-center project.

Huang has described a different approach. He says the goal is to draw in more than $500 billion in third-party capital from firms such as Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, using direct loans, equipment leasing and financing tied to usage. Under that setup, the main credit risk would sit with those investors, not with Nvidia.

If you’re watching Nvidia’s balance-sheet exposure, that distinction matters. Markets have calmed, backers see private capital as the practical route to fund new capacity, and Nvidia says its own role would be limited, at most, to residual-value support rather than broad guarantees on entire projects.

Still, the structure is something to keep watching through Huang’s public remarks and the market coverage around them. Skeptics argue that complicated financing can start to resemble bubble-era engineering if demand or revenue growth weakens, even as Nvidia keeps saying it doesn’t want to turn into the lender of last resort for its customers.