Huang Says Nvidia's AI Investments Aren't Circular Financing
At a Goldman Sachs conference, Jensen Huang said Nvidia's money for AI labs brings back far more than it costs. Investors still ask how independent the demand it helps create really is.
TL;DR
- Nvidia CEO Jensen Huang denied that Nvidia's investments in its own customers amount to circular financing: "I put in one, and a hundred comes back" [^1].
- The figure was rhetorical, not a disclosed return, and as Nvidia's stock fell, investors weighed how intertwined AI demand and Nvidia's money had become [^1].
Background
Nvidia sells most of the chips used to train and run AI models, and it also invests in the companies that buy them. Critics call this circular: Nvidia funds customers who then spend the money on Nvidia hardware. Supplier financing is not new; aerospace, telecom and industrial companies have long helped customers buy expensive equipment [^1].
What happened
Speaking at Goldman Sachs' Communacopia + Technology Conference, Huang said: "It's not circular because we put a little bit of money in, and a lot of money comes back" [^1]. According to Invezz, Nvidia has invested about $50 billion in AI labs and agreed to provide guarantees of up to $105 billion linked to OpenAI's Ohio data-center project, where it is the exclusive chip supplier; a financing platform aims to mobilize roughly $500 billion of third-party capital [^1].
Nvidia's stock fell 2.37 percent to $218.36 that day, its third consecutive decline, though the Nasdaq also dropped as Treasury yields approached 5 percent [^1]. Views were split. "Not worried about it at all," David Wagner of Aptus Capital Advisors told The Washington Post, noting that institutions such as BlackRock, Blackstone, Apollo and Goldman Sachs supply much of the capital [^1]. Gary Tan of Allspring Global Investments told the Los Angeles Times that "capital is increasingly being used to fund future AI customers and infrastructure deployments" [^1].
Why it matters
What sets Nvidia apart is scale and interconnection: it can invest in an AI company, back financing for its data center, supply the GPUs inside it and benefit as it grows [^1]. That is a powerful flywheel as long as AI spending rises. If it slows, heavily leveraged cloud operators that need strong utilization to service debt would feel it first [^1]. Investor Dan Niles adds another caveat: under scarce supply, big customers "all double order," so even huge orders may overstate real demand [^1].
Practical example
Say Nvidia invests $1 billion in an AI startup, which then signs a data-center deal backed by Nvidia's guarantee and filled with Nvidia GPUs. Nvidia books the chip revenue, and the startup's growth increases its stake's value [^1]. The question investors are asking is what happens to all three if the startup's own revenue never catches up with its spending.