Nvidia Needs Loans to Sell Chips, Still Projects 70% Growth

Nvidia, the $3 trillion semiconductor giant currently valued somewhere between Costco and the GDP of France, announced it projects 70% revenue growth in its next fiscal year—a number that would make any Series B founder weep with envy. The stock duly rose in after-hours trading, because investors have apparently decided that projections from the world's most successful chip company deserve the same credibility as a pre-revenue AI startup's TAM analysis. What was not discussed at length: the company's simultaneous, almost sheepish admission that it is now providing "capital solutions" to customers unable to afford its products. Translation: Nvidia is lending money to people so they can buy Nvidia chips. This is not a feature. This is a warning light.

For context, Nvidia manufactures GPUs—the silicon that powers everything from video games to the large language model that will eventually replace your job. The company has gone from a respectable semiconductor play to the de facto tax collector of the AI boom, with customers ranging from hyperscalers like OpenAI and Google to startups burning through venture capital faster than they can say "unit economics." Nvidia's current dominance is so complete that buyers have essentially no alternative; the company's gross margins are somewhere in the stratosphere, and demand has been treated as infinite. Until, apparently, it wasn't.

The irony would be delicious if it weren't so perfectly on-brand for the current moment. Nvidia is effectively admitting that the entire AI gold rush—the rush that has made it a $3 trillion company—has exhausted available capital at the customer level. Startups and enterprises want the chips. They want them desperately. They just cannot pay for them without Nvidia's help. This is what happens when a company's growth story depends entirely on customers who are themselves burning through venture money and public markets capital at unsustainable rates.

The company's defense of its decision to provide "capital solutions" was notably muted in Axios's coverage. Nvidia has not, to date, positioned itself as a fintech provider or capital markets participant. It manufactures semiconductors. Yet here it stands, functionally operating as both vendor and lender—a model that works until it doesn't, which is usually around the time your primary customer cohort (AI startups) stops raising money. The language itself—"capital solutions"—is pure venture speak, the kind of euphemism that translates to "we discovered we couldn't hit our growth numbers without financing our own sales."

What could go wrong? Consider the straightforward risk: if the AI market softens, funding dries up, or customers decide they've bought enough GPUs for now, Nvidia's loan book becomes toxic and its revenue projections become fiction. A $3 trillion company with a meaningful portion of its growth dependent on lending to customers who may not be able to repay is not diversifying risk—it's consolidating it. The playbook here resembles less a semiconductor company and more a subprime auto lender during the pre-2008 years.

More broadly, this move signals that the AI economy has entered a new phase: the desperation phase. When the vendor is financing the buyer, when growth projections require the vendor to become a shadow bank, the market has stopped asking "can we build this?" and started asking "how long can we pretend demand is real?" Nvidia's 70% growth projection may well come true. But the fact that the company needs to bankroll its own customers to achieve it suggests the bellwether is not ringing as clearly as investors think.

In 2028, when Nvidia reports $120 billion in revenue, investors will celebrate. In 2032, when the company's loan book turns sour and customer repayment rates crater, they'll call it a black swan event.