Jensen Huang is gone, and suddenly everyone wants to know whether a single chief executive’s personal scandals can take down the American economy. The question, posed by a Brookings regulation expert with admirable deadpan, deserves a straighter answer than the media is giving it. No, one man’s sex life cannot crater GDP. But the fact that we’re even asking reveals something far uglier about where we’ve landed: an economy so dependent on a single company, a single narrative, and a single leather-jacketed prophet that his resignation moves the needle at all.
Let’s dispense with the obvious first. Nvidia’s stock fell on the news; the broader market barely shrugged. That’s the market telling you, correctly, that CEO identity is not the marginal variable here. What should worry you isn’t the departure — it’s what the calm response papers over. Nvidia has become the load-bearing wall of this expansion. Its chips fill the data centers, its balance sheet finances the customers who buy those chips, and its market capitalization anchors the index funds that anchor retirement accounts that anchor consumer confidence that anchors the whole fragile edifice. When the economy’s “primary driver of growth” — I’d say only driver, and I’m hardly alone — is a single semiconductor company with circular revenue arrangements, you don’t have an AI revolution. You have a monoculture.
We’ve seen this film before, and the ending never changes. Railroads in the 1870s. Radio in the 1920s. Cisco in 1999, when the “internet’s plumb” was measured by a company that, like Nvidia today, sold the infrastructure and was celebrated for it. The pattern is always identical: a genuine technological transformation arrives, capital floods in beyond any rational measure of eventual returns, a handful of names become the market, and then reality delivers its invoice. The technology survives; the valuations don’t. Nobody remembers that, of course, because the people writing about it now were in elementary school in 2000.
The circularity deserves particular scorn. Nvidia invests in the very data center developers and AI startups that purchase its GPUs — OpenAI’s gargantuan commitments being the crowning example. If I lent you money to buy my product and booked your purchase as revenue, my accountant would be having a very bad week. Yet Wall Street, which spent two decades lecturing us about governance and cash flow quality, has waved this through because the numbers go up. Depreciation schedules for GPUs are now a matter of theological debate: hyperscalers are quietly extending useful lives of servers that may be obsolete in three years, all to flatter current earnings. When accounting assumptions become the load-bearing structure of a bull market, you’re no longer investing — you’re participating in a confidence game with extra steps.
And here’s what the Huang saga actually exposes. The AI buildout was never just about chips; it was about a narrative, and Huang was the narrative’s chief custodian. He was rhetorically irreplaceable — the rare executive who could stand on a stage, invert a hypothetical power grid, and convince institutional allocators that spending hundreds of billions on depreciation-heavy infrastructure was a value proposition rather than a capital destruction exercise. Markets run on stories, and this market ran on one storyteller. Now the storyteller is gone, and his successors will be administrators, not evangelists. Narratives don’t transfer smoothly. Ask anyone who watched General Electric after Jack Welch, or Apple in the interregnum before Jobs returned — except this time, the departing figure isn’t leaving a company; he’s leaving the economy’s central myth.
The macro arithmetic is where the complacency becomes dangerous. The AI capex boom has contributed an outsized share of GDP growth over the past two years. Strip out data center construction, chip orders, and the second-order spending they generate, and the “resilient” American economy looks considerably less resilient — a consumer strained by cumulative inflation, a labor market cooling beneath the headline averages, and a federal deficit running at levels that historically appear only during recessions or wars. We are, in effect, running a fiscal and monetary experiment in which private capital expenditure on speculative infrastructure substitutes for broad-based economic strength. That works magnificently until the capex slows, and capex always slows, because no buildout is infinite. Railroads stopped being laid. Fiber stopped being dug. The question is never whether, only when and how violently.
The Fed, meanwhile, sits in its customary position of fighting the last war. Having spent the tightening cycle insisting AI-driven equity wealth effects didn’t matter for policy, officials now face the opposite problem: an economy whose growth engine depends on valuations the central bank cannot control and would privately admit it cannot value. If AI capex falters, the Fed will cut — but you cannot ease your way out of a malinvestment problem. Cutting rates in 2001 didn’t make fiber optic cable valuable. It just cushioned the fall on the way to nowhere.
None of this means the AI technology is fraudulent. It isn’t — the productivity gains in code generation, drug discovery, and automation are real and will compound for decades. That’s precisely the trap. Genuine transformation and catastrophic overinvestment are not mutually exclusive; they’re historically correlated. The internet was real, and pets.com still went to zero. The semiconductor buildout of the 2000 bust left us with cheap capacity that eventually powered the smartphone era — but only after trillions in paper wealth evaporated and Cisco spent a decade digging out from inventory write-offs. The winners of the eventual AI economy may well be customers buying cheap, distressed compute, not today’s owners of it.
Which brings us back to Huang. His resignation is not the catalyst for a reckoning; it’s merely a mirror held up to one. The stock barely moved because the market understands the machine no longer requires the man. But that’s colder comfort than it sounds. An economy that treats a single vendor as its growth engine, a single narrative as its justification, and a leadership cult as its risk management framework is not resilient — it’s brittle, wearing resilience as a costume. The real test won’t come from a CEO scandal. It will come from the first quarter when hyperscalers collectively blink, capex guidance slips, and everyone discovers simultaneously that the revenue behind the revenue was always the same revenue.
When that day arrives, no leather jacket will save anyone.
