Nvidia says there’s no AI bubble. Investors agree — for now

• StockStory Top Pick NVDA -0.97%

On Wednesday afternoon, Jensen Huang looked straight at the market’s favorite anxiety and waved it off. “There’s been a lot of talk about an AI bubble. From our vantage point, we see something very different,” the Nvidia CEO told analysts on the quarter’s earnings call as he walked through $57 billion in quarterly revenue, a $65 billion forecast, and sold-out cloud GPUs. By the time he hung up, the stock was up over 6% in after-hours trading, analyst blasts were landing in inboxes with subject lines such as “AI Bubble Who?” with notes that they’d raised their earnings forecast (again), and even the people paid to worry about excess for a living sounded oddly relieved to let Huang win the argument — at least for one more quarter.

Or so it seemed.

Thursday afternoon came, and the mood snapped in half. Nvidia flipped from up big at the day’s open to down nearly 3% mid-afternoon, and the S&P 500 and Dow Jones Industrial Average fell with it. Nvidia may have put up numbers that made any AI bubble talk sound unserious, but the market is still behaving like something beneath the surface hasn’t been settled at all.

How the numbers landed

On paper, Nvidia delivered exactly what nervous AI investors needed: a monster quarter. Revenue jumped 62% year over year. Data-center sales — the heart of the AI boom — clocked in around $51.2 billion. Gross margins pushed toward 75%. Guidance landed $4 billion ahead of the Street’s consensus expectation. No analyst seemed surprised by yet another beat-and-raise quarter from a company that keeps delivering nothing but. Wedbush analyst Dan Ives called it another “monster quarter from the Godfather of AI,”(aka Huang) raising his target to $230 and arguing that the $500 billion in Blackwell and Rubin revenue through 2026 now looks conservative. BNP Paribas lifted its earnings forecasts and said fears of an “AI infrastructure bubble” look overblown.

Inside the call, the subtext was louder. Nvidia reaffirmed that its next two years of supply are effectively spoken for. Chief financial officer Colette Kress pointed to a 63% sequential jump in supply commitments — an absurd number in a normal hardware cycle, but a fitting one in an industry where every hyperscaler is fighting for floor space, power, and delivery windows. Analysts who cover the broader hardware stack signaled a clear read-through: TSMC stays full, component pricing stays tight, and the companies selling cables, optics, memory, and power infrastructure keep riding Nvidia’s slipstream.

The vibe across the sell-side was the same: Whatever you think about AI froth, the growth itself isn’t imaginary.

David Wagner, a portfolio manager and head of equities at Aptus Capital Advisors, said Nvidia’s earnings do “more for the market than [they do] for the stock,” because the past few weeks have been one long wobble — tech rolling over, the hype cycle sagging under its own weight. Nvidia’s quarter didn’t just beat numbers; it told him the AI capex boom is still barreling. “Let’s get real here,” he said. “There are good bubbles and bad bubbles. So could it be a pocket here? Absolutely. Is it anytime in the near term? Probably not.” To him, this is a macro story, not a micro one, and “this capex train is going to continue down the hill.”

“There’s so much worry out there in this market,” he said. “Nvidia keeps proving the naysayers wrong.”

So: Bubble who, right?

“It’s hard to say there’s a bubble,” said Motley Fool Senior Investment Analyst David Meier. But he also said nobody knows the true return on Nvidia-scaled AI investment yet. “Anybody who’s purchasing these chips needs to generate a return on that investment, right? And we’re not going to know. Right now, nobody is telling us how much incremental revenue they’re generating from AI.” Still, the demand is real, he said, and the investment dollars are as real as they come.

The trouble is that “AI bubble” means two very different things, and Nvidia may have only killed one of them. The version Huang swatted away — the idea that demand is peaking or that revenue is illusory — doesn’t survive a quarter like this. The other version, the one that keeps investors flinching at every macro wobble, is about whether the buildout can sustain this pace: the power loads, the hyperscaler arms race, the financing loops, the handful of companies holding up the entire trade.

By Thursday afternoon, all of the Nvidia-related thrill had somewhat faded, and the market largely gave back its early gains and slipped into the red. Leveraged ETFs that had lined up to ride Nvidia’s “pop the champagne” moment, as Wedbush put it, suddenly found themselves looking at volatility instead. Analysts from Direxion, which runs some of those funds, admitted as much, describing Nvidia’s earnings as “hyped to a level that suggests the broad market hinges on their success,” then turned around and reminded clients that its products can cut both ways if volatility stays high.

So even after a monster, amazing, sensational — or whatever other word for “huge” analysts want to use — quarter, AI worries remain. The same notes and analysts who praised Nvidia also noted that valuations across the AI complex are still rich, the circular AI economy is a big concern, the Fed still sounds hawkish, and the rest of the economy is still showing late-cycle jitters. Nvidia may have bought the AI trade more time, but it clearly hasn’t cured a nervous market’s broader anxiety.

What tomorrow depends on

Nvidia’s today is solid, but that doesn’t mean everyone stops worrying about tomorrow.

It can be easy to forget how much of Nvidia’s future still depends on things it doesn’t control. The clearest is electricity. Today’s grid isn’t built for what Nvidia, the hyperscalers, and the GPU-cloud providers are promising. Yvette Schmitter, co-founder & CEO of IT consulting firm Fusion Collective, noted that Nvidia is “selling futures contracts backed by electricity that won’t flow through transmission lines for another 3–7 years.” She’s not wrong. Between permitting, substations, and transmission fights, the infrastructure running behind this boom is moving at political speed, not semiconductor speed.

Kevin Cook, a senior stock strategist at Zacks Investment Research said, “You can’t keep scaling models without scaling electricity; that’s the choke point.” He added, “The compute demand curve is vertical. The grid isn’t.” Still, he said that Nvidia gives investors the clearest read on AI demand anywhere, and that “this stack is going to add hundreds of basis points to GDP.”

Huang isn’t wrong when he says he sees something very different from an AI bubble. The demand he describes is real — present tense. Hyperscalers are still increasing data-center capex. Networking and ancillary products are growing instead of dragging on results. China, once a major growth engine, remains largely off limits for leading-edge chips, and Nvidia’s sales everywhere else have roughly doubled without it. And the bubble talk lives in that gap: Everyone can see that Nvidia’s numbers are real, and everyone can see how fragile it is to have so much index-level performance riding on the same small circle of companies.

Even in the cleanest version of this world, where AI demand keeps rising and adoption broadens, the fragilities sit in plain sight. Second-tier GPU lessors are borrowing to keep up with hyperscalers. Utilities are scrambling to greenlight massive industrial loads. Regulators are eyeing concentration in both cloud markets and chip pipelines. Power-hungry deployment schedules are stacked on networks that weren’t designed for it. None of these issues contradict Nvidia’s numbers. They sharpen them. Demand is so strong, so compressed, and so front-loaded that the weakest links — not Nvidia — will dictate how long this cycle can run at its current pace.

And that’s why the market loves and fears Nvidia at the same time. It’s the star of the story, the strongest hand in the trade, and the signal that the entire index trades off. But it’s also the reminder that this era is built on infrastructure, cash flows, and physical constraints that don’t care about hype cycles. Huang can wave away bubble questions because his quarter gave him the right to. Investors can agree with him because the numbers let them. Neither can guarantee that the world around those numbers holds up.

Nvidia’s print didn’t deflate the bubble argument. It just pushed it further out. For now, that’s enough.