Nvidia beat everything — and the stock fell. The market stopped grading the quarter.
Record $81.6B revenue, data center up 92%, a $91B guide, an $80B buyback — and Nvidia's stock slid. When you're priced for perfection, the question isn't the quarter. It's the half-life of the AI buildout.
Nvidia just reported the kind of quarter that used to send a stock up 10%: record revenue of $81.6 billion, up 85% from a year ago, with data-center revenue of $75.2 billion, up 92%. It guided next quarter to $91 billion, comfortably above the ~$87 billion analysts expected. It authorized an $80 billion buyback and raised its dividend 25-fold. And the stock fell.
That gap — blowout numbers, falling price — is the whole story.
The numbers
For the quarter that ended April 26, 2026 — which Nvidia, whose fiscal year runs almost a year ahead of the calendar, labels Q1 of fiscal 2027 — it did $81.6B in revenue at a ~75% gross margin: a hardware company printing software margins. Data center, now effectively the entire business, nearly doubled year over year. The $91B guide says demand isn't cooling. Management paired it with an $80B repurchase authorization and a dividend hike from a token penny to 25 cents a share. One quiet line matters: the outlook assumes zero data-center compute revenue from China.
Why the stock fell anyway
When a company is priced for perfection, beating expectations isn't the bar — durability is.
The market has stopped grading Nvidia on the quarter and started grading it on the half-life of the AI buildout. Every figure here is a record; none of them answers the only question that moves the stock now: how many more quarters does this compound? A 92% data-center jump is dazzling and also a brutal comp to lap a year from now. The guide beat, but by less than the whisper number some had penciled in — and at Nvidia's valuation, "great, but" is a sell.
The $80B buyback is the most interesting tell, and it cuts both ways. It signals confidence and hands cash back to shareholders — but a buyback is also what a company does when it generates more cash than it can productively reinvest. The first faint sign of a hypergrowth company turning into a cash-return story is a giant repurchase and a real dividend. Nvidia just announced both.
My read
The results are staggering and slightly beside the point. Nvidia is no longer a bet on this quarter; it's a bet on how long its customers keep spending — and that's where it gets uncomfortable. The same AI capex funding Nvidia's records is the capex those customers are straining to justify. Meta just cut 8,000 jobs to pay its AI bill (see Meta is cutting 8,000 jobs to pay for AI); the AI labs are racing to go public partly to fund compute. Nvidia sells the shovels, and the $91B guide says the diggers are still buying. The risk isn't this quarter. It's the first quarter a major customer decides its AI spend isn't paying off — because Nvidia's revenue is the sum of everyone else's capex conviction.
The China line matters too. Guiding with zero China data-center revenue both removes a risk and quietly concedes a market. If it reverses, that's upside; if it hardens, it's the new baseline.
Watch two things, not the headline beat: whether the year-over-year growth rate starts to decelerate (it must, eventually), and whether the hyperscalers' capex commentary stays confident. Nvidia's own stock already told you what it's watching. The quarter was perfect. The market wanted to hear about the next ten.
