Nvidia Posts a $57 Billion Quarter and Dares Skeptics to Call It a Bubble
Revenue up 62% and a $65 billion guide put fresh numbers behind Nvidia's argument that AI demand is real, just weeks after it hit $5 trillion.
Key takeaways
- Nvidia's fiscal Q3 revenue hit a record $57.0 billion, up 62% year over year, with data center at $51.2 billion.
- The company guided the next quarter to roughly $65 billion, a 65% jump from a year earlier.
- The print lands in the middle of a loud debate over whether AI spending is outrunning AI revenue.
Nvidia reported fiscal third-quarter revenue of $57.0 billion on November 19, a record and 62% above the same quarter a year earlier. It also told investors to expect about $65 billion next quarter. For a company that had crossed a $5 trillion market value only weeks before, the numbers were the answer to a question Wall Street kept asking: is anyone actually buying all this?
What Nvidia reported
The quarter ended October 26. Total revenue reached $57.0 billion and diluted earnings per share came in at $1.30, up 67%. The data center business, which sells the GPUs and networking gear that fill AI server racks, produced $51.2 billion, a 66% increase and roughly 90% of everything Nvidia sold.
That concentration is the story. Not long ago Nvidia was a gaming-graphics company with a side business in servers. In the fiscal second quarter, data center brought in more than $41 billion against $4.3 billion for gaming. A year before that, the gap was $26.3 billion to $2.9 billion. The third quarter widened it again.
The guidance matters as much as the quarter. A $65 billion forecast implies Nvidia expects to add another eight billion dollars of quarterly sales in a single step, on a base that was already enormous. Markets had been priced for exactly that kind of acceleration, so the debate shifted from whether growth would continue to whether it could keep beating expectations that were already sky-high. The company also used the report to show off new, cheaper-to-run AI chips, a signal that it intends to keep pressing the pace of its product roadmap while rivals and its own customers build alternatives.
The bubble argument, in plain terms
Skeptics do not dispute that Nvidia is selling chips. Their worry is the loop around it. Cloud providers and AI labs buy GPUs with money raised from investors or borrowed against future AI revenue, and Nvidia's customers are not yet earning enough from AI to obviously justify the bill. If the buyers pause, Nvidia's growth pauses with them.
Chief executive Jensen Huang rejected that framing in late October, telling Bloomberg Television he did not believe the industry was in an AI bubble. He pointed to demand for the Blackwell generation, which the company expects to ship in far greater volume than the roughly 4 million Hopper chips of the prior cycle. The quarterly results give that claim a receipt, but not a verdict.
Why it matters beyond one stock
Nvidia is the index-weight anchor of the AI trade. Its earnings function as a quarterly referendum on whether hundreds of billions of dollars in data center plans are sensible. A strong print props up suppliers, power and cooling vendors and the hyperscalers whose own capex plans depend on a continued supply of accelerators.
The flip side is dependence. When one supplier represents roughly nine dollars of every ten in a segment that drives the market, any stumble in orders, any export restriction or any customer going quiet gets amplified. That is the risk that bubble-watchers point to, and a record quarter does not remove it. It only postpones the test.
Huang told Bloomberg Television in October that he does not believe the industry is in an AI bubble.
The numbers behind the numbers
Earnings per share rose 67%, slightly faster than revenue, which suggests the company is not buying growth with margin. Shares had gained nearly 39% over the prior year by mid-January, a strong run but far smaller than the headline $5 trillion milestone implies, because much of the climb came in a handful of months. That context matters for the bubble debate: a stock that has already repriced for years of growth has less room for disappointment than the underlying business does.
It is also worth separating two questions that often get blurred. One is whether Nvidia's revenue is real, and the quarter answers that decisively. The other is whether the spending that generates it will be sustained at current levels, which depends on the finances of customers rather than Nvidia itself. Investors who own Nvidia are, in effect, making a bet on a handful of cloud and AI-lab budgets.
What to watch
- Whether Nvidia actually delivers the roughly $65 billion it guided to, and how much of it is data center.
- Hyperscaler capex plans for 2026, which are the demand side of this equation.
- Any change in export rules for China, which the guidance treats cautiously.
- Signs that AI customers are monetizing: cloud growth, software attach rates and lab revenue.
