The short version
● Revenue $115.8 billion for the quarter ended 25 October, up 103% year over year.
● Data Center $107.6 billion, up 110% - the segment that supplies almost every frontier model you use.
● Next quarter $130.0 billion plus or minus 3%, and that forecast assumes no Data Center compute revenue from China at all.
● China Interim CEO Colette Kress confirmed Nvidia is no longer shipping licensed Hopper data center products there, writing off a $2.4 billion Q3 line.
Nvidia reported third-quarter fiscal 2027 results on 17 November, the first quarter reported by Colette Kress as interim chief executive while remaining CFO. The headline numbers beat expectations. The number that matters to anyone buying AI compute is the forecast, because Nvidia has removed an entire market from it and still raised the figure.
The quarter
| Measure | Q3 FY2027 | Year over year |
|---|
| Total revenue | $115.8B | +103% |
| Data Center revenue | $107.6B | +110% |
| Edge Computing revenue | $8.2B | +42% |
| GAAP diluted EPS | $2.70 | — |
| Non-GAAP diluted EPS | $2.68 | — |
| GAAP gross margin | 74.8% | — |
| Non-GAAP gross margin | 74.9% | — |
Data Center is now 93% of the business. A 74.8% gross margin at $115.8 billion of revenue is the clearest available measure of how little pricing pressure Nvidia faces on AI accelerators, which is the same thing as saying inference costs are not falling because hardware got cheaper.
China is out of the forecast, not just out of the quarter
Kress told the earnings call that Nvidia is no longer shipping licensed Hopper data center products to China, and did not explain why. Bloomberg put the size of that business at $2.4 billion of licensed Hopper sales to Chinese customers in the third quarter.
The forward guidance goes further than pausing shipments. Nvidia's own release states plainly that it “is not assuming any Data Center compute revenue from China in its outlook”, and Kress confirmed the $130 billion figure contains no licensed Hopper data center product at all.
So the $130.0 billion number, plus or minus 3%, is a China-free forecast that is still 12% above a record quarter. Read against the $2.4 billion written out of it, that is Nvidia saying demand everywhere else more than covers the loss.
What this changes for anyone buying AI
- API prices are not about to fall from cheaper hardware. Margins near 75% with guidance rising mean the accelerator supply stays tight and expensive. Price cuts you see on model APIs are coming from efficiency and competition, not from Nvidia.
- Chinese model vendors have a harder compute path. Licensed Hopper leaving the market is a constraint on the labs behind several open-weight models now in production use.
- The guidance is the bull signal, not the quarter. Doubling revenue was expected. Raising the next quarter while deleting a market was not.
Kress summarised it on the call as momentum being real and Nvidia being an exceptional technology leader powered by the relentless innovation of its people. The guidance is the part that carries evidence.
Context
This lands the day before Anthropic's listing, in which Nvidia is an anchor investor, and a week after Nvidia opened GR00T 2 and 100,000 hours of robot training video. Its AI-factory monitoring product Minerva shipped last month. Everything else from today is in the 18 November roundup.
FAQ
How much revenue did Nvidia report in Q3 FY2027?
$115.8 billion for the quarter ended 25 October 2026, up 103% year over year. Data Center accounted for $107.6 billion of that, up 110%.
What is Nvidia's guidance for next quarter?
$130.0 billion plus or minus 3%. Nvidia states the outlook assumes no Data Center compute revenue from China, so that figure is a China-free number.
Has Nvidia stopped selling AI chips to China?
Interim CEO Colette Kress said on the earnings call that Nvidia is no longer shipping licensed Hopper data center products to China, and gave no reason. Bloomberg reported $2.4 billion of such sales in the third quarter.
Who is running Nvidia now?
Colette Kress, previously and still chief financial officer, is interim chief executive. This was her first earnings report in that role.
Does this mean AI API prices will go up?
Not directly, but it means they are unlikely to fall because of cheaper hardware. A 74.8% gross margin alongside rising guidance indicates accelerator supply remains tight, so the price cuts appearing on model APIs are coming from efficiency gains and competition between labs instead.