On this page · 8 sections
- When NVIDIA reports, and what will be published
- The leadership change, from NVIDIA's own releases
- The $108 billion guide against the record
- Four signals that matter more to GPU buyers than revenue
- What each signal means for your next GPU decision
- For buyers in India and the UK
- How eCorpIT can help
- References
Summary. NVIDIA reports results for the third quarter of fiscal 2027, which ended on 25 October 2026, on Tuesday 17 November 2026 at about 1:20 pm PT, with the conference call at 2 pm PT (5 pm ET, 10 pm in London and 3:30 am on 18 November in India). It is the first report since Jensen Huang resigned on 12 November and Colette Kress, the chief financial officer, became interim CEO while keeping the CFO role. NVIDIA said the results would come out as previously scheduled. The guide is revenue of $108.0 billion, plus or minus 2%, so a range of $105.84 billion to $110.16 billion, with gross margin of 74.0% and no data centre compute revenue from China assumed. The baseline is the second quarter: revenue of $96.2 billion, up 106% year on year, and data centre revenue of $89.0 billion, up 117%. NVIDIA beat the midpoint of its own revenue guide by 4.6% to 5.7% in each of the last four quarters. For GPU buyers the headline number matters less than four signals: the margin path as memory costs rise, how much Vera Rubin ships, supply limits into 2027, and what the interim leadership says about the roadmap.
About this article. This preview is for CTOs, infrastructure leads and procurement teams in the US, UK and India who buy GPU capacity, either as cloud instances or as racks. We checked every figure on 15 November 2026 against NVIDIA's own press releases and its second-quarter CFO commentary, and used one secondary report of the August earnings call, which we attribute. No third-quarter result is stated here, because none has been published. This is not investment advice, and no vendor paid for inclusion. Researched with AI assistance and checked against the primary sources listed in the references.
When NVIDIA reports, and what will be published
NVIDIA set the date on 28 October. Three things arrive on 17 November, in this order:
- The press release, at about 1:20 pm PT (4:20 pm ET, 9:20 pm GMT, 2:50 am IST on 18 November).
- Written CFO commentary from Kress, posted on NVIDIA's investor site straight after the release. In August this document carried the segment split, inventory, purchase commitments and the outlook, so it is the most useful file for buyers.
- The conference call at 2 pm PT, webcast in listen-only mode, with questions limited to financial analysts and institutional investors. A replay stays up until the fourth-quarter call.
The quarter covers 27 July to 25 October 2026. Anything a supplier tells you about November pricing or allocation sits outside these numbers and will show up, if at all, in the fourth-quarter outlook.
The leadership change, from NVIDIA's own releases
We keep this section to what NVIDIA has published.
- On 8 November the board said it was opening an independent investigation, overseen by a committee of independent directors with outside counsel, after media reports that Huang had been the target of a foreign intelligence effort. The board said it had no reason to believe national security or NVIDIA secrets were compromised.
- On 10 November the independent directors said they had retained Joon H. Kim, a former acting US Attorney for the Southern District of New York, and his firm Wachtell, Lipton, Rosen and Katz to run that investigation.
- On 12 November NVIDIA announced that Huang had resigned as president, chief executive and director with immediate effect. Kress, CFO since September 2013, became interim CEO and remains CFO until a permanent successor is named. The board will consider internal and external candidates and appointed Harvey C. Jones, a former Synopsys chief executive and executive chairman, as its first chair.
The board said it expects operations, customer commitments and the product roadmap to continue without interruption. For buyers the practical question on 17 November is whether management repeats that on the call with specifics: Vera Rubin shipment timing, allocation to named cloud partners, and any change to the cadence beyond Rubin. Our GTC 2027 briefing on Rubin Ultra and Kyber covers the roadmap that the next CEO will inherit.
The $108 billion guide against the record
The outlook NVIDIA gave on 26 August sets the bar. The figures below come from the second-quarter release and CFO commentary.
| Metric | Q2 FY27 actual | Q3 FY27 guide | What the guide implies |
|---|---|---|---|
| Revenue | $96.2 billion | $108.0 billion, plus or minus 2% | $105.84 billion to $110.16 billion; about 12% sequential growth at the midpoint |
| Gross margin (GAAP and non-GAAP) | 75.0% | 74.0%, plus or minus 0.5 points | One point below the second quarter |
| Operating expenses | $8.4 billion GAAP, $8.2 billion non-GAAP | About $9.2 billion GAAP, $9.0 billion non-GAAP | Q2 rises came from compute infrastructure and pay |
| China data centre compute | Hopper shipments under 1% of data centre revenue | None assumed | Any China sale is upside, not in the plan |
At the midpoint, $108.0 billion would be 89% more than the $57.0 billion NVIDIA reported for the same quarter a year earlier. The arithmetic: 108.0 divided by 57.0 is 1.89. Against the second quarter, 108.0 divided by 96.2 is 1.12.
NVIDIA has finished above the midpoint of its own revenue guide every quarter for the past year. The figures in the next table come from NVIDIA's results releases for each quarter, using the guide from each prior release.
| Quarter | Guide midpoint | Reported revenue | Beat |
|---|---|---|---|
| Q3 FY26 (to October 2025) | $54.0 billion | $57.0 billion | 5.6% |
| Q4 FY26 (to January 2026) | $65.0 billion | $68.1 billion | 4.8% |
| Q1 FY27 (to April 2026) | $78.0 billion | $81.6 billion | 4.6% |
| Q2 FY27 (to July 2026) | $91.0 billion | $96.2 billion | 5.7% |
If the same pattern held, revenue would land between about $113.0 billion (108.0 × 1.046) and $114.2 billion (108.0 × 1.057). That is arithmetic on past behaviour, not a forecast: a change of chief executive, a product transition and memory costs all make this quarter less typical than the four before it.
Four signals that matter more to GPU buyers than revenue
1. Gross margin and memory costs
The guide of 74.0% is a point below the second quarter's 75.0%. On the August call, according to Zacks' report of it, Kress said margin would bottom at 71% to 72% in the fourth quarter and settle at 72% to 73% in fiscal 2028, blamed memory prices that were higher than expected, and said pricing actions would take effect in the fiscal first quarter, which on NVIDIA's calendar starts in early 2027. NVIDIA's purchase commitments rose from $119 billion to $279 billion in one quarter, mainly for memory.
For buyers this is the most direct signal. If the third-quarter margin lands near 74% and the fourth-quarter guide repeats the 71% to 72% trough, expect system and cloud prices to carry higher memory costs into 2027. Our analysis of HBM4E supply and GPU pricing in 2027 explains why memory sets the floor.
2. How much Vera Rubin ships
NVIDIA said on 31 May that Vera Rubin was ramping into full production and that production shipments would begin this autumn. On 26 August it said racks were running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius. The CFO commentary said inventory rose from $25.8 billion to $31.6 billion, up 22%, to prepare for the introduction of Vera Rubin in the third quarter, and that Blackwell was still the vast majority of revenue.
Watch whether the new commentary puts any figure or share on Rubin revenue, whether inventory falls as racks ship, and which clouds are named as live. Those three facts tell you whether Rubin instances will be broadly rentable in early 2027 or still allocated to a short list of labs. Our guide to Vera Rubin cloud instances covers what to ask a provider.
3. Supply limits and who gets capacity
Zacks reported that Kress guided fiscal 2028 revenue growth of about 70% and called that outlook supply constrained, with customer forecasts pointing to demand roughly doubling. On the same call, Zacks reported, she said the AI clouds, industrial and enterprise segment (ACIE) would make up roughly half of data centre revenue. In the second quarter that segment was $40.3 billion, or 45% of data centre revenue, and grew 25% in the quarter against 13% for hyperscale.
The split matters if you rent rather than build. Growth in ACIE means more capacity flowing through specialist GPU clouds and sovereign projects, not only the largest public clouds. Days sales outstanding also rose from 45 to 60, which NVIDIA put down to extended payment terms on large, multi-quarter agreements with certain investment-grade customers. Our reading: the best terms go to large, creditworthy buyers with long commitments, not to a small buyer placing one order.
4. China, financing and the fourth-quarter outlook
The guide assumes no data centre compute revenue from China, and Hopper shipments there were under 1% of data centre revenue in the second quarter. NVIDIA also announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at mobilising more than $500 billion of third-party capital for AI infrastructure over time, subject to definitive agreements. Look for progress on those platforms in the release: financing that reaches mid-sized clouds can widen the pool of capacity buyers can rent.
What each signal means for your next GPU decision
| Signal on 17 November | If it comes in strong | If it comes in weak |
|---|---|---|
| Gross margin and Q4 margin guide | Memory pressure contained; less risk of a 2027 price rise | Expect higher rack and instance prices from early 2027 |
| Named Rubin revenue or live clouds | Plan Rubin capacity for the first half of 2027 | Keep Blackwell Ultra in the plan for longer |
| Inventory movement | Racks are shipping; allocation should ease | Rubin volume still building; reserve early |
| Fiscal 2028 growth and supply comments | Supply loosening; shorter reservations may work | Supply tight; lock in capacity blocks now |
| Leadership and roadmap statements | Roadmap unchanged; keep current plans | Ask suppliers for written delivery dates |
None of this changes what a well-run GPU budget looks like. Price per useful token, not price per GPU hour, is the measure that survives a product transition, as our FinOps note on GPU spend explains.
For buyers in India and the UK
At the ECB reference rates of 13 November 2026, the $108.0 billion guide equals about ₹10.18 lakh crore or £81.65 billion. We crossed the euro rates for the US dollar, the rupee and sterling, which gives ₹94.24 and £0.76 per dollar, rounded to two decimals; the totals use the unrounded cross rates. NVIDIA prices in dollars, so for Indian buyers a weaker rupee adds to any memory-led price rise. Cloud capacity in India is quoted in rupees by domestic providers and in dollars by the hyperscalers; our comparison of GPU cloud rental prices in India shows the spread. For UK teams the timing matters most: the release lands at 9:20 pm GMT on 17 November, so plan any supplier conversations for the morning of 18 November.
How eCorpIT can help
eCorpIT is a Gurugram-based technology organisation, founded in 2021, assessed at CMMI Level 5 and MSME certified, with senior-led engineering teams working across AWS, Microsoft and Google platforms. Our artificial intelligence and machine learning services team helps you size GPU capacity, compare cloud and on-premises options and measure cost per token before you commit. Talk to us at /contact-us/.
Last updated: 15 November 2026.
