AI-generated analysis, not investment advice. The articles are written by AI, edited, and checked against company filings — but the judgements are opinions and the figures go stale. How this is made · Terms
◆ What the Market Isn't Pricing In
Nvidia (NVDA) — the variant view
Eleven times forward earnings on 103% growth is not the price of euphoria but the price of disbelief — and the market now has three fresh reasons for it.
📈 NVDA valuation, revenue & earnings — P/E, P/S, revenue, EPS →Everyone knows Nvidia is expensive. Look at what people actually pay, though, and the truth is stranger: measured against its own history and its own growth, Nvidia has rarely been cheaper. In November 2026 it reported revenue up a hundred and three percent to $115.8 billion, data-center sales up a hundred and ten, a gross margin of nearly seventy-five percent, and guided the very next quarter to $130 billion1 — and the shares traded at roughly seventeen times trailing earnings and about eleven times the year ahead2. A few years earlier, growing more slowly, the same stock fetched seventy and eighty times. The multiple has collapsed even as the business exploded, because earnings have compounded faster than the price. So the fashionable warning — that Nvidia is 'priced for perfection' — is very nearly backwards. What follows is a note on what a de-rated multiple on a hypergrowth business may be missing, in both directions, and it is not a price forecast.
The multi-year record is almost comic in its asymmetry. Nvidia's earnings per share went from seventeen cents in fiscal 2023 — a gaming-and-crypto trough that produced that very hundred-and-twenty-times multiple — to $1.19, then $2.94, then $4.90 by fiscal 2026, and $9.31 on the trailing year now3: more than fifty-fold in under four years. Revenue over the same span ran $27 billion, then $61 billion, $131 billion, and $216 billion4. The share price rose a great deal too, of course — but nowhere near thirty-fold, which is the whole point. When the denominator grows faster than the numerator, a 'high' multiple quietly becomes a low one, and that is precisely what happened here.
A forward multiple of eleven on a company growing a hundred and three percent is not the price of euphoria; it is the price of disbelief. The market is not extrapolating the boom — it is quietly betting the boom rolls over. Embedded in that number is the assumption of a cyclical peak: that the handful of hyperscalers pouring hundreds of billions into AI infrastructure will one day decide they have built ahead of demand, that orders will fall as fast as they rose, and that a maker of picks and shovels feels the end of a gold rush more sharply than anyone. That is a coherent fear, and I would not wave it away. The whole arrangement is reflexive in a way that should sober any owner: Nvidia's revenue is its customers' capital budget, and capital budgets are the first thing cut when returns disappoint.
If you want a lesson in how quickly a slice of this business can vanish, the same quarter supplied one — though not through the cycle. Nvidia's data-center sales to China went from billions a quarter to essentially zero under export controls5, and the company simply guided the line out of its forecast altogether. That is a different mechanism from a spending bust — it is policy, not demand — but it teaches the same humility: a large, profitable market can close by decree, overnight, with no appeal. An owner who assumes the whole world stays open to Nvidia's best chips is assuming something governments have already shown they will not always allow.
The other thing the cheap multiple only partly credits is that Nvidia's richest customers are also its most motivated future rivals, and in 2026 that stopped being theoretical. Google began shipping its own TPUs into other companies' data centers and advertising a generation it said offered far better performance per dollar6 — a customer's captive silicon turning into a merchant product aimed at Nvidia's own market. CUDA and the pace of Nvidia's roadmap still hold the frontier, and there is a real counter-signal I take seriously: over the past year the hyperscalers grew their data-center purchases 71% while every other kind of buyer grew 195%7, the rest coming from sovereign, enterprise, and AI-cloud buyers who cannot build their own. But the direction of the customer-chip threat runs one way, and it presses hardest on exactly the high-volume, commoditized work where Nvidia's margins are fattest.
And yet the very skepticism that makes the multiple cheap may be underrating the durability of the demand. Two things argue the build-out is more than a one-time frenzy. The first is inference — running AI models in production rather than training them — which is recurring in a way a capacity build-out is not, and grows with usage rather than with a single wave of orders. The second is that widening base of buyers beyond the reflexive hyperscaler capex, into enterprises, nations, and industries only beginning to spend. If artificial intelligence is a lasting shift on the order of electricity rather than a passing mania, then a mid-twenties forward multiple on the one company selling the essential tool is not expensive at all — it is a market refusing, out of scar tissue, to believe its own arithmetic.
So which is it — a cheap multiple correctly warning of a peak, or a cheap multiple wrong out of fear? I do not claim to know, and anyone who says they do is selling something. But the valuation history is the tell, and it points the opposite way from the cliché. The market has already placed its cautious bet; the euphoria has largely been wrung out of the price even as it lingers in the headlines. That means the real unpriced variables are not 'is the crowd too excited' but the ones the multiple handles clumsily. Two are long-standing: whether policy closes more markets the way it closed China, and whether the customers' own chips claim the commodity middle while Nvidia keeps the frontier. Two arrived in November. The first is the margin, guided down from 74.8% to between 71.0% and 72.0% in a single quarter on the Rubin ramp and the price of memory8 — suppliers taking a visible share of the best hardware margin in the industry. The second is that the company has no permanent chief executive: Jensen Huang resigned after thirty-three years, effective immediately and including his board seat, the chief financial officer is holding both jobs, and an independent committee of the board is conducting an investigation Nvidia has not described9. A business compounding at this rate is also one where a year of hesitation is expensive. Resolve all four the durable way and today's multiple looks timid; resolve them badly and it looks generous. The gap worth watching is not between Nvidia and perfection — the market long ago stopped paying for perfection — but between a business still compounding at a pace almost no one believes can last, and a price that has quietly decided it cannot.
- ReportedQ3 FY2027: revenue up 103% to $115.8 billion, Data Center up 110%, gross margin 74.8%, Q4 guided to $130.0 billion.NVIDIA, Q3 FY2027 financial results (Form 8-K, exhibit 99.1) — revenue of $115.8 billion for the quarter ended 25 October 2026, up 20% sequentially and 103% year on year; Data Center revenue of $107.6 billion, up 21% sequentially and 110% year on year; Edge Computing $8.2 billion, up 42%; GAAP gross margin 74.8% and non-GAAP 74.9%; GAAP operating expenses $9,156M and non-GAAP $8,841M; GAAP operating income $77.5 billion; GAAP net income $65,287M, up 105%; GAAP diluted EPS $2.70 against non-GAAP $2.68; nine-month revenue $293.7 billion against $147.8 billion; a record $28.1 billion returned to shareholders in the quarter; the board approved a $150 billion increase to the repurchase authorisation in September 2026, leaving about $227.3 billion; next quarterly dividend $0.25 a share. Outlook for Q4 FY2027, which includes an additional week: revenue of $130.0 billion plus or minus 3%, with NVIDIA not assuming any Data Center compute revenue from China; GAAP and non-GAAP gross margins of 71.0% to 72.0%; GAAP operating expenses of approximately $11.0 billion and non-GAAP approximately $10.7 billion; full-year tax rate 16.0% to 18.0%. The forward-looking statements name 'NVIDIA's leadership transition and the search for a new chief executive officer' and 'the ongoing investigation by an independent committee of NVIDIA's board of directors'. Colette Kress, interim CEO and executive vice president and chief financial officer: 'Our momentum is real... Vera Rubin is off to an explosive start, our roadmap is on track and our ambition has never been greater.' — Q3 FY2027 (quarter ended 25 October 2026) · publ. 2026-11-17 · source ↗
- Third-party estimateRoughly seventeen times trailing and about eleven times forward earnings at the close before the Q3 FY2027 report.Market data for NVDA at the close of 17 November 2026, before the Q3 FY2027 release — share price $155.46, market capitalisation about $3.75 trillion on 24.15 billion shares, trailing P/E 16.70 and forward P/E 11.01, trailing revenue $361.81 billion and trailing net income $226.26 billion, trailing diluted EPS $9.31, 52-week range $148.80 to $243.37 (so about 36% below the high and about 4% above the low) — 17 November 2026 · publ. 2026-11-18 · source ↗
- ReportedDiluted EPS $0.17 (FY2023) → $1.19 → $2.94 → $4.90 (FY2026), split-adjusted.NVIDIA Forms 10-K (FY2023–FY2026) — revenue & diluted EPS series — Fiscal years 2023–2026 · publ. 2023–2026 · source ↗
- ReportedRevenue $27B → $61B → $131B → $216B (FY2023–FY2026).NVIDIA Forms 10-K (FY2023–FY2026) — revenue & diluted EPS series — Fiscal years 2023–2026 · publ. 2023–2026 · source ↗
- ReportedChina data-center revenue fell to essentially zero under U.S. export controls and was guided out of the forecast.U.S. Commerce Department (BIS) — export controls on advanced AI chips to China (Oct 2022 & Oct 2023 rules, since tightened) — 2022–2026 · publ. 2022–2026 · source ↗
- ReportedGoogle began offering TPUs into third-party data centers, claiming large performance-per-dollar gains.Google Cloud — TPU program (external availability; latest-generation performance-per-dollar claims; 2026 push into third-party data centers) — 2025–2026 announcements · publ. 2025–2026 · source ↗
- ReportedOver the year to Q3 FY2027 Hyperscale data-center revenue grew 71% while AI Clouds, Industrial & Enterprise grew 195%.NVIDIA Q3 FY2027 CFO commentary (Form 8-K, exhibit 99.2) — Data Center revenue of $107.6 billion splits into Hyperscale $60.1 billion (up 71% year on year, up 23% sequentially) and AI Clouds, Industrial & Enterprise $47.5 billion (up 195%, up 18%), growth 'driven by the introduction of our Rubin infrastructure, our fastest architecture ramp ever, standalone Vera CPUs and the continued growth of Blackwell Ultra'; shipments of Data Center Hopper products to China during the quarter were $2.4 billion; Edge Computing $8.2 billion, up 42% and up 14% sequentially, as 'Strong Blackwell workstation sales and edge compute for Physical AI applications drove the increases, while higher memory and system prices continue to affect consumer PC demand'; reportable segments Compute & Networking $106.0 billion (up 108%) and Graphics $9.8 billion (up 62%); GAAP gross margin 74.8% against 75.0% in Q2 FY2027 and 73.4% in Q3 FY2026, higher year on year 'on improved mix from Blackwell Ultra' and down sequentially 'reflecting the initial Rubin ramp'; GAAP operating expenses up 57% year on year and 9% sequentially — Q3 FY2027 (quarter ended 25 October 2026) · publ. 2026-11-17 · source ↗
- ReportedGross margin is guided down from 74.8% to between 71.0% and 72.0% in a single quarter, on the Rubin ramp and the price of memory.NVIDIA, Q3 FY2027 financial results (Form 8-K, exhibit 99.1) — revenue of $115.8 billion for the quarter ended 25 October 2026, up 20% sequentially and 103% year on year; Data Center revenue of $107.6 billion, up 21% sequentially and 110% year on year; Edge Computing $8.2 billion, up 42%; GAAP gross margin 74.8% and non-GAAP 74.9%; GAAP operating expenses $9,156M and non-GAAP $8,841M; GAAP operating income $77.5 billion; GAAP net income $65,287M, up 105%; GAAP diluted EPS $2.70 against non-GAAP $2.68; nine-month revenue $293.7 billion against $147.8 billion; a record $28.1 billion returned to shareholders in the quarter; the board approved a $150 billion increase to the repurchase authorisation in September 2026, leaving about $227.3 billion; next quarterly dividend $0.25 a share. Outlook for Q4 FY2027, which includes an additional week: revenue of $130.0 billion plus or minus 3%, with NVIDIA not assuming any Data Center compute revenue from China; GAAP and non-GAAP gross margins of 71.0% to 72.0%; GAAP operating expenses of approximately $11.0 billion and non-GAAP approximately $10.7 billion; full-year tax rate 16.0% to 18.0%. The forward-looking statements name 'NVIDIA's leadership transition and the search for a new chief executive officer' and 'the ongoing investigation by an independent committee of NVIDIA's board of directors'. Colette Kress, interim CEO and executive vice president and chief financial officer: 'Our momentum is real... Vera Rubin is off to an explosive start, our roadmap is on track and our ambition has never been greater.' — Q3 FY2027 (quarter ended 25 October 2026) · publ. 2026-11-17 · source ↗
- ReportedJensen Huang resigned after thirty-three years, effective immediately and including his board seat, the chief financial officer is holding both jobs, and an independent committee of the board is conducting an investigation Nvidia has not described.NVIDIA, Q3 FY2027 financial results (Form 8-K, exhibit 99.1) — revenue of $115.8 billion for the quarter ended 25 October 2026, up 20% sequentially and 103% year on year; Data Center revenue of $107.6 billion, up 21% sequentially and 110% year on year; Edge Computing $8.2 billion, up 42%; GAAP gross margin 74.8% and non-GAAP 74.9%; GAAP operating expenses $9,156M and non-GAAP $8,841M; GAAP operating income $77.5 billion; GAAP net income $65,287M, up 105%; GAAP diluted EPS $2.70 against non-GAAP $2.68; nine-month revenue $293.7 billion against $147.8 billion; a record $28.1 billion returned to shareholders in the quarter; the board approved a $150 billion increase to the repurchase authorisation in September 2026, leaving about $227.3 billion; next quarterly dividend $0.25 a share. Outlook for Q4 FY2027, which includes an additional week: revenue of $130.0 billion plus or minus 3%, with NVIDIA not assuming any Data Center compute revenue from China; GAAP and non-GAAP gross margins of 71.0% to 72.0%; GAAP operating expenses of approximately $11.0 billion and non-GAAP approximately $10.7 billion; full-year tax rate 16.0% to 18.0%. The forward-looking statements name 'NVIDIA's leadership transition and the search for a new chief executive officer' and 'the ongoing investigation by an independent committee of NVIDIA's board of directors'. Colette Kress, interim CEO and executive vice president and chief financial officer: 'Our momentum is real... Vera Rubin is off to an explosive start, our roadmap is on track and our ambition has never been greater.' — Q3 FY2027 (quarter ended 25 October 2026) · publ. 2026-11-17 · source ↗
- Nvidia Form 10-K filings — Business & Risk Factors (SEC EDGAR)
- Nvidia annual financials (stockanalysis.com)
- Nvidia valuation history — P/E & P/S by year (stockanalysis.com)
- Nvidia investor relations — results, filings & events