Nvidia sells the chips, networking and software that nearly every large AI model is trained and run on. Revenue doubled again this year and margins sit at 75%. The question is not the business but how long this level of spending lasts, and the price already assumes it lasts a long time.
Today’s close against the report’s numbers, updated every weekday. The analysis is as of October 20, 2026; the rating as of October 22, 2026. Both change only when the report is revised.
The cash-flow model’s bear-to-bull range, the probability-weighted value, the price and its 52-week range. The copper line marks the base case; the fine dotted line is today’s close.
Written on October 20, 2026, before its fiscal third-quarter results in November. Figures are from the company’s filings and releases unless noted. Not investment advice. The author holds NVDA shares.
Nvidia designs graphics processors and the systems built around them: full racks of GPUs, CPUs and networking that data centers buy as one unit, plus CUDA, the software layer most AI developers write for. It does not own factories; TSMC makes its chips.
Data centers are now almost the whole company: $89.0 billion of $96.2 billion in revenue last quarter. Its biggest customers are the cloud companies (Microsoft, Google, Amazon, Oracle, Meta) and a fast-growing group of specialist AI clouds. Several of those customers also design their own chips, which is the main long-run competitive threat alongside AMD.
Takeaway. The lead rests on three things competitors have to match at once: the chip, the networking that ties thousands of them together, and the software developers already use.
In the quarter to July, revenue rose 18% from the previous quarter and 106% from a year earlier. Operating income was $63.7 billion, about two-thirds of revenue. Management guided the next quarter to $108 billion, assuming no data-center sales to China, with gross margin of about 74%.
The product cycle is moving to Vera Rubin, which the company says is in full production and ramping at cloud partners. Nvidia also returned about $26 billion through buybacks and dividends in the quarter and has about $99 billion left under its buyback authorization.
| FY2022 | FY2023 | FY2024 | FY2025 | FY2026 | TTM Jul 2026 | |
|---|---|---|---|---|---|---|
| Revenue | $26.9B | $27.0B | $60.9B | $130.5B | $215.9B | $303.0B |
| Operating margin | 37.3% | 15.7% | 54.1% | 62.4% | 60.4% | 65.2% |
| Net income | $9.8B | $4.4B | $29.8B | $72.9B | $120.1B | $192.9B |
| Operating cash flow | $9.1B | $5.6B | $28.1B | $64.1B | $102.7B | $134.4B |
| Capital spending | $1.0B | $1.8B | $1.1B | $3.2B | $6.0B | $7.4B |
| Free cash flow | $8.1B | $3.8B | $27.0B | $60.9B | $96.7B | $127.0B |
Revenue was flat in fiscal 2023, then took off with ChatGPT. Margins rose with it: operating margin went from 16% to above 60%, because each new chip generation sold at a higher price than it cost to make. Free cash flow follows profit closely; the business needs little capital because TSMC builds the factories.
Debt is small next to a $5 trillion market value, and cash and investments roughly cover it, so the model ignores net debt. Almost all cash returns come through buybacks; the dividend is token. The bigger balance-sheet items are supply commitments: Nvidia pre-pays and commits to suppliers to secure chip and memory capacity, which would become a burden if demand fell quickly.
Read. No financial risk; the exposure is operational, through supply commitments made in a boom.
The base year is fiscal 2027 (to January 2027), estimated at $404 billion from the first two quarters, guidance and a modest fourth quarter. The model then assumes another strong year and a long fade.
| Assumption | Base case | Basis |
|---|---|---|
| Revenue, FY2027 | $404B | Two reported quarters, Q3 guidance, a higher Q4 |
| Growth, next two years | 40%, then 15% | The Rubin cycle, then normalising spending |
| Growth after that | Fades to 3% by year 10 | AI spending matures |
| Operating margin | 63% falling to 55% by year 7 | Competition and custom chips take some pricing |
| Tax rate | 17% | Company guidance of 16–18% |
| Reinvestment | $1 for every $2.50 of new revenue | Inventory, supply prepayments and capital |
| Discount rate | 12.3% | 5.15% Treasury, beta 1.58, 4.5% premium; almost no debt |
| Terminal growth | 3.0% | With a 25% return on new capital |
| Bear | Base | Bull | |
|---|---|---|---|
| Growth, next two years | 20% / 5% | 40% / 15% | 55% / 25% |
| Long-run operating margin | 45% | 55% | 60% |
| Discount rate | 13.3% | 12.3% | 11.3% |
| Value per share | $82 | $166 | $306 |
| Probability | 25% | 50% | 25% |
Weighted 25/50/25, the cases give $180.
| Discount rate | 9.0% | 10.0% | 11.0% | 12.3% |
|---|---|---|---|---|
| Value per share | $258 | $221 | $193 | $166 |
Read. The discount rate decides this one. At 10%, a common choice for a company this size, the price is about fair; at the 12.3% that Nvidia’s beta implies, it is not.
Every figure in this valuation comes from one model script, rerun and checked against this page each time the site is built. The XLSX model rebuilds it in live formulas: change any blue input and every year’s cash flow, the bridge to the value, the scenarios and the sensitivity table recalculate.
| To reach today’s price, you need | Instead of |
|---|---|
| Revenue growth of about 84% next year, then 15% | 40%, then 15% |
| A long-run operating margin near 73% | 55% |
| A discount rate of about 10.1% | 12.3% |
None of these is impossible for Nvidia; each has been beaten before. But the price needs the AI build-out to keep growing well past 2027 without customers’ own chips taking meaningful share.
| Risk | Why it matters |
|---|---|
| Customer spending slows | A handful of cloud companies buy most of the output. If their returns on AI disappoint, orders can fall quickly. |
| Custom chips | Google, Amazon, Microsoft and Meta all design their own accelerators and would rather not pay Nvidia’s margins. |
| China and export rules | Guidance assumes no data-center sales to China; policy can move either way. |
| Supply chain | Dependence on TSMC and on memory supply; a disruption in Taiwan is the tail risk. |
Upside: AI moving from training into everyday use (inference), robotics and sovereign AI could extend the cycle further than the model allows.
Nvidia earns more than almost any company in history and still has most of the AI accelerator market. The model values it below the price at its own cost of capital, but by less than the market as a whole is valued below its price on the same assumptions.
What would change the value: Next-year guidance pointing to growth well above 40% would raise the value; signs that cloud AI spending is peaking would lower it. Fiscal third-quarter results in November, against $108 billion of guidance, are the next test.
Ranges, not three cases. The expected value is the average of 4,000 runs of this report’s own model, with every input drawn from a range: the base case is the middle, and the bear and bull cases are the 10th and 90th percentiles. Business inputs move together; the discount rate moves on its own. The 25/50/25 weighting in the valuation section gives $180. How the ranges work
Two different questions. The rating is relative: it asks whether the stock is cheaper or dearer than the S&P 500 on the same assumptions. The absolute view asks whether it is worth its price at all: on its own the stock trades above our estimated value, which is 20% below the price.
On the same assumptions every report uses (a 5.15% Treasury yield and a 4.5% equity premium), the S&P 500 itself is worth about 68% of its price. The rating compares each stock with that: Buy if it is at least 15% cheaper than the market on this basis, Sell if at least 15% dearer, Hold in between. It is the same rule for every company, whoever owns it. The rating depends on that assumption: it would be Buy if the index grows 6% a year and Hold at 10%, against 8% here. How ratings work
| Measure | Nvidia | Industry median | Rank | |
|---|---|---|---|---|
| Revenue growth, last fiscal year | 65.5% | 15.3% | 2 of 15 | Better than the median |
| Operating margin | 65.2% | 26.0% | 2 of 15 | Better than the median |
| Price to earnings | 26.7× | 46.4× | 5 of 14 | Better than the median |
| Free cash flow yield | 2.5% | 2.5% | 9 of 15 | Worse than the median |
Peers: the 14 other S&P 500 companies in Semiconductors with complete data (ADI, AMD, AVGO, FSLR, INTC, MCHP, MPWR, MRVL, MU, NXPI, ON, QCOM, SWKS, TXN). Last twelve months where reported, from SEC filings; prices from the latest close. A lower price-to-earnings ratio ranks higher. These figures update with each data run; the rating does not.
Nvidia earns more than almost any company in history; the price already counts on that continuing.
| What | Figures | Source |
|---|---|---|
| Latest results | $96.2B (Revenue, Q2 FY2027), +106% (Growth from a year earlier), $89.0B (Data center revenue, +117%), 75.0% (Gross margin) and others | Q2 FY2027 results |
| The numbers (five-year table) | Five years of annual figures and the last twelve months | SEC XBRL company facts; latest: 10-K for the period to 2026-01-25, 10-Q for the period to 2026-07-26 |
| Price at writing | $216.47 (close) | Yahoo Finance; the time of the quote is stated beside the price |
| Discount rate inputs | 10-year Treasury 5.15%, equity risk premium 4.5% | Fixed for every report so they can be compared; How ratings work |
| Valuation | Every value, case and sensitivity | This site’s model: XLSX, CSV |
| Checked by hand | 10 figures on October 23, 2026 | All match. Audit log |
| Input | Bear (10th percentile) | Base (median) | Bull (90th percentile) |
|---|---|---|---|
| Growth, first forecast year | 20.0% | 40.0% | 55.0% |
| Growth, second forecast year | 5.0% | 15.0% | 25.0% |
| Operating margin, near term | 60.0% | 63.0% | 65.0% |
| Operating margin, long run | 45.0% | 55.0% | 60.0% |
| Discount rate | 13.3% | 12.3% | 11.3% |
| Simulation result | |
|---|---|
| Expected value (average of 4,000 runs) | $173.08 |
| Likely range, 10th to 90th percentile | $97.95 to $257.77 |
| Median | $163.89 |
| Chance the value is above the price | 22% |
| Chance it is cheaper than the S&P 500 | 61% |
Each input is drawn from a split normal: its bear and bull values are the 10th and 90th percentiles, so the two sides can differ. Business inputs share a common factor (correlation 0.6); the discount rate moves on its own. Inputs not listed are held at their base value. How the simulation works.