🖥️🤖 Business Overview
🎯Key Metrics
Total: 12.5/18
- +2 ✅ Projected Operating Margin: 55%
- +2 ✅ Projected 5-Year Revenue CAGR: 28.7%
- +2 ✅ Last 5-Year ROIC: 94.2%
- +1 ✅ Estimated Cost of Capital: 11.99% (lower than ROIC)
- +1 ✅ Last 5-Year Shares Outstanding CAGR: -0.25%
- +2 ✅ Projected 5-Year EPS CAGR: 26.8%
- +1 ✅ Projected 5-Year Dividend CAGR: ~15% (dividend reset to $1.00 a year in May 2026, ~12% of core EPS; a constant payout would track Base EPS CAGR of ~27%, but the policy is new, so I take ~15%)
- +1.5 ✅ Moody’s Rating: Aa1
- +2 ✅ Morningstar Moat: Wide
- -2 ❌ Morningstar Uncertainty: Very High
Nvidia designs the GPUs, networking and software that power AI data centers, and sells them to cloud providers, AI labs and enterprises. Data Center is 92% of revenue (earnings release, TTM Oct 2026).
- Q3 FY2027 revenue was $115.8B (+103% year over year) with a 74.8% gross margin, above the $108B guided; Q4 guidance is $130B for a 14-week quarter, excluding any China data center sales (earnings release, Nov 17).
- Jensen Huang resigned as CEO on November 11 and CFO Colette Kress is interim CEO; the release also cites an ongoing investigation by an independent board committee (8-K, Nov 12).
- Reported EPS includes $25.5B of gains on its stakes in other companies over nine months. I use core EPS and value those stakes separately.
- The main risks are customer concentration (hyperscalers), custom chips from its customers, memory costs and China export rules.
I expect growth to stay high through FY2028 and then slow quickly, with margins slipping from today’s peak.
📖 My Narratives
Bad: Customers digest their spending, custom chips take share, or the leadership change and investigation distract execution. Growth drops to 8% after year 1 and the operating margin falls to 40%.
Base: Supply-limited growth continues through FY2028, then slows to 20%. Competition and memory costs pull the operating margin from 66% to 55%.
Good: Demand for AI compute keeps compounding (30% a year) and the software and networking moat holds the margin near 62%.
| Bad | Base | Good |
|---|---|---|---|
Revenue growth (Year 1 / Years 2-5) | 55% / 8% | 70% / 20% | 80% / 30% |
Operating margin (initial → terminal) | 62% → 40% | 66% → 55% | 68% → 62% |
Sales / Capital (initial → terminal) | 2.2 → 1.5 | 2.7 → 2.0 | 3.2 → 2.5 |
Terminal ROIC | 15% | 30% | 45% |
Terminal cost of capital | 11.5% | 10.6% | 10.55% |
Net margin (Year 5) / Shares CAGR / P/E (Year 5) | 43% / -0.2% / 16 | 51% / -0.7% / 24 | 54% / -1.0% / 30 |
Methods: DCF 65%, EPS Growth 35%, because Nvidia grows fast, reinvests and pays almost no dividend. I leave out dividend models (dividends are 1.5% of free cash flow to equity over 5 years) and historical multiples (the business changed completely with AI).
The Base narrative drives each valuation. Bad and Good set the low and high end of each Monte Carlo range.
📈Business Valuation
Revenue Growth

Revenue grew 65% in FY2026 and 93% over the last twelve months; the 10-year median is 47% (chart above). Management guides $130B for Q4 (14 weeks) and said in August it expects about 70% growth in FY2028 (earnings releases and call).
💡 Year 1: 70% (Bad 55% · Good 80%). Years 2-5: 20% (Bad 8% · Good 30%). Year 1 follows the Q4 guide and the FY2028 outlook. After that growth fades well below the 10-year median, because the revenue base is now large; Bad assumes a digestion phase, Good an AI market that keeps compounding.
Cost of Capital
Inputs: the Q3 earnings release (the 10-Q is not filed yet), the 10-year US bond (5.12%) and revenue by geography. Debt rating: Moody’s Aa1 (the tool’s closest bucket is Aa2/AA). Total debt is $38.7B (borrowings plus operating leases). Interest expense, current leases and the geography mix are estimated from the Q2 10-Q.


Cost of Capital: 11.99% (industry average in the US: ~10.55%).
💡 It starts at 11.99% and converges to 10.6% (Bad 11.5% · Good 10.55%). As growth fades it should look like the industry, which is the floor.
Try it yourself: Cost of Capital – The Fair Value Journal (free).
Operating Margin
Nvidia reports two segments with operating income and revenue by billing region. The Q3 release has no segment or region tables, so those charts end at the Q2 10-Q (TTM Jul 2026), except the market split.
Geographic Segments
(10-Q and 10-K, TTM Jul 2026) United States: 70.9% of revenue. Taiwan: 21.4%. China (incl. Hong Kong): 6.1%. Other: 1.6%. No operating income by region. Taiwan is mostly system makers building for US clouds. China fell from 19% of revenue in FY2025 to 6%, and Q4 guidance assumes none.

Business Segments
(10-K, 10-Q and Q3 release) Compute & Networking: ~91% of revenue, operating margin 71% (data center GPUs, networking and software). Graphics: ~9%, margin 44% (gaming, workstation, auto; Edge Computing in the new market split). Data Center is 92.4% of TTM revenue; the mix keeps moving to the higher-margin side.


Overall Operating Margin

💡 Starts at 66% (Bad 62% · Good 68%) and converges to 55% (Bad 40% · Good 62%). The last twelve months were 66% and the 5-year average is 46%. I stop at 55%, above that average, because Data Center is now 92% of sales at a 71% segment margin. Q4 gross margin is guided down to 71-72% (earnings release), and customers’ own chips should force price concessions over time.
Sales / Capital ratio (Reinvestment)
Revenue generated per dollar of invested capital. Investments can take a few years to turn into sales.

💡 Starts at 2.7 (Bad 2.2 · Good 3.2) and converges to 2.0 (Bad 1.5 · Good 2.5). The last twelve months were 2.7, inside the 1.4 to 3.5 range of recent years; more capacity makes it drift lower.
ROIC

💡 Terminal ROIC 30% (Bad 15% · Good 45%), above the cost of capital. The last 5 years averaged 94%, which I don’t expect to last as competitors narrow the gap. Good keeps 45% because the CUDA software base and networking make switching costly (Morningstar’s wide moat).
Discounted Cash Flows (Weight: 65%)



Where the inputs come from:
- Total Revenue: trailing twelve months to October 2026 (earnings release).
- Cash: cash plus marketable debt securities (Oct 25, 2026).
- Total Debt: borrowings plus operating leases.
- Non-operating Assets: its listed and private stakes in other companies ($44.2B marketable equity plus $65.5B non-marketable), at balance sheet value (earnings release).
- Terminal Revenue Growth: the risk-free rate (5.12%), as a proxy for long-term economic growth.
- Tax Rate: 16.5% rising to 18%, around the 16.5% effective rate of the last twelve months.
- Everything else: the assumptions above.
DCF fair value: $343.11.
The Monte Carlo simulation runs 1,000,000 DCFs, drawing each input between its Bad and Good values (Base is the most likely):



At $155.46, the price is below the P10 ($225.07): Nvidia looks undervalued (fewer than 1 in 10 simulations are lower).
Try it yourself: DCF – The Fair Value Journal
EPS Growth (Weight: 35%)
TTM EPS: $9.31 reported (earnings release, TTM Oct 2026). I use core EPS of $8.28, which removes $31.7B of other income (mostly gains on its stakes in other companies).
Revenue Growth (CAGR)
1 Year | 3 Years | 5 Years | 10 Years |
|---|---|---|---|
65.47% | 100.05% | 66.90% | 45.70% |
💡 28% a year (Bad 16% · Good 38%). It matches the DCF path; the history is far above it but can’t repeat on a base this large.
Shares Outstanding (CAGR)

1 Year | 3 Years | 5 Years | 10 Years |
|---|---|---|---|
-0.80% | -0.69% | -0.25% | +1.15% |
💡 -0.7% a year (Bad -0.2% · Good -1.0%). Buybacks ($65B a year, $227B authorization left) outrun stock compensation.
Net Profit Margin (Average)

| 1 Year | 3 Years | 5 Years | 10 Years |
|---|---|---|---|---|
Reported | 55.60% | 53.43% | 42.55% | 35.51% |
Ex. non-operating items | 51.21% | 50.11% | 38.46% | 31.14% |
💡 51% in year 5 (Bad 43% · Good 54%). It follows the DCF’s year-5 operating margin (about 61% Base, 52% Bad, 65% Good, since the DCF moves margins linearly over 10 years) after about 17% tax. The last twelve months were 56% (core).
P/E ratio (Average)
On core EPS ($8.28), the current P/E is 18.8.
Core EPS | 1 Year | 3 Years | 5 Years | 10 Years |
|---|---|---|---|---|
P/E | 34.73 | 36.71 | 60.77 | 53.40 |
💡 24 in year 5 (Bad 16 · Good 30). Today’s 18.8 is below every average above. By year 5 growth is slower, so I use a multiple a little above today’s level but well below the history.
Scenarios (5 years)
| Bad | Base | Good |
|---|---|---|---|
Revenue growth (CAGR) | 16% | 28% | 38% |
Net profit margin | 43% | 51% | 54% |
Shares outstanding (CAGR) | -0.2% | -0.7% | -1.0% |
P/E | 16 | 24 | 30 |
EPS in year 5 | $13.65 | $27.15 | $42.51 |
EPS CAGR | 10.5% | 26.8% | 38.7% |
The Base scenario, discounted at the 11.99% cost of capital:


EPS Growth fair value: $369.92.
Monte Carlo between the Bad and Good scenarios:


At $155.46, the price is below the P10 ($244.40): Nvidia looks undervalued (fewer than 1 in 10 simulations are lower).
Try it yourself: EPS Growth – The Fair Value Journal
EPS Scenario Returns



The Base scenario returns 33.2% a year, well above the 11.99% cost of capital; the Bad scenario returns 7.0% and the Good one 52.3%.
Try it yourself: EPS Scenario Returns – The Fair Value Journal
✍️Summary
Each method enters with its weight. Monte Carlo methods use their P10/P20 (buy) and P80/P90 (sell) values; historical methods use 20%/10% below and above their fair value.

Try it yourself: Weighted Valuation – The Fair Value Journal
I would buy or add below $231.84, because despite my conviction in the company there’s too much uncertainty and risk associated with it right now.
A fair value is always wrong to some degree, decimals included 😮
Being a long time Nvidia shareholder, I’m still in the positive despite the recent drop in share price, and that goes to show how much this business has return to its shareholders over time.
At $155.46, Nvidia looks undervalued, and I’m again gradually increasing my position in it.
Fair Value: 321.14 dollars.


