$NVDA.US Let's start with the uncomfortable part: using the conservative methods of this approach, NVDA appears to be priced for a decade of growth that no traditional discounted cash flow model can justify today. The current FCF Yield (~2.3%) is below its own 10-year Treasury (~4.9-5.0%) — this isn’t a technical detail; it’s the market saying it’s paying for duration, not for present cash.
## 0. Actual Data Collected — NVDA | Issuer: Domestic (10-K/10-Q) | Date: 21/Sep/2026
| Indicator | Value | Period | Source |
|---|---|---|---|
| Net Revenue (LTM) | US$ 302,97 bn | up to 26/Jul/2026 | 10-K FY26 + 10-Q Q1/Q2 FY27 |
| FY2026 revenue (fiscal year) | US$ 215,94 bn (+65,5%) | closed 25/Jan/2026 | 10-K, access 0001045810-26-000021 |
| GAAP Net Income (LTM) | US$ 192,88 bn | up to 26/Jul/2026 | derived (see note) |
| Diluted GAAP EPS (LTM) | US$ 7,91 | up to 26/Jul/2026 | 10-K + quarterly releases |
| Diluted Adjusted EPS (Non-GAAP, LTM) | US$ 7,01 | up to 26/Jul/2026 | quarterly releases |
| SBC (FY2026) | US$ 6,39 bn (3,0% of revenue) | FY2026 | 10-K, DFC |
| CFO (LTM) | ~US$ 220 bn (implied) | up to 26/Jul/2026 | derived |
| FCF (LTM) | US$ 126,91 bn | up to 26/Jul/2026 | derived from quarterly releases |
| FCF Yield | ~2,3% | 21/Sep/2026 | calculated |
| Cash + Marketable Securities | US$ 56,6 bn | 26/Jul/2026 | Q2 FY27 release |
| Gross Debt | ~US$ 33,4 bn (ST+LT, after issuing US$25 bn in notes) | 26/Jul/2026 | Q2 FY27 release |
| Net Cash (cash position, not net debt) | ~US$ 23,2 bn | 26/Jul/2026 | calculated |
| Supply commitments (off-balance sheet) | US$ 279 bn (from US$ 119 bn in the prior quarter) | 26/Jul/2026 | 10-Q Q2 FY27 |
| Diluted Shares Outstanding | ~24,15 bn | Sep/2026 | Yahoo/Google Finance |
| Current Quote | US$ 230,36 (most recently verified close) | 04/Sep/2026 | Yahoo Finance / Google Finance |
| Beta | 2,22 | Sep/2026 | Google Finance |
| P/E TTM | 29,1x | Sep/2026 | Yahoo Finance |
| 10-Year Treasury Yield | ~4,95% (reached 5,04% on 15/Sep, 16-year high) | Sep/2026 | CNBC/CME/FRED |
| Moody's Aaa Corporate Bond Yield | 5,76% | Jul/2026 (most recent monthly data located) | FRED |
| Implied ERP (Damodaran) | 4,20% | Jul/2026 | Damodaran/NYU Stern |
⚠️ [Likely] The US$ 230,36 quote is from 4/Sep/2026. Alternative sources (tokenized trackers) suggest a range of US$ 213-227 between 10-19/Sep — divergence up to ~7%, above the 5% threshold that the methodology requires to declare. I didn’t find a verifiable official Nasdaq close for the last 2-3 trading days. I use US$ 230 as the central reference; if you have today’s quote, tell me and I’ll recalculate the multiples.
⚠️ [Assumption] Some items (current EPS, exact LTM EBITDA, quarterly D&A) were estimated by the difference between quarters/consistency with published market multiples (P/B, EV/EBITDA), because the full quarterly balance sheet for 26/Jul/2026 didn’t appear in the search results in full. Marked below.
---
## 1. Investment Thesis (summary)
NVDA dominates the AI accelerator market (CUDA as a software moat, ~90% share in training frontier models) and turned that dominance into GAAP margins of ~55-71% and ROE >100%. The bull thesis doesn’t need my help — it’s everywhere on Wall Street’s research desks. My job here is the opposite: to test whether the price has already surpassed the thesis.
## 2. Fundamental Data
- Revenue growth: FY2024→FY2026 was 125,9% → 114,2% → 65,5%. **The growth rate is structurally decelerating**, even with record absolute numbers each quarter — this is basic math of a bigger base, not a sign of weakness, but it contradicts the “infinite acceleration” narrative.
- GAAP operating margin: 60,4% (FY26) → 65,6% (Q1FY27) → 66,2% (Q2FY27). In expansion.
- Segment: Data Center is 90% of revenue (real business concentration, not hypothetical).
- Geography: U.S. 69%, Taiwan 20% (proxy for Asian customers), China fell from 19% to 9% of revenue — direct effect of H20 export controls.
### 2a. GAAP vs. Non-GAAP — and here the distortion goes in the opposite direction of the usual
[Sure] In the last two quarters (Q1 and Q2 FY27), GAAP Net Income was **above** Non-GAAP — the opposite of the typical pattern. Reason: US$ 15,9 bn (Q1) and US$ 7,8 bn (Q2) of mark-to-market gains from the company’s own investment portfolio (non-marketable equity securities), which Non-GAAP excludes because they are not operational. That’s why I use the **Adjusted EPS (US$ 7,01)**, not GAAP (US$ 7,91), for valuation — GAAP is inflated by an item unrelated to chip sales.
### 2b. Dilution and Capital Return
[Sure] Diluted shares outstanding **fell** from 24.940 bn (FY24) to 24.514 bn (FY26) — buybacks (US$ 40,1 bn in FY26) exceeded dilution from SBC (US$ 6,4 bn). This is the opposite of the alert pattern that tends to show up in small/mid-cap tech. A point in its favor.
### 2c. Debt Structure — here something doesn’t add up
[Sure] In Q2 FY27, NVIDIA issued **US$ 25 bn in senior notes**, increasing long-term debt from US$ 7,5 bn to US$ 32,4 bn — **in the same quarter** it returned US$ 26 bn to shareholders (buyback + dividend) and had US$ 56,6 bn in cash and securities. A company sitting on this cash volume doesn’t need to borrow for “general corporate purposes.” This isn’t leverage in the credit-risk sense (the company still has positive net cash), but it is a signal of financial engineering — possibly to avoid repatriating/realizing cash outside the U.S., possibly to lock in rates before the Treasury rises further. Worth monitoring, not a reason to panic.
---
## 3. Valuation — Multiple Methods
| Method | Estimated Fair Price | Suggested Weight | Note |
|---|---|---|---|
| Graham (revised) | US$ 259,70 | Low | [Assumption] g=20% per year — the Graham formula wasn’t designed for hypergrowth; I treat it as a fragile reference |
| Bazin | N/A | Zero | Symbolic dividend (payout 3,5%), increased 25x in May/2026 — no real 3-year history at the new level |
| Gordon | N/A | Zero | Not an established dividend grower |
| DCF | **US$ 110,10** | High | Ke=CAPM=14,27%; WACC≈14,2% (immaterial debt in the mix); terminal g 3,5%; see assumptions below |
| Multiples (P/E, EV/EBITDA, EV/Sales) | **US$ 185,10** (average) | High | Peers: TSM, AVGO, QCOM, TXN |
| **Weighted Consensus** | **≈ US$ 166,00** | — | DCF 45% / Multiples 35% / Graham 20% |
| **Current FCF Yield** | **~2,3%** | ← validation | Below the 10-year Treasury (~4,95%) |
**DCF assumptions (explicit, for you to challenge):** FCF LTM US$ 126,9 bn growing 45%→30%→20%→12%→8% over the next 5 years, terminal g=3,5%, WACC 14,2%. That’s already generous versus the consensus CFO slowdown (guidance of ~70% revenue growth for FY28, then declining afterward). Even so, the present value is ~52% below the current price.
[Likely] The dispersion between methods is huge (US$ 110 to US$ 260 — more than 2x) — this isn’t a calculation mistake; it’s the picture of an asset where the market is pricing a 10+ year duration narrative that any “conservative” DCF framework simply can’t accommodate without heroic terminal growth assumptions or artificially low WACC.
---
## 4. Margin of Safety and Buy Price
| | |
|---|---|
| Current Price | US$ 230,36 (04/Sep) |
| Fair Price (Consensus) | US$ 166,00 |
| Margin of Safety applied | 28% |
| **Buy Price** | **US$ 119,50** |
| Upside/downside vs fair value | **-28%** |
I classified it as intermediate between “solid large cap” (15-25%) and “cyclical/levered” (35-50%) — the company doesn’t have problematic debt, but it has cycle risk (AI capex), geopolitical concentration (China/export) and a price that already bakes in perfect growth for years.
**[I disagree with the “market” here, with the structure you asked for]:** I disagree with the thesis that “the current price is justified because AI is only just starting.” Here’s what I would do instead: treat a position in NVDA today as a bet on narrative/momentum, not value — and size the position according to that, not according to the size that a fundamentalist conviction alone would justify. The risk of this “optimistic without margin of safety” approach is clear: if the AI capex cycle slows down (signals are already showing in the ecosystem’s circular financing — US$ 366 bn in off-balance-sheet commitments, guarantees up to US$ 105 bn for a single campus), the repricing to the numbers of this report is not a remote hypothesis.
## 5. Technical Analysis
- **Support (technical)**: ~US$ 205-210 (MA50 region)
- 52 weeks: low US$ 164, high US$ 236,54 (14/May/2026)
- Recent sequence: 7 trading days in a row of decline (-7,5%) before the Q2 FY27 results, followed by a rally of +8,7% on the results day (26-27/Aug)
- Technical support: ~US$ 205-210 (MA50 region)
- Resistance: ~US$ 236 (historical high)
- [Assumption] Precise RSI/MACD not located in this search — N/D. I won’t make it up.
- Suggested Stop Loss: below US$ 205
- Relevant macro context for timing: the 10-year Treasury hit 5,04% on 15/Sep/2026 (16-year high) — discount pressure on high-multiple growth stocks is an active headwind now, not a hypothetical one.
## 6. Risks and Scenarios
- **Customer concentration**: 3 customers = 30%+18%+16% of accounts receivable (Apr/2026)
- **Geopolitical/export risk**: China revenue fell from 19% to 9% of revenue within a year due to export controls (H20/Hopper banned)
- **Circular financing of the AI ecosystem**: supply commitments jumped from US$ 119 bn to US$ 279 bn in one quarter; guarantee of up to US$ 105 bn for a single data center (SB Energy/PORTS-Pike) — exposure to a concentrated counterparty in customers who, in turn, depend on heavy external financing
- **Macro risk**: rising yields pressure long-duration multiples
- **FX risk**: if you report in BRL, USD/BRL is a risk separate from the stock’s risk itself — I don’t mix the two here
- **Liquidity**: not a constraint (average volume >130m shares/day)
| Scenario | Description | Implied Price |
|---|---|---|
| Optimistic | AI cycle supports 40%+ growth for another 3-5 years, higher terminal g | US$ 280-350+ |
| Base | Deceleration in line with the company’s own guidance, converging to the consensus of this report | US$ 160-190 |
| Pessimistic | Digestion of AI capex, escalation of export controls, multiple compression | US$ 100-120 |
## 7. Conclusion
**Recommendation: WAIT** (do not BUY at the current price under this methodology). This is not an anti-NVIDIA thesis — it’s an anti-entry-price-without-margin-of-safety thesis. If you already hold a position, this is not an automatic sell signal (the operating fundamentals remain excellent); it’s a sign that adding here, without a downside plan, is betting on the continuity of the narrative, not on value-math.
If you disagree with my `g` of 20% in the Graham model or with the DCF growth rates because they’re too conservative for a company in this cycle, tell me the numbers you would use — I’ll recalculate, but I won’t simply raise assumptions until the fair value matches the current quote just to “close the gap.”
---
*Sources: 10-K FY2026 (access 0001045810-26-000021) and 10-Q Q1/Q2 FY2027 via SEC EDGAR; NVIDIA earnings releases IR; Yahoo/Google Finance; FRED (Treasury, Moody's Aaa); Damodaran/NYU Stern (ERP); GuruFocus (sector multiples). This analysis does not constitute an offer to buy or sell. Past profitability does not guarantee future results.*