NVIDIA is the most-debated single stock in value investing right now. A 70%+ gross margin and a 63% ROIC look like the textbook quality compounder. A P/E of 46 and a P/B of 35 look like the textbook bubble. Same business, opposite reads, depending on which page of the value-investing canon you open first.
So we did the thing the canon never lets you do: we ran NVDA through all seven named-investor frameworks on invest-like.com in parallel and printed the scores side by side. This post is the result. Every number below is pulled from the production scoring tables on 25 May 2026, the same scores that power /buffett/nvda/ and every public API response for ticker NVDA.
TL;DR - NVDA on all 7 frameworks
| Framework | Score | Grade | Verdict | Criteria met |
|---|
| Fisher (growth quality) | 93 | A | yes | 9 / 10 |
| Greenblatt (Magic Formula) | 91 | C | unclear | 2 / 5 |
| Graham (defensive) | 89 | A | yes | 4 / 7 |
| Smith (Fundsmith) | 89 | A | yes | 10 / 12 |
| Munger (mental models) | 87 | C | unclear | 10 / 11 |
| Buffett (moat + quality) | 72 | C | unclear | 8 / 9 |
| Lynch (GARP) | 63 | C | unclear | 6 / 6 |
Cross-framework average: 83.4 out of 100. Minimum score: 63 (Lynch). All seven scores clear the 60 pass-threshold the invest-like data study uses, which means NVDA is one of the 47 US-listed stocks that pass all 7 value-investing frameworks at a B-minus or better, alongside names like Alphabet, Microsoft and Visa. Source: the cross-framework consensus study, data snapshot 25 May 2026.
Headline numbers that drive every framework:
- Price: $215.33. Market cap: $5.22 trillion.
- ROIC trailing twelve months: 62.9%. Five-year average ROIC: 49.2%.
- Gross margin: 71.1%. Operating margin: 60.4%. Net margin: 55.6%.
- P/E TTM: 45.6. P/B: 34.8. EV/EBIT: 42.0.
- PEG ratio: 0.69.
- Five-year revenue CAGR: 68.3%. Five-year EPS CAGR: 88.6%.
- Debt to equity: 0.07. Current ratio: 3.9. Net debt to EBITDA: 0.0.
- Owner-earnings yield: 1.77%. Earnings yield: 2.19%.
- Dividend yield: 0.02% (effectively none).
- AAOIFI halal status: compliant. Interest-bearing debt to market cap: 0.14%.
The rest of this post walks through each framework, explains why it scored NVDA the way it did, and surfaces where the seven authors agree and where they part ways.
Why NVIDIA is the most-debated stock in value investing right now
In 2020, NVDA was a $300 billion graphics-chip company growing at the pace of the gaming and crypto cycles. Five years later, it is a $5.2 trillion business that designs the accelerators every hyperscaler is competing to buy at industrial scale. Revenue compounded at 68% per year over five years. EPS compounded at 89%. ROIC nearly tripled.
The framework debate falls out of one observation: value investing was written when "wonderful business" and "fair price" did not pull in opposite directions for businesses this large. Coca-Cola in 1988 had a 30% ROIC, a 15% P/E and a 13% growth rate. NVIDIA in 2026 has a 63% ROIC, a 46% P/E and an 89% growth rate. The frameworks were calibrated against the Coca-Cola world. Now they have to render a verdict on a business that is more profitable than anything Buffett saw at See's, growing faster than anything Lynch saw at Dunkin Donuts, with a multiple Graham would have called speculative on principle.
Run NVDA through the frameworks anyway and the seams show.
Buffett-fit: 72 out of 100, verdict unclear
The invest-like Buffett scorer marks NVDA as 8 of 9 criteria passing. The one failure is the criterion Buffett built his career on: owner-earnings yield. Buffett's "5% minimum" rule, derived from his preference for a 5 to 8% earnings yield as a fair-value anchor, requires NVDA's owner-earnings yield to be at least 5%. It is 1.77%. That is one failure, but it is the failure he weights heaviest.
What passes:
- ROIC of 62.9% blows past the 15% minimum. NVDA's ROIC is more than four times the Buffett floor and roughly six times the S&P 500 median.
- Gross margin of 71.1% passes the 40% bar with room to spare.
- 5-year profitability: yes, never a loss year in the window.
- Net debt to EBITDA: 0.0x. Buffett's "fortress balance sheet" test is satisfied trivially.
- Revenue CAGR 68% and EPS CAGR 89% both clear the 8% growth floor by an order of magnitude.
- FCF to net income at 0.80 clears the 0.80 cash-quality floor.
- Debt to equity at 0.07 clears the 1.0 ceiling.
What this score means in plain English: NVIDIA is a wonderful business that fails the price half of Buffett's test. The Buffett Brain breaks the overall score into five pillars, and the split is dramatic: moat 100, durability 70, management 49, valuation 55, financial health 61. A perfect moat score and a 55-out-of-100 valuation pillar is exactly the shape you would expect for a wide-moat business trading at a premium multiple.
The Buffett scorer's German-language headline on the most recent run reads "NVIDIA zeigt beeindruckende ROIC von 62,9%, aber die Bewertung mit einem KGV von 45,6 ist eine Herausforderung." Roughly: impressive ROIC, but the 45.6 P/E is a challenge. That is the framework speaking.
Graham: 89 out of 100, verdict yes (the most surprising score)
Most readers will guess Graham would hate NVDA. The pure defensive screen (low P/E, low P/B, current ratio above 2x) is what most people think of when they hear "Graham". On those three lines, NVDA fails:
- P/E of 45.6 against the 15.0 Graham ceiling: fail.
- P/B of 34.8 against the 1.5 Graham ceiling: fail (catastrophically).
- Earnings yield of 2.2% against the 7% Graham floor: fail.
So how does NVDA score 89? Because the invest-like Graham scorer applies the enterprising-investor weighting on top of the defensive criteria, which gives partial credit for the four things NVDA does pass: a current ratio of 3.9x, a debt-to-equity of 0.07, five years of profitability, and revenue growth. Those four "balance sheet and durability" passes outscore the three "valuation" fails because the scorer weights them as the foundation of the company surviving long enough for the multiple to compress.
This is the cleanest case in the seven-framework set where the headline score and the lay reading of the framework diverge. If you ran a strict five-rule Graham defensive screen, NVDA would fail every valuation line and the verdict would be a hard no. The invest-like implementation gives NVDA an A because the structural pillars (no debt, profitable, growing, liquid) are pristine. A reader using NVDA as a Graham test case should know which version they are reading. We document both in /methodology/.
The summary the scorer prints is honest about the split: "Meets 4/7 Graham criteria - classic deep-value candidate." That phrasing is generous; NVDA is not a deep-value candidate by anyone's reading. It is a quality candidate that clears the structural half of Graham's test while failing the price half outright.
Fisher: 93 out of 100, verdict yes (the highest score)
Phil Fisher is the framework Tech bulls always wanted to invoke and rarely had the data to defend. The invest-like Fisher scorer tests for ten things: gross margin above 50%, revenue CAGR above 10%, EPS growth that outpaces revenue (operational leverage), ROIC above 15%, operating margin above 15%, net margin above 10%, conservative balance sheet, FCF quality, no share-count dilution, and a P/E tolerance ceiling at 35.
NVDA passes 9 of those 10. The one fail is the P/E ceiling at 35: NVDA's 45.6 is over the line. Everything else is a beat:
- Gross margin 71.1% against the 50% Fisher floor.
- Revenue CAGR 68% against 10%.
- EPS CAGR 88.6% outpaces revenue CAGR of 68% by a wide margin - the operational leverage signal Fisher built his "scuttlebutt" career around.
- ROIC 62.9% against 15%.
- Operating margin 60.4%, net margin 55.6%, both deep above their floors.
- Net debt to EBITDA 0.0x. FCF to net income 0.80x.
- Share count growing at -0.7% per year (i.e., shrinking through buybacks).
Fisher is the framework most willing to pay up for genuine pricing power, durable growth and management that compounds. NVDA scores 93 because almost every test Fisher would have applied through his scuttlebutt research is a beat: the gross margin says pricing power is real, the EPS-outpacing-revenue line says operating leverage is intact, the share count says management is capital-disciplined, and the R&D-to-revenue ratio of 8.6% is the kind of reinvestment Fisher would actively look for. The 45.6 P/E is the only line item the scorer dings, and Fisher historically tolerated higher multiples than Graham or Buffett for businesses with this profile.
Lynch: 63 out of 100, verdict unclear (the lowest score)
The Lynch score is the surprise in the other direction. NVDA passes 6 of 6 Lynch criteria the scorer tests for, including the famous PEG ratio at 0.52 (well under the 1.0 Lynch ceiling), EPS CAGR of 89% (over the 15% growth floor), revenue CAGR of 68% (over the 10% revenue floor), positive net margin, manageable debt and positive free cash flow. By the line-item count, NVDA looks like a perfect Lynch GARP play.
So why is the score 63 and the verdict "unclear"? Because the invest-like Lynch scorer applies a structural penalty on top of the criteria checklist. NVDA at a $5.2 trillion market cap is the largest stock by market value in the world. Lynch's GARP framework was explicitly designed around mid-cap discovery - businesses where the analyst community is thin enough that the average investor can read the same 10-K and form an edge. NVIDIA has dozens of dedicated sell-side analysts and is covered by every macro outlet daily. The "mid-cap sweet spot" Lynch optimised around is not where NVDA lives, even if every quantitative criterion passes.
There is also the GARP-vs-growth nuance. PEG at 0.52 looks compelling on the surface, but the growth rate in the denominator (88.6% EPS CAGR over five years) is unsustainable for a multi-trillion-dollar company over the next five years. Lynch warned readers explicitly about extrapolating extraordinary growth, and the invest-like scorer carries that warning into the verdict. The 6-of-6 criteria check is met, but the framework's spirit cautions against the trade.
This is the cleanest "score does not match verdict" output in the seven-framework set. We surface it on every Lynch page so users know the criteria count and the framework spirit do not always point the same direction.
Greenblatt (Magic Formula): 91 out of 100, verdict unclear
Joel Greenblatt's Magic Formula ranks every stock on combined earnings yield and ROIC, then buys the top decile. NVDA passes 2 of 5 line-item criteria:
- ROIC of 62.9% against the 25% floor: pass, by a wide margin.
- Gross margin of 71.1% against the 30% floor: pass.
- EBIT/EV yield of 2.4% against the 10% floor: fail.
- EV/EBIT of 42.0 against the 12 ceiling: fail.
- Earnings yield of 2.2% against the 6% floor: fail.
By line-item count, this is a 2/5 fail. But the rank-based Magic Formula is different from line-item screening. The Magic Formula ranks NVDA's ROIC against the entire US universe; NVDA's 62.9% ROIC ranks in the top 1 to 2% of the 6,621-stock apples-to-apples cohort. The earnings-yield rank lands in the bottom quintile. Combined, the rank-based formula puts NVDA roughly in the top decile on quality, the bottom quintile on price, with a combined rank that is borderline-acceptable on a 12-month hold.
The invest-like scorer compresses that into a 91/100 score because the ROIC excellence dominates the formula by design. The "unclear" verdict carries the cost-of-price warning: yes, Magic Formula will hold NVDA, but only because of the ROIC pillar, and only for one year before re-ranking.
This is the framework most explicitly built to tolerate cyclicality, but it has not been stress-tested against a semiconductor cycle since 2010. Magic Formula readers should understand that the framework rotates by design, so a high score today does not mean Greenblatt would hold NVDA through a cycle turn.
Munger: 87 out of 100, verdict unclear
Munger's mental-models filter is the framework closest in spirit to Buffett's, with three small differences: a slightly higher ROIC floor (18% vs 15%), an explicit "willingness to pay up for quality" with a softer P/E ceiling of 30, and a focus on capital intensity and balance-sheet conservatism.
NVDA passes 10 of 11 Munger criteria:
- ROIC 62.9% against 18%, 5-year ROIC 49.2% against 15%: both pass deeply.
- Gross margin 71.1% against 40%, operating margin 60.4% against 18%, net margin 55.6% against 12%, ROE 5y 58.8% against 15%: every quality test clears its floor.
- Net debt to EBITDA 0.0x against 1.5x, FCF/NI 0.80x against 0.80x: pass.
- Share count flat (down 0.7%/yr) and EPS CAGR 88.6% against 8%: pass.
- P/E 45.6 against 30: fail.
Munger's "willing to pay up for quality" softens the P/E test (Buffett's floor at the Wonderful-Fair-Price test is 25; Munger lets it stretch to 30), but NVDA is at 45.6. The score is 87 because the 10 passes dominate the math; the verdict is "unclear" because the one fail is the criterion Munger spent the second half of his career talking about (Daily Journal Meeting transcripts, repeatedly). "Pay up for quality" was never license to pay any price, and 45 times earnings for a cyclical-tinged business is the line Munger flagged in his own writing as the boundary of the principle.
T. Smith (Fundsmith): 89 out of 100, verdict yes
Terry Smith's Fundsmith framework is the strictest of the three "quality compounder" tests in the set. It uses a 20% ROCE floor (versus Buffett's 15%), a 45% gross margin floor (versus Buffett's 40%), a 95% FCF-to-net-income floor (versus Buffett's 80%), and a P/E ceiling of 35.
NVDA passes 10 of 12 criteria:
- ROCE 62.9% against 20%: pass, big.
- Gross margin 71.1% against 45%, operating margin 60.4% against 20%, net margin 55.6% against 15%: every margin test passes.
- 5y ROCE 49.2% against 18%, 5y ROE 58.8% against 18%: sustained quality, pass.
- Net debt to EBITDA 0.0x against 1.5x: pass.
- Revenue CAGR 68.3% against 7%, EPS CAGR 88.6% against 10%: growth passes.
- Share count flat: pass.
- FCF / NI 0.80 against the strict 0.95 floor: fail.
- P/E 45.6 against 35: fail.
The two fails are the same ones that show up across the other frameworks: cash conversion is good but not Fundsmith-grade (the 0.80 ratio is fine for Buffett but tight for Smith), and the multiple is above the price ceiling Smith would have set. The 10-of-12 pass count drives the 89 score and the "yes" verdict. The two fails are why this is a quality-compounder with caveats, not a Smith-style core-portfolio pick.
Where the frameworks agree on NVDA
Despite seven different lenses, four things show up in every scoring breakdown:
- The business is excellent. ROIC, margins, balance sheet, cash quality and growth all clear every framework's floor by a wide margin. The moat pillar of the Buffett Brain prints 100/100 because nothing in the data argues otherwise.
- The balance sheet is fortress-grade. D/E of 0.07, net debt-to-EBITDA of 0.00, current ratio of 3.9, halal debt ratio of 0.14%. Every framework that includes a balance-sheet test passes NVDA on it.
- Capital allocation is owner-friendly. Share count is shrinking (-0.7%/yr), R&D investment is at 8.6% of revenue (high but not bloated for this industry), there is no aggressive M&A footprint clogging the income statement, and management has not chased acquisitions or buybacks at peak.
- Growth is real and broad-based. 68% revenue CAGR and 89% EPS CAGR over five years are not artifacts of one quarter or one product cycle. Fisher's "EPS outpaces revenue" test passes by 20 percentage points, which is exactly the operational-leverage signal he built his methodology around.
Where the frameworks disagree on NVDA
Three points of disagreement:
- Valuation. Buffett, Munger, Smith and Greenblatt all flag the multiple. Fisher tolerates it. Graham technically passes the score because of the balance-sheet weighting, but lay-reading Graham would reject the P/E and P/B outright. Lynch's PEG passes mathematically (0.52) but the framework spirit cautions against extrapolating the growth rate.
- Cyclicality. Buffett and Graham implicitly weight semiconductor cyclicality more heavily than Lynch and Greenblatt. The numbers in the snapshot are five years of unbroken growth, but the industry's 30-year history is cyclical. The frameworks that prize multi-decade durability (Buffett, Munger) discount more for that than the frameworks that re-rank periodically (Greenblatt) or accept higher growth volatility (Fisher, Lynch).
- Size. Lynch is the only framework where the $5.2 trillion market cap is a structural negative. The other six frameworks are size-agnostic by design. This is why NVDA's Lynch score (63) sits 20 points below the cross-framework average (83.4) even though it passes every line-item criterion.
Does NVDA make the 47-stock all-7-frameworks cohort?
Yes. NVDA's minimum framework score is 63 (Lynch), which clears the 60 pass-threshold by 3 points. Its cross-framework average is 83.4 out of 100, which is the highest cohort average among the four mega-caps in the consensus list (NVDA 83.4, GOOGL 68.3, MSFT and ADBE close behind).
The full 47-stock consensus list is in our cross-framework data study. The headline finding there is that 0.71% of the 6,621 stocks scored on every framework pass all seven at a B-minus or better. NVDA is the largest company in that 47-stock cohort by market cap, which is itself a reframing of "is NVDA overvalued": the cross-framework consensus, applied to 6,621 stocks, lists NVIDIA as one of the 47 names that clears every test, multiple included.
That is not a recommendation to buy. It is the data answer to a question the value-investing canon implicitly poses: when all seven of these authors are asked the same question, do they all say yes? On NVDA, with the 60-point B-minus threshold, the answer is yes by a 3-point margin on the strictest framework.
Halal compliance: is NVDA AAOIFI Standard 21 compliant?
Yes. invest-like.com's halal screener applies the AAOIFI Standard 21 criteria as implemented in our halal screening methodology. NVDA's interest-bearing debt to market cap ratio is 0.14%, which is essentially zero and well below the 30% AAOIFI ceiling. The business activity (semiconductor design) is not on the prohibited-industry list. The non-permissible income ratio (interest income from cash holdings) is below the 5% threshold.
NVDA shows up on /halal/nvda/ as compliant, and the halal status field in the database (halal_status = 'compliant') is consistent with that. For halal-mode users, the framework results above are unchanged - the AAOIFI test sits on top of the value-investing test as an additional filter, not a replacement.
What this means for an investor
The seven-framework breakdown is not a "buy NVDA" or "sell NVDA" signal. It is a structured way to see which value-investing lenses agree on the business and which disagree on the price. Three observations a reader might draw, none of which are advice:
First, every framework agrees the underlying business is high-quality. If your edge is buying quality and holding, the business half of NVDA's test is uncontroversially passed.
Second, the disagreement is concentrated on valuation. Six of the seven frameworks explicitly flag the multiple (Buffett, Munger, Smith, Greenblatt, Lynch by spirit, Fisher by one line item) and only Graham passes it through the balance-sheet weighting. If you trust the canon's collective view of price, NVDA is expensive. If you discount the canon's view because the businesses it was calibrated against were smaller and slower, you will weight the bullish reads more.
Third, the cyclicality risk is the bear case that does not show up explicitly in any framework's score because none of the seven frameworks has a semiconductor-cycle test built in. Reading the frameworks correctly requires layering that history on yourself.
The Boardroom feature on /boardroom/nvda/ runs the four-investor debate in long form and is the natural follow-on to this post. It lets the four authors argue the cyclicality question directly rather than letting the scorers paper over it.
FAQ
Does NVIDIA pay a dividend?
Yes, but trivially. NVDA's current dividend yield is 0.018% - roughly $0.04 per share annualised at the current ~$215 price. The company has paid 20 dividends over the past 5 years, but the total dollar amount is negligible against the share price. NVDA is fundamentally a capital-appreciation story, not a dividend story. Investors looking for value-fit dividend names should browse /best/dividend-yield/ or /fit/dividend/ instead.
Is NVIDIA overvalued?
It depends on which framework you trust. By Graham's strict defensive rules (P/E above 15, P/B above 1.5), yes - by a wide margin. By Buffett's owner-earnings yield test (above 5%), yes - NVDA's yield is 1.77%. By Greenblatt's earnings-yield ranking, NVDA is in the bottom quintile of the universe on price. By Lynch's PEG (under 1.0), no - NVDA's PEG is 0.52 because earnings growth has run faster than the multiple expansion. By Fisher's "tolerable for quality" P/E ceiling of 35, NVDA at 45.6 is moderately expensive but not extreme. The seven-framework view: six say expensive, one (Fisher) says tolerable, one (Lynch by PEG) says actually cheap relative to growth.
Is NVDA a Buffett stock?
It is a partial Buffett fit. The invest-like Buffett scorer rates NVDA 72 out of 100 with a "C" letter grade and verdict "unclear" - 8 of 9 criteria pass, but the one fail is owner-earnings yield, the criterion Buffett weights most heavily. The Buffett Brain breaks this into pillars: moat 100, durability 70, management 49, valuation 55, financial health 61. A perfect moat and a 55-out-of-100 valuation pillar is the Buffett analyst's classic "wonderful business at a premium price" verdict - bullish on quality, cautious on entry price.
What are good NVDA alternatives in the same sector?
The invest-like 47-stock cross-framework consensus cohort contains several semiconductor names with stronger valuation profiles, though all are smaller. Broadcom (AVGO) at a market cap of $1.96T trades at a higher P/E (80.7) but with lower ROIC (17.6%). TSMC (TSM) trades at a P/E of 30.5 with a 25.8% ROIC - a more conventional Buffett shape. Micron (MU) trades at a P/E of 35.5 with a 27.7% ROIC but is a memory-chip business with much heavier cyclicality. None of these is a like-for-like substitute for NVDA's AI-accelerator position. The cleanest pure peer comparisons are on the NVDA vs AMD, NVDA vs TSM and NVDA vs AVGO pages.
Why does Graham give NVDA an "A" if P/E and P/B both fail?
Because the invest-like Graham scorer weights the balance-sheet and durability criteria (current ratio, debt-to-equity, 5-year profitability, revenue growth) alongside the price criteria (P/E, P/B, earnings yield), and the score is a weighted aggregate. NVDA's 4-of-7 pass count is dominated by balance-sheet excellence (current ratio 3.9, D/E 0.07, no loss years) that compensates for the valuation fails. A reader looking for the pure 5-rule Graham defensive screen would reject NVDA on P/E alone. We document this nuance on the methodology page at /methodology/.
Is NVIDIA halal?
Yes, by AAOIFI Standard 21 as implemented on invest-like. NVDA's interest-bearing debt to market cap is 0.14%, well below the 30% ceiling. Its business activity (semiconductor design and manufacturing) is not prohibited. Non-permissible income from cash interest is below the 5% threshold. The halal status field returns "compliant" on every API response for ticker NVDA. Halal-mode users can include NVDA in their watchlist with no purification required beyond the trivial portion attributable to interest income.
How often do these scores update?
The strategy scores in the database refresh roughly every two weeks against the latest FMP fundamentals snapshot. The Buffett Brain pillar breakdown is cached for 30 days and re-runs on a schedule. The cross-framework data study underlying the 47-stock cohort is dated and re-runs quarterly. Specific number citations in this post are stable against the 25 May 2026 snapshot but will drift over time as new earnings get ingested.
Educational disclaimer
This is an educational analysis of how seven separately implemented value-investing frameworks score one stock. It is not investment advice, not a buy or sell recommendation, and not a substitute for reading the original Buffett shareholder letters, Graham's Intelligent Investor, Fisher's Common Stocks and Uncommon Profits, Lynch's One Up on Wall Street, Greenblatt's Little Book that Beats the Market, Munger's Poor Charlie's Almanack, or Smith's Investing for Growth. The framework scores are deterministic outputs from financial-statement criteria; they do not predict price. Past performance of NVIDIA stock (or any stock) is not a forecast.
All scores cited in this post come from invest-like.com's strategy_scores and buffett_analyses production tables, snapshot 25 May 2026. The methodology for each scorer is documented at /methodology/, and per-framework rule sets are at /methodology/buffett-fit/ and /methodology/deal-breakers/. The same scoring logic powers every verdict on the production site, including the per-ticker page at /buffett/nvda/ and the public API response at /api/public/verdict/NVDA.
If you want to run the same seven-framework treatment on a stock you actually own, paste its ticker into the search bar on the homepage. The free tier gives three full verdicts a week. The 47-stock all-seven-frameworks cohort that NVDA sits inside is browsable at /blog/12500-stocks-7-frameworks-cross-framework-consensus/.