NVDA Is Up 108% in a Year. Our Model Says It's Still Cheap.
As of April 19, 2026
At $201.68 per share, NVIDIA's market cap stands at $4.9 trillion. The trailing P/E is 41x. The P/FCF is 84x. The price-to-sales ratio is 22.7x.
By traditional value metrics, this looks like an expensive stock. So I ran it through FactorForge.
Five of six lenses returned a verdict of Cheap. The sixth landed on Fair. That result deserves a close look — because the model isn't ignoring the expensive multiples. It's showing you exactly where the disagreement lives.
The Six-Lens Verdict
FactorForge scores stocks across six independent investing frameworks — you can read about how FactorForge scores stocks for full methodology detail. Each lens applies its own sector-calibrated thresholds and doesn't know what the others are doing.
| Lens | Score | Verdict |
|---|---|---|
| Balanced | 0.43 | Cheap |
| Value | 0.15 | Fair |
| Growth | 0.52 | Cheap |
| Quality | 0.47 | Cheap |
| GARP | 0.40 | Cheap |
| Momentum | 0.59 | Cheap |
All six lenses carry HIGH confidence — the model has sufficient data to form a firm opinion across every framework.
FactorForge scores stocks across six independent investing frameworks: value, growth, quality, GARP (growth at a reasonable price), momentum, and a balanced composite. Each lens applies its own sector-calibrated thresholds and doesn't know what the others are doing. Verdicts — Cheap, Fair, or Expensive — reflect each lens's aggregate score, not a single metric.
The One Dissenting Lens: Value
The value lens lands on Fair, and the data behind that verdict is straightforward.
Every static multiple scores at its worst possible level. P/E is 41.2x, above the model's unfavorable threshold of 40x for technology. P/FCF is 84.3x against a threshold of 50x. P/S is 22.7x against a threshold of 12x. EV/EBIT is 37.6x against a threshold of 20x. On those four metrics, the model sees no ambiguity — all four score at the floor.
But the model doesn't stop there.
NVIDIA carries zero debt (D/E of 0.00, Net Debt/EBITDA of 0.0x). ROE is 101.5%. The PEG ratio — which weights P/E against forward earnings growth — sits at 0.72, below the model's favorable threshold of 1.0. With a 3-year EPS CAGR of 50% and revenue growing at 73.2%, the model sees a company paying a large valuation premium but generating enough growth to partially offset it.
The value lens reads Fair rather than Cheap because the static multiples are too far above their thresholds for the growth offsets to fully compensate. The balance sheet and profitability metrics help. The multiples don't.
Growth: The Deceleration Signal
The growth lens scores Cheap — 73.2% revenue growth and a 50% EPS CAGR are hard to argue with.
But the model flagged something worth examining: revenue acceleration shows -48.7 percentage points over three years. That means NVDA's revenue growth rate was even higher in prior periods. NVDA isn't slowing down in absolute terms — it's growing at 73%. But the pace of acceleration has been declining, and the model weights that signal at 2.5 in the growth lens. It's the primary drag keeping the growth score from being higher.
That distinction matters. "Is this company growing?" and "Is growth speeding up or slowing down?" are different questions. NVDA passes the first one easily. The second one the model views more cautiously.
Quality: Where NVDA Dominates
The quality lens scores Cheap at 0.47, and the underlying data is exceptional across the board.
ROE is 101.5%. ROIC is 62.1%. Profit margin is 55.6%. Gross margin is 71.1%. Interest coverage is 503.4x — the model's favorable threshold is 3x. The quality lens is doing what it's designed to do: rewarding operational excellence.
One signal pulls the other direction: Earnings Consistency. The model measures this as the standard deviation of three-year EPS growth rates. NVDA's reads 227.8% — well above the 40% unfavorable threshold. That doesn't indicate bad earnings; it indicates highly variable ones. A company that went through an AI-driven revenue explosion will show enormous variability by construction. The model scores what it can measure.
What the Implied Growth Rate Tells You
The model surfaces a display metric I find useful context: the implied EPS growth rate — the annual growth rate that would justify the current stock price under a 10-year DCF framework, assuming a 10% discount rate.
At $201.68, that implied rate is 17.5% per year.
The 3-year historical EDGAR EPS CAGR is 50.0%. The gap is 32.5 percentage points.
The model scores what has happened, not what will. But that gap tells you something about what's embedded in the price: the market isn't pricing in a continuation of NVDA's recent growth trajectory. Whether NVDA delivers 17.5%, the 50% of recent history, or something in between is outside the model's scope.
What the Model Doesn't Capture
The model uses trailing GAAP earnings from EDGAR filings. The forward consensus P/E is 17.9x — surfaced in the model as a display metric, not a scored signal. The model's unfavorable threshold for P/E in technology is 40x; trailing GAAP P/E sits at 41.2x and scores at the floor. The forward figure reflects both analyst growth expectations and the GAAP/non-GAAP distinction: NVIDIA reports non-GAAP earnings that exclude substantial stock-based compensation, producing a materially different earnings number than GAAP. The model sees only the GAAP figure.
NVDA's scores also reflect trailing EDGAR data more broadly. The model doesn't know about Blackwell chip ramp timing, hyperscale data center demand signals, or competitive pressure from AMD and custom silicon (Google TPUs, Amazon Trainium, Microsoft Maia). All of those factors matter to NVDA's trajectory. None appear in the score.
FCF Margin shows N/A — the EDGAR capex data needed to compute it isn't populated in NVDA's most recent filing. The model surfaces P/FCF at 84.3x using market data as a proxy, but the FCF Margin metric is absent.
The Piotroski F-Score is 4/9 — mid-range between the model's favorable threshold of 7 and unfavorable threshold of 3. One contributing factor: the Piotroski penalizes a declining gross margin trend, and NVDA's shows -1.6pp over three years. For a company with 55% net margins and 101% ROE, that score is worth examining.
Beta is 2.33 — the model surfaces this as context, not a scored signal.
What Would Change This Rating?
The value lens is the interesting case here, since it's the sole dissenter.
If trailing P/E fell below 40x, the value lens score would tick up — but the P/FCF at 84.3x and EV/EBIT at 37.6x are further above their thresholds and would need meaningful compression to shift the verdict from Fair to Cheap.
The growth and GARP lenses would shift toward Fair if revenue growth decelerated significantly below the model's 12% favorable threshold. At 73.2%, there's substantial buffer, but the existing revenue deceleration signal (-48.7pp) indicates the model is already tracking that direction.
The quality lens would face pressure if the Earnings Consistency signal worsened — though at 227.8% volatility it is already scoring poorly and the strong profitability signals currently outweigh it.
Summary
At $201.68 with a $4.9T market cap, NVIDIA carries trailing multiples that score at their worst possible level on four of the value lens's most heavily weighted metrics. P/E of 41.2, P/FCF of 84.3, EV/EBIT of 37.6x — none of those are close to favorable territory.
But a PEG of 0.72, a 3-year EPS CAGR of 50%, revenue growing at 73.2%, and a balance sheet with zero debt present a different picture. Five of six lenses process that combination as Cheap. The value lens holds at Fair.
The reverse DCF adds context: the model estimates approximately 17.5% annual EPS growth is priced into the current share price over a 10-year horizon. The trailing historical EDGAR trajectory is 50%. The distance between those two numbers is the debate.
All scores and metrics are generated by the FactorForge scoring engine as of April 19, 2026. FactorForge classifies NVDA under Technology and applies sector-calibrated scoring thresholds accordingly. This is not financial advice. You are solely responsible for any investment decisions you make. Always consult a qualified financial advisor before acting on investment decisions. Do not rely solely on FactorForge for financial decisions.