CapyStudio CapyStudio

Compute Valuation Snapshot — Semiconductor Basket

Date: Tuesday, May 12, 2026
Generated: 2026-05-12 09:17 PDT
Category: Compute
Universe: KLAC, AMAT, LRCX, SMTC, TSEM, INTC, AMD, NVDA, ASML, MU, ARM
Data source: Yahoo Finance / yfinance snapshot, current as of generation time. Metrics are approximate and may drift intraday.

Executive Summary

The semiconductor selloff creates a tempting “buy the dip” setup, but on traditional valuation metrics most of this basket is still not obviously cheap. The exceptions are MU and NVDA, for very different reasons:

Most attractive on traditional + growth-adjusted valuation:

  1. NVDA — best quality-adjusted value.
  2. MU — cheapest statistically, but highest cyclicality risk.
  3. AMD — acceptable if AI accelerator estimates keep rising, but less compelling than NVDA.

Ranking: Relative Cheapness

1. MU — Cheapest Statistically, Highest Cycle Risk

Snapshot:

Verdict: Statistically cheap, cyclically dangerous.

MU is the clear quantitative value winner. A 7.2x forward P/E with a 0.26 PEG is rare in this basket. But this is memory-cycle math: HBM/DRAM pricing and margins are booming. If the cycle stays tight, MU is cheap; if supply catches up or pricing weakens, the forward multiple is a mirage.

Best use case: high-conviction bet that AI memory tightness persists.

2. NVDA — Best Growth-Adjusted Value

Snapshot:

Verdict: Not optically cheap, but very cheap relative to growth and margins.

NVDA still looks expensive on P/S, but every other quality-adjusted metric is strong: sub-20x forward P/E, 73% revenue growth, 56% net margin, 65% operating margin, and a sub-1 PEG. Compared with the rest of the semiconductor basket, NVDA is still the cleanest combination of scale, profitability, and growth.

Best use case: core AI compute exposure when the market sells semis indiscriminately.

3. AMAT — Most Reasonable Equipment Name

Snapshot:

Verdict: Highest-quality “not too crazy” equipment exposure, but not cheap.

AMAT is the most reasonable of the equipment group on forward P/E and P/S, but revenue is currently shrinking. It can work if capex expectations rebound, but by traditional metrics it is more “acceptable” than cheap.

4. LRCX — Strong Growth, Rich Multiple

Snapshot:

Verdict: High quality, not a bargain.

LRCX has better current growth than AMAT/ASML/KLAC, but the market is already paying for it. Forward P/E above 35x and EV/EBITDA near 47x leave limited valuation cushion.

5. ASML — Elite Moat, Fair-to-Rich Valuation

Snapshot:

Verdict: Buy for moat, not for cheapness.

ASML remains a unique monopoly-like asset in EUV lithography. But traditional metrics are not cheap: nearly 49x trailing earnings, 31x forward earnings, and 17x sales for low-teens revenue growth.

6. KLAC — Excellent Margins, Expensive Versus Growth

Snapshot:

Verdict: Great business, expensive stock.

KLAC has fantastic profitability and ROE, but growth is modest relative to valuation. It is a quality compounder, not a value candidate.

7. AMD — Reasonable Forward Setup, Weak GAAP Optics

Snapshot:

Verdict: Not cheap on GAAP, acceptable if AI accelerator growth keeps scaling.

AMD is a bet on future margin expansion and AI accelerator share gains. The forward P/E and PEG are tolerable, but the current GAAP P/E and margins are much weaker than NVDA’s. If the goal is traditional valuation discipline, AMD is less attractive than NVDA despite a stronger “catch-up” narrative.

8. SMTC — Turnaround, Not Value

Snapshot:

Verdict: Requires a turnaround thesis.

SMTC has negative GAAP profitability and a rich forward multiple. It does not screen cheap on traditional metrics.

9. INTC — Turnaround Speculation, Not Traditional Cheapness

Snapshot:

Verdict: Not cheap, despite legacy-value perception.

Intel is expensive on forward earnings, unprofitable on GAAP, and free-cash-flow negative. It may work as a turnaround/foundry/geopolitical subsidy story, but it is not a traditional value stock today.

10. TSEM — Expensive for the Quality Profile

Snapshot:

Verdict: Avoid on valuation.

TSEM does not offer enough growth or profitability to justify its current valuation. High P/E, high P/S, weak FCF yield, and low ROE make it unattractive on traditional metrics.

11. ARM — Strategic Asset, Very Expensive Stock

Snapshot:

Verdict: Most expensive stock in the basket by traditional metrics.

ARM has a powerful royalty model and huge strategic relevance across mobile, AI edge, and custom silicon. But that quality is more than priced in. At ~241x GAAP earnings and ~47x sales, it is not cheap.

Comparative View

Rank Ticker GAAP P/E Fwd P/E P/S Rev Growth Net Margin PEG FCF Yield Verdict
1 MU 34.4x 7.2x 14.1x 196.3% 41.5% 0.26 0.4% Cheapest, cyclical
2 NVDA 44.6x 19.3x 24.5x 73.2% 55.6% 0.68 1.1% Best growth-adjusted value
3 AMAT 43.0x 29.0x 11.8x -2.1% 27.8% 2.07 1.3% Reasonable equipment exposure
4 LRCX 52.9x 35.4x 16.2x 23.8% 30.9% 1.72 1.2% Quality, not cheap
5 ASML 48.7x 30.9x 17.0x 13.2% 29.7% 2.36 1.4% Moat, not bargain
6 KLAC 49.8x 35.3x 17.5x 11.5% 35.7% 1.98 1.3% Great margins, expensive
7 AMD 146.5x 33.8x 19.0x 37.8% 13.4% 1.10 1.0% AI upside, weaker GAAP quality
8 SMTC NM 43.8x 11.3x 9.3% -3.8% 1.21 1.2% Turnaround needed
9 INTC NM 77.1x 11.1x 7.2% -5.9% 1.36 -1.4% Not cheap; turnaround/spec
10 TSEM 109.6x 43.9x 16.2x 13.7% 14.1% 5.07 ~0.0% Expensive for profile
11 ARM 240.7x 67.7x 46.6x 20.1% 17.1% 2.39 0.4% Strategic, very expensive

Investment Conclusion

Buy-the-dip candidates

NVDA is the cleanest candidate. It has the best combination of growth, margins, earnings power, and valuation. If the selloff is sector-wide rather than NVDA-specific, NVDA is still the best “quality at a reasonable growth-adjusted price” name.

MU is the most statistically undervalued, but should be sized as a cyclical trade rather than a sleep-well compounder. The thesis depends heavily on sustained AI memory tightness and HBM pricing power.

Watchlist / selective entries

AMD is interesting, but only if one believes its AI accelerator roadmap will drive major margin expansion. It is not cheap on GAAP.

AMAT / ASML / LRCX / KLAC remain strong companies, but the current multiples do not offer much margin of safety. Of these, AMAT screens most reasonable.

Avoid on valuation discipline

ARM, INTC, TSEM, SMTC do not currently satisfy traditional cheapness tests. ARM is an excellent strategic asset but priced like perfection. Intel is a turnaround, not a value stock. TSEM and SMTC do not offer enough profitability/FCF support for their multiples.

Final Ranking

Best relative value:

  1. NVDA — strongest growth-adjusted value.
  2. MU — cheapest statistically, but most cyclical.
  3. AMD — acceptable if AI growth accelerates; not GAAP-cheap.

High-quality but not cheap:

  1. AMAT
  2. ASML
  3. LRCX
  4. KLAC

Not cheap / valuation-risky:

  1. SMTC
  2. INTC
  3. TSEM
  4. ARM

This is research, not financial advice. Metrics are point-in-time estimates and should be refreshed before execution.