Playbook › Themes
Semiconductor materials, consumables & subsystems
Claim
The thesis under test was: materials and consumables are less cyclical than equipment, because they're consumed per wafer rather than bought per fab. The mechanism is real but was stated backwards in the first pass.
Materials consumption tracks fab utilization / wafer starts. Equipment orders track fab capex. These are different cycles, and which one is falling determines which layer is defensive:
| Downturn | Materials | Equipment | Type |
|---|---|---|---|
| 2019 | −1.1% | −7.3% | capex downturn → materials defensive ✅ |
| 2023 | −8.2% | −1.3% | utilization downturn → thesis inverts ❌ |
A demand shock hits utilization first; capex lags because equipment ships against backlog. So in the downturn that usually matters most, materials is the more exposed layer, not the less. SEMI named CMP among the largest 2023 contractions — precisely the line ENTG and Qnity dominate.
Evidence — July 2026 drawdown (SOX peak close 6/22 → 7/29)
| Name | Drawdown | Net leverage |
|---|---|---|
| ENTG (75% unit-driven — the thesis incarnate) | −41.8% | 3.6x |
| MKSI | −38.7% | 3.5x |
| LRCX | −38.4% | — |
| KLAC | −36.8% | — |
| AMAT | −31.8% | — |
| Q (Qnity) | −29.7% | ~2.0x |
| ^SOX | −28.6% | — |
| ASML (equipment) | −19.5% | — |
Entegris was the worst performer in the group — worse than every equipment name — and not on earnings, since it doesn't report until Aug 4. The best performer was an equipment company. The equipment/materials axis explained almost nothing; leverage and AI-narrative beta explained far more — the two most levered names were the two worst.
Ranking
Qnity > Entegris on this layer. Q's 10-K states >90% consumable/unit-driven (vs ENTG's ~75%), leverage ~2.0x vs 3.6x, guidance raised, smallest drawdown in the cohort. ENTG has now declined in revenue two consecutive years while WFE grew, and its apparent 2023 resilience was a CMC Materials acquisition artifact, not organic.
Exclusions
- 🔴 CBT is not a semiconductor company. CMP slurries left Cabot in 2000 (spun as Cabot Microelectronics → CMC Materials → acquired by Entegris 2022). Cabot today is ~63% carbon black for tires. It held up (+0.9%) because it has no AI narrative to compress. Stripping it moves the group drawdown from ~−27% to ~−33%, converging on equipment — its presence was flattering the whole thesis. See [[pitfall-cbt-is-not-a-semi-company]].
- ROG is not semi materials — power-electronics packaging, zero wafer-start exposure.
- ESI trades ~31% below its implied deal value ($34.79 vs ~$50.10) — but the gap is Solstice's own share price collapsing, making ESI a leveraged SOLS proxy rather than a merger-arb spread.
What would falsify this
A capex-led (not demand-led) downturn in which materials names outperform equipment by the 2019 margin. That would confirm the original thesis conditional on downturn type, which is the version worth holding.
🚩 Unresolvable gaps
- SEMI materials data for 2001, 2002 and 2008–2010 is behind a paid subscription. The two deepest downturns in semiconductor history cannot be used to test this at all. Anyone asserting "materials held up in 2009" is doing so without a number.
- ENTG FY2025 margins are corrupted in both Yahoo and roic.ai — the widely quoted "41% EBITDA margin" is fiction; true figure is 13–17%. See [[pitfall-entg-margins-corrupted-in-vendor-feeds]].
- A circulating wire-service claim that MKS reported Q2 revenue of $920M on June 24 is false. ENTG, Q and MKSI all report Aug 4–5; every Q2 figure for them is still guidance.
History
- 2026-07-29 — swept. Initial conclusion: "materials is the less cyclical layer."
- 2026-07-30 — mechanism corrected and ranking reversed after primary-filing research. Utilization-vs-capex distinction established; ENTG revealed as worst performer; CBT excluded; Qnity promoted above Entegris. The original conclusion was directionally wrong and is preserved above as the falsification record.