The Wafer Process Layer Recovery: Why Our BUY Signals Are Clustering in Equipment and Materials
Tradestie Research — September 10, 2026
The Signal Nobody Is Watching
If you scanned our scoring system this week expecting the BUY list to read like an AI headline reel — it doesn't. The famous names are mostly absent. Instead, the densest cluster of BUY signals in the entire coverage universe sits in the wafer process layer: the chemicals, materials, inspection tools, and capital equipment that make chips before a chip is a chip.
The Semi/Wafer cluster currently has 7 of 24 companies rated BUY — a 29% hit rate, the highest-density semiconductor cluster in our system:
- KLAC (KLA) — inspection & metrology
- LRCX (Lam Research) — etch & deposition equipment
- ENTG (Entegris) — process chemicals & filtration
- MKSI (MKS Instruments) — process control subsystems
- NVMI (Nova) — metrology
- AMKR (Amkor) — OSAT / advanced packaging
- AXTI (AXT) — compound semiconductor substrates
Secondary clusters reinforce the picture: Semi/ASIC (AMBQ, MCHP, avg score 63.9), Semi/RF-analog (ADI, MTSI), and the Physical AI "Boring BOM" group (BA, DOV, NTAP, RIVN, TXT). But the wafer layer is where the density is.
This is a signal-driven observation. When a scoring system that evaluates each company independently starts agreeing on one supply-chain layer, the interesting question is why.
Why Now: Follow the Memory Money
The most likely driver sits in our sector money-flow data. As of September 9, Memory & DRAM is the top inflow sector in the entire system, with a flow score of 66.4 — well clear of second place (Oil & Gas E&P at 50.9). Under the hood the DRAM basket shows 100% CMF breadth and 100% momentum breadth: every holding is seeing accumulation, simultaneously.
Here is the transmission mechanism:
- HBM demand pull. AI accelerators are memory-bound; every new GPU/ASIC generation ships with more high-bandwidth memory stacks.
- DRAM capex response. Memory makers add wafer capacity — and memory fabs are voracious consumers of etch, deposition, metrology, and process chemicals. DRAM's 3D scaling roadmap (4F2, vertical transistors, hybrid-bonded HBM base dies) is increasing process intensity per wafer.
- Equipment and materials orders. That capex lands as revenue at LRCX (etch is disproportionately important in memory), KLAC and NVMI (more layers = more inspection steps), ENTG and MKSI (chemicals and subsystems scale with wafer starts), and AMKR (HBM and advanced packaging assembly).
The important nuance: this is memory-driven demand for process tools, a second engine distinct from the logic/GPU inference cycle everyone already owns. The market prices the GPU cycle daily. It re-prices the wafer-process layer slowly, in lumps, usually after the equipment companies guide up.
The Numbers
Live data from our database (prices Sep 9, 2026 close):
| Ticker | Price | Fwd PE | Rev Growth (QoQ YoY) | EPS Growth | Op Margin | Score | Rating | Street Target |
|---|---|---|---|---|---|---|---|---|
| KLAC | $182.91 | 34.4x | +15.2% | +13.8% | 42.5% | 63.6 | STRONG BUY | $233.77 |
| LRCX | $315.84 | 33.2x | +30.0% | +34.8% | 37.4% | 67.8 | BUY | $370.87 |
| ENTG | $143.03 | 29.2x | +11.5% | +75.1% | 19.3% | 57.7 | STRONG BUY | $173.36 |
| MKSI | $268.05 | 10.4x | +28.3% | +162% | 20.6% | 57.3 | STRONG BUY | $417.57 |
| AMKR | $51.35 | 16.9x | +25.6% | +218% | 10.5% | 58.9 | BUY | $76.40 |
| AXTI | $68.91 | 26.0x | +164.8% | −57.1% | 21.9% | 59.8 | BUY | $91.60 |
| NVMI | $373.85 | 26.2x | +15.9% | +2.9% | 29.9% | 57.6 | STRONG BUY | $532.00 |
Two things stand out. First, growth is already showing up — LRCX +30% revenue, MKSI +28%, AMKR +26% — this is not a hope trade. Second, the valuations are not stretched for the growth on offer: MKSI at ~10x forward earnings with 162% EPS growth, AMKR at ~17x with earnings tripling. These are cyclicals priced as cyclicals while the cycle is turning up.
The Technical Setup: Bottoming, Not Breaking Out
The tape confirms why our bottoming composite likes this group. Across the cluster the pattern repeats: RSI in the 40s, price at or below the 50-day SMA, but MACD histograms turning positive.
| Ticker | RSI(14) | MACD Hist | Price vs SMA50 | Read |
|---|---|---|---|---|
| KLAC | 45.8 | +1.33 | −9.8% | Basing below SMA50, momentum curling up |
| LRCX | 51.9 | +1.14 | −0.9% | Reclaiming SMA20, first through the base |
| ENTG | 52.2 | −0.21 | +1.6% | Already above SMA50 — furthest along |
| MKSI | 43.8 | +1.42 | −13.7% | Deep base, strongest histogram turn |
| AMKR | 48.4 | +0.66 | −11.0% | Basing, options flow strongest in cluster (69) |
| NVMI | 46.5 | +4.02 | −9.4% | Sharp momentum inflection |
| AXTI | 52.1 | −0.20 | +9.7% | Extended vs SMA50 — the one that already moved |
Nobody is chasing here. These are early-turn setups, which is precisely when a bottoming-composite system should fire — and it is firing on seven of them at once.
The Near-Miss Watchlist
Just behind the BUY cluster sits a group with bottoming setup scores of 90+ that hasn't crossed the BUY threshold yet. The common thread is the same: wafer processing and interconnect.
| Ticker | What They Do | Bottoming Setup | Score | RSI | Fwd PE | Why It Matters |
|---|---|---|---|---|---|---|
| INTT | Semiconductor test equipment | 100 | 58.8 | 43.8 | 21.6x | Highest setup score in our entire universe |
| FN | Optical networking / photonics mfg | 96 | 64.8 | 40.9 | 12.8x | +44.6% revenue growth at 13x forward |
| AIP | Interconnect (Arteris) | 95 | 65.6 | 39.9 | — | +46.2% revenue growth, chiplet interconnect IP |
| ICHR | Wafer fab fluid subsystems | 94 | 63.6 | 42.9 | 16.7x | Pure-play on wafer starts recovering |
| PLAB | Photomasks | 92 | 70.9 | 45.4 | 13.3x | Every chip needs one; 10x trailing PE, completely unpriced |
If the wafer-layer thesis is right, these are the names that upgrade next. One flag outside the cluster: AVGO sits at a 95 bottoming setup (score 59) in the ASIC group — if it crosses, Semi/ASIC becomes the second-densest BUY cluster and the thesis broadens.
How This Fits Our House Theses
This is a refinement of the Boring Bill of Materials thesis (August 2026), which covered the mature-foundry, power, and facility buildout around AI datacenters. This goes one layer deeper — inside the fab itself. Not what powers the building; what processes the wafer.
It also converges with two other house views: the 3D Stacking / hybrid bonding thesis (July 2026 — BESI, ASMPT, Disco, and again advanced packaging) and Chokepoint Durability on packaging. AMKR appears in multiple theses independently, which is worth noting: when separate frameworks keep landing on the same OSAT, the packaging bottleneck is probably real.
How This Could Be Wrong
An honest thesis names its failure modes:
- Memory capex is the most boom-bust line item in all of tech. DRAM makers have a decades-long habit of overbuilding into every demand signal and then slashing capex 40%+ when prices roll. If HBM demand plateaus or DRAM spot prices crack, the equipment orders driving this thesis get pushed out fast — and the stocks will front-run the cancellations.
- WFE guidance risk. The whole group trades on wafer-fab-equipment spending forecasts. One cautious guide from LRCX or KLAC on a memory-capex pause would hit the entire cluster at once. Correlated positions mean correlated drawdowns; this is one trade, not seven.
- China export-control exposure. KLAC and LRCX each derive a large share of revenue from China. A new round of export restrictions — or Chinese domestic substitution ramping faster than expected in deposition/etch — is a persistent, headline-driven risk that our scoring system cannot see coming.
- Signal-density is not causality. Seven independent BUYs in one cluster may reflect a shared factor exposure (all bottoming after the same sector drawdown) rather than a shared fundamental driver. The DRAM money-flow link is our best explanation, not a proven one.
Bottom Line
The market is paying up for the visible layer of the AI trade and slowly repricing the layer that makes it physically possible. Our system — evaluating each company on its own — is finding the wafer process layer bottoming in unison, at cyclical-trough valuations, just as memory money flow turns decisively positive. Watch the near-miss list for upgrades; watch DRAM flow scores for the invalidation.
Data sourced from Semi Supply Chain · Market Pulse · Top 50 Growth