Best-in-class etch monopoly riding a once-a-decade memory/AI WFE upcycle into record margins — but at ~69x guided FY26 GAAP EPS the tape, not the business, is the risk; the moat is real, the multiple is borrowed from the future.
| Date | Type | What happened | Source |
|---|
| 2026-07-25 | editorial note | Valuation figure revised: $385.63 → 21%Valuation moved from $385.63 (deep-dive-2026-06-18.md) to 21% (deep-dive-2026-07-25-refresh.md). | dossier |
| 2026-07-25 | editorial note | Capex figure revised: $600M → $20.5BCapex moved from $600M (deep-dive-2026-06-18.md) to $20.5B (deep-dive-2026-07-25-refresh.md). | dossier |
| 2026-07-25 | editorial note | Margin figure revised: $2.911B → 35.0%Margin moved from $2.911B (deep-dive-2026-06-18.md) to 35.0% (deep-dive-2026-07-25-refresh.md). | dossier |
| 2026-07-25 | editorial note | Revenue figure revised: $2.22B → $5.841BRevenue moved from $2.22B (deep-dive-2026-06-18.md) to $5.841B (deep-dive-2026-07-25-refresh.md). | dossier |
| 2026-07-25 | editorial note | Verdict changed: Best-in-class etch monopoly riding a once-a-decade memory/AI WFE upcycle into record margins — but at ~69x guided FY26 GAAP EPS the tape, not the business, is the risk; the moat is r…Before (deep-dive-2026-06-18.md): Best-in-class etch monopoly riding a once-a-decade memory/AI WFE upcycle into record margins — but at ~69x guided FY26 GAAP EPS the tape, not the business, is the risk; the moat is real, the multiple is borrowed from the future. After (deep-dive-2026-07-25-refresh.md): (no verdict) | dossier |
Primary sources
SEC filings
Earnings calls
Source documents — open to read in full
Filings — NOTHING NEW.
--form 10-Q --limit 2) matched only the two already-ingested filings (Q3 FY26 report 2026-03-29, filed 2026-04-23; Q2 FY26 report 2025-12-28) — both skipped as existing. This is expected: FY2026 ended 2026-06-28, so the next SEC filing is the FY26 10-K (typically ~mid-August), not a 10-Q. [primary: EDGAR submissions, CIK 0000707549]--limit 5) returned only filings dated 2026-04-22 and earlier — i.e. no 8-K has been filed since the Q3 earnings release (2026-04-22). Confirms no material corporate event was 8-K-disclosed in the refresh window; the next Form 8-KA filing for something that happened between reports and matters enough to tell shareholders now. will be the 2026-07-29 Q4 earnings release. Newly written to disk: 8-k-2026-q2.md (the Apr-22 Q3 results Form 8-KA filing for something that happened between reports and matters enough to tell shareholders now. — cover only; the numbers live in Exhibit 99.1). [primary: EDGAR submissions]Transcript — NEW, fills a prior gap.
transcripts/2026-q3.md — the Q3 FY26 earnings call, held 2026-04-22 5:00pm ET (Archer + Bettinger), 63.5 KB, from fool.com [primary: transcript 2026-Q3, source fool.com]. The 2026-06-18 dossier explicitly flagged "No transcripts on disk — Lens 6 is web-only." Lens 6 is now primary-grounded (see §5). This is the latest call (Q4 has not yet reported), so it is the current management-sentiment anchor.No new financials/segments/guidance CSVs populated (still header-only — carried-forward open item).
| Metric | 2026-06-18 (prior) | 2026-07-24 (now) | Δ |
|---|---|---|---|
| Price | ~$385.63 | $305.21 | −21% |
| Market capitalisationThe share price multiplied by the number of shares. What the market says the equity is worth. | ~$482B | $381.7B | −$100B |
| Trailing P/E | ~73x (on TTM EPS ~$5.29) [prior estimate] | 57.7x | de-rated |
| Forward P/E | ~69x FY26E [prior estimate] | 40.0x | de-rated |
| EPS (TTM) | ~$5.29 [prior estimate] | $5.29 | unchanged |
| 52-wk range | — | $90.94 – $438.50 | new 52-wk high set, then −30% off it |
What happened (mechanics, ``): after roughly doubling / +150% in H1 CY2026, LRCX made a 52-week high ($438.50) and then a crowded WFE trade unwound violently through early-mid July. Notable single-session drops on the sector: Jul 2 −10.2% (LRCX close $351.41; same day KLA ≈−12%, AMAT ≈−7%, Teradyne ≈−14%, Axcelis ≈−19%), then further down-days (Jul 7 −7.5%, Jul 13 −5.3%, Jul 15 −6.3%, Jul 16 −3.6%). Cumulative >15% over consecutive sessions.
Stated drivers of the de-rate: (1) fear that AI-infrastructure capex is closer to a peak than a floor; (2) valuation compression off a >70x trailing multiple; (3) memory-cyclicality warnings — one bear model has Lam system-shipment growth decelerating to ~3% in CY2026 vs +82% in CY2025; (4) renewed China-exposure reassessment (34–35% of revenue). This is precisely the "the multiple is the position" bear scenario the 2026-06-18 dossier laid out — the market began pricing the cycle/multiple tail. Note the business news that triggered it was macro/sentiment, not a Lam-specific negative print.
Sell-side (now ``): ~36 analysts, 29 Buy / 6 Hold / 1 Sell; average target clusters ~$315–$368 depending on source (S&P Global avg ~$368 "Strong Buy"; a 41-analyst set $315). Recent action: Evercore ISI raised to $355 from $300 (Outperform). Reframe vs prior dossier: on 2026-06-18 the median target ($315) sat below spot ($385) — the tape had run past the optimists. Now spot ($305) sits at/below most targets — the tape has come back to (or under) the analysts. The sign of the gap has flipped. (Disregard stray "$105→$120 Argus" / "$170→$200 Mizuho" hits in search — those are stale pre-de-rate or older-cycle prints; do not use.)
No new quarter has printed since the last dossier, so the hard financials (Q3 FY26, ended 2026-03-29) are unchanged and carried forward: revenue $5.841B (+24% YoY, +9% QoQ), GM 49.9%, op margin 35.0%, diluted EPS $1.47 non-GAAP / $1.45 GAAP, first-ever $2.0B+ CSBG quarter [primary: filings/10-q-2026-q1.md; transcript 2026-Q3]. What the newly-ingested transcript adds is the finer memory mix and the forward memory-capex narrative — directly relevant to this week's HBM/DRAM theme:
Lam's own memory numbers (Q3 FY26 systems-revenue mix) [primary: transcript 2026-Q3]:
Forward memory-capex read (Lam management, [primary: transcript 2026-Q3]):
External corroboration of the memory-WFE upcycle:
The gap worth watching: Lam's $140B CY2026 WFE call (with "bias to the upside") sits ABOVE SEMI's ~$135B WFE and well above Morgan Stanley's ~$128B [primary: transcript 2026-Q3; web: SEMI/MS, 2026]. Lam is the most bullish voice on the number; the July de-rate is partly the market questioning whether $140B is a peak. Net read for this week's HBM/DRAM theme: the equipment layer confirms memory (esp. DRAM/HBM) is the fastest-growing WFE pocket in 2026, and Lam is the most memory/etch-dep-intensity-levered way to express it — which is exactly why it is also the most exposed to a peak-cycle re-rate. [synthesis of primary + web above]
The imminent catalyst — Q4 FY26 print, 2026-07-29 after-close (the June quarter; fiscal-year-end):
Catalyst list (updated):
{EARNINGS, "Q4 FY26 print — tests $6.6B/$1.65 raise, June-qtr China decline, and CY2027 WFE framing into a de-rated tape", 2026-07-29 after-close} — 4 days out, dominant near-term catalyst.{MACRO, "Memory (DRAM/HBM/NAND) capex trajectory — Lam's highest-beta driver; the July de-rate is a bet that CY2026 WFE ~$140B is a peak", CY2026-27}{REGULATORY, "Further BIS China tightening / affiliate-rule scope; enforcement continuing (AMAT $252M Feb settlement precedent) — but annual license approvals preserve a limited maintenance/upgrade channel", ongoing CY2026}{LAUNCH, "Advanced-packaging +50% CY2026; NAND $40B conversion pulled forward to <end-2027; DRAM 1C / Stryker ALD ramp", CY2026-27} [primary: transcript 2026-Q3]Prior dossier had Lens 6 as `` only. With the ingested Q3 FY26 transcript, the sentiment read is now primary [primary: transcript 2026-Q3]:
Tone = confident, AI-anchored, explicitly refusing to call a peak. Recurring frame: rising etch/deposition intensity from 3D-scaling architecture inflections drives SAM expansion + share gain ("multiyear outperformance setup"). Asked directly (Deutsche Bank) whether "this memory cycle is different" and how they de-risk cyclicality, Archer declined to call a peak "given the tremendous demand," and leaned on operational flexibility (Dextro cobots, Equipment Intelligence, footprint) to reduce fixed-cost scaling — "if you look at our track record, in those periods we've also outperformed." Bettinger's tell on discipline: "everybody is just kind of lugging into where demand is… disciplined investment." Other primary signals: installed base >100,000 chambers; utilization "as full-out as it can be" (spares/service growth stabilizing at highs, not accelerating from here); customer conversations now extend into 2027-2028 fab openings (longer visibility, no down-payments needed); customer down-payments at a ~4-year low (correlated with smaller China customers — a China-demand tell, not a broad-demand tell). Management committed to updating its long-term margin/opex model "later in the year" (currently running above the old 35% op-margin target).
No new financial statements since the last dossier, so the forensic read is unchanged; re-stating only the live watch-items with any new color:
Lenses 1 (overview), 2 (supply chain), 3 (moats), 9 (management), 11 (forward projection), 12 (bull/bear), 13 (devil's advocate), 14 (management questions) are unchanged — see the previous dossier. Headlines still valid: etch #1 (~45% share) / deposition #2; WFE oligopoly (AMAT/LRCX/TEL ~70% of dep/etch/clean); ~72% ROIC; ≥85% FCF returned; top-10 customers ~93% of FY25 revenue; two customers ~32%; China ~34%; CEO Tim Archer / CFO Doug Bettinger long-tenured, clean disclosure. The forward-projection EPS ladder is unchanged (FY26E ~$5.60; base-FY27E ~$6.20; bull-FY28E ~$8.00 / bear ~$4.20) — but the multiple applied to it has moved a lot: at $305, that ladder now implies ~54x FY26E, ~49x base-FY27E, versus ~69x / ~62x at the prior $385. The de-rate compressed the valuation gap the prior dossier called "the bear case in one row" without changing the earnings estimates.
our figures, our figures, our figures, our figures, our figures) — populating them (esp. the systems/CSBG + memory-submix time series now available from the transcript) would make future refreshes diff-able. Carried-forward gap.kb/hardware/wiki/*.md pointers unresolved) — Phase A still leans on filings + web. Carried-forward gap.Refresh generated 2026-07-25. Grounding: 3 carried-forward SEC filings + newly-ingested Q3 FY26 transcript (primary) + web market/consensus/WFE data (labeled). No new filings existed to ingest (next filing = FY26 10-K, ~August). Own view: NOT ASSESSED — data + mechanics only.
Every dossier we have written on Lam Research, newest first.
Best-in-class etch monopoly riding a once-a-decade memory/AI WFE upcycle into record margins
Covered in the Knowledge Base
Hardware — Memory & Compute Architecture
| Industry | Semiconductors |
| Size | Public Company |
Where Lam Research sits against the other names we cover on this beat. Each line is that company’s verdict, not a summary of it.
The Q2 beat was not a peak — July revenue accelerated to +44.7% YoY, the board appropriated a single-tranche US$29.4B of capacity capital, and H1 cape…
The Q2 print did the one thing the bear case could not survive
Cash $4.7B
The thesis got WIDER and the price got WORSE.
Cash $19.6B
The de-rate the June dossier warned about arrived early and for the exact reason it named
Cash $1.4B
LONG — MEDIUM (conviction DOWN one notch from the prior MEDIUM-HIGH, on wider risk, not a worse business).
Cash $25.0B