Lam Research
Capital-equipment maker selling etch + deposition systems to chip fabs, with a high-margin recurring installed-base services/spares/upgrades annuity (CSBG); revenue is cyclical and gated to customer fab capex
The thesis on this name
State of the Memory Supercycle
The most memory-levered of the WFE toll-booths and the highest-quality way to own the supply side rather than the volatile commodity. Roughly a third of revenue is memory equipment, and management noted DRAM hit a company record 23% of systems in FQ2'26 on HBM3E/4 and 1b/1c node migrations (Futurum, Q2 FY26). Lam dominates the etch + deposition steps that gate 3D-NAND layer stacking and the high-aspect-ratio DRAM capacitor — the steps that scale with bit-density, not just wafer count. WFE is guided to ~$135B for cal-2026 (Lam mgmt) with etch/dep share expanding. As a toll-booth Lam earns on every maker's capex (Micron's FY26 capex +45% to ~$20B; SK Hynix's $8B ASML commitment) regardless of which oligopolist wins HBM4 — a more diversified claim on the cycle than any single memory maker.
State of the Memory Supercycle
The purest memory-WFE toll-booth — etch & deposition gated to DRAM/NAND; DRAM hit a company-record 23% of systems on HBM3E/4 and 1b/1c node migrations, with Akara 3D-DRAM etch and Aether dry-resist tool-of-record wins.
State of the Memory Supercycle
Purest memory-WFE; DRAM record 23% of systems, 3D-DRAM etch + dry-resist wins. Highest cycle beta of the trio.
Earnings, margins, COGS & capex
Lam is in a sharp up-cycle: FY2025 revenue rose 23.7% to $18.44B (10-K) and the Mar-2026 quarter set a record at $5.84B (+24% YoY) with gross margin at a near-record 49.9% non-GAAP (fact). The mix has shifted hard toward memory — DRAM hit a record 27% of systems on HBM and 1c-node spend — and the CSBG service annuity crossed $2B/quarter for the first time, structurally lifting margins (fact). Management guided the June-2026 quarter to $6.6B revenue and 50.5% gross margin, and raised its 2026 WFE forecast to $140B with upside bias (fact).
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~48¢ is cost of goods and ~14¢ operating expense, leaving ~37¢ of operating profit (~1¢ net).
Revenue trend
Margins
up
up
up
flat/up
COGS structure
COGS is dominated by purchased components and materials for etch/deposition systems (RF power, chambers, gas/flow, electrostatic chucks, vacuum, precision machined parts) plus factory labor/overhead; the high-fixed-cost system build means gross margin levers on volume, product mix (newer tools richer), and the rising CSBG/services share. FY2025 gross profit was $8,979M (~48.7%); mix shift to services + factory efficiency drove margin to 49.9% in Q3 FY26 (fact).
Capex
Asset-light: capex funds R&D/lab cleanrooms, the Malaysia/Korea/India manufacturing and engineering footprint, and the global service network — low-single-digit % of revenue (estimate). The bulk of investment is R&D, not plant.
Latest earnings
Beat, and the guide was the bigger story. Revenue $6.72B came in above the ~$6.65B consensus and non-GAAP EPS $1.82 above ~$1.68; non-GAAP gross margin of 52.0% landed above the guided 50.5% +/-1pt range. Fourth straight record revenue quarter, with record operating margin and record EPS. The shares rose 17.98% the next session, their largest one-day gain since 1999
Sept-2026 quarter (Q1 FY2027): revenue $8.10B +/- $400M, gross margin 52.0% +/- 1pt, operating margin 39.5% +/- 1pt, EPS $2.15 +/- $0.15, on a 1.255B diluted share count — roughly $1B (~14%) above consensus. On the call management raised its calendar-2026 WFE forecast to 'the low $150 billion range' from the prior $140B, described 2027 as 'an extraordinary setup for WFE growth' with '8 to 10 new fabs coming online', and said China overall WFE is 'flat to slightly up'
- Q4 FY2026 revenue
- $6,722.2M (+30% YoY, +15.1% QoQ)
- Q4 FY2026 non-GAAP gross margin
- 52.0% (+210bps QoQ) — highest in 20 years
- Q4 FY2026 non-GAAP operating margin / EPS
- 38.4% / $1.82 (GAAP 37.4% / $1.81)
- FY2026 revenue / GAAP diluted EPS
- $23,232.7M (+26.0%) / $5.76 (+38.8% vs $4.15)
- CSBG (customer support-related revenue)
- $2,472.4M — a record, +42.6% YoY and +17.1% QoQ
- Memory as % of systems revenue
- 46% — a record, up from 39% in March; NAND 23% (from 12%), DRAM 23% (from 27%)
- China as % of revenue
- 26%, down from 34% in the March quarter; Taiwan 27%, Korea 20%
- Sept-2026 quarter guidance
- Revenue $8.10B +/- $400M; EPS $2.15 +/- $0.15
- Capital returned
- $571.9M in Q4 (buybacks $246.6M + dividends $325.3M) = 45% of Q4 FCF; $5,122.0M for FY2026 = 105% of FY FCF
- Deferred revenue
- $2.43B, up from $2.22B; plus ~$490.2M of estimated future revenue from Japan shipments held in inventory
Growth drivers
- HBM3E/HBM4 high-bandwidth memory — DRAM hit a record 27% of systems (Q3 FY26); etch/deposition intensity per HBM wafer is rising (fact)
- DRAM node migration to 1b/1c — Aether dry-resist EUV and Akara conductor etch winning tool-of-record positions (fact)
- NAND conversion super-cycle — ~$40B of NAND upgrade spend now pulled forward to before end-2027, a tailwind for Lam's etch-heavy NAND share (fact)
- 3D-DRAM transition — Akara is extendible to 4F2/3D-DRAM, deepening Lam's etch share as DRAM goes vertical (fact, forward)
- CSBG installed-base annuity — first $2B+ quarter, mid-teens services growth; recurring, less cyclical, margin-accretive (fact)
- Gate-all-around (GAA) logic/foundry ramp — Akara + ALD/ALTUS Halo molybdenum fill add etch/dep content at leading-edge foundry (fact)
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-08-07. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’26
Bull & bear
Every load-bearing part of the read got confirmed at once: memory hit a record 46% of systems, the service annuity hit a record $2.47B, gross and operating margins hit records, and management guided the September quarter ~$1B above consensus while raising 2026 WFE to the low $150B range. Meanwhile the stock is ~32% below its June level, so the valuation objection that led the bear case has substantially deflated.
- A guide, not just a print: $8.10B +/- $400M for the September quarter against ~$7.1B consensus — a ~14% beat on the outlook — with operating margin guided to 39.5% and EPS to $2.15, implying the up-cycle is accelerating rather than digesting
- Memory leverage is now demonstrated, not asserted: memory reached a record 46% of systems revenue (from 39%), and the mix is broadening — NAND jumped to 23% from 12% while DRAM held near its record at 23%, so both legs of the memory build are now paying
- The annuity is compounding faster than the systems business: customer support-related revenue was a record $2,472.4M, +42.6% YoY versus systems' +23.6%, taking the recurring share to 37% of revenue and structurally lifting the through-cycle margin floor
- Margins are at 20-year highs with more guided: 52.0% non-GAAP gross margin beat the guided range, operating margin hit a record 38.4%, FY2026 gross margin rose to 50.5% from 48.7% — and management put a 'mid 50%' long-term gross-margin target on the table
- The top-down number moved up, not down: calendar-2026 WFE raised to 'the low $150 billion range' from $140B, with 2027 framed as 'an extraordinary setup' on 8-10 new fabs — the single signal we said would flag a cycle peak is still rising
- The China risk shrank in the numbers while revenue grew 30%: China fell to 26% of revenue from 34%, so the record quarter was NOT China-funded — Taiwan (27%) and Korea (20%) carried it, and management sees China WFE flat to slightly up rather than collapsing
- Valuation de-rated into the beat: ~50.9x trailing GAAP EPS but only ~34x the annualized September guided EPS run-rate, against ~63x trailing when we last wrote this — on a net-cash balance sheet (~$1.84B) that returned 105% of FY2026 FCF
This is still a deeply cyclical equipment maker printing peak-looking numbers at the top of an AI-memory capex wave — and July showed exactly how the market re-prices that: a 42% peak-to-trough drawdown in a month on nothing but a change of mind about AI capex. The quarter's own cash flow shows the strain, and the whole guide leans on a WFE forecast management itself just moved by $10B+.
- The market has already demonstrated the de-rating mechanism: the shares fell from a $433.33 Jun 30 close to $252.35 on Jul 29 — roughly 42% — with no company-specific news, purely on AI-capex-peak fears that hit every WFE name. The 18% earnings pop recovered only part of it
- Cash conversion is lagging the income statement badly: FY2026 net income $7,265.4M against just $4,891.3M of free cash flow, on a $1,913.9M working-capital build, with receivables at $5,339.7M versus $3,378.1M a year earlier — revenue is being recognized well ahead of collection
- The guide rests on a forecast, not a backlog: 2026 WFE was just moved from $140B to 'the low $150 billion range', and 2027 commentary is qualitative ('extraordinary setup', 8-10 fabs) with no number. A top-down forecast that can be raised $10B+ in one quarter can be cut by as much
- Peak-margin risk compounds peak-revenue risk: at a 20-year-high 52.0% gross margin and a record 38.4% operating margin, both volume AND rate are at cycle highs, so an air-pocket hits earnings twice over
- The memory mix that drove the quarter rotated inside memory: DRAM slipped to 23% of systems from a record 27% while NAND leapt to 23% from 12%. NAND conversion spend is the more discretionary, more pull-forward-prone of the two — the ~$40B of conversion spend pulled into 2026-27 is demand borrowed from later years
- China at 26% of revenue is lower but still the largest single exogenous exposure, and the drop from 34% in one quarter shows how quickly that line can move on policy rather than demand
- Buybacks slowed sharply into the drawdown: $246.6M of repurchases in Q4 versus $1,162.8M in Q3, so capital return fell to 45% of FCF from 139% — the company bought less stock at $250-300 than it did at ~$211
What it is worth
Peer-relative (vs AMAT, TEL, KLA, ASML) + reverse-DCF sanity check on the implied growth/margin
~$250-315 (multiple compresses to a cyclical ~20-25x on a memory digestion air-pocket or broadened China controls; ~$253 24/7 Wall St. / ~$315 low-target zone) (fact: low-end targets)
~$340-380 (consensus zone
mid-cycle multiple normalization toward high-20s/low-30s forward P/E as growth stays strong but cyclicality is priced) (fact: avg analyst target)
~$480 (BofA target
sustained beat-and-raise, 50%+ GM, DRAM mix compounds, WFE > $140B into 2027) (fact: analyst target)
At ~$401 and ~63x trailing P/E the price implies continued beat-and-raise through-cycle — i.e. the market is treating WFE as structurally AI-driven, not cyclical; that holds if memory mix and 50%+ gross margin persist, but de-rates fast on any cycle wobble or China step-up.
SWOT
Strengths
- Duopoly-grade leadership in plasma etch and thin-film deposition — the two process steps that scale fastest as memory goes 3D/vertical (fact)
- Outsized memory leverage — when DRAM/NAND capex turns up, Lam's mix benefits more than peers — DRAM a record 27% of systems (Q3 FY26)
- $2B+/quarter CSBG installed-base service annuity — recurring, margin-accretive, dampens cyclicality (fact)
- Tool-of-record wins on the cutting edge — Aether EUV dry-resist and Akara conductor etch locked into leading-memory-maker advanced DRAM (fact)
- Near-record 49.9% gross margin with operating leverage still building (50.5% guided) (fact)
Weaknesses
- Deeply cyclical — revenue swung from $18.7B-class to $14.9B (FY24 trough) and back; earnings are capex-gated, not annuity-gated
- Heavy China exposure (recent quarters 30-42% of revenue) sits under U.S. export-control risk (fact)
- Customer concentration — a handful of memory + leading-edge foundry buyers (Samsung, SK hynix, Micron, TSMC) drive the order book
- Lithography gap — Lam has no scanner; it depends on ASML's EUV roadmap to pull its dry-resist/etch content
- Limited pricing power vs. a few sophisticated buyers who dual-source etch/dep where they can
Opportunities
- 3D-DRAM transition multiplies etch steps per wafer — Akara extendibility to 4F2/3D-DRAM is a multi-year content-gain vector (fact, forward)
- ~$40B NAND conversion spend pulled into 2026-27 — Lam is etch-share-heavy in NAND (fact)
- HBM4 and beyond raise deposition/etch intensity per DRAM bit (fact)
- GAA + advanced-packaging (backside power, hybrid bonding) open new served-available-market in logic/foundry
- Equipment Intelligence / services upgrades expand the recurring-revenue mix further (fact)
Threats
- U.S.-China export controls tightening or broadening could strand a large slice of demand (regulatory; fact-risk)
- Memory is the most cyclical end-market — an HBM/DRAM digestion air-pocket would hit Lam disproportionately
- Applied Materials and Tokyo Electron compete directly in etch/deposition and can take share at node transitions
- Customer in-housing or dual-sourcing of process steps to de-risk supply
- A WFE down-cycle (the $140B 2026 forecast could roll over) compresses both volume and margin
Moats, dependencies & bottlenecks
Moats
Once a tool is qualified into a node (Aether dry-resist, Akara etch), switching cost is enormous — requalification risks yield; wins persist across node generations (fact)
Lam is the etch/dep leader exactly where memory is scaling fastest (3D-NAND, HBM, 3D-DRAM); content per wafer rises with verticality
$2B+/qtr of spares, upgrades, Equipment Intelligence on a large installed base — recurring, high-margin, sticky (fact)
R&D and process know-how compound, but two well-funded peers (AMAT, TEL) sustain credible competition
Deep joint roadmap work with memory/foundry leaders embeds Lam early — but the same intimacy enables dual-sourcing leverage
Dependencies
Revenue is gated to customer fab investment; the most cyclical end-market in semis (fact)
A handful of buyers drive the order book; their HBM/DRAM/NAND roadmaps set Lam's growth (fact)
30-42% of recent revenue from China; rule changes can strand demand (fact-risk)
Dry-resist (Aether) and etch content gains pull through EUV adoption cadence Lam doesn't control (fact)
precision machining, gases, chucks) Complex multi-tier supply chain for system builds; shortages compress lead times/margins
Advantages
- Purest memory-WFE leverage — DRAM a record 27% of systems; mix benefits most when the AI-memory wave turns up (fact)
- Tool-of-record positions on cutting-edge DRAM (Aether dry-resist, Akara etch) that compound across node generations (fact)
- Etch/deposition franchise sits in the process steps whose intensity rises as memory goes 3D/vertical
- $2B+/qtr high-margin CSBG service annuity that lifts the through-cycle margin floor (fact)
- Near-record 49.9% gross margin with operating leverage still building toward the guided 50.5% (fact)
- Strong balance sheet + aggressive capital return (139% of FCF in Q3) near net-cash (fact)
Weaknesses
- Deep cyclicality — FY24 revenue troughed at $14.91B before the recovery; earnings are capex-gated (fact)
- Outsized China exposure (30-42% recent) under export-control risk (fact)
- Customer concentration in a few memory/foundry buyers
- No in-house lithography — dependent on ASML's EUV cadence
- Premium valuation (~63x P/E) above average analyst targets leaves little margin of safety (fact)
- Limited pricing power against sophisticated, dual-sourcing customers
Bottlenecks
- WFE-cycle gating — Lam can only grow as fast as memory/foundry customers invest; no annuity large enough to fully decouple from capex
- China export-control ceiling — a regulatory cap on a large served market that Lam cannot engineer around
- Lithography dependence — content gains pace ASML's EUV insertion, not Lam's own roadmap
- Customer dual-sourcing — buyers deliberately keep AMAT/TEL qualified to cap Lam's etch/dep pricing/share
- Talent + cleanroom R&D capacity — process-engineering depth is the rate-limiter on new tool-of-record wins
Top signals & trends
Top signals
Direct gauge of HBM/1c memory-mix leverage; further gains confirm the AI-memory thesis (fact)
The top-down demand signal; a roll-over here would flag cycle peak (fact)
An acceleration; Lam is etch-share-heavy in NAND (fact)
30-42% recent China mix is the single biggest exogenous risk to the order book (fact-risk)
The recurring, margin-accretive line that dampens cyclicality (fact)
Margin is near record; sustaining 50%+ confirms mix/efficiency gains are structural, not transient (fact)
Trends
Raises DRAM/deposition/etch intensity per wafer; the core driver of the record memory mix (fact)
Multiplies etch steps per wafer; Akara is extendible to 4F2/3D-DRAM — multi-year content gain (fact, forward)
High-aspect-ratio etch is Lam's strength; ~$40B conversion spend pulled forward (fact)
New patterning content category; production tool-of-record at a leading memory maker (fact)
Caps a large served market and adds policy whiplash to the order book (fact-risk)
Opens incremental etch/deposition SAM beyond memory (fact)
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
RF/microwave power, gas/flow, vacuum and motion subsystems for etch/dep tools
Plasma RF/DC power-delivery systems for etch and deposition chambers
Advanced materials, filtration, fluid-handling and specialty gases for the tools and the fab
Gas and chemical fluid-delivery subsystems — a major Lam outsourced module supplier
Critical subsystems and contract assembly for semiconductor capital equipment
Photonics, lasers and precision optics used in process and metrology subsystems
Top memory (DRAM/NAND) + foundry customer; HBM and 1b/1c DRAM buyer (non-US listed)
Leading HBM3E/HBM4 and DRAM maker — central to Lam's record DRAM mix (non-US listed)
U.S. memory leader (DRAM/HBM/NAND); key Lam etch/dep customer
Leading-edge foundry; GAA logic + advanced packaging etch/deposition demand
Logic/foundry customer ramping advanced nodes and backside power
NAND-flash maker — buyer of Lam's high-aspect-ratio etch as NAND converts (SanDisk US-listed)
Largest WFE company and Lam's primary direct competitor in deposition and etch; broader portfolio, less memory-concentrated
Japanese WFE leader; direct etch + deposition + coat/develop competitor, strong in memory and a key litho-track partner to ASML (non-US listed)
Not a direct etch/dep competitor but the gating litho supplier; controls the EUV roadmap Lam's dry-resist/etch content pulls through
Process control/metrology leader — adjacent, not direct, but competes for the same fab capex wallet
China's leading domestic etch/deposition maker; rising indigenization threat inside China, accelerated by export controls (non-US listed; analysis only, not a recommendation)