Lam Research

Phase A — Understand the business

Lens 1 · Company Overview

Lam Research makes the machines that etch and deposit the atomic-scale features inside every advanced chip. It is a wafer-fabrication-equipment (WFE) supplier — specifically the global leader in dry etch and a top-two player in thin-film deposition, plus clean and a fast-growing advanced-packaging business. HQ Fremont, CA; incorporated Delaware; fiscal year ends late June (FY2026 ends 2026-06-28, a 52-week year). ~20,600 employees as of 2026-03-29.

How it makes money — two engines:

• Systems revenue — new leading-edge etch/deposition/clean tools. Q3 FY26 $3.73B of $5.84B total (64%).
• Customer support-related revenue (CSBG) — service, spares, upgrades, and non-leading-edge (Reliant® line). Q3 FY26 $2.11B (36%). This is the annuity: it grows with the installed base, not with the capex cycle, and is the structural shock-absorber that has kept margins from collapsing in past downturns.

Customers: the world's leading memory, foundry, and integrated-device makers. The two named largest are Samsung Electronics and TSMC. End-market mix is the real story (Q3 FY26): Foundry 54%, Memory 39%, Logic/IDM 7% — vs FY25 full-year Foundry 45% / Memory 42% / Logic-IDM 13%. The mix is rotating toward foundry (leading-edge GAA logic) while memory re-accelerates (DRAM/HBM).

Contract structure: equipment is largely book-and-ship with customer down-payments (deferred profit $2.09B on the balance sheet); CSBG is recurring. Deferred revenue backlog disclosed at $2.22B as of Q3 FY26, ~68% recognizing within a year. There is no take-or-pay lock-in — demand is the customers' capex budget, which is cyclical. One reportable segment (wafer-processing equipment + service).

Lens 2 · Supply Chain

Upstream (Lam's suppliers) → Lam → end customer → demand sink. Name the chain:

• Upstream inputs: precision-machined chambers, RF power generators, gas-delivery and vacuum subsystems, robotics, and electronic subassemblies. Lam is a heavy buyer from semicap-subsystem specialists — MKS Instruments, Advanced Energy, Ichor Holdings, Ultra Clean Holdings (UCT), VAT Group, Entegris (materials/filtration). Inventory $4.00B (raw materials $2.51B) shows how component-heavy the model is. Supply-chain constraints and material/tariff costs are flagged as a margin risk in the filing.
• Lam (the node): integrates hardware + process + materials + software + process-control. Core competency is plasma/etch and deposition chemistry. Manufacturing increasingly in Asia (Malaysia, Korea, Taiwan) to be close-to-customer and cost-competitive; PP&E rose to $2.85B net as the company builds capacity.
• Direct customers (the buyers of the tools): TSMC (foundry), Samsung and SK Hynix (memory + foundry), Micron (memory), Intel (foundry/IDM), plus domestic Chinese fabs (SMIC, CXMT, YMTC and affiliates — increasingly export-restricted). Two customers = 17% + 15% of FY25 revenue (=~32%).
• End-demand sink: the chips those fabs build go into AI accelerators, HBM/DRAM, NAND, smartphones, PCs, servers, autos. The ultimate pull is hyperscaler AI capex (Microsoft, Google, Amazon, Meta, plus NVIDIA's roadmap) and the memory shortage it is creating.

Chokepoints / single-source dependencies:

1. Lam is itself a chokepoint — for high-aspect-ratio dielectric etch (the deep holes in 3D NAND and DRAM capacitors) there is no real substitute at the leading edge. That is the moat in supply-chain terms.
2. Customer concentration — top-10 customers = 93% of FY25 revenue. Lam needs them more than any one of them needs Lam in a given quarter, but at the leading edge the dependency is mutual.
3. Geographic chokepoint = China policy. China was 34% of FY25 revenue (was 42% FY24, 26% FY23); Q3 FY26 still 34%. A pen-stroke in Washington (BIS) can remove a third of the revenue base — the single largest non-operating dependency.

Lens 3 · Competitive Advantages (moats)

Lam sits inside the WFE oligopoly: Applied Materials, Lam, and Tokyo Electron together control ~70% of the deposition/etch/clean market; ASML monopolizes EUV lithography; KLA dominates process control. Within that structure Lam's moats:

1. Process leadership in etch (the deepest moat). Lam is the undisputed etch leader, ~45% global share, and clear #2 in deposition. Etch is where the hardest physics lives — the move to 3D NAND (now 300+ layers), DRAM capacitor scaling, and gate-all-around (GAA) logic all increase the number and difficulty of etch/deposition steps. Process complexity is Lam's tailwind: every architecture inflection adds steps that need its tools.
2. Switching costs / cycles-of-learning. Once a tool is qualified into a customer's process flow, swapping it risks yield — the most expensive thing in a fab. Lam compounds this with its installed base (the CSBG annuity) and its "Velocity" strategy of co-developing process recipes with customers, deepening lock-in.
3. IP + tacit process knowledge. Patents matter but the filing itself says success "depends principally upon our research and development, engineering... skills" rather than patents alone — i.e. the moat is institutional know-how, harder to copy than a patent.
4. Bargaining power — asymmetric. Against suppliers (MKS, UCT, Ichor): strong — Lam is a large, sophisticated buyer. Against customers (TSMC, Samsung): more balanced — concentration cuts both ways, but at the leading edge there is no second source, so Lam holds pricing on the newest tools and discounts the trailing edge (Reliant).

Where the moat is thinner: in mature/trailing-edge equipment (the Reliant line sold heavily into China), competition from domestic Chinese toolmakers (AMEC/Naura) is real and growing. The moat is leading-edge-deep, trailing-edge-shallow.

Lens 4 · Segments

Lam reports one segment, so the meaningful breakouts are systems-vs-CSBG, end-market, and geography — all ``.

Systems vs. Customer Support (Q3 FY26 vs 9M):

End markets (% of equipment+upgrade revenue):

Geography (Q3 FY26): China $2.00B (34%), Korea $1.35B (23%), Taiwan $1.32B (23%), Japan $0.47B (8%), US $0.36B (6%), SE Asia $0.21B (4%), Europe $0.14B (2%). China is the swing factor and the policy risk; Korea + Taiwan (Samsung/SK Hynix/TSMC) are the durable leading-edge core.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print — Q3 FY26, ended 2026-03-29)

A clean beat-and-raise on an accelerating tape.

• Vs. consensus / guidance: EPS $1.45 GAAP ($1.47 non-GAAP reported) beat the ~$1.36 consensus by ~8%. The prior-quarter guide had been ~$5.3B rev / ~49.5% GM, so the print cleared it.
• Drivers: sequential growth led by systems revenue into DRAM plus expanding CSBG (spares/upgrades). Foundry softened sequentially on lower mature-node spend; memory strengthened across DRAM and NVM.
• Margin: GM up on "improved factory efficiencies"; 9M GM 49.9% vs 48.2% prior-year, helped by customer mix, partly offset by tariff-related spend. The low 9.3% tax rate (SBC excess benefits + released reserves) flattered EPS this quarter — normalize to ~12-13%.
• Balance-sheet flags: clean. Cash $4.75B; net cash-positive (LT debt $3.73B; repaid $750M 2026 Notes at maturity in the quarter). Watch items: AR rose to $4.13B (a $756M 9M build, outrunning revenue growth slightly) and deferred profit fell to $2.09B (~$300M of customer down-payments unwound) — a mild tell that some forward demand was pulled in. Inventory down to $4.00B (healthy).
• Cash flow: 9M operating cash flow $4.40B on $4.99B net income — OCF/NI ~0.88, dragged by the working-capital build, not by earnings quality.
• Forward guide (June qtr, Q4 FY26): $6.6B ± $400M revenue (a further sequential acceleration off $5.84B), GM 50.5%, op margin 36.5%, EPS $1.65 ± $0.15. Management raised its CY2026 WFE forecast to ~$140B (above SEMI's ~$123B). Tone: confident, AI/memory-led.
• Market reaction: stock rose on the print; LRCX has compounded a ~180-300% one-year move into the upcycle — the market is pricing the raise and the next one.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts/ empty) — this lens is ``. Across the last ~3-4 calls the arc is unambiguous: rising confidence, AI-anchored. Recurring management phrases: "technology inflections," "gate-all-around," "advanced packaging," "dry resist/EUV patterning," "NAND layer conversions (moly/carbon gapfill)," "SAM expansion." Lam now frames itself as the AI-era process enabler, repeatedly guiding its served available market (SAM) to the high-30s% of WFE and calling advanced packaging +50% in CY2026. What they say less of: the China-mature-node story (de-emphasized as policy tightens). The "Velocity" operational-efficiency narrative (Archer's signature) has shifted from cost-out to margin-expansion-while-growing — Q3 GM 49.8% and a 50.5% guide validate it. Sentiment trend = steadily more bullish over four quarters, with the only hedge being explicit acknowledgment of China/tariff volatility.

Lens 7 · Comps

Lam vs. the WFE oligopoly. **Multiples are , dated; market caps approximate.** Hard fundamentals for LRCX are .

¹ My arithmetic: at ~$385.63 on FY25 GAAP diluted EPS $4.15 = ~93x; on TTM ~$5.29 = ~73x; on FY26 est ~$5.60 = ~69x. The web's headline "P/E ~61x" likely uses non-GAAP or an earlier price.

Read: Lam is the most expensive name in the group on a P/E basis except possibly ASML — and it trades at a meaningful premium to AMAT (which has the broadest WFE exposure and arguably better diversification). The premium is justified if you believe etch/deposition step-intensity grows faster than litho in the AI/3D era; it is hard to justify on cycle-normalized earnings. The comp table is the bear case in one row.

Lens 8 · Stock-Price Catalysts (>5% moves, ~5yr)

Pattern is ``. What actually moves LRCX:

1. Earnings + the WFE-forecast raise/cut — the single biggest lever. The +8% Q3 FY26 beat and the $140B WFE raise drove the latest leg.
2. Memory cycle inflections — NAND/DRAM capex turning up (2026) or down (the 2023 trough) swings the stock more than foundry, because Lam is memory-levered.
3. China export-control headlines — the October 2022 BIS rules, and the June 2026 affiliate-rule tightening, each produced sharp drawdowns. This is the dominant downside catalyst.
4. AI/hyperscaler-capex sentiment — LRCX now trades as an AI-infrastructure beta; NVIDIA/hyperscaler capex commentary moves it intraday.
5. Analyst re-ratings — e.g. the Morgan Stanley downgrade to Underweight (a low-$92 target, now far below spot) and the wave of target hikes into 2026. Conclusion: the market reacts to the WFE forecast and the China headline far more than to the in-quarter print. Own the WFE cycle, hedge the China tail.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO Timothy Archer — President & CEO since Dec 2018; joined Lam 2012 as COO; 18 years prior at Novellus (etch/deposition) in tech-development/leadership roles. A process-engineer-operator archetype, not a founder. Credited with the "Velocity" strategy (compressing time-to-high-volume-manufacturing) and disciplined cycle management. Track record: navigated the 2023 memory trough without margin collapse, then delivered record revenue and ~50% gross margin into the 2026 upcycle.
• CFO Douglas Bettinger — long-tenured, well-regarded for capital discipline and clear guidance. COO Patrick Lord; CTO Vahid Vahedi (also Sustainability).
• Capital allocation — the standout strength. ROIC ~72%, vs ~60% 5-yr average and an ~11% market average. The company commits to returning ≥85% of free cash flow and over-delivered (~139% of FCF in the March quarter). 9M FY26: $3.60B buybacks (incl. ASRs), $0.95B dividends. Dividend raised to $0.26/qtr (from $0.23). $10B buyback authorized May 2024, $4.29B remaining as of Q3 FY26. Caveat: the buyback is pro-cyclical — average price paid rose $105.67 → $153.62 → $210.57 across the last three quarters, i.e. Lam is buying its most expensive stock with peak-cycle cash. Returns-accretive only if the upcycle persists.
• Skin in the game: professional-manager-level insider ownership (modest), typical for a large-cap semicap. Recent insider selling ~$8.6M over 3 months — small relative to the cap, but a mild caution signal at the highs.
• Red flags: none structural. Clean disclosure, long-tenured team, no related-party noise. The only critique is the pro-cyclical buyback timing.

Lens 10 · Forensic Red Flags

Forensic read of the income statement, balance sheet, and cash flow. Verdict: clean books, a few cycle-driven watch items, no accounting red flags.

• Revenue recognition: book-and-ship with customer down-payments; deferred profit $2.09B. Management explicitly notes it "exercises discretion... as to timing and prioritization of... deliveries, which has impacted... the timing of revenue recognition" — i.e. it can pull/push quarters within reason. Not fraud, but it means any single quarter's beat can be partly a timing artifact. Deferred profit fell ~$300M (down-payment unwind) — watch whether forward demand was pulled in.
• Receivables vs. revenue: AR +$756M over 9M (to $4.13B) outpaced the 9M revenue growth modestly — a yellow flag worth tracking, though consistent with a fast-ramping top line and large customers on standard terms.
• Inventory: down to $4.00B (raw materials $2.51B). Healthy — inventory is not outrunning revenue.
• SBC: $282M over 9M (~1.7% of revenue) — modest, well-covered by GAAP earnings; non-GAAP add-backs are not flattering results unduly. The 9.3% Q3 tax rate (SBC excess benefits + reserve release) is the bigger "quality of EPS" caveat — normalize tax to ~12-13%.
• Goodwill/intangibles: $1.88B, small vs $20.8B assets — low impairment risk.
• Cash vs. earnings: 9M OCF $4.40B vs NI $4.99B (0.88x) — the gap is the working-capital build (AR up, deferred profit down), not earnings quality.
• Leverage: net cash; well-laddered debt; $2.0B commercial-paper backstop. No concern.

Regulatory findings (required).

• SEC Litigation Releases / AAERs: None. regulatory/regulatory-findings.md (fetched 2026-06-18 via SEC EDGAR EFTS, LR + AAER, period 2021-06-18 → 2026-06-18) returned 0 findings.
• 10-K Item 3 / 10-Q Legal Proceedings: "While the Company is not currently a party to any legal proceedings that it believes material..." — only ordinary-course IP and commercial matters; no material accrual.
• Non-SEC (web): the material regulatory exposure is not enforcement but export policy — U.S. BIS controls on shipments to Chinese customers. The June 2026 "50% affiliate rule" tightening is expected to cost ~$200M in the June quarter and ~$600M in CY2026, pushing China below 30% of revenue. The IRS is examining FY2019-2021 returns with no significant adjustments proposed to date. No FTC/DOJ/FDA actions found.
• Bottom line: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K/10-Q legal-proceedings disclosure as of 2026-06-18. The China export regime is a policy risk, not a compliance failure.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Built bottom-up from filings + guidance. Every output ``; inputs labeled.

Anchor (actuals): 9M FY26 diluted EPS $3.95. Q4 FY26 (June qtr) guide $1.65 ± $0.15.

• FY2026E diluted EPS ≈ $5.60. (TTM through Mar 2026 ≈ $5.29.)
• FY2026E revenue ≈ $23.1B, +~25% YoY vs FY25 $18.44B.

FY2027 / FY2028 paths (FY ends ~late June):

. The spread ($4.20 bear → $8.00 bull FY28) is the cyclicality made explicit — and why a single trailing multiple is dangerous here.

Valuation context: at ~$385.63, that is ~69x FY26E, ~62x base-FY27E, ~56x base-FY28E. Even three years out on the base case, the stock is not cheap. The bull case gets you to ~48x FY28 — still a premium-to-peers multiple on peak-ish earnings. (Per --watchlist rules, no forecast.ts create logged — base case noted here for a future human-gated /thesis pass.)

Lens 12 · Bull vs Bear

Bull case. Lam owns the deepest moat in the highest-step-intensity part of chipmaking at the exact moment AI is forcing the hardest architecture inflections in a decade. Etch and deposition steps scale super-linearly with 3D NAND layer counts, DRAM capacitor scaling, gate-all-around logic, and advanced packaging (HBM, +50% in CY2026). Lam's SAM is expanding to the high-30s% of WFE, it is memory-levered into a structural memory shortage (IDC: DRAM/NAND supply growth below historical norms), and it converts it all at ~50% gross margin / ~35% operating margin with ~72% ROIC and ≥85% of FCF returned. Management raised the CY2026 WFE call to $140B. This is a toll booth on the AI build-out with monopoly economics in its core product.

Bear case (permanent-impairment-grade risks).

1. Cyclicality is not abolished. WFE is the most cyclical layer of semis. Bull 2026 growth estimates (+82%) have already compressed toward +3% in some models. A 2027 capex-digestion air-pocket takes EPS to the ~$4-5 bear band — and a ~69x multiple does not survive a down-cycle.
2. China is a third of revenue and a policy variable. A further BIS tightening (beyond the −$600M CY2026 hit) or Chinese retaliation could remove far more than the modeled drag — and the domestic-Chinese WFE champions (Naura, AMEC) are climbing the trailing-edge, structurally shrinking Lam's Reliant/mature TAM.
3. The multiple is the position. At ~69x FY26 GAAP / 62x base-FY27, the stock is priced for the bull case to compound with no cycle. The median sell-side target ($315) sits below spot — the tape has run past even the optimists.

Pre-mortem (18 months out, thesis broke): It's late 2027. Memory capex peaked in CY2026 and digested; foundry GAA spend front-loaded into 2026 and air-pocketed; a 2027 BIS/affiliate-rule escalation took China toward 20% of revenue; Lam kept buying stock at $200-380 into the top, so per-share book didn't cushion the de-rate. EPS reset to ~$4.50, the multiple compressed to ~25-30x, and the stock halved — not because the business broke, but because the price had borrowed three years of growth.

Contrarian view (what the market refuses to see): Both tails. Bulls refuse to price the cycle/China tail at 69x; bears (MS at $92) refuse to price how much step-intensity has structurally raised Lam's mid-cycle earnings power — this is a higher-margin, more-CSBG-annuitized company than the 2019 version. The truth is a great business at a demanding price: the right call is direction-up on the franchise, conviction-capped by the multiple.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the money machine? A memory-capex roll. Lam is memory-levered; NAND/DRAM capex is violently cyclical, and 2026 is plausibly a peak year (record WFE $140B). The short thesis is simply "you are paying 69x for peak-cycle earnings."
• Revenue concentration: top-10 = 93%; two customers = 32%; China = 34%. Any one of: a Samsung/TSMC capex pause, or a China escalation, hits a huge slice at once.
• Why the moat may be weaker than bulls think: it is leading-edge-deep but trailing-edge-shallow. ~36% of revenue is CSBG/Reliant (service + mature tools), and the mature-tool franchise is exactly what Chinese domestic toolmakers are eating. The "annuity" is partly exposed to import-substitution.
• Most dangerous underestimated competitor: Naura / AMEC (China). Policy that locks Lam out of China simultaneously subsidizes the domestic champions to climb the stack — a double hit.
• Worst capital-allocation move: buying back stock at $210/sh average and rising with peak-cycle cash — value-destructive if the cycle turns. ~139% of FCF returned at the highs.
• What must hold for today's price: WFE stays ≥$140B and grows; memory shortage persists into 2027-28; China loss stays bounded at ~$600M; ~50% GM holds through tariffs. Break any one and ~69x is indefensible.
• −20-30% growth shock: apply the bear path — FY27 EPS ~$4.80 and falling; at a cycle-normal 25-30x that's a ~$120-145 stock, ~60-70% below spot. The downside is a multiple problem layered on a cyclical problem.
• Single permanent-impairment scenario: a hard US-China decoupling that (a) zeroes China demand and (b) hands the domestic Chinese WFE market permanently to Naura/AMEC. Plausibility: low-to-moderate and rising — the most important risk to size.

Lens 14 · Management Questions (ordered by information value)

1. At CY2026 WFE of ~$140B (a record), how much of current revenue do you believe is mid-cycle vs. peak, and what is your honest estimate of Lam's trough-year EPS in the next down-cycle?
2. With the June-2026 affiliate rule pushing China below 30% of revenue, what is your base/bear range for total China revenue over the next 3 years, and how much of the Reliant/mature TAM do you concede to Naura/AMEC permanently?
3. You're returning ~139% of FCF and buying stock at a rising $210+ average — what valuation or cycle signal would make you slow the buyback, and why is buying near all-time highs the best use of peak-cycle cash vs. retaining flexibility?
4. Advanced packaging +50% in 2026 — what is the absolute revenue base, the multi-year CAGR, and how defensible is your share against AMAT/TEL as it scales?
5. How much of Q3's sequential beat was delivery-timing discretion vs. underlying demand, and how should we read the ~$300M deferred-profit (down-payment) unwind?
6. Where exactly does your SAM expansion to high-30s% of WFE come from — etch steps, deposition, dry-resist/EUV-patterning, packaging — and which is most at risk if GAA/3D-NAND roadmaps slip?
7. Gross margin printed 49.8% and you guide 50.5% despite tariff costs — what is the structural ceiling, and how much is mix (memory vs. foundry) vs. durable factory efficiency?
8. AR grew $756M over 9M, outpacing revenue — any change in customer payment terms or concentration of credit risk we should watch?
9. The NAND layer-count upgrade cycle (moly/carbon gapfill) — how many quarters of runway, and is this a one-time conversion or a recurring multi-node tailwind?
10. With the IRS examining FY19-21 and your 9.3% Q3 effective tax rate (excess SBC benefits + reserve release), what is the normalized go-forward tax rate under BEPS 2.0 / OBBBA?
11. Manufacturing is shifting to Asia (Malaysia/Korea) — how does that change your tariff exposure, margin, and geopolitical risk profile?
12. If a customer asked for a second-source on a leading-edge etch tool, how real is that threat, and what does it imply for pricing power?
13. Capex rose to ~$0.78B over 9M — what capacity are you building for, at what utilization assumption, and what happens to it in a down-cycle?
14. Succession: the team is long-tenured — what is the CEO/CFO succession plan, and how is the next generation being developed?
15. What is the single risk on the horizon you think the market is most underpricing for Lam specifically?

---