Lasertec

Phase A — Understand the business

Company Overview

Lasertec makes the machines that check whether the most expensive stencils in human industry are perfect. It is a fab-lite / R&D-oriented ("fabless-style") designer and integrator of inspection and metrology systems for the semiconductor industry, founded 1960, headquartered in Yokohama, Japan, listed on the Tokyo Stock Exchange (Prime) under 6920 ``.

What it actually sells. The crown jewel is EUV photomask inspection — the tools that find killer defects on the masks (reticles) and mask blanks used in extreme-ultraviolet lithography at 7nm/5nm/3nm/2nm and below. Two franchises matter:

• ACTIS series (A150 → A300 → A200HiT) — actinic patterned-mask inspection: it inspects the finished mask at the actual 13.5nm EUV exposure wavelength, so it catches "printable" phase defects that DUV-based tools physically cannot see. Lasertec's ACTIS is the world's first and only commercial actinic patterned-mask inspection system ``.
• Mask-blank inspection (ABICS / MAGICS / MATRICS lineages) — inspects the blank before it's patterned. Plus wafer inspection, advanced-packaging metrology, SiC / compound-semi (energy & environment), FPD photomask inspection, and hybrid laser microscopes ``.

Business model = razor + very expensive razor + blades. ASPs run to ¥8–15bn per tool on the flagship EUV systems . On top of unit sales, an **installed-base service annuity** compounds: EUV inspection tools require ongoing maintenance, calibration and upgrades, so revenue accretes after delivery. That annuity is now visibly inflecting — **FY2025 service sales grew +48.3% YoY to ¥42.96bn** even as new-tool orders collapsed . This is the structurally attractive part of the story: as the fleet of shipped EUV tools grows, the recurring, higher-margin service base grows with it regardless of the order cycle.

Customers, suppliers, competitors (detail in Lenses 2–3): customers are the ~three companies on earth running leading-edge EUV fabs — TSMC, Samsung, Intel — plus the mask-blank/photomask supply chain (Hoya, AGC, Toppan, DNP). Its closest competitor is KLA; Zeiss plays in adjacent actinic mask imaging (AIMS EUV).

Contract structure / payment terms. Big-ticket, long-lead capital equipment (multi-quarter build cycles), typically with customer advances/progress payments and revenue recognised on delivery/acceptance — the exact recognition mechanics are the crux of the Lens 10 forensic discussion and were the core of the 2024 short thesis. Backlog is the visibility instrument: ~¥316bn at FY2025 year-end ``, i.e. well over one year of revenue booked.

Supply Chain

Lasertec sits at a single, narrow chokepoint inside the broader EUV value chain. Map it end to end, named:

Upstream inputs into Lasertec's tools → Lasertec → the customer that needs the tool → the end product:

• Optics & EUV light sources: high-precision EUV/DUV optics and 13.5nm light-source components. Carl Zeiss SMT is the reference name for EUV-grade optics across the industry (it supplies ASML's scanner optics and is itself an actinic-imaging player); Lasertec's actinic tools require an EUV source and reflective optics of comparable pedigree. This is a single-source-class dependency and was a specific Scorpion allegation ("defective light-source technology") — see Lens 13 ``.
• Precision stages, detectors, vacuum, control electronics — specialist Japanese/global sub-suppliers (not individually disclosed at scale).
• Lasertec (Yokohama): integrates the above into ACTIS / mask-blank / wafer inspection systems. Fab-lite → it outsources heavy fabrication and concentrates on optics integration, algorithms and calibration.
• Direct customers (the mask + fab layer):
• Mask-blank & photomask makers: Hoya, AGC, Toppan, Dai Nippon Printing (DNP) — they buy blank-inspection tools to certify defect-free blanks.
• Leading-edge foundries/IDMs: TSMC, Samsung, Intel — they buy patterned-mask (ACTIS) and wafer inspection tools to run their EUV mask shops and fabs ``.
• The parallel dependency — ASML. Lasertec does not sell to ASML, but its entire addressable market is derivative of ASML's EUV installed base and roadmap. No EUV scanners shipping → no EUV masks to inspect → no ACTIS demand. ASML plans ~60 High-NA EUV units in 2026 and ~80 in 2027, with High-NA mass production 2027–28 ``. High-NA masks create new actinic-inspection demand — the "silent bottleneck" thesis (Lens 11).
• End product: leading-edge logic (TSMC N2/A16, Samsung SF2, Intel 18A/14A) and increasingly DRAM — i.e. the compute substrate for the AI build-out.

Chokepoints / single-source dependencies:

1. Lasertec IS the chokepoint for actinic patterned-mask inspection (≈100% share) — everything above it flows through one vendor.
2. EUV light source / optics into Lasertec — a narrow, high-barrier input (Zeiss-class).
3. Customer concentration is itself a supply-chain fragility in reverse: three buyers (§Lens 1) drive the order book, so a single customer's capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. pause (e.g. ACTIS cancellations in Q4 FY2025) swings orders violently.

This lens is web-derived; the specific per-supplier contract terms are not disclosed and are n/a.

Competitive Advantages (moats)

Lasertec has one of the cleaner technology-monopoly moats in hardware — but it is a narrow moat with a concentrated customer set, which is exactly the tension the whole thesis turns on.

The moat, ranked by durability:

1. Actinic-inspection process/IP monopoly (deep). ACTIS is the only commercial tool that inspects patterned EUV masks at 13.5nm. Building an actinic source + reflective-optics + detection stack that resolves EUV phase defects is a decade-class engineering problem; KLA leads e-beam / die-to-database mask inspection, but that is a different physical modality that cannot see everything actinic sees at the exposure wavelength . Near-**100% share in actinic**, ~**90%+ in EUV mask inspection** overall .
2. Switching costs + qualification lock-in (deep). Mask shops and fabs qualify an inspection recipe/tool into a production flow; re-qualifying on a rival tool is slow, risky and rarely done mid-node. Once ACTIS is the reference, it stays the reference for that node.
3. Installed-base service annuity / data flywheel (compounding). Every shipped tool feeds a maintenance+upgrade stream and a proprietary defect-signature dataset that improves detection — the +48% FY2025 service growth is the moat monetising itself.
4. Roadmap coupling to ASML High-NA (secular, but borrowed). As long as EUV advances, mask complexity rises and actinic inspection becomes more necessary (High-NA → thinner films, tighter defect budgets). ACTIS A200HiT (Oct 2025) is explicitly the High-NA answer ``.

Bargaining power — who needs whom more?

• vs. customers: Strong at the tool level (you cannot certify an EUV mask without it), weak at the order-timing level (three customers control when they buy; TSMC in particular has the leverage of scale and, per Scorpion, at one point slowed purchases — Lens 13). The moat protects price/share, not volume cadence.
• vs. suppliers: Moderate — dependent on a thin set of EUV optics/source suppliers; not a price-setter over Zeiss-class inputs.

Where the moat is thinner than bulls claim: it is a single-product-line monopoly on a single, cyclical, customer-concentrated demand driver. That is a very different risk profile from, say, ASML's monopoly (which every leading-edge chip on earth requires). Lasertec's monopoly is real but its demand is neither diversified nor non-cyclical.

Segments

No our figures data exists on the shelf — every figure ``; treat as approximate and reconcile against the FY2025 tanshin on next ingest.

By product line (FY2025, year ended June 2025):

• Semiconductor-related equipment: ¥202.97bn, +11.7% YoY — the bulk of revenue; EUV mask-blank + patterned-mask (ACTIS) + wafer inspection ``.
• Service: ¥42.96bn, +48.3% YoY — the standout accelerator; the installed-base annuity inflecting ``.
• (Smaller lines: FPD photomask inspection, SiC/energy-environment, laser microscopes — not separately material at group scale.)
• Group total FY2025 net sales: ¥251.48bn, +17.8% YoY; operating income ¥122.84bn, +51.0% YoY — both record highs ``.

Read on the trend: the mix is quietly shifting toward service, which is the healthiest possible direction (higher-margin, recurring, order-cycle-insensitive). The equipment line's +11.7% is revenue (recognised backlog burning down), which masks the real leading indicator — orders fell 61.4% YoY to ¥105.2bn with ACTIS cancellations in Q4 — see Lens 5/8. Revenue is a rear-view mirror here; the order book is the windshield, and the windshield turned ugly in FY2025 even as the P&L printed records.

By geography (FY2025): the United States was the largest region, followed by Taiwan and South Korea ``. The US-first mix is notable — it reflects Intel's leading-edge push and US-based mask/fab activity, not just the TSMC/Samsung Asia base. Precise geographic percentages are n/a (behind Statista paywall).

---

Phase B — Measure performance

Earnings Result (latest print: Q2 FY2026, reported Feb 2026)

Fiscal year ends June 30. FY2026 = the year ending June 2026, so as of July 2026 the last full print is Q2 FY2026 (Oct–Dec 2025), reported ~2 Feb 2026; H2 FY2026 (through June 2026) has likely just closed but full-year FY2026 actuals are not yet in the shelf.

Q2 FY2026 (reported Feb 2026):

• Net sales ¥54.1bn — +36.8% QoQ, but −41.3% YoY ``.
• Net income ¥26.7bn — +40.0% QoQ, −22.4% YoY ``.
• The YoY declines are the backlog-burn hangover: FY2025's record revenue pulled forward, and the FY2024→FY2025 order collapse is now flowing through the top line.

Q1 FY2026 (reported ~31 Oct 2025), for context:

• Revenue ¥54.2bn, +48% YoY; net income ¥19.1bn, +113% YoY; net margin ~35% (vs 24% prior-year Q1) ``.

FY2026 full-year guidance (as revised):

• Net sales ¥220bn (−12.5% YoY) — but raised ¥20bn vs the prior forecast.
• Operating income ¥100bn, operating margin ~45.5% ``.
• Guidance raise, absolute decline. The nuance that moved the stock (Lens 8): management raised the forecast on AI-driven spending, but the level still implies a down year off the FY2025 record — and the raise was smaller than a market priced for an AI-capex melt-up wanted.

Margins. Operating margin sits in the 45–49% band with ROE ~44–48% `` — extraordinary for capital equipment, and the empirical core of the bull case and the short case (bulls: monopoly rent; shorts: too good to be true, hence "fraud"). Both can't be right; the Special Investigation Committee (Lens 10) sided with "it's real."

Balance-sheet flags. FY2025 (June 2025): total assets ¥329.6bn, shareholders' equity ¥207.9bn, net income attributable to owners ¥84.65bn . Equity ratio ~63% — conservative, effectively **net cash** (a separately-sourced "¥30.66bn cash / ¥17.56bn OCF" figure is inconsistent with group scale and is flagged unreliable / partial — n/a — not reliably sourced; reconcile on ingest). The forensic tension is not leverage — it's inventory and revenue-recognition timing (Lens 10).

Market reaction: the Feb 2026 outlook-day reaction was a ~10% drop — "raised outlook less than investors anticipated in the face of an AI-related surge in spending" ``. Classic expectations-vs-reality gap on a high-multiple name.

Unusual vs. its own history: the orders/revenue divergence is the flag — FY2025 delivered record revenue and profit while orders fell 61% and backlog fell 32%. A print that looks like a peak because it is one on the recognised-revenue line, while the leading indicator already rolled.

Earnings Calls (sentiment trend)

No transcripts/ on the shelf; sentiment reconstructed from IR presentation scripts/FAQs ``.

• Through FY2025 (mid-2024 → mid-2025): defensive + cyclical-trough messaging. Management repeatedly framed orders as bottoming "in Q2 or Q3 FY2025," recovering only in Q4 — and then Q4 delivered cancellations instead of recovery. Tone: cautious, backlog-reassuring ("¥300bn+ backlog = multi-year visibility"), emphasis shifted to the service annuity as the stable leg while equipment orders wobbled.
• The governance overhang (mid-2024 onward): denial + process. Post-Scorpion, the register hardened into legalistic denial ("we clearly deny improper accounting") plus the Special Investigation Committee process — a defensive posture that persisted into the disclosure cadence.
• Into FY2026 (late-2025 → early-2026): cautious optimism, AI-capex tailwind. The Q1/Q2 FY2026 message pivoted to AI-driven demand and High-NA as the re-acceleration story, culminating in the guidance raise — but management kept the absolute number below FY2025, i.e. optimism tempered by realism about the down year.

Recurring phrases: "monopoly / only actinic," "installed base / service growth," "multi-year backlog," "High-NA readiness (A200HiT)." Things they stopped saying: the confident double-digit dividend growth guidance — FY2026 dividend is now guided flat at ¥329 vs prior double-digit hikes ``, a quiet tell that management sees the near term as a plateau, not a launchpad.

Comps

Peer set: EUV/inspection & leading-edge WFE. All multiples `` with source; where I cannot source cleanly, n/a. No multiple is fabricated.

Read. Lasertec at ~52x P/E and ~35x EV/EBITDA is priced at or above ASML — the semiconductor industry's only indispensable-to-every-chip monopoly — and at a large premium to KLA (~30x fwd), its nearest functional peer, and to AMAT (~22x). The premium is partly earned by best-in-class ~45–49% operating margin and ~44–48% ROE. But the market is capitalising Lasertec's monopoly rent as if its demand were as durable as ASML's, when in fact its demand is (a) more cyclical, (b) far more customer-concentrated (77% top-3), and (c) currently in an order down-cycle (−61% orders). That is the valuation crux: a monopoly multiple on a non-monopoly demand profile.

Stock-Price Catalysts (moves >5%, last ~5 years)

The tape reveals what the market actually reacts to for this name. Pattern: (1) EUV-capex/AI-cycle sentiment, (2) order/guidance surprises, and (3) the governance scar — not day-to-day product news.

• 2020–May 2024 — the ascent: rode the EUV-adoption + AI-infrastructure narrative to an all-time high of ¥45,490 on 15 May 2024 ``. A high-beta, high-multiple expression of "EUV is the bottleneck."
• 5 June 2024 — Scorpion Capital short report: 334-page "colossal fraud" allegation; sharp drop, then the stock entered a multi-quarter derating as investor confidence broke ``. The single most important non-fundamental catalyst in its history.
• Jul 2024 — Special Investigation Committee clears the company: "no irregularities" — stabilised sentiment but did not restore the premium; the derate continued on fundamentals (the order cycle), decoupling from the fraud question ``.
• Through Aug 2025 — cyclical trough: fell to ~¥15,790 on 13 Aug 2025, −65% from the peak ``, tracking the FY2025 order collapse (−61%) + ACTIS cancellations, not the short report.
• Late 2025 → mid 2026 — AI-capex re-rating: a violent recovery to a 52-week high of ¥58,580 and ~¥49,700–55,100 by late Jun/early Jul 2026 (market capMarket capitalisationThe share price multiplied by the number of shares. What the market says the equity is worth. back to ~¥5tn) on the AI/High-NA demand narrative ``.
• 2 Feb 2026 — outlook-day −10%: guidance raise judged insufficient vs. AI-capex euphoria ``.

What the market reacts to: order momentum and the AI/EUV-capex narrative, amplified by a high multiple (so small expectation gaps → big moves), with a persistent governance-risk discount that lingers post-Scorpion. It does not meaningfully react to incremental product launches unless they change the demand math (High-NA).

---

Phase C — Judge people & books

Management

• Track record. Osamu Okabayashi — President & Representative Director since 1 July 2009 (joined 2001) . A **long, stable, ~15+ year tenure** through the entire EUV build-out — Lasertec went from a niche microscope/inspection maker to a ¥5tn actinic-EUV monopoly on his watch, which is a genuinely elite value-creation record *by outcome*. (Note: some sources list a separate CEO title/appointment in 2023 — Japanese board structures split "President/Representative Director" from "CEO"; Okabayashi is consistently the operating principal. Reconcile exact titles on ingest — , mild ambiguity.)
• Skin in the game / insider ownership: n/a. Japanese executive equity stakes are typically modest vs. US founder-CEOs; specific insider holdings not sourced. This is a gap to close before sizing any position.
• Capital allocation. Conservative and shareholder-friendly, not empire-building. Net-cash balance sheet; FY2025 dividend ¥329 (35.1% payout) plus a buyback of up to 1m shares / ¥12bn . Low capex (fab-lite: ~¥2.4bn) → **very high ROE (~44–48%)** and high FCF conversion. No value-destroying M&A on record. The one *soft* flag: FY2026 dividend held **flat at ¥329** after years of double-digit growth — a signal of near-term caution, not misallocation.
• Red flags. No related-party or promotional-behaviour smoking gun proven — but the Scorpion report specifically attacked governance/oversight and the whole episode raised durable questions about disclosure quality and working-capital transparency (Lens 10/13). The Special Investigation Committee found no irregularities, which is exculpatory but was company-commissioned (independent members, but company-appointed) — a governance-purist would still discount it somewhat.
• Archetype. Professional-manager stewardship of a technology monopoly, not a visionary-founder story. For a mature monopoly harvesting an installed base, that is the right archetype — steady hands, capital discipline, defend the moat. The risk of the archetype is complacency + weak external-communication instincts (evident in the terse, legalistic Scorpion response).

Forensic Red Flags

Forensic lens. No filings on the shelf → this is built from web-sourced financials + the public record of the 2024 short thesis and the investigation. Every figure ; the primary tanshin should be ingested to convert these to .

The structural forensic question for Lasertec is timing, not solvency. It is net-cash with a 63% equity ratio, so there is no leverage/going-concern risk. The risk lives in long-lead capital-equipment revenue recognition and inventory/working-capital build — precisely where the short thesis aimed.

• Revenue recognition (income statement). Big-ticket tools built over multiple quarters with customer advances; revenue recognised on delivery/acceptance. Aggressive timing (pulling acceptance forward, or booking on shipment vs. acceptance) is the classic manipulation vector here. Scorpion alleged exactly this — fraudulently inflated revenue/margins via the recognition mechanics — and predicted a large impairment . The **Special Investigation Committee (Jul 2024) found no evidence** of revenue-recognition or inventory-accounting irregularities . Net: the specific fraud allegation is unproven and formally rebutted, but the sensitivity of reported profit to recognition timing is real and inherent to the model — a bear should stress-test it.
• Inventory / working capital (balance sheet). The short thesis leaned hard on inflated inventory. In a down-order environment (orders −61% in FY2025), inventory and contract balances relative to forward demand are the number to watch — a build against a falling order book is the yellow flag. Specific inventory-days / contract-liability figures are n/a (not disclosed in the aggregated web data; the tanshin has them). This is the single highest-value gap in this dossier — do not opine on the QoE without the actual inventory and contract-liability lines from the FY2025/H1-FY2026 tanshin.
• Cash flow vs. earnings. With ~45–49% operating margin and low capex, FCF conversion should be high — if earnings are real and working capital behaves. A divergence (earnings up, operating cash flow lagging on inventory/receivables build) would be the tell that vindicates part of the short case. The aggregated cash-flow figures I retrieved are inconsistent / partial and explicitly flagged n/a — not reliably sourced — a real QoE read requires the primary statement.
• SBC / non-GAAP flattering: not a material feature (Japanese GAAP; SBC is not the profit-flatterer it is for US tech). Low concern.
• Goodwill/intangibles: minimal (organic, fab-lite). Low concern.

Regulatory findings (required sub-section):

• SEC (EDGAR): None possible — Lasertec has no CIK and does not file with the SEC. Per the on-disk regulatory/regulatory-findings.md (generated 2026-07-06, total_sec_findings: 0): no EDGAR LR/AAER search is applicable ``.
• Non-SEC / Japan: Web search on "Lasertec" (FTC OR DOJ OR FDA OR settlement OR fine OR penalty) enforcement surfaces no government enforcement action. The only "official" scrutiny stemming from the Scorpion episode is US plaintiff-side shareholder-investigation solicitations (Bronstein Gewirtz & Grossman; Pomerantz) announced June 2024 — these are law-firm investigation notices targeting ADR holders, not regulatory findings and not filed suits with adverse outcomes ``. No Japanese FSA/SESC enforcement or restatement on record.
• Company's own disclosure (Item-3 equivalent): No 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. to quote; the analogous disclosure is the FY2024 Annual Securities Report and the Special Investigation Committee report, which disclosed the allegations and the "no irregularities" conclusion ``.
• Verdict: No material regulatory or legal findings — verified via (i) SEC EDGAR non-applicability (no CIK), (ii) web search for non-SEC enforcement, and (iii) the public investigation record, as of 2026-07-06. Caveat: an active short-seller fraud allegation + US shareholder-investigation notices constitute a governance-reputation overhang even though no enforcement or restatement resulted. Not a clean bill of health; a cleared-but-scarred one.

---

Phase D — Project & stress-test

Forward Projection

Bottom-up from FY2025 actuals + FY2026 guidance. Fiscal years end June. Every input or; outputs with arithmetic. Share count for EPS: not cleanly sourced → I project **operating income / net income levels** and give EPS only as a labelled derivation. Shares outstanding ≈ **90–91m**.

Anchors:

• FY2025 (actual): net sales ¥251.5bn, OP ¥122.8bn (48.8% margin), net income attributable ¥84.65bn ``.
• FY2026 (guided): net sales ¥220bn (−12.5%), OP ¥100bn (45.5% margin) ``.
• Backlog ¥316bn at FY2025-end (visibility) ``; orders the swing factor.
• Structural driver: High-NA EUV ramp 2027–28 (ASML ~60 units 2026 → 80 units 2027) creates a second wave of actinic-inspection demand + the compounding service annuity (+48%/yr currently).

Three-year net-income path (FY2026E → FY2028E) — net income attributable, ¥bn:

**EPS derivation (base, ):** FY2026E NI ~¥78bn ÷ ~90.9m sh ≈ **~¥858 EPS**; FY2028E base ~¥112bn ÷ ~90.9m ≈ **~¥1,232 EPS** . At ~¥55,000, FY2026E base P/E ≈ ~64x; FY2028E base P/E ≈ ~45x `` — i.e. even on a constructive 2-year-out base case, the multiple only compresses to today's ASML level, and only if High-NA delivers on schedule.

The base call is fundamentally a bet on the High-NA timing — orders re-accelerating into 2027–28 and Lasertec holding ~100% actinic share into that wave. If that timing slips (TSMC has said it will not adopt High-NA for its 1.4nm node ``, deferring one large buyer), the bear path dominates.

(our model create skipped — unattended --watchlist run per SKILL; log a Brier forecast only on a committed base case in an interactive our position log pass. Suggested forecast to log later: "6920.T FY2027 net income attributable ≥ ¥90bn," p≈0.5, resolves 2027-08-15.)

Bull vs Bear

Bull case. Lasertec owns a true, ~100%-share actinic-EUV inspection monopoly at a physical chokepoint that every leading-edge EUV mask must pass through — and that necessity intensifies as the industry moves to 2nm, then High-NA (thinner films, tighter defect budgets make actinic inspection more, not less, essential). It converts that monopoly into ~45–49% operating margins and ~44–48% ROE on a fab-lite, net-cash, low-capex model that gushes FCF and returns it (dividend + buyback). Underneath the cyclical equipment line, a service annuity compounding ~+48%/yr is building an order-cycle-insensitive earnings floor. The AI-infrastructure build-out is the secular tailwind; the ASML High-NA ramp (2027–28) is a second demand wave the market is only beginning to price. And the biggest tail risk — the fraud allegation — was formally investigated and dismissed, so the 2024 derate was, in hindsight, a gift.

Bear case (2–3 things that could permanently impair or de-rate it).

1. Customer concentration is an existential fragility, not a footnote. Three customers ≈ 77% of revenue ``. Any one of TSMC/Samsung/Intel slowing capex (as already happened — ACTIS cancellations, orders −61% in FY2025) swings the order book violently. TSMC's decision to skip High-NA at 1.4nm removes/defers the largest buyer from the marquee up-cycle catalyst — the bull's own second-wave thesis has a hole in it.
2. Monopoly multiple on non-monopoly demand. At ~52x P/E / ~35x EV/EBITDA — above ASML — the price already capitalises the monopoly rent and a smooth High-NA re-acceleration. If orders stay soft into FY2027, there is no valuation cushion; this is a stock that can round-trip 65% (and just did, 2024→2025).
3. The governance scar hasn't fully healed. Even cleared, the Scorpion episode leaves a durable discount risk: any future working-capital wobble, inventory build against soft orders, or disclosure stumble reopens the "quality-of-earnings" debate instantly, and the market's memory is short but its trust is now conditional.

Pre-mortem (18 months out, thesis broke — what happened?): Orders failed to re-accelerate through 2026–27 because (a) TSMC's High-NA deferral + a broader AI-capex digestion pushed the mask-inspection wave right, and (b) a competitor (KLA e-beam, or a Zeiss/JV actinic effort) qualified an alternative on the High-NA node, cracking the 100% share narrative at exactly the moment the multiple needed it to hold. The stock de-rated from ~52x toward KLA's ~30x on a lower earnings base → a 40–50% drawdown, and a fresh disclosure question reignited the QoE overhang.

Are multiples too high? On today's earnings, yes — a monopoly-plus multiple that only looks reasonable if you underwrite the FY2028 High-NA up-leg with confidence. The margin of safety is thin.

Contrarian view (what the market refuses to see). The bull consensus treats Lasertec's monopoly as ASML-like (indispensable, durable, deserves the premium). The contrarian read: the monopoly is real but the demand is not ASML-like — it is cyclical, hyper-concentrated, and its flagship growth catalyst (High-NA) just lost its biggest customer (TSMC) at the leading node. The market is paying for monopoly durability it is actually only getting on share, not on volume cadence. The other contrarian point cuts bullish: the service annuity is under-appreciated — if you value the recurring installed-base stream separately from the lumpy equipment line, the "quality" of the earnings is higher than the order-cycle bears think. Both contrarian reads point to the same trade: own the service-annuity compounder at a cyclical-trough order print and a de-rated multiple — not at a fresh all-time high on a guidance-raise-that-disappointed.

Devil's Advocate (short-seller)

Channeling the actual short case — Scorpion Capital, June 2024 — plus the structural bear.

• What could structurally break how it makes money: the entire P&L is derivative of ~three customers' EUV capex timing at the leading edge. It does not control its own demand cadence. A leading-edge capex pause, a node delay, or a shift in mask-buying patterns (e.g. more mask reuse, fewer new tape-outs) directly compresses the order book — and orders already fell 61% in FY2025 with ACTIS cancellations, proving the fragility is live, not theoretical.
• Revenue concentration → what happens if it shifts: 77% top-3. Scorpion's specific claim was that TSMC had ceased buying its EUV machines and that Intel/Samsung were "struggling, exasperated, or furious" with the tools ``. Even if overstated, any directional truth here is devastating: lose or slow one of three buyers and the growth story inverts.
• Why the moat may be weaker than bulls think: it's a single-product-line monopoly. KLA leads e-beam/die-to-database mask inspection and is a far larger, more diversified process-control franchise; if KLA (or a Zeiss-led effort) closes the actinic gap on the High-NA node — where the mask ecosystem is being rebuilt anyway — the 100% share that justifies the premium is exactly what erodes. The most dangerous competitor bulls underestimate is KLA, because it has the balance sheet, the customer relationships, and the incumbency in adjacent mask inspection to fund a multi-year actinic push.
• Worst capital-allocation / governance: the substance is thinner here (no proven related-party looting), but the Scorpion allegations of inflated inventory, aggressive revenue recognition, and a "hoax" A300 next-gen product + a deserted Yokohama Innovation Park (investigators' ~20 site visits Dec-2023→May-2024 allegedly found no visible R&D/production) `` are the short's core. The company's response was terse and legalistic, and the exonerating investigation was company-commissioned — a skeptic discounts both.
• Assumptions that must hold for today's price (~¥55k, ~52x): (1) orders re-accelerate 2026→2028; (2) High-NA arrives on schedule and drives actinic demand despite TSMC's 1.4nm High-NA skip; (3) Lasertec holds ~100% actinic share through the High-NA transition; (4) margins stay ~45%+; (5) no fresh QoE/inventory scare. That is a demanding, serially-dependent stack.
• If growth disappoints 20–30%: on a high-fixed-cost-of-multiple name, a 20–30% revenue/order miss vs. the recovery path likely triggers a de-rate from ~52x toward ~30x (KLA) on a lower base → a 40–55% drawdown, rhyming with the 2024→2025 −65% round-trip.
• Single scenario that permanently impairs the business: a credible actinic competitor qualifies on the High-NA node (KLA/Zeiss) — breaking the 100% share that is the entire investment case — combined with a structurally lower leading-edge tape-out cadence. Plausibility: low-to-moderate over 3 years, rising thereafter. The moat is deep but narrow; narrow moats are the ones that get flanked.

Management Questions (ordered by information value)

1. Provide the exact revenue-recognition policy for EUV inspection tools (shipment vs. installation vs. customer acceptance), and quantify what fraction of FY2025 revenue was recognised in the final 30 days of the fiscal year. (Directly tests the core short allegation.)
2. Give the inventory balance, inventory-days, and contract-liabilities (customer advances) for the last 8 quarters, and explain any inventory build against the 61% order decline in FY2025. (The single highest-value QoE gap in this dossier.)
3. What is the precise revenue share of your top 1 / top 3 customers in FY2024 and FY2025, and how do you de-risk a single-customer capex pause? (Sizes the concentration risk.)
4. TSMC has said it will not adopt High-NA EUV at its 1.4nm node. Quantify the actinic-inspection demand you expect from High-NA excluding TSMC leading-edge, and when. (Tests the up-leg thesis's biggest hole.)
5. Bridge the FY2025 order collapse (−61%) and ACTIS cancellations to your FY2026 guidance raise — what specifically re-accelerated, and is it orders or backlog burn?
6. What is the replacement/competitive threat from KLA e-beam and Zeiss actinic on the High-NA node specifically — where the mask ecosystem is being re-qualified anyway?
7. Decompose the service segment (+48% FY2025): how much is contractual/recurring vs. one-off upgrades, and what is its steady-state margin vs. equipment?
8. What is the A200HiT High-NA qualification status at each of your three lead customers, and what is the design-win-to-revenue lag?
9. What specific governance and disclosure changes did you implement after the 2024 short report and Special Investigation Committee — beyond the "no irregularities" finding?
10. What is insider/management ownership, and how is executive compensation tied to ROIC / long-term share performance vs. reported earnings?
11. How exposed is the light-source / EUV-optics supply chain (single-source dependencies), and what is your second-source strategy?
12. What is your China exposure (DUV mask inspection) and the export-control risk to it?
13. Why hold the FY2026 dividend flat at ¥329 after years of double-digit growth — is this a demand signal or a capital-allocation shift?
14. What would peak leading-edge mask-inspection TAM look like at full 2nm+High-NA industry adoption, and what share do you underwrite holding?
15. What single operational or market development would most change your own view of the 3-year outlook — bull or bear?

---