ASML Holding

Phase A — Understand the business

Lens 1 · Company Overview

ASML Holding N.V. (Veldhoven, Netherlands; Nasdaq: ASML; nominal value €0.09/share) is the world's sole supplier of extreme-ultraviolet (EUV) lithography systems — the machines that pattern the smallest features on the most advanced logic and memory chips. It is, in the most literal sense the word allows, a monopoly: no other company on earth can build an EUV scanner, and every leading-edge chip (every advanced AI accelerator, every leading-edge CPU, every cutting-edge DRAM bit) is printed on an ASML machine or it is not printed at all.

Business model. Two revenue engines ``:

• Net system sales — €24,474.3M in FY2025 (74.9% of total), selling lithography systems (EUV 0.33 NA "NXE", EUV 0.55 NA "High-NA / EXE", DUV immersion "NXT", DUV dry, and metrology/inspection). A single High-NA EXE system lists in the hundreds of millions of euros; a 0.33-NA EUV NXE in the ~€180–220M range ``.
• Net service & field option sales (installed-base management) — €8,193.0M (25.1%), +26.2% YoY — recurring service contracts, productivity upgrades, and field options on the growing installed base. This is the annuity, and it is the fastest-growing line ``.

One reportable segment. ASML reports as a single "holistic lithography" segment; the CODM (CEO + CFO jointly) manages on consolidated US-GAAP net income and sales ``.

Customers (extreme concentration). Four customers each individually exceeded 10% of total net sales in 2025, together €20.0B = 61.2% of revenue (2024: 53.8%) . These are, by every external account, **TSMC, Samsung, SK Hynix, and Intel/Micron** . End-markets: Logic (leading-edge foundry, the AI-accelerator supply) and Memory (HBM + DDR5 for AI). In Logic, net sales rose +€2.9B in 2025 on leading-edge foundry AI capacity; Memory dipped −€0.2B off a strong 2024 but remains HBM/DDR5-driven ``.

Suppliers. Critically single-sourced upstream (see Lens 2): Carl Zeiss SMT (optics — ASML holds a stake), Cymer (light source, ASML-owned), TRUMPF (CO2 lasers for EUV), VDL and a deep Dutch/German supplier base.

Contract structure — the revenue-quality crown jewel. Customers buy under volume purchase agreements covering up to 5 years, with large up-front down-payments. Contract liabilities (customer cash paid in advance) stood at €19,372.6M (€16,006.3M current + €3,366.3M non-current) at year-end ``. Demand is pre-funded by the buyer — the opposite of channel risk.

Lens 2 · Supply Chain

Upstream inputs → ASML → end customer, every named link:

Tier-1 critical suppliers (single-source chokepoints):

• Carl Zeiss SMT (Oberkochen, Germany) — the EUV optics. Mirrors polished to sub-atomic tolerances; the single hardest input in the machine. ASML holds an equity-method stake; equity-method investments €822.6M at YE2025 ``. No alternative exists.
• Cymer (San Diego, ASML-owned since 2013) — the EUV light source (laser-produced plasma; tin droplets hit 50,000×/sec by a CO2 laser).
• TRUMPF (Ditzingen, Germany) — the high-power CO2 lasers driving the Cymer source.
• VDL Groep, Settels, Neways, and ~700 Tier-1 suppliers, concentrated in the Brainport Eindhoven cluster (NL) and southern Germany.

ASML integrates ~100,000 parts per EUV scanner in Veldhoven.

Downstream (the buyers): TSMC (Hsinchu + Arizona GigaFab, $52–56B 2026 capex ), **Samsung** (memory + foundry, ~$40B capex), **SK Hynix** (HBM — sold out 2026 output ), Intel (Oregon/Arizona, lead High-NA customer, but capex DOWN <$18B in 2026 ``), Micron.

Chokepoints / single-source dependencies:

1. Zeiss optics — the binding constraint on EUV output. ASML can only ship as many EUV systems as Zeiss can supply mirror sets.
2. The plasma light-source supply chain (Cymer + TRUMPF) — high-power-source yield gates productivity.
3. ASML itself is the chokepoint for the entire semiconductor industry — it is the single point of failure for Moore's Law. There is no second EUV vendor and there will not be one this decade.

This lens is the whole thesis in one diagram: a monopoly fed by a handful of irreplaceable European suppliers, selling to four buyers who have no alternative and pay in advance.

Lens 3 · Competitive Advantages (moats)

ASML's moat is among the widest in public markets — arguably the single widest in hardware.

• Absolute technological monopoly in EUV. Nikon and Canon compete in mature DUV but have no EUV product and have abandoned the race. EUV took ASML ~20 years and >€6B of R&D (much of it co-funded by Intel, Samsung, TSMC as customer-shareholders in the 2012 "Customer Co-Investment Program"). The barrier is not a patent — it is two decades of accumulated process knowledge across ASML + Zeiss + Cymer + TRUMPF that cannot be cloned with capital. China's SMEE is a generation-plus behind even in DUV; an indigenous EUV tool is not a 2020s event ``.
• Switching costs / lock-in. Once a fab qualifies a process on ASML tools (months of work), it is captive — the installed base then throws off 25%+ of revenue as high-margin service ``.
• Bargaining power — ASML needs its customers less than they need it. Down-payments of ~€19.4B sit on the balance sheet ``; customers fund ASML's working capital. The constraint on ASML's growth is its own (and Zeiss's) output, not demand.
• The roadmap moat. High-NA (0.55 NA, EXE series) extends the monopoly into the sub-2nm era; Hyper-NA is on the roadmap beyond. By the time anyone could conceivably replicate 0.33-NA EUV, ASML is two nodes ahead.
• Brand / perceived value. Among chipmakers ASML is not a vendor but a partner — the 2024 Investor Day framing of "lithography at the heart of customer innovation" is, unusually, literally true.

The one genuine constraint on this moat is not competitive — it is political (export controls; see Lens 10/13). The market cannot take ASML's customers; governments can restrict where ASML sells.

Lens 4 · Segments (revenue, geography, end-market)

ASML has one reportable segment, so the breakdowns are by technology, end-use, and geography ``.

By revenue line (€M):

¹ service derived as total − system . New systems €23,898.6M / used €575.7M in 2025 .

Units: 327 lithography systems sold in 2025 (2024: 418); 48 EUV recognized (4 EXE + 44 NXE; 2024: 44 EUV = 2 EXE + 42 NXE); DUV 279 units (2024: 374) ``. Read that carefully: unit count FELL 22% but revenue ROSE 15.6% — the mix shifted decisively to high-ASP EUV and high-margin service. This is the most important single fact about FY2025.

By geography — total net sales (€M) and the China normalization ``:

The story: China fell from 36.1%→29.1% as the 2023–24 DUV pre-buy (front-running export controls) unwound; Taiwan exploded 15.4%→25.5% on TSMC's AI/leading-edge ramp; Korea rose to 25.0% on HBM/DRAM. The demand baton passed cleanly from China stockpiling to AI leading-edge — a healthier, more durable mix.

Geographic trend = accelerating (AI-driven Taiwan/Korea) offsetting decelerating (China policy-driven). The cause is structural, not cyclical.

---

Phase B — Measure performance

Lens 5 · Earnings Result

FY2025 full-year P&L (€M, U.S. GAAP) ``:

Drivers of the print: Revenue +€4.4B = system sales +12.4% (higher EUV + DUV-immersion, partly offset by lower ArF-dry/KrF/i-line volumes) and service +26.2% (growing installed base + NXE field upgrades). Gross margin expanded to 52.8% on a favorable NXE product mix and richer service, partly offset by the dilutive impact of EXE (High-NA) systems recognized in sales — High-NA is margin-dilutive today as it ramps, a near-term headwind that flips to a tailwind by 2030 (the €44–60B / 56–60% GM model) ``. Operating leverage was real: opex grew ~9% on 15.6% revenue → +2.7pp operating margin.

Balance-sheet flags (all clean → positive):

• FCF €11,027.3M (CFO €12,658.5M − PP&E €1,573.6M − intangibles €57.6M) ``.
• Accounts receivable FELL to €3,023.0M from €4,477.5M while revenue rose — receivables down on rising sales is the single cleanest quality signal there is ``.
• Inventories €11,429.3M (up modestly from €10,891.5M) — ~35% of sales, but structural for 12–18-month-build capital equipment, not a stuffing flag ``.
• Net cash ~€8,931M ``.
• Contract liabilities €19,372.6M — buyers pre-funding €19B of future shipments.

Guidance / outlook (tone strengthened sharply): FY2026 €34–39B / GM 51–53% / ETR ~17% (set at the 20-F); Q1'26 €8.2–8.9B . **The market then got better news:** Q1'26 actual €8.8B (beat ~€8.7B), EPS €7.15, and management **raised FY2026 to €36–40B** on AI demand .

Latest print (Q1 2026): net sales €8.8B, EPS €7.15, guidance raised — but stock fell ~6-7% on the MATCH Act China-DUV-ban headline despite the beat ``. The fundamentals beat; the politics dictated the tape.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts are in the research layer (transcripts/ empty), so this is ``-grounded against the CEO/CFO annual-report letters + reported call coverage.

Trajectory of management tone (4-period arc):

• Early 2025: cautious/mixed — "AI benefiting only a small number of customers," geopolitical/tariff uncertainty, soft non-AI (PC/smartphone) demand ``.
• Mid-2025: inflecting — AI demand "began to spread to more customers," all three DRAM players seeing AI uptick.
• Late 2025 / FY report: confident — "new and significant demand for AI… fueling capacity build-up across our broad customer base… we believe will continue in 2026 and beyond" (CEO Fouquet) ``. Outlook "strengthened significantly in the final months of 2025."
• Q1 2026 call: outright bullish — raised FY26 guide to €36–40B, "unprecedented demand," building ≥60 EUV systems in 2026 (+36%) ``.

Recurring phrases (rising): "AI," "broadening customer base," "High-NA maturity," "holistic lithography," "lithography at the heart of innovation." What they stopped emphasizing: quarterly net bookings — ASML stopped publishing bookings as a quarterly metric ``, having grown frustrated that a lumpy order number drove the stock more than fundamentals. Bulls read this as confidence (backlog visibility makes the metric noise); bears read it as removing a demand barometer right as China DUV orders roll off. Net tone: steadily, materially more positive across the year, peaking on the Q1'26 raise.

Lens 7 · Comps

Peer table — ASML vs. the WFE complex.

5-yr avg ROE: n/a as a clean 5-yr series; FY2025 ROE ≈ 49% — exceptional, and structurally high because the down-payment model finances the balance sheet.

Read: ASML trades at a ~55x TTM P/E vs a ~28–33x peer band — a 70–100% premium, against a 15-year average of ~34x ``. The premium is not irrational (it is the only monopoly in the group, with the widest moat and the cleanest balance sheet), but it leaves zero margin for disappointment. You are paying a monopoly multiple on top of a cyclical-peak earnings base.

Lens 8 · Stock-Price Catalysts (what actually moves >5%)

The 5-year pattern is unambiguous: ASML trades on two things — AI/EUV demand signals (up) and China export policy (down). Earnings beats alone rarely hold if a China headline lands the same week ``.

• 2021–H1'22: ran with the post-COVID semi supercycle; EUV backlog ballooned. Multiple expansion.
• Oct 2022: US October-7 export controls on advanced China chipmaking — first major China-policy drawdown.
• 2023: Dutch government formalized DUV-immersion export licenses (Sep 2023); China then pre-bought DUV aggressively to front-run the rules → China hit 36%+ of revenue, inflating 2024.
• 2024: lumpy bookings prints drove violent moves; a weak Q3'24 booking number triggered a sharp single-day drop — the episode that ultimately led ASML to stop reporting quarterly bookings.
• 2025: AI demand broadened; backlog €38.8B at YE; Q4'25 net bookings €13.2B incl. €7.4B EUV . Stock re-rated hard (52-wk low $683 → high $1,914) .
• Apr 15, 2026 (latest): +16% into the print, then −6-7% out of it — beat + raised guide, but the MATCH Act (bipartisan US bill banning DUV-to-China, naming ASML in the text) dominated ``. China system sales fell 36%→19% Q4'25→Q1'26.

What the market reacts to: (1) anything that changes the China DUV revenue line (policy — the dominant negative catalyst); (2) EUV/High-NA demand and the AI capex narrative (the dominant positive); (3) the single-customer signals from TSMC capex. It does NOT meaningfully react to routine GM/opex beats — those are assumed.

---

Phase C — Judge people & books

Lens 9 · Management

CEO: Christophe Fouquet — President, CEO and Chair of the Board of Management ``. A ~30-year ASML insider (ran the EUV business unit, then CTO/Chief Business Officer) who succeeded Peter Wennink in April 2024. Archetype: the deep operator-technologist, not a parachuted-in professional manager — exactly the profile you want running a company whose moat IS its process knowledge. Continuity over disruption.

CFO: Roger Dassen — long-tenured, ex-Deloitte global board; ASML's reputation for clean, conservative, award-winning reporting (and ESRS-leading disclosure) is largely his ``. Notably, Dassen took the board seat on Mistral AI, signaling how strategically ASML treats that bet.

1. Track record (quantified): Under the Wennink→Fouquet continuum, ASML grew from ~€11B (2018) to €32.7B (2025) revenue , delivered EUV from lab curiosity to industry-standard, and shipped first full-spec High-NA (EXE:5200B) after ~10 years of development . Operating margin 31.9%→34.6% in a single year on Fouquet's watch.
2. Tenure & skin in the game: Fouquet ~30 yrs at ASML; insider ownership is modest in absolute terms (typical for European large-cap, equity-comp-driven) — insider-transactions.csv not present, so precise holdings n/a. The alignment comes from career-long tenure, not a founder stake.
3. Capital allocation — strong and shareholder-friendly: €8.5B returned in FY2025 (€2.6B dividends + €5.9B buyback) ; treasury shares grew to €(2,252.9)M from €(476.0)M; share count fell 393.3M→385.4M . A new €12B buyback (2026–2028) + authorization for up to 10% of capital through Oct 2027 ; dividend raised to €7.50 (+17%) . ROE ~49% ``. This is textbook disciplined allocation: reinvest in R&D (€4.7B), then return the rest.
4. Red flags: Minimal. The one item to watch is the €1.3B Mistral AI investment for ~11% (ASML is now Mistral's biggest shareholder) `` — a strategic-but-unusual move for a litho monopoly to take a venture stake in an LLM lab. Bull read: cheap optionality + a real internal-R&D accelerant. Bear read: scope creep / "AI-washing" capital outside the core. It is small (€1.3B of an €11B+ FCF base) and defensible, but it is the first thing a skeptic flags.
5. Founder vs professional manager: Professional managers, but of the rare insider-lifer kind whose incentives and knowledge are fused to the asset. For a company where the moat is institutional, that is the correct archetype.

Lens 10 · Forensic Red Flags

Forensic pass across income statement, balance sheet, cash flow. The headline: this is one of the cleanest large-cap books in the market.

• Revenue recognition: Standard ASC 606; revenue on delivery/installation under 5-yr volume purchase agreements. Remaining performance obligations €46.5B (2024: €43.3B), 65% to be recognized within 12 months ``. No aggressive percentage-of-completion games — recognition is event-driven (ship/install). Cumulative catch-up adjustments disclosed and immaterial.
• Cash flow vs earnings — converges, doesn't diverge: CFO €12,658.5M > net income €9,609.4M (1.32x) ``. Earnings are over-covered by cash. No red flag — the opposite.
• Receivables vs revenue: Receivables FELL (€4,477.5M→€3,023.0M) on +15.6% revenue ``. Clean.
• Inventory: €11,429.3M (~35% of sales) is high in the abstract but normal for a maker of €200M+ machines with 12–18-month build cycles; grew far slower than the order book. Watch only for a demand air-pocket that strands WIP.
• SBC / non-GAAP: ASML reports primarily on US-GAAP net income (the only non-GAAP measure it leans on is free cash flow, fully reconciled) ``. It does not run a heavily-adjusted "non-GAAP EPS" that flatters away stock comp — a meaningful integrity point vs US semi peers.
• Goodwill / intangibles: Goodwill €4,588.6M flat (no impairment); other intangibles only €540.1M ``. Minimal acquisition froth; the balance sheet is overwhelmingly real assets + cash.
• Tax: ETR 17.7% (2025) vs Dutch statutory 25.8%, bridged primarily by the Dutch Innovation Box (qualified R&D income taxed at 9.0%; €1,055.8M benefit; agreement confirmed with Dutch authorities through 2028) ``. This is a legitimate, durable, disclosed structural advantage — not a contrivance — but it is a modeling input (a future Innovation-Box change would lift the tax rate).
• Related parties — the one item to actually watch: Non-current loans receivable to related parties €1,647.3M (up from €1,440.8M); related-party accounts payable €1,085.8M; equity-method investment in Carl Zeiss SMT ``. These flow from the deep Zeiss/strategic-supplier relationships and are normal for ASML's structure, but related-party loans of €1.6B warrant ongoing attention. Not a flag today; a thing to keep eyes on.
• Contingencies: Management states ASML "does not have any material contingencies which are considered probable or reasonably possible" as of 2025-12-31 ``.

Regulatory findings (required sub-section):

• SEC Litigation Releases: None for ASML in the search period (2021-06-18 → 2026-06-18) ``.
• SEC AAERs: None ``.
• 10-K/20-F Item — Legal proceedings (Note 17): No material litigation or contingencies disclosed; standard IP-indemnification language for customers, no active material case ``.
• Non-SEC enforcement (web): No material FTC/DOJ/FDA/CFPB enforcement actions or consent decrees against ASML found ``. The dominant "regulatory" exposure is export control / sanctions — Dutch + US licensing of EUV and certain DUV-immersion tools, and the prospective MATCH Act — which is a trade-policy constraint on the business, not an accounting/enforcement finding (covered fully in Lens 13). Historic trade-secret matters (the 2018–19 XTAL/Dongfang case) are resolved and absent from this filing.
• Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 20-F Note 17 as of 2026-06-18.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028 EPS)

Built bottom-up from FY2025 actuals + management guidance. Currency: EUR (filing basis). Outputs `` with arithmetic. Base anchors: FY2025 EPS €24.73; FY2025 net margin 29.4%; ~385M shares falling ~1.5%/yr on buyback.

Revenue path:

• FY2026: management guides €36–40B (raised post-Q1) ``. Base €38.0B (+16.3%).
• FY2027: AI leading-edge + High-NA ramp + memory recovery; toward the 2030 €44–60B model (€52B moderate by 2030 ≈ ~9% CAGR off 2025) . Base **€43.5B** (+14.5%) .
• FY2028: continued High-NA HVM (Fouquet: HVM begins 2027–28) . Base **€48.5B** (+11.5%) .

Margin path: GM 51–53% guided FY26 (High-NA dilution caps near-term), rising toward 56–60% by 2030. Net margin: hold ~29% FY26 (High-NA drag), ~30% FY27, ~31% FY28 ``.

EPS bridge ``:

(FY2026 cross-check: Q1'26 actual EPS was €7.15 ``; ×4 naively = €28.6, consistent with ~€29 base.)

Scenarios (FY2026 EPS):

• Bull ~€31 — revenue €40B (top of raised guide), GM 53%, China DUV holds longer than feared.
• Base ~€29 — midpoint above.
• Bear ~€26 — MATCH Act passes mid-year and removes ~5% of revenue (€2B DUV); revenue ~€36B, GM 51%.

Forecast NOT logged (forecast.ts create skipped per --watchlist breadth-loop rule). For the record, the trackable base call would be: "ASML FY2026 (calendar, ends 2026-12-31) US-GAAP basic EPS ≥ €28.0, p≈0.72."

Lens 12 · Bull vs Bear

Bull case. ASML is the single non-substitutable node in the entire AI build-out: you cannot make a leading-edge AI chip without an ASML machine, full stop, and that will be true for the rest of the decade. Demand is no longer one-customer-deep — it has broadened across foundry AND all three DRAM players , so the cycle has a wider, more durable base than 2021's. **High-NA extends the monopoly two nodes forward** (Intel 14A, imec sub-2nm), pulling the 2030 model to €44–60B at 56–60% gross margin . The financials are a fortress: ~49% ROE , €11B FCF, net cash, €19B of customer down-payments pre-funding growth, €12B buyback in flight. Earnings surprises skew up — management just **raised** FY26 mid-cycle. This is a compounding monopoly with a 5-year demand runway visibly contracted into a €46.5B RPO .

Bear case (risks that could permanently impair or de-rate):

1. The China DUV cliff is now legislative, not just regulatory. The MATCH Act names ASML in its text and would ban DUV-immersion sales to China — ~5% of group revenue (€5-7B of the higher-margin-than-feared but still meaningful DUV line) ``. China already fell 36%→19% of system sales in two quarters. A blanket ban is a permanent revenue subtraction, not a timing issue.
2. Cyclical peak baked into a peak multiple. ~55x TTM P/E vs ~34x 15-yr average `` on top of record earnings. If AI capex digests in 2027 (TSMC/Samsung/SK Hynix all spending at record levels now), both E and the multiple compress simultaneously — the classic semi-cap double-whammy.
3. High-NA adoption risk + customer concentration. TSMC is delaying High-NA — bypassing it for A14, staying on 0.33-NA to 2028 . ASML's largest customer is in no hurry for its flagship next-gen tool. If High-NA economics don't close for the broad base, the 2030 model's EUV mix is at risk. And 61.2% of revenue sits with four buyers — any one capex pause is felt hard.

Pre-mortem (18 months out, thesis broke): It's late 2027. The MATCH Act passed and a reciprocal Chinese rare-earth/gallium counter-measure pinched ASML's own supply ``. AI capex took a breather as hyperscalers digested 2025-26 buildouts; TSMC pushed High-NA further right; ASML printed a flat-to-down 2027 and the multiple compressed from 55x to 35x — the stock halved despite the monopoly being fully intact. Nothing broke the business; the cycle and the politics broke the price.

Multiples too high? On a 2-3 year view, no — €40+ FY28 EPS `` at even a 35x multiple is ~€1,430/share, near today. On a 12-month view, the multiple offers no cushion: you are underwriting flawless execution + benign China policy.

Contrarian view (what the market refuses to see): The bears are fighting the last war on China. The China DUV roll-off is already in the numbers and the guidance (36%→19%, China explicitly guided to ~20%) — the MATCH Act mostly codifies a decline ASML already absorbed while still raising 2026 guidance. The genuine, under-priced risk is not China — it is AI-capex digestion timing in 2027, which nobody wants to model because the current demand is too good. The monopoly is not the question; the cycle is.

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

Dismantling the bull case.

• Where revenue is concentrated, and the break: 61.2% of sales in four customers; 29.1% in one country (China) that is being legislated away ``. The bull's "broadening" is real but thin — it broadened to the same four giant buyers' AI lines. If hyperscaler AI capex (which funds TSMC/Samsung/SK Hynix orders) pauses even two quarters, ASML's order book — now hidden because they stopped reporting bookings — could roll over with no public warning. Removing the bookings disclosure right as China rolls off is, to a skeptic, exactly what you'd do to obscure a softening order rate.
• Why the moat may be weaker than bulls think — at the margin: The moat protects that chips need EUV, not how many tools sell in a given year. ASML is a monopoly on a cyclical — monopoly pricing does not exempt it from the capex cycle, and a monopoly trading at 55x is a monopoly priced as if the cycle were abolished. It wasn't.
• Most dangerous thing bulls underestimate: TSMC's High-NA delay. Your single most important customer is publicly skipping your flagship next-generation system for its A14 node ``. If 0.33-NA "good enough" persists, the High-NA super-cycle the 2030 model depends on slips years right, and the EXE order book the premium is built on thins.
• Worst capital-allocation move: The €1.3B Mistral stake — a litho-monopoly buying 11% of an LLM startup whose valuation is itself a bubble candidate ``. If Mistral marks down, it's a visible "what were they thinking" write-down; even if it doesn't, it's €1.3B of FCF outside the core, and "AI-washing" by a company that already IS the AI supply chain.
• What must hold for today's price: FY26 €36-40B delivered, GM holding 51%+, NO blanket China DUV ban, High-NA broadening beyond Intel, and AI capex NOT digesting before 2028. That's four-to-five things all going right, at 55x.
• Growth disappoints 20-30%: If FY26 comes in at €30-31B (flat-ish) instead of €38B base, EPS drops to ~€23 and the multiple compresses to mid-30s → stock ~€800-900, roughly −45-50% ``. The downside is a re-rate, and it's large because the multiple is large.
• Single scenario that permanently impairs: A genuine, lasting AI-demand reset (over-built 2025-26 capacity → multi-year digestion) layered on a full China DUV ban + Chinese rare-earth retaliation hitting ASML's own inputs ``. Plausibility: low-to-moderate — each piece is real, the combination in 2027 is the tail. The business survives intact; the equity does not survive the de-rate unscathed.

Lens 14 · Management Questions (ordered by information value)

1. With quarterly bookings no longer disclosed, what specific forward metric should investors use to detect an order-rate inflection, and will you commit to it?
2. If the MATCH Act passes in its current form, what is the precise FY2026/27 revenue and gross-margin impact, and how much is recoverable via non-China demand reallocation?
3. TSMC is bypassing High-NA for A14 — what is your conviction (and evidence) that High-NA reaches broad high-volume manufacturing in 2027-28 rather than slipping to 2029+?
4. Walk us through the path from 52.8% gross margin to the 56-60% 2030 model — how much is High-NA crossing breakeven vs. mix vs. service, and when does High-NA stop being dilutive?
5. How do you stress-test the order book for an AI-capex digestion in 2027, and what early indicators would change your capacity plan?
6. What is the strategic logic — and the financial return hurdle — for the €1.3B Mistral stake, and under what circumstances would you write it down or add to it?
7. Zeiss optics is the binding constraint on EUV output — what is the realistic ceiling on annual EUV system shipments through 2028, and how are you de-risking that single-source dependency?
8. China retaliation on rare earths/gallium is a named risk — how exposed is your own supply chain (not just your China sales) to Chinese counter-measures?
9. At ~49% ROE and net cash with a €12B buyback, why not lever the balance sheet to accelerate returns — what is the conservatism buying you?
10. How much of the €46.5B RPO is firm/non-cancellable vs. subject to volume-purchase-agreement modification, and what's the historical cancellation rate through a downturn?
11. What is the realistic timeline and capex for Hyper-NA, and does it extend the monopoly or does High-NA prove to be the last node where litho is the bottleneck?
12. Customer concentration is rising (61.2% in four) — does the AI buildout structurally entrench that, and how do you think about a single-customer capex shock?
13. Service is now 25% of revenue and growing 26% — what is the steady-state ceiling on installed-base revenue as a share of the total, and its margin?
14. How does the Innovation Box (9% R&D tax rate, agreed through 2028) renew beyond 2028, and what's the ETR sensitivity if it changes?
15. On capital allocation, how do you weigh a step-up in the dividend vs. buybacks at today's ~55x multiple — are you comfortable repurchasing stock at this valuation?

---