ASML Holding (US-listed NY registry shares)
Capital-equipment OEM: sells advanced lithography scanners (EUV / High-NA EUV / DUV) to leading-edge logic and memory fabs, plus a high-margin recurring Installed Base Management (service, upgrades, spares) annuity on a fleet of ~6,500 systems.
The thesis on this name
State of AI Compute
The single most irreplaceable asset in compute — sole EUV/High-NA source, 10+yr moat, >1yr backlog, +26% service annuity — but verify says trade the franchise not the High-NA option and NOT at the 25-Jun multiple: the base-EUV franchise (10yr) and the High-NA upgrade leg (Intel/Samsung-dependent after TSMC's A14 skip) are two separate cash-flow streams; acc…
State of AI Compute
Long ASML over 18–36 months as the sole-source EUV/High-NA monopoly compounds a >€38B backlog into structural 10–14% revenue CAGR to 2030, entered as a starter now and scaled on the cyclical drawdown — edge is structural (irreplaceable single-source moat the market prices but cannot durably arbitrage), NOT a call on n…
State of AI Compute
Sole-source EUV/High-NA monopoly, but ~flat expected return at a near-ATH entry with Q1-26 book-to-bill already ~0.94 and TSMC High-NA deferred to 2029+. 1.0% patient starter; first add only on a confirmed 2nd sub-1.0 b…
State of the Memory Supercycle
US-listed ADR of the EUV monopoly that increasingly gates DRAM supply: SK Hynix was first to install High-NA EUV (TWINSCAN EXE:5200B) in DRAM and committed ~$8B to ASML EUV by 2027; Samsung is taking multiple High-NA tools for memory + foundry (TrendForce, 2025–26). As DRAM scaling to 1c-node adopts more EUV layers, ASML's per-DRAM-wafer content rises structurally — litho becomes a memory toll-booth, not just a logic one. Netherlands name expressed via the US ADR; sized medium because EUV is heavily logic/foundry-driven and the memory contribution, while growing, is still the smaller share — it is a context/demand name being upgraded to a real memory-supply gate, not a pure memory play.
State of the Memory Supercycle
The EUV monopoly that now gates DRAM — memory orders +186% to EUR2.54B (Q3 2025), >50% of Q1 2026 shipments are memory, SK Hynix installing 20 low-NA EUV for HBM; the ultimate memory-supply choke point, expressed via the US-listed shares.
State of the Memory Supercycle
EUV now gates advanced DRAM — memory bookings +186%, >50% of 2026 shipments memory. Held via US-listed NY registry shares; non-US-domiciled caveat.
Earnings, margins, COGS & capex
ASML earns from selling a small number (~400–600/yr) of extremely high-ASP lithography systems plus a fast-growing service/upgrade annuity; FY2025 set records of €32.7B sales (+16%) and €9.6B net income at 52.8% gross margin (fact). Growth re-accelerated into 2026 on AI-driven EUV demand: Q1 FY26 was €8.8B (+13%) at 53.0% GM, and management raised FY26 guidance to €36–40B (fact). The model is gated by EUV build capacity, not demand — a €38.8B year-end-2025 backlog (fact) extends visibility into 2027. The long-term model targets €44–60B revenue at 56–60% gross margin by 2030 (fact, Investor Day reaffirmed).
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~46¢ is cost of goods and ~17¢ operating expense, leaving ~37¢ of operating profit (~31¢ net).
Revenue trend
Margins
up — high end of guide; FY25 52.8% vs FY24 ~51% (fact)
flat-to-up with EUV/service mix
up vs FY25 ~29%
lumpy — driven by customer down-payment timing, not deterioration (fact)
COGS structure
COGS is dominated by purchased high-precision components from a concentrated supply base — Zeiss SMT optics modules, Cymer (ASML-owned) light sources, Trumpf CO2 lasers, plus precision mechatronics and thousands of suppliers; an EUV system carries >100,000 parts. Mix is the key margin lever: EUV/High-NA and the high-margin Installed Base Management (service/upgrades) carry richer margins than DUV, so gross margin rises as EUV and service mix grows toward the 56–60% 2030 target.
Capex
Asset-light vs customers: ASML's own capex (~5–6% of sales, est) funds EUV/High-NA manufacturing capacity expansion (target ~90 EUV + High-NA units/yr by 2025–26), the Veldhoven and Berlin campuses, and R&D facilities — not fabs. The capital intensity of the industry sits with ASML's customers (TSMC/Samsung/Intel), whose multi-hundred-billion fab capex is ASML's demand signal.
Latest earnings
Beat on both lines management guides. Net sales €9,326.5M topped the €8.4-9.0B guide and gross margin 54.0% topped the 51-52% guide - the CEO attributes both 'primarily to higher than expected Installed Base Management sales', i.e. the annuity, not system timing. Shares closed +2.23% at $1,815.27 on results day (15 Jul). Note ASML no longer publishes a quarterly bookings number (discontinued from Q1 2026), so there is no order print to beat or miss against.
Q3 FY26: €11.0-12.0B net sales (+46% to +60% YoY) of which Installed Base Management ~€2.9B; gross margin 55-57%; R&D ~€1.2B; SG&A ~€0.4B. FY2026 RAISED to €43-45B at 54-56% gross margin - the second raise of the year, from €36-40B at 51-53% in April; annualized effective tax rate ~17%. Capacity: adding 30% to the ~65-unit 2026 low-NA EUV capacity for 2027 and investigating another 30% for 2028, with the same treatment for the ~130-unit DUV immersion capacity. Management says it is 'pretty much already close to receive all the orders we need for 2027'. The standing 2030 model (€44-60B revenue, 56-60% GM, set at the Nov 2024 Investor Day) will not be revised until the 10 Jun 2027 Capital Markets Day.
- Operating margin (Q2 FY26)
- 37.1% - record; disclosed in the filing, not estimated
- Installed Base Management (Q2 FY26)
- €2,761.7M (+31.8% YoY); guided ~€2.9B for Q3
- FY2026 revenue guide
- €43-45B, raised from €36-40B in April (+32% to +38% vs FY2025)
- Free cash flow (Q2 FY26)
- €1,317M (14.1% margin) - but H1 FY26 is still -€1,291M
- Lithography systems sold (Q2 FY26)
- 91 units (86 new + 5 used), vs 79 in Q1 FY26
- 2026 capacity base disclosed
- ~65 low-NA EUV and ~130 DUV immersion units/yr, both +30% planned for 2027
- China share of net sales (Q2 FY26)
- ~20% (management on the call), vs ~33% for FY2025
- EUV growth expected FY2026
- ~+45%; memory revenue ~+75%, advanced logic ~+25% (management)
Growth drivers
- AI-accelerator demand pulling forward leading-edge logic capacity (TSMC N2/A16, Intel 18A/14A, Samsung SF2) — customers accelerating buildouts cited as the reason for the Q1 FY26 guide raise (fact)
- EUV layer expansion — more single-patterning EUV 0.33-NA and High-NA 0.55-NA layers in both logic and DRAM → double-digit EUV-spend CAGR 2025–2030 (fact, company model)
- High-NA EUV ramp (EXE platform) — two High-NA systems recognized in Q1 FY26 revenue (fact); next-gen ASP step-up
- DRAM/HBM EUV adoption as AI memory demand drives EUV into advanced DRAM nodes
- Installed Base Management annuity compounding on a growing ~6,500-system fleet — high-margin service, spares, and upgrade business (€2.488B in Q1 FY26, fact)
- Backlog conversion: €38.8B year-end-2025 backlog (fact) underpins multi-year revenue visibility
Bull & bear
The monopoly stopped being a 2030 story and became a 2026 one. ASML raised FY2026 revenue guidance twice this year to €43-45B - the bottom of the €44-60B range it set for 2030 - printed a record 37.1% operating margin, guided Q3 gross margin to 55-57% (inside the 56-60% 2030 band), and is adding 30% EUV and DUV capacity for 2027 with essentially all of 2027's orders already in hand. The share price fell 11.5% over the same window, so the forward multiple compressed from ~47x to ~29x on rising numbers.
- The 2030 model is being met four years early: FY2026 guided €43-45B against the Nov-2024 Investor Day's €44-60B/2030 scenario, and Q3's 55-57% gross-margin guide already reaches into the 56-60% 2030 band. The 10 Jun 2027 Capital Markets Day exists to raise the long-term frame, not defend it (all company-stated)
- Record profitability, and it is a filed number not an estimate: 37.1% operating margin in Q2 FY26 on 54.0% gross margin, with R&D+SG&A down to 16.9% of sales from 19.1% a year ago - the mix shift to EUV and the service annuity is showing up exactly where the thesis said it would
- Supply, not demand, is the constraint, and management is spending to relieve it: +30% to the ~65-unit low-NA EUV capacity for 2027 (investigating another +30% for 2028) and the same for the ~130-unit DUV immersion line, with 2027 orders 'pretty much' complete and 2028 EUV orders already arriving (fact, CEO)
- The annuity is now the beat engine, not the swing factor: Installed Base Management was €2,761.7M in Q2 (+31.8% YoY) and is guided to ~€2.9B in Q3; management named it as the reason both revenue and gross margin cleared guidance. That is the least cyclical euro in the model growing fastest
- High-NA crossed from pilot to production - Intel has qualified the High-NA process option on select Intel 18A product layers, the first high-volume logic insertion (fact, IR deck) - which is the validation the next ASP and margin leg required
- The stock got cheaper as the business got better: $1,629.00 on 31 Jul vs $1,841.18 on 25 Jun, with forward P/E at 29.37. Consensus FY2026 revenue (€42.96B) still sits below the company's own €43-45B guide
China's lithography indigenization stopped being a slide in the risk section: a Huawei-linked Shanghai firm is reported to have begun producing immersion DUV tools, and the stock lost ~8% in a session on it. Layer on ~20% China revenue (~€9B) still exposed to the MATCH Act, a €43-45B guide that now sets an extremely high bar for 2027, and H1 free cash flow that is still negative, and the risk is that 2026 is the cycle peak being extrapolated as a plateau.
- The China indigenization threat has a date and a unit count now, not just a narrative: Shanghai Yuliangsheng (tied to Huawei's SiCarrier) reportedly began making immersion DUV tools, ~5 units in 2026 and ~20 in 2027. Small against ASML's ~130-unit DUV immersion capacity - but this is the first time the long-tail erosion argument has had physical machines attached to it
- ~20% of 2026 revenue (~€9 billion) is still China, and the MATCH Act - which would restrict DUV tools AND servicing of the installed base into China - remains live and unpassed. Servicing is the exposure that matters most, because it hits the highest-margin, most-defended line in the model
- The guide sets a punishing 2027 comparison: FY2026 at €43-45B is +32-38%, with Q3 alone guided +46% to +60% YoY. Consensus already models €53.84B for 2027. Semi-cap has never sustained that rate for long, and a digestion year off this base would be a violent reset even with the monopoly fully intact
- You can no longer see the cycle turn coming: ASML discontinued the quarterly bookings disclosure from Q1 2026 (verified absent from all three Q2 exhibits), so the single best leading indicator of an order-book roll is gone. Investors now hold management's qualitative '2027 is nearly booked' in its place
- Cash conversion is not keeping up with reported profit: H1 FY26 free cash flow is -€1,291M despite €5,674.3M of net income, and cash + short-term investments fell from €13,322M to €7,582M. It is down-payment timing rather than deterioration - but it means the earnings quality is only verifiable annually
- Customer concentration is unchanged and now more levered: memory is guided ~+75% in 2026 against advanced logic ~+25%, so the 2026 upside is disproportionately an HBM/DRAM capex bet - the most cyclical customer cohort ASML has
What it is worth
Comps (forward P/E vs semi-cap peers) cross-checked with a reverse-DCF on the 2030 company model (€44–60B revenue, 56–60% GM)
$1,000–1,300
a broad DUV-to-China ban (~5% revenue hit) plus AI-capex digestion compresses the multiple toward the peer median on a cyclical down-leg
$1,750–1,850
low-teens revenue CAGR + mid-50s GM toward the moderate (€52B) 2030 case; multiple gradually normalizes as growth is delivered (~consensus ~$1,740)
$2,300–2,350
hitting the high end of the 2030 model (€60B, ~60% GM) sustains the premium multiple (BofA $2,345 target, fact)
At ~$1,841 / ~$698B cap, ~60x trailing and ~47x forward P/E (fact) — a monopoly premium to the ~37x semi-cap median that prices in flawless execution of the 2030 model and AI demand persisting; the multiple, not the business, is the risk.
SWOT
Strengths
- Sole-source monopoly on EUV and High-NA EUV lithography — no qualified competitor; advanced logic and DRAM below ~7nm cannot be made without ASML (fact)
- Recurring high-margin Installed Base Management annuity (~€2.5B/quarter) on a ~6,500-system fleet smooths the equipment cycle (fact)
- €38.8B backlog (year-end 2025) gives multi-year revenue visibility into 2027 (fact)
- Net-cash balance sheet (€8.4B cash/STI end-Q1 FY26) funds R&D, buybacks (€1.1B in Q1) and a rising dividend (fact)
- Structural margin expansion path to 56–60% gross margin by 2030 as EUV/High-NA and service mix grows (fact, company model)
Weaknesses
- Extreme customer concentration — a handful of fabs (TSMC, Samsung, Intel, SK Hynix, Micron) drive the order book; one customer's capex pause swings the cycle
- Cyclical and lumpy — revenue/FCF swing on customer capex timing and system-recognition timing (Q1 FY26 FCF was −€2.6B) (fact)
- Deep dependence on a concentrated, hard-to-replace supplier base (Zeiss optics, Cymer/Trumpf light source) that caps annual EUV build rate
- Heavy China exposure being structurally curtailed — China falling from ~33% (FY25) to ~20% (FY26 guide) and at risk of further cuts (fact)
- Long, capital-intensive R&D cycles (High-NA took ~20 years) — innovation lead time is a decade, not quarters
Opportunities
- AI-accelerator super-cycle driving leading-edge logic capacity (TSMC N2/A16, Intel 18A/14A, Samsung SF2) and pulling EUV demand forward (fact)
- High-NA EUV (EXE) adoption ramping — higher ASP, expands the single-patterning frontier and the moat
- EUV penetration into advanced DRAM/HBM as AI memory demand surges → new layer-count growth
- Installed-base upgrades/services compounding as the fleet grows — annuity with attractive incremental margins
- Reaching the €44–60B / 56–60% GM 2030 model would roughly double revenue from FY25 (fact)
Threats
- US export controls escalating — the MATCH Act would ban DUV exports to China and pressure Dutch alignment; a broad DUV ban est. ~5% revenue hit (Quilter Cheviot est, fact)
- China indigenization (SMEE and domestic DUV/EUV efforts) eroding the long-tail DUV market and, eventually, threatening the EUV moat
- Semiconductor cyclicality / AI-capex digestion — a leading-edge capex pause compresses orders sharply
- Single-supplier fragility (Zeiss, Cymer/Trumpf) — a disruption at one supplier caps system output
- Geopolitical use of ASML as a policy chokepoint (export-license withholding, allied alignment demands) introduces regulatory tail risk to demand
Moats, dependencies & bottlenecks
Moats
very high (5–10+ yrs) Sole producer of EUV; High-NA EXE platform extends the lead. Built on a ~20-yr R&D program + the Zeiss optics partnership no rival can replicate (fact).
Owns Cymer (light source) and Berliner Glas; 25% stake + exclusive optics from Zeiss SMT. Rivals cannot source the core EUV subsystems.
~6,500-system fleet generates recurring high-margin service/upgrades (€2.488B Q1 FY26) — sticky, switching-cost-protected revenue (fact).
Fab processes are co-developed and qualified around ASML tools over years; re-qualifying around a competitor is effectively impossible at the leading edge.
~€4B+/yr R&D and an ecosystem of thousands of suppliers create an entry barrier measured in decades, not quarters.
Dependencies
Concentrated, hard-to-replace subsystem suppliers; their throughput ceiling caps how many EUV systems ASML can build per year.
A handful of customers drive the order book; their capex timing is ASML's revenue cycle. Concentration cuts both ways.
MATCH Act / BIS rules govern what can ship to China; China already cut to ~20% of FY26 sales and a broad DUV ban est. ~5% revenue hit (fact).
Demand follows leading-edge capacity additions; an AI-capex digestion phase would compress orders sharply.
Margin and growth model assume High-NA ramps on schedule and EUV layer-count keeps rising; slippage delays the 2030 model.
Advantages
- The only company on earth that can make EUV/High-NA lithography — a literal single point of control for advanced chips
- Pricing power: high-ASP systems with rising mix toward 56–60% gross margin by 2030 (fact)
- Multi-year backlog (€38.8B) buffers the cycle and signals demand depth (fact)
- Vertically secured supply (owns Cymer/Berliner Glas, 25% of Zeiss SMT) protects the core IP
- Net-cash balance sheet self-funds a decade-long R&D moat while returning capital (buyback + dividend)
- Recurring service annuity grows with every system shipped, decoupling part of revenue from the equipment cycle
Weaknesses
- Severe customer concentration — a few fabs swing the entire order book
- Lumpy, cyclical revenue and FCF (Q1 FY26 FCF −€2.6B) make quarterly results noisy (fact)
- Structural China revenue erosion under export controls (~33%→~20% FY25→FY26) with further downside risk (fact)
- Supply-side build ceiling limits how fast ASML can convert demand to revenue
- Premium valuation (~60x trailing P/E) amplifies downside on any cyclical or regulatory shock (fact)
- Long innovation lead times mean a roadmap stumble is expensive and slow to fix
Bottlenecks
- EUV system build rate capped by Zeiss optics + Cymer/Trumpf light-source throughput (target ~90 EUV/High-NA units/yr by 2025–26) — a supply, not demand, ceiling
- High-NA (EXE) ramp/yield maturity — early units (2 recognized in Q1 FY26) must scale before the ASP step-up flows through
- Export licensing — what can ship to China gates a meaningful revenue slice and is set by Washington/The Hague, not ASML
- Skilled-labor and clean-room manufacturing capacity at Veldhoven/Berlin to expand output
- Customer fab-readiness — systems recognize as revenue only as customers' cleanrooms are ready, creating timing lumpiness (Q1 FY26 FCF −€2.6B)
Top signals & trends
Top signals
Q4 FY25 bookings were €13.2B incl. €7.4B EUV (fact); booking strength is the leading indicator of the cycle — watch each print.
Cleared House Foreign Affairs Committee Apr 2026 but not yet law; a broad DUV ban est. ~5% revenue hit (fact). Binary regulatory catalyst.
Two High-NA units in Q1 FY26 revenue (fact); pace of adoption validates the next ASP/margin leg.
Customers accelerating buildouts drove the Q1 FY26 guide raise (fact); a capex pause would reverse it.
Falling ~33%→~20% (FY25→FY26, fact); further compression or indigenization signals a structural demand headwind.
53.0% in Q1 FY26 (fact); sustained progression validates the EUV/service mix-shift thesis.
Trends
Drives leading-edge logic + HBM/DRAM capacity, pulling EUV demand forward; cited as the reason for the FY26 guidance raise (fact).
Double-digit EUV-spend CAGR 2025–2030 for logic and DRAM in ASML's model (fact) — structural unit + ASP growth.
MATCH Act + BIS rules curtail China sales; China cut to ~20% of FY26 (fact). Persistent, escalating headwind.
Erodes the DUV long tail near-term; a long-dated threat to the EUV moat if Chinese EUV matures.
Expands EUV's addressable layer count beyond logic into advanced memory — incremental demand pool.
Intel Ohio/Arizona, TSMC Arizona, Samsung Taylor add EUV-equipped capacity in allied geographies, diversifying ASML's demand base.
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.
Exclusive supplier of EUV/High-NA optics; ASML owns 25% and co-funds R&D. The single most critical supplier.
EUV light-source (LPP) division; ASML acquired it to secure the source. Build-rate bottleneck.
High-power CO2 lasers that drive the EUV light source — essential, single-source-class.
Optical/ceramic modules; acquired to vertically secure precision components.
Dutch precision-mechatronics and module suppliers in ASML's deep European supply chain.
Largest customer; N2/A16 leading-edge ramp is the single biggest EUV demand driver.
Logic (SF2 foundry) + DRAM/HBM EUV buyer; co-invested in ASML's EUV R&D.
Lead High-NA EUV customer (18A/14A at Ohio/Arizona); co-invested in EUV R&D.
Memory/HBM leader adopting EUV in advanced DRAM as AI memory demand surges. (Korea-listed.)
US memory maker bringing EUV into advanced DRAM nodes.
Competes in DUV (ArF/KrF) lithography but has no EUV product; share has eroded badly at the leading edge.
DUV lithography + pushing nanoimprint (NIL) as an alternative patterning path; unproven at scale vs EUV.
China's domestic litho champion; DUV-focused, EUV years behind — a long-dated indigenization threat, not a current rival. (Named for analysis; not a recommendation.)
Not a litho rival — adjacent WFE giant (deposition/etch). Competes for fab capex wallet, not for EUV.
Etch/deposition leader; complement and capex-wallet competitor, not a lithography substitute.
Broad WFE (coater/developer track tightly paired with ASML scanners, plus etch/deposition); partner-competitor for fab budget.