ASML — Research Report (2026-07-15)

TL;DR

• What it is: The Dutch company that makes the machines that print advanced computer chips — and the only firm on Earth that makes the EUV machines every cutting-edge chip needs.
• My take: The purest bottleneck monopoly in the global economy (business 10/10), but after a ~140% run in 18 months the price already pays for most of the 2030 bull case — thin margin of safety.
• FY2025 revenue €32.7B, +15.6% (growth reaccelerating, not just holding).
• Net income €9.6B, ~29% net margin (hardware firm, software margins).
• FCF ~€11.1B, ~85% of EBITDA (profit converts to real cash).
• Backlog €38.8B ≈ 100% of 2026 guided revenue (next year is pre-sold).
• Strongest bull point: 100% EUV share — TSMC/Samsung/Intel must buy or fall behind; 2030 model €44–60B revenue (JPMorgan PT $2,200).
• Strongest bear point: ~55x trailing earnings while China (33% of 2025 sales) sits under tightening US/Dutch export bans — a disciplined fair-value lands ~35–40% below today's price.
• One thing to watch next: Q2 2026 results Wed July 15, 2026 — the first read with no bookings number to anchor on (ASML retired that disclosure in 2026).

Estimate labels used throughout: (reported) = primary source (ASML PR / SEC filing) · (secondary) = credible data provider or press · (est.) = my own computation, method shown.

One-line verdict: the strongest bottleneck monopoly in the global economy — sole maker of the EUV machines every advanced chip needs — but after a ~140% ADR run in 18 months the price already pays for most of the 2030 bull case. Business 10/10; entry price offers thin margin of safety.

1. Business overview

So what: ASML makes the single most critical machine in chipmaking — and for the EUV type every advanced chip needs, it has a 100% monopoly.

Every chip is made by "printing" microscopic circuit patterns onto silicon wafers — lithography (an insanely precise projector shining a pattern onto light-sensitive material). ASML makes the projectors. Based in Veldhoven, NL; trades as a NASDAQ ADR + on Euronext Amsterdam.

• What it sells: EUV scanners (0.33-NA "Low-NA" NXE, ~€180–220M each; 0.55-NA "High-NA" EXE, ~€380–400M each — TrendForce, Tom's Hardware); DUV scanners; metrology/inspection; and Installed Base Management (service + upgrades) — a recurring ~€8.3B/yr stream (est.) that ran €2.49B in Q1 2026 alone (Q1 2026 PR).
• 2025 system mix (reported — 2025 annual report financials): EUV €11.6B (48 systems, +39% y/y), DUV €12.0B (279 systems, −6%), metrology €825M (+28%).
• To whom: the world's leading-edge fabs. TSMC is the largest at roughly a quarter of sales; TSMC + Samsung ≈ 38%, Intel third (secondary — TechMarketBriefs). Extreme concentration, but mutual lock-in: a monopolist selling to an oligopsony (a market with only a few buyers) that has no alternative supplier.
• Geography: China was 33% of 2025 sales (36% in Q4) — all DUV/older tools, zero EUV ever shipped there — and ASML expects China to fall materially in 2026 as pulled-forward backlog clears (reported via Caixin). (Note: the "~20% China in 2026" figure some press cites is forward guidance, not the 2025 actual.)
• Disclosure change that matters: from 2026 ASML no longer reports quarterly net bookings — CFO Dassen calls them lumpy (Moomoo). The market's best early-warning gauge is gone; expect bigger surprises both ways.

2. Financials — FY2021–2025 (+ Q1 2026)

So what: Revenue nearly doubled in four years, margins are rising, and profit converts almost fully into cash — with a net-cash balance sheet.

Sources: FY2025 Q4-2025 PR + 2025 annual report 6-K (OCF €13,826.5M reported); FY2024 Q4-2024 PR; FY2021 PR; OCF/capex 2021–24 (secondary — mlq.ai); FCF 2025 €11.1B (secondary — Alpha Spread).

• Revenue CAGR 2021→25: 15.1% (est.) — why it matters: faster than the chip-equipment market, so ASML is gaining share of every fab's capex (EUV intensity grows each node).
• The 2024 flat year (+3%) — why it matters: proof this is still cyclical. Chipmakers pause spending; 2024 was digestion before the 2025 AI reacceleration.
• EBITDA not headlined by ASML; est. FY2025 ≈ €13B, consistent with ~43x TTM EV/EBITDA third parties compute (KoalaGains). Cash conversion (FCF/EBITDA) ≈ 85% (est.) — why it matters: profit here isn't an opinion, it turns into cash.
• Net debt: none — net CASH ≈ €10B (est.): €13.3B cash + short-term investments vs ~€2.7–3.7B gross Eurobond debt (end-2024 primary: €3,677M LT + €1,010M current, SEC interim report). Why it matters: funds ~€5B/yr R&D through any downturn — R&D scale is the entry barrier.
• Q1 2026 (reported): sales €8.8B, GM 53.0%, net income €2.8B, EPS €7.15; FY2026 guidance raised to €36–40B at 51–53% GM (Q1 2026 PR).
• One-off flagged: €1.3B Mistral AI stake (Sept 2025) sits in investing outflows — a strategic investment, not capex (Bloomberg).

3. Unit economics — the number the business runs on

So what: ASML sells ~300 machines a year, and the one number that runs the business is EUV average selling price — up 61% in four years with zero competition.

Blended EUV ASP: ≈ €242M per system and rising (est.: €11.6B EUV revenue ÷ 48 systems, 2025). In 2021 it was ≈ €150M (€6.3B ÷ 42 systems, reported — FY2021 PR). +61% ASP growth in four years with zero competitive pressure — that is monopoly pricing power in one number. High-NA (~€380–400M/unit) mechanically extends the staircase: 4 EXE systems recognized in 2025 vs 2 in 2024, expected to reach ~¼ of EUV revenue by 2028 (TrendForce).

Contrast: DUV ASP ≈ €43M (est.: €12.0B ÷ 279) — the competitive-ish, China-heavy end. And every machine sold feeds the ~€8.3B/yr installed-base service annuity across a 20+ year field life — why it matters: each sale becomes a two-decade annuity, and that recurring stream is exactly what China curbs threaten (curbs can block service/parts to installed Chinese tools).

4. Metrics that matter — and why, here

So what: Every quality metric is strong or improving — the entire debate is price, not business.

5. Ownership, management, insider signals

So what: No controlling shareholder, takeover-proof, management pay is equity-heavy and long-horizon, and the CEO bought stock at half today's price.

• Ownership: ~100% free float, no controlling shareholder; institutions dominate — BlackRock, Vanguard, State Street, Capital Group among the largest (secondary — TIKR). Takeover-proof in practice: ~€590B size, EU-strategic-asset status, and a Dutch Stichting (foundation) preference-share anti-takeover option in the 20-F.
• Management: CEO Christophe Fouquet (since Apr 2024, ex-EVP for EUV). Incentives (reported — 2025 Remuneration Report): base €1.125M; cash bonus capped at 150% of base; long-term equity up to 350% of base vesting over 3 years + 2-year hold — overwhelmingly equity-linked. Aligned.
• Insider signal: Fouquet bought ~2,000 ADRs at ~$748–751 in Jan 2025 (secondary — TIKR) — small, but a buy at half today's price; no notable selling surfaced (Dutch AFM registry not systematically checked — flagged).
• Promise-keeping check: guided 2025 to ~€30–35B / ~52% GM → delivered €32.7B / 52.8% ✓. Warned honestly in July 2025 that 2026 growth wasn't guaranteed (DCD) → then guided up twice. Conservative-then-beat: credible.
• Capital returns: 2025 dividend €7.50/share, +17% y/y; new buyback up to €12B through 2028; €1.1B repurchased in Q1 2026 alone (reported — Q4-2025 PR). Combined shareholder yield ≈ 1.2%/yr (est.).

6. M&A track record

So what: Every acquisition has locked up a supply-chain chokepoint; the one off-pattern deal (Mistral AI) is small but worth watching.

Disciplined, supply-chain-locking, well-integrated (reported): Cymer 2013 (light sources — the EUV source problem bought and solved), HMI 2016 (e-beam metrology), 24.9% of Carl Zeiss SMT 2016 (the optics monopoly inside the monopoly), Berliner Glas 2020. The 2025 €1.3B for ~11% of Mistral AI (largest shareholder, board seat — Bloomberg) is the first off-pattern deal — a strategic AI-sovereignty bet, not a tuck-in; small against €10B+ FCF, but watch for mission creep.

7. Market & value-chain position — bottleneck or commodity?

So what: ASML is the purest bottleneck in the global economy — 100% EUV share, no second source this decade — and the only parties with power over it are non-commercial (Zeiss, and governments).

Chip designers (Nvidia, Apple) ── need advanced chips
        │
Foundries (TSMC, Samsung, Intel) ── need EUV to make them
        │
   ►►► ASML ◄◄◄  (100% EUV monopoly = THE chokepoint)
        │
ASML's suppliers: Zeiss (optics, quasi-exclusive), Cymer (ASML-owned), Trumpf (lasers)

100% share in EUV; no credible second source this decade (Nikon/Canon exited; China's SMEE remains years behind even in immersion DUV — Asia Times). Every AI-chip roadmap runs through ASML's installed base. Company's own 2030 model: revenue €44–60B at 56–60% gross margin (reported — Investor Day 2024).

Who holds pricing power over ASML? Three parties, none commercial:

1. Zeiss SMT — sole EUV-optics supplier, a bilateral monopoly ASML manages via its 24.9% stake; a partner, not a squeezer.
2. Governments — the real boss. Dutch/US export controls already zeroed China EUV and cap DUV; 33% of 2025 revenue sits where policy, not ASML, sets the ceiling.
3. Customer concentration — TSMC (~¼ of sales) can delay orders but cannot substitute. Monopoly vs oligopsony ends in negotiated truce — and ASML keeps raising ASP per generation; that's the scoreboard.

8. Valuation view

So what: ASML is a wonderful business at a rich price — ~55x trailing earnings vs a semicap sector at ~38–48x and its own ~35x history; a disciplined buyer's fair value sits ~35–40% below the tape.

Current price (secondary — StockAnalysis): ADR ~$1,804; market cap ~$686–696B (≈ €590B, est.), early July 2026 — 2.4× Fouquet's Jan-2025 buy ($748).

Peer multiple range (secondary, mid-2026 — KoalaGains): trailing P/E — AMAT ~38x, KLA ~40x, LRCX ~48x, ASML ~55x+; ASML EV/EBITDA ~43x TTM. Sector context: semi-equipment peers trade ~38–48x P/E vs the broad semis sector ~20–35x (source: sector screens / Macrotrends) — the whole complex is inflated by AI-capex hopes; ASML carries the top premium on monopoly grounds. Which multiple & why: ASML is highly profitable and net-cash, so P/E is the right primary lens (EV/EBITDA just confirms it). Peers aren't true comps — nobody else sells leading-edge litho — they're the sentiment gauge.

What a reasonable buyer would pay (a strategic acquisition is impossible — size + EU strategic asset + Stichting defense — so the "buyer" is a long-term investor underwriting the 2030 model; all est., assumptions shown):

• Mid: €52B rev at 58% GM → ~€18B net income → EPS ~€49 on ~375M shares. At 27x = ~€1,320 in 2030; discounted ~9%/yr ≈ €900–1,000 today — 35–40% below the current ~€1,540 equivalent.
• High (€60B / 60% GM): EPS ~€61; even at 30x = ~€1,830 in 2030 → from today only a mid-single-digit IRR + ~1% yield.
• Conclusion: today's price already pays for the 2030 bull case. A disciplined buyer pays €900–1,100; the marginal AI-momentum buyer pays $1,804. The gap is sentiment, not arithmetic (own estimate; the market has disagreed for a year).

9. PE/quality lens

So what: Near-perfect business quality; the only real fears are macro/political, never internal.

• Moat: absolute — 100% EUV share, ~40 years of R&D nobody can shortcut, the Zeiss optics lock, a 20+ year service annuity, embedded in every leading fab's process.
• A buyer would love: the ASP staircase (each generation +50–100%), ~85% cash conversion, net-cash balance sheet, a pre-sold year of revenue, aligned management.
• A buyer would fear: ~43x EBITDA for a still-cyclical business (2024 grew 3%); geopolitics as permanent tail risk both ways (China loss / Taiwan conflict at the biggest customer); the bookings-disclosure blackout; Hyper-NA bets paying off past 2035.

10. Bull / base / bear — named institutions (vs ~€1,540/sh · $1,804 ADR today)

So what: Sell-side is near-unanimously bullish (9 Buy / 0 Hold / 0 Sell); the credible bear is about price and China policy, not the franchise.

🐂 Bull — JPMorgan (Overweight, PT raised to $2,200, June 2026; MarketBeat): the AI-capex supercycle runs through the decade, DRAM adopts EUV at scale, High-NA ramps at Intel 14A. 2030 lands €60B/60% GM → EPS ~€61, multiple holding >30x → ~€1,900–2,100 by 2030. Morgan Stanley (Overweight, PT €1,830, July 2026 — GuruFocus) and UBS (Buy, May 2026) concur; avg. 12-mo target ~$2,120. Note: even the bull mostly defends today's price rather than beating it (est.).

⚖️ Base — analyst consensus (~$1,775 avg. target; MarketScreener): 2026 delivers the guided €36–40B; growth digests toward the €52B 2030 mid-case; the multiple drifts from ~53x toward ~35x → price roughly flat over 4 years → low-single-digit IRR + ~1.2% yield (est.). The wide target range ($1,172–$2,500) signals real disagreement on the multiple.

🐻 Bear — Simply Wall St / Motley Fool (valuation + China; Simply Wall St, Motley Fool): shares flagged ~93% above one fair-value estimate. 2027 becomes a digestion year (litho is late-cycle; AI datacenter pauses hit foundry capex with a lag) just as China falls from 33% toward the low-20s%; the proposed US MATCH Act could restrict even DUV tools and service to China (quietly high-margin); TSMC is deferring High-NA. EPS stalls near €29, multiple reverts to low-30s → ~€870–950/share, about −40%, without the franchise being impaired. Stock fell ~8% in April 2026 on the China-bill headline.

Catalysts: (1) Q2 2026 results July 15 — first read with no bookings anchor; (2) TSMC's High-NA insertion decision + Intel 14A ramp in 2027; (3) DRAM/HBM EUV layer expansion (Samsung, SK hynix, Micron); (4) US/Dutch China policy moves; (5) buyback pace at 50x+ earnings.

What would prove the thesis wrong: Quality thesis breaks if: EUV revenue falls 2+ consecutive years; TSMC extends Low-NA multi-patterning instead of inserting High-NA; GM guidance drops <51%; SMEE ships production-worthy immersion DUV at scale; or Mistral-style allocation becomes a pattern. *Valuation caution (my actual conclusion) breaks if:* 2026 prints near €40B **and** 2027 is guided up again — the cycle not pausing — while the multiple holds >45x. Then the market was right and the €900–1,000 entry never arrives.

📚 What this company teaches

1. The bottleneck captures the margin, not the end-product. Nvidia gets the headlines, but ASML sits at the one chokepoint every chipmaker must pass through — and a monopoly at the bottleneck earns more durable economics than the glamorous companies downstream. (The value-chain lesson: drill down to the monopoly bottleneck, don't just buy the leader.)

2. A great business and a great investment are different questions. ASML's quality is nearly flawless — but at ~55x earnings the price already pays for years of it. Separate "is this a good company?" (yes) from "is this a good price?" (debatable) — they are not the same test.

3. Recurring revenue is worth more than one-time sales — and is the softest political target. ASML's installed-base service turns each machine into a 20-year annuity the market rewards with a premium multiple. But that same sticky, high-margin stream is exactly what export controls can sever — the most valuable revenue is often the most politically exposed.

Self-test question: ASML has 100% EUV share and ~15% growth yet trades near 55x earnings while Applied Materials trades near 38x. Give one reason that premium is justified and one reason it could be dangerous — and say which risk would hurt a 55x stock more: a cyclical AI-capex pause, or a slow loss of EUV monopoly?

📖 Glossary

• Lithography — "printing" circuit patterns onto silicon wafers with light; the core step ASML's machines perform.
• EUV (Extreme Ultraviolet) — lithography using 13.5nm light to print the tiniest chip features; required for all leading-edge chips. ASML is the only maker.
• DUV (Deep Ultraviolet) — older, longer-wavelength lithography for larger features; competitive market (Nikon, Canon); the China-exposed segment.
• High-NA EUV — the newest, higher-resolution EUV generation (larger "numerical aperture" lens); machines cost ~€380–400M; Intel adopting first, TSMC deferring.
• NA (Numerical Aperture) — a lens's light-gathering measure; higher NA prints smaller features. "0.33-NA" = Low-NA, "0.55-NA" = High-NA.
• Foundry — a company that manufactures chips designed by others (e.g. TSMC). ASML's core customers.
• Fab — a chip fabrication plant; ASML's machines live inside fabs.
• Installed base / Installed Base Management — the fleet of machines already sold, plus recurring service/upgrade/parts revenue earned on them for 20+ years.
• Backlog — the euro value of orders received but not yet delivered; future revenue already contracted.
• Net bookings — new orders taken in a period (ASML stopped reporting this quarterly in 2026).
• ASP (Average Selling Price) — average price per machine; rises as mix shifts to EUV/High-NA.
• P/E (price-to-earnings) — share price ÷ earnings per share; dollars paid per $1 of annual profit. High = expensive or high growth expected.
• EV/EBITDA — enterprise value (market cap + debt − cash) ÷ earnings before interest, tax, depreciation, amortization; a debt-neutral whole-business multiple.
• EBITDA — a rough proxy for operating cash generation before financing and accounting non-cash items.
• FCF (Free Cash Flow) — operating cash flow minus capex; the cash truly left for dividends, buybacks, debt. Hardest number to fake.
• OCF (Operating Cash Flow) — cash generated by the core business before investment spending.
• Cash conversion (FCF/EBITDA) — how much operating profit becomes real free cash; ~85% here is very high.
• Capex (Capital Expenditure) — money spent on physical assets. Low capex/sales signals an asset-light model.
• Net cash / net debt — cash minus debt; "net cash" means more cash than debt (a fortress balance sheet).
• Gross / net margin — profit as a % of sales, before costs (gross) and after everything (net). Rising margins signal pricing power.
• ROE / ROIC vs WACC — return on equity / invested capital vs the cost of that capital; ROIC ≫ WACC means each euro reinvested creates value — the moat in one number.
• CAGR (Compound Annual Growth Rate) — the smoothed annual growth rate over multiple years.
• Oligopsony — a market with only a few buyers (ASML's customers); the mirror image of a monopoly (one seller).
• Stichting — a Dutch foundation; ASML's can issue preference shares as an anti-takeover defense.
• MATCH Act — a proposed 2026 US bill that could extend export restrictions to DUV tools and services sent to China.
• Cyclical business — one whose sales rise and fall with a broader spending cycle (here, chipmakers' capex); the opposite of steady/defensive.

Prepared per equity-research workflow §3/§3b/§4/§4b · primary sources: ASML press releases + SEC 6-K/20-F filings · all estimates labeled · not financial advice; feeds the [[investment-research]] loop before any position.

Two notes on the deliverable:

• The vault already holds a 2026-07-12 report (investing/ASML-2026-07.md) that is more current than my initial web searches — it correctly has China at 33% of 2025 sales (my search surfaced the 2026 guidance of ~20%), plus Q1 2026 actuals, the Mistral stake, and the CEO's insider buy. I built this report on that verified content rather than replacing it with shallower figures, then added the required TL;DR / "So what:" / teaching / glossary layers. Saving the updated file was blocked (write permission declined this session) — the report exists only in this output.
• DONE (report) / BLOCKED (vault save — grant write access to investing/ASML-2026-07.md and I'll persist it + update the logs).