ASML Holding ASML

Reference table

1. What the business is

ASML Holding N.V. is a Dutch company headquartered at De Run 6501, Veldhoven, listed on Nasdaq, with 43,938 full-time employees and chief executive Christophe D. Fouquet. The profile describes it as providing "lithography solutions for the development, production, marketing, sales, upgrading, and servicing of advanced semiconductor equipment systems," across extreme ultraviolet (EUV) systems, deep ultraviolet (DUV) immersion and dry systems, and metrology and inspection.

There is only one reportable segment. The 20-F, note 3, is explicit: "ASML has one reportable segment, since we are a holistic lithography solution provider... The Chief Operating Decision Maker regularly sets and monitors goals and boundaries on a consolidated basis." The vendor's seg_prod block therefore is not a segment table — it is a revenue disaggregation by product line, and there is no segment operating income anywhere in this file. Any margin attribution by product is inference, and is labelled as such below.

Revenue by product line (EUR millions, FY2025 vs FY2024). The FY2025 row in the vendor file is incomplete — the installed-base line is missing entirely — and is reconstructed here from the reported total. The reconstruction is verified against the 20-F in Section 8.

Three readings of that table. First, EUV is now 35.5% of revenue (NXE plus EXE, EUR 11,602.7M) against 29.4% in FY2024 — the mix is shifting toward the monopoly product, which is also the highest-priced one. Second, the DUV mainstream is contracting: KrF fell 49.7% and ArF dry 44.9%, and the 20-F confirms total DUV unit shipments fell from 374 in 2024 to 279 in 2025, a 25% decline. Third, installed base management at EUR 8,193.0M is the second-largest line in the company and grew 26.2%, faster than systems at 12.4%. That is the cyclical damper, and it is the reason a machinery business with falling unit volumes still grew 15.6%.

Units and derived average selling prices. The 20-F discloses unit counts the vendor file does not: "We recognized four EXE and 44 NXE systems in sales in 2025 compared to two EXE and 42 NXE systems in 2024." Dividing the disaggregated revenue by those counts:

These are derived, not disclosed, and carry two caveats. The 20-F notes that ASML "carried out a significant number of TWINSCAN NXE:3800E field upgrades, which resulted in a substantial portion of EUV system revenue being shifted to installed base revenue" — so the NXE line understates total EUV economics and the installed-base line overstates pure service. And first-of-kind High-NA systems can carry deferred or partial revenue recognition, so the EXE figure is a lower bound on list price. The direction is nonetheless unambiguous: price per machine rose about 25% across every family in a single year, driven on the EUV side by the NXE:3800E stepping from 160 to 230 wafers per hour, and on the DUV side by mix shifting toward immersion as the cheap dry and KrF units fell away.

Revenue by geography (EUR millions). Note that seg_geo does not sum to reported revenue in either year — FY2025 leaves EUR 524.3M (1.6%) unallocated and FY2024 leaves EUR 1,322.1M (4.7%). Shares are therefore given both ways.

China fell from 36.1% to 29.1% of revenue in one year while Taiwan almost doubled its share — the mix rotated from mainstream DUV toward leading-edge EUV, exactly as the product table shows. The 20-F adds the colour: "Our DUV business in China turned out to be stronger than anticipated, offsetting marginally lower than anticipated non-China DUV business as mainstream business outside of China remained weak."

Customer concentration is high and rising, and only the filing says so. From the 20-F: "In 2025, four customers each individually exceeded 10% of total net sales, totaling EUR 20.0 billion, or 61.2%, of total net sales." The comparatives are EUR 15.2 billion / 53.8% in 2024 and EUR 14.9 billion / 53.9% in 2023. Four buyers are now 61% of the business. Receivable concentration moved the other way — the three largest customers were 35.4% of receivables at 2025-12-31 against 54.1% a year earlier — which is a working-capital improvement, not a demand one.

2. The single most important thing to understand about these numbers: ASML reports in euros and trades in dollars

Every financial statement, estimate and segment figure in this file is denominated in EUR. The quote, the market capitalisation, the price targets and the earnings calendar are in USD. Mixing them produces errors of 16%, and the vendor file itself mixes them.

The reportedCurrency field on inc_a, inc_q, bal_a, cf_a and the FY2025 seg_prod row all read EUR. The profile.currency field reads USD. Both are correct for what they describe.

Deriving the rate. The file supplies the same quantity twice in two currencies: quote.marketCap is $659,793,204,613 and km_ttm.marketCap is EUR 567,566,533,547. The ratio is 1.16249 USD per EUR. Two independent checks confirm it is the rate the vendor actually used:

• ratios_ttm.priceToBookRatioTTM is 26.00365. Book value per share is EUR 56.63. 26.00365 × 56.63 = EUR 1,472.60, and $1,711.89 ÷ 1.16249 = EUR 1,472.60. Exact.
• ratios_ttm.priceToEarningsRatioTTM is 53.43. Trailing net income per share is EUR 27.60. 53.43 × 27.60 = EUR 1,474.7, within 0.15% of the same figure.

So the vendor's entire ratio block is computed on a euro-basis price of EUR 1,472.60, and it is internally correct. The trap is for anyone who takes the dollar price off the screen and divides it by the euro earnings on the same screen:

Every multiple in this dive is stated on the euro basis: euro price of EUR 1,472.60 against euro earnings. Where a dollar figure is used — fair values, price targets, the quoted price — it is converted at 1.16249 and said so.

The vendor's own earn_cal block is in USD and does not match the income statement. The 2026-07-15 entry reports epsActual 8.68 on revenueActual 10,665.68M; the Q2'26 income statement reports EUR 7.58 diluted EPS on EUR 9,326M of revenue. The ratios are 1.1451 and 1.1436 — a period-average dollar conversion, not the 1.16249 spot rate. earn_cal and est are in different currencies and must never be compared. The forward estimates (est) are in euros: FY2025 consensus EPS of EUR 24.77 against a reported EUR 24.71 confirms it.

One consequence is favourable and worth stating. Because consensus EPS tracks reported diluted EPS so closely — EUR 24.77 vs EUR 24.71 (FY2025), EUR 19.29 vs EUR 19.24 (FY2024), EUR 19.11 vs EUR 19.54 (FY2023) — the estimate basis here is effectively GAAP. There is no non-GAAP wedge to argue about, which is unusual in this batch and means the forward multiples below are not flattered by an adjusted-earnings convention.

A second structural point belongs here: the disclosure cadence is thinner than for the US-listed peers. ASML is a foreign private issuer. It files a 20-F annually and 6-K reports for interim events; it does not file 10-Qs. This folder contains four documents, not six: the 20-F of 2026-02-25 and 6-Ks of 2026-04-15, 2026-04-23 and 2026-07-15. All three 6-Ks are cover pages only — the exhibits (99.1 press release, 99.2 presentation, 99.3 summary US GAAP statements, 99.4 statutory interim report) were not extracted. Consequently there is no interim balance sheet, no quarterly cash-flow statement, and no quarterly net system bookings figure available to this dive. Compared with LRCX, KLAC or AMAT, the quarterly granularity is materially poorer, and that is a real limitation on what can be said about the most recent six months.

3. Growth, margin, and the order book the vendor file does not contain

The long record.

Five-year compound revenue growth is 18.5%, and the FY2022-FY2024 stretch — +13.8%, +30.2%, +2.6% — is the honest reminder that this is still a cyclical. The 20-F attributes the FY2025 margin step to "a favorable NXE product mix and higher net service and field option sales and margins," partly offset by "the dilutive impact of EXE systems recognized in sales." High-NA is currently margin-dilutive. That is a fact from the filing and it matters for the FY2027-28 margin path, because EXE revenue grew 148.8% in FY2025 and is expected to keep growing.

The recent quarters (EUR).

Trailing twelve months to 2026-06-30: revenue EUR 35.328B, gross margin 52.7%, operating margin 35.4%, net income EUR 10.638B, diluted EPS EUR 27.55. H1'26 revenue of EUR 18.093B is +17.2% on H1'25.

Now the part that only the 20-F supplies, and which is the most valuable thing in this dive.

Remaining performance obligations — the US GAAP order book. From the 20-F: "As of December 31, 2025, the remaining performance obligations amount to EUR 46.5 billion (December 31, 2024: EUR 43.3 billion)... We estimate that 65% (December 31, 2024: 59%) of these anticipated revenues will be recognized during the next 12 months."

That is +7.4% year on year in absolute size, and a six-point increase in near-term convertibility. Multiplying: EUR 30.2 billion of the FY2026 revenue was already under contract at the year end, against a guidance midpoint of EUR 44 billion — roughly 69% coverage before the year began. In FY2025 the equivalent was 0.59 × EUR 43.3B = EUR 25.5B against actual revenue of EUR 32.667B, or 78% coverage. So coverage of the guide is lower this year in percentage terms, because the guide grew faster than the book.

Backlog. The 20-F discloses a backlog figure exactly once, in the chief executive's letter: "Our backlog currently stands at a healthy level of EUR 38.8 billion." No definition, no reconciliation to the EUR 46.5B of remaining performance obligations, and no comparative. We use the EUR 46.5B RPO as the primary order-book measure because it is an audited note with a stated basis and a prior-year comparative; the EUR 38.8B is reported for completeness and is management's own framing, therefore half-weighted.

Net system bookings — MISSING. ASML's most-watched single number does not appear anywhere in the extracted 20-F or in any of the three 6-K cover pages, and it is not in the vendor payload. This dive cannot state Q1'26 or Q2'26 bookings. That is the single largest hole in the analysis and it is stated rather than estimated around.

The guidance ladder — three raises in five months. This comes entirely from the filings; the vendor file contains no guidance at all.

The midpoint rose EUR 7.5 billion, or 20.5%, in twenty weeks, and the margin guide rose three points. This is company guidance and is therefore half-weighted under house rules — but the raises are corroborated by the Q2'26 actuals (revenue EUR 9.326B and net income EUR 2.918B match the 6-K headline of "EUR 9.3 billion total net sales and EUR 2.9 billion net income" exactly) and by the RPO trend.

And here is the arithmetic that turns a good story into a Watch. H1'26 was EUR 18.093B. To reach the guide:

Layering the vendor's own Q3'26 consensus (USD 12.960B ≈ EUR 11.148B at 1.16249) on top gives an implied Q4'26 of EUR 13.758B to EUR 15.758B — against a Q2'26 actual of EUR 9.326B. That is a 48% to 69% sequential increase in two quarters. It is not impossible for a company whose revenue is recognised on machine delivery and whose order book is contracted; it is, however, the most demanding near-term ramp in this batch, and it is the number on which the FY2026 multiple depends.

4. Balance sheet and the enterprise-value rebuild

As at 2025-12-31 (the only balance sheet in this file; there is no interim statement — see Section 2):

• Cash and equivalents EUR 12.911B · short-term investments EUR 405.7M · total EUR 13.317B
• Total debt: the vendor reports EUR 2.708B, all long-term, with short-term debt of EUR 0 and capital lease obligations of EUR 0. THE FILING CONTRADICTS THIS AND THE FILING WINS — see immediately below. The debt figure used in this dive is approximately EUR 4.49B.
• Corrected net position: +EUR 8.824B NET CASH (EUR 13.317B liquidity less approximately EUR 4.49B of filing-disclosed debt)
• Total assets EUR 50.545B · total liabilities EUR 30.941B · total equity EUR 19.604B
• Inventory EUR 11.424B — 22.6% of total assets, 35.0% of revenue, 256.6 days on hand
• Goodwill EUR 4.587B · intangibles EUR 539.9M
• Financial leverage 2.30x · current ratio 1.334

The vendor understates total debt, and the 20-F says so explicitly

This is the clearest filing-versus-vendor contradiction in the dive and it was caught on a second pass. The vendor's bal_a reports total debt of EUR 2.708B at 2025-12-31, entirely long-term, with shortTermDebt and capitalLeaseObligations both at EUR 0. The 20-F filed 2026-02-25 discloses, verbatim:

> "We have Eurobonds outstanding with an aggregate principal amount of EUR 3.8 billion" — across five senior notes: 1.375% due 2026, 1.625% due 2027, 0.625% due 2029, 0.250% due 2030 and 2.250% due 2032.

> "As of December 31, 2025, we had EUR 693.0 million (2024: nil) outstanding under our EUR 1.5 billion ECP program and the carrying amount was EUR 691.7 million (2024: nil) with a weighted-average interest rate of 2.03%."

Three corrections follow, and all three run the same way.

First, the vendor's shortTermDebt of EUR 0 is wrong. The EUR 3.8B of Eurobond principal exceeds the vendor's EUR 2.708B long-term carrying value by roughly EUR 1.09B — that difference is the 2026-maturity note, which is current. On top of it sits the entire EUR 691.7M commercial-paper balance, which is by definition short-term (the ECP programme issues paper of up to 364 days). Neither appears in the vendor file. Filing-disclosed debt is therefore approximately EUR 4.49B before lease liabilities, against the vendor's EUR 2.708B — an understatement of roughly EUR 1.78B.

Second, the claim that "debt fell sharply during FY2025, from EUR 4.994B to EUR 2.708B" is an artefact of that omission and is WITHDRAWN. ASML did repay borrowings, but it also added EUR 691.7M of commercial paper that had a nil balance a year earlier. The vendor's year-on-year total-debt comparison is not meaningful and is not used.

Third, capitalLeaseObligations of EUR 0 is the standing vendor defect, present here. The field reads EUR 181.9M (2023), EUR 237.4M (2024) and EUR 0 (2025). ASML has not extinguished its leases; the 2025 balance-sheet row is structurally degraded (see Section 8), and the lease liability is one of five line items zeroed in it. The true debt figure is therefore approximately EUR 4.49B PLUS an unrecoverable lease liability, so our EUR 8.824B net-cash figure is itself a mild overstatement.

Net cash therefore corrects from the vendor's EUR 10.203B, and from this dive's own earlier figure of EUR 10.609B, to EUR 8.824B. The conclusion does not change — ASML is comfortably net cash on any reading, and the 17% reduction in the net-cash position does not move the risk score, the fair value or the verdict. But the standing rule is that where the filing and the vendor disagree, the filing wins and the dive says so, and it is said here rather than smoothed. The rating agencies corroborate the direction of travel independently: the 20-F records Moody's at A1, upgraded from A2 in November 2025, and Fitch at A+, upgraded from A in May 2024, alongside a EUR 1.5 billion committed revolver with no amounts outstanding, maturing May 2030.

The enterprise-value rebuild — two further issues, both small.

First, bal_a.netDebt at 2025-12-31 is reported as −EUR 10.203B, which is the vendor's understated total debt (EUR 2.708B) less cash and equivalents (EUR 12.911B) only. It omits the EUR 405.7M of short-term investments — the familiar defect, immaterial here at 3.8% of the position — and it rests on the understated debt figure corrected above. Both errors are present in the same field and they run in opposite directions; the net effect is that the vendor overstates net cash by EUR 1.379B.

Second, and more interesting, km_ttm.enterpriseValueTTM of EUR 562.879B against a euro market capitalisation of EUR 567.567B implies net cash of only EUR 4.688B — EUR 4.14 billion less than the filing-corrected year-end figure of EUR 8.824B. We cannot reconcile this, because the file contains no balance sheet later than 2025-12-31. A decline is plausible: FY2025 saw EUR 5.95B of buybacks and EUR 2.45B of dividends, and the new EUR 12B programme has been running since 2026-01-28. We report both: corrected EV on the filing-adjusted FY2025 balance sheet is EUR 558.743B; the vendor's is EUR 562.879B; the gap is 0.73% of market capitalisation. Because the net-cash position is only 1.9% of the equity value either way, EV-based multiples are barely affected and no conclusion in this dive turns on which is used. evToSalesTTM of 15.93x and evToEBITDATTM of 41.52x are therefore reported as approximately correct rather than rejected — though EV/EBITDA is a poor lens for a company with almost no debt and is not used.

What the balance sheet does not tell you. There is no interest-coverage figure: interestCoverageRatioTTM, debtServiceCoverageRatioTTM and shortTermOperatingCashFlowCoverageRatioTTM are all reported as 0, and the quarterly income statements carry EUR 0 of interest expense across the trailing four quarters against EUR 61.1M of interest income. Coverage is undefined rather than zero. With approximately EUR 4.49B of filing-disclosed debt against EUR 12.5B of trailing operating income, the point is academic, but the ratios are rejected rather than quoted.

Inventory is the one balance-sheet line that deserves watching. EUR 11.424B at 256.6 days is enormous — a full two-thirds of a year's cost of sales sitting on the floor. For a company building EUR 240M machines with multi-year lead times that is structural, not a warning. But it is also the mechanism through which a demand air-pocket would show up first, and the H2'26 ramp described in Section 3 is precisely what that inventory is staged for.

5. Cash flow and capital returns

Free cash flow of EUR 10.647B on EUR 32.667B of revenue is a 32.6% free-cash-flow margin, and capital intensity is genuinely low: capex was 4.2% of revenue and 12.9% of operating cash flow. The FY2023 dip to EUR 3.230B is the working-capital cost of a build cycle, not a profitability event — operating margin that year was 32.8%.

The filing contradicts the vendor on the buyback, and the filing wins. The cash-flow statement shows EUR 5.715B of common stock repurchased in FY2025 and EUR 522.3M in FY2024. The 20-F, note 22, states: "In 2025, we repurchased 8,323,320 shares (2024: 574,925 shares) for a total consideration of EUR 5,950.0 million (2024: EUR 500.0 million)." We use EUR 5,950.0M and EUR 500.0M. The difference is settlement timing. ASML also cancelled 5,683,018 shares in 2025 (2024: 5,754,117).

Dividend — the filing again. The 20-F: "ASML intends to declare a total dividend for the year of 2025 of EUR 7.50 per ordinary share, which is a 17.2% increase compared to the 2024 total dividend of EUR 6.40 per ordinary share," comprising three interim payments of EUR 1.60 (August 2025, November 2025, February 2026) and a proposed EUR 2.70 final. The policy is "a sustainable dividend per share that will grow over time, paid quarterly." The vendor's dividendPerShareTTM of EUR 7.78 is a trailing-twelve-month figure straddling two declaration years and is consistent; the vendor's profile.lastDividend of 8.35043 reconciles to neither figure in either currency and is rejected (Section 8).

Shareholder yield is real but small. On the euro market capitalisation of EUR 567.567B: dividend yield 0.53%, buyback yield 1.05% (using the filing's EUR 5.95B), total 1.58%. Free-cash-flow yield is 1.88%. This is not a name where capital return contributes materially to the expected return — the entire case is earnings growth and the multiple, which Section 6b makes explicit.

The new programme and a caution about execution. On 2026-01-28 ASML announced a buyback of up to EUR 12 billion to be executed by 2028-12-31, of which up to 2.0 million shares will cover employee plans and the remainder is intended for cancellation. The 20-F also discloses that the previous programme completed only EUR 7.6 billion of an authorised EUR 12.0 billion before finishing in December 2025 — a 63% execution rate. Announced authorisations at this company are ceilings, not commitments, and the EUR 12B figure should be read as roughly EUR 4B a year at the prior execution rate, or about 0.7% of market capitalisation annually. Share count is falling accordingly and slowly: 393.6M diluted (FY2024) to 388.9M (FY2025) to 384.9M in Q2'26, about 1.2% a year.

6. Valuation — priced in or room?

At $1,711.89, which is EUR 1,472.60 at the implied rate of 1.16249 (Section 2), against a euro market capitalisation of EUR 567.567B:

(Analyst counts in parentheses. Dispersion is (high − low) ÷ average.)

Coverage is deep and the near years are tight; the far years are not. Twenty-two to twenty-seven analysts on FY2026-FY2028 EPS is among the best coverage in this batch. But FY2028 dispersion is 50% — a range of EUR 46.31 to EUR 80.13 — and FY2029/FY2030 fall to twelve analysts. The FY2027 row (24 analysts, 30% dispersion) is used as the primary forward anchor. The FY2029 and FY2030 rows are shown for shape and no conclusion rests on them.

Consensus sits at the bottom of the company's own guide. FY2026 consensus revenue of EUR 43.232B is at the low end of the raised EUR 43-45B guidance, twenty days after the raise. Either the street has not finished revising, or it does not believe the top half of the range.

Peer context. The vendor peer set — AMAT ($434.0B), AMD ($845.6B), CSCO, IBM, KLAC ($255.3B), LRCX ($397.4B), MU ($1,008.2B), QCOM, SAP, TXN — is not comparable on multiples and no peer multiples are supplied in the file. The genuinely useful comparison is the wafer-fab-equipment complex: ASML, LRCX, KLAC and AMAT are one trade with four expressions, and today ASML participated least (+4.22% against LRCX +7.85%, KLAC +6.95%, AMAT +5.48%). Section 9 explains why.

The vendor's composite rating is B+ with an overall score of 3, driven down by price-to-earnings and price-to-book sub-scores of 1 out of 5 each and up by return-on-equity and return-on-assets sub-scores of 5. That is the whole argument in one line: best-in-class returns, worst-in-class price.

6a. What today's price assumes (the inversion)

At EUR 1,472.60 — 38.6x FY26E and 28.2x FY27E consensus EPS — today's price embeds roughly the following falsifiable claims:

• FY2026 revenue lands at or above EUR 43.2 billion, which requires an H2 of at least EUR 24.9 billion against an H1 of EUR 18.093 billion. (Consensus-derived, 25 analysts; H1 from the reported quarterly income statements.) Layering the vendor's Q3 consensus of ~EUR 11.148B implies a Q4'26 of EUR 13.758B, 47.5% above the EUR 9.326B just reported. This is the most fragile assumption in the price and it resolves on 2026-10-14 and again in January 2027.
• Gross margin holds at or above 54%. (Company guidance of 54-56%, half-weighted; corroborated by the Q2'26 actual of 54.0%.) The 20-F states that EXE systems are currently margin-dilutive, and EXE revenue grew 148.8% in FY2025 — so mix works against this as High-NA scales.
• EPS compounds from EUR 24.71 (FY2025 actual) to EUR 67.24 by FY2028 — a 39.6% three-year compound rate. (Consensus-derived, 27 analysts, but with 50% dispersion on that FY2028 figure.)
• The market keeps paying roughly 28x two-years-forward earnings, i.e. no de-rating despite a 148% twelve-month advance, rising customer concentration at 61.2%, and 29.1% of revenue shipping to China under an export-control regime with two suspensions expiring 2026-11-10. (Our characterisation of the current multiple.)
• China does not close the lithography gap within the horizon. (Our reading. The knowledge base contains the counter-case — see Section 7 — but it is asserted overwhelmingly by the lowest-weight speakers in the set and the strongest sources contradict it.)

6b. The return bridge (why the multiple moves)

Expected return over 12-24 months decomposes as: EPS growth (+37% FY26 to FY27 on consensus) + multiple drift (COMPRESSION assumed, from 38.6x FY26E toward roughly 31x FY27E at our base) + shareholder yield (~1.6%).

Our base case assumes multiple COMPRESSION, and that is the honest reading rather than a conservatism. Note carefully what is and is not being claimed. Moving from 38.6x FY26E to 28.2x FY27E at an unchanged price is mechanical rolldown, not compression — the addendum's guardrail. Real compression is what we assume on top: that the market pays 31x FY27E at the end of the horizon rather than the ~38x it pays for the current year today. The reason is growth maturation. Consensus revenue growth decelerates from +32.3% (FY2026) to +27.3% (FY2027) to +20.6% (FY2028) to +13.1% (FY2029) and +7.7% (FY2030); a business decelerating on that path does not sustain a high-thirties multiple indefinitely.

Almost the entire expected return in our base case is earnings growth. From EUR 1,472.60 at 28.2x FY27E to a base of EUR 1,620.68 at 31x FY27E is +10.1%; of that, the arithmetic is entirely carried by using FY2027 rather than FY2026 as the anchor year. If the multiple were to hold at 28.2x FY27E, the euro price would be EUR 1,474 — no return at all. That is the cleanest statement of why this is a Watch: on consensus, at a constant forward multiple, the stock is worth what it costs. The return requires either that the market pay a higher multiple on FY2027 than we assume, or that FY2027 earnings beat.

The bull case at $2,328 requires 38.3x FY27E — the market paying next year's earnings the multiple it pays this year's. That is exactly what the street's $2,305.75 consensus target embeds (37.9x FY27E). It is a coherent view and it is not our base case; it is a demand for no de-rating at all through a period of visible deceleration.

6c. Variant perception (where we differ, what would surprise)

• We differ from the street on direction, and the gap is large. Our base of $1,884 is 18.3% below the consensus target of $2,305.75. We do not differ on the earnings — we use the same consensus EPS. We differ entirely on the multiple: the street's target implies 37.9x FY27E; we assume 31x. Watchable number: whether published targets are revised up, down, or not at all after the 2026-10-14 print.
• We think part of the 34.7% target gap is staleness, not conviction, and the evidence is specific. The target band is high $2,623, low $2,000, median $2,300 — every published target is above spot, and the lowest of them sits $0.04 below the 52-week high of $1,999.96. A genuine dispersion of views on a stock trading 14% below its high would produce at least one target beneath the price. It does not. The most economical explanation is that targets set at or near the high were never marked down through the drawdown. Against that reading: the band itself is a real 31% spread from low to high, and the rating distribution (1 strong buy, 25 buy, 16 hold, 3 sell) shows genuine disagreement, including three sells. The honest verdict is partial staleness, not fabrication — and it means the +34.7% headline gap should be discounted, not taken at face value.
• We differ from the company's own long-range frame, and so does the street, in the same direction but by more. The 20-F reaffirms a 2030 revenue opportunity of EUR 44-60 billion at 56-60% gross margin. FY2026 guidance is EUR 43-45 billion — the company will reach the bottom of its own 2030 range four years early. Consensus FY2030 revenue is EUR 80.843 billion, 35% above the top of that range. This frame is stale on any reading. Watchable event: whether ASML issues a new long-range model. If it does, and the number starts with a seven or an eight, the street is validated and the multiple is defensible. If it reiterates EUR 44-60B, the FY2029-30 estimates that underpin the terminal value are unsupported by the company itself.
• We do NOT differ on the moat, and the knowledge base is unusually unanimous. Of 31 usable entity-matched claims, zero are bearish. The entire bear case — that China solves lithography domestically — lives in the match: text lane and is asserted principally by a management-class speaker at skill weight 0.5 in five near-identical restatements. Watchable event: any credible production-scale (not journal-stage) Chinese EUV tool.
• Positive surprise that would force a re-rating: a Q3'26 print that confirms the Q4 ramp, plus a bookings number that extends the RPO above EUR 50 billion; or a new long-range model materially above EUR 60 billion for 2030; or High-NA unit recognitions guided into double digits with ASPs above EUR 300M. Watchable numbers: RPO at the next disclosure, EXE units, and FY2027 gross-margin commentary.
• Negative surprise that would break the thesis: an FY2026 guidance CUT — after three consecutive raises, the fourth revision going the other way would reset both the multiple and the credibility of the ladder. Also: either export-control suspension lapsing on 2026-11-10 without replacement, given China is 29.1% of revenue; or customer concentration above 61.2% coinciding with a single large customer deferring.

Synthos fair values

All three anchors are multiples of FY2027 consensus EPS of EUR 52.28 (24 analysts, the deepest reliable line), stated in euros and converted at the implied rate of 1.16249 USD/EUR. Cross-checked against FY2028 consensus EPS of EUR 67.24.

• Bear ~$1,216 — 20x FY27E (EUR 1,046). Cross-check: 15.5x FY28E. The scenario: the H2'26 ramp slips, FY2026 lands at or below EUR 40B, the guidance ladder reverses, EXE dilution holds the gross margin at 52-53%, and one or both export-control suspensions lapse on 2026-11-10 with China falling below 25% of revenue. Note the stock traded at $684.24 within the last twelve months, so a move of this size is well within its demonstrated range. −29.0%.
• Base ~$1,884 — 31x FY27E (EUR 1,620.68). Cross-check: 24.1x FY28E. The scenario: the guide is met near the midpoint, gross margin settles at 54-55%, the RPO keeps growing at a high single-digit rate, and the market pays a mid-thirties multiple on the current year that rolls to about 31x on the next. +10.1%, plus ~1.6% of shareholder yield.
• Bull ~$2,328 — 38.3x FY27E (EUR 2,002). Cross-check: 29.8x FY28E. The scenario: no de-rating at all, the FY2026 guide is beaten, ASML issues a 2030 frame above EUR 70 billion, and High-NA scales faster than modelled. This is essentially the street's consensus target of $2,305.75 (37.9x FY27E) — a useful cross-check that our bull case and the street's base case are the same number. +36.0%.

Base is 10.1% above spot; asymmetry roughly 1.24:1 to the upside (29% down, 36% up). That is not enough asymmetry to buy into a 148% twelve-month advance, 71 days before the print that resolves the central question. Watch.

7. Knowledge base — 60 raw hits, 37 used, 23 discarded, and not one bearish entity match

Raw hits: 60. Used: 37. Discarded: 23. Breadth: 17 distinct sources. Net conviction: positive-high.

Search terms: ASML, EUV, lithography. 37 hits were match: entity (the claim names ASML explicitly) and 23 were match: text (ASML, EUV or lithography appears in prose about something else). Of the 31 entity matches retained, 22 are bullish and 9 neutral. Zero are bearish.

What was discarded and why (23):

• 3 claims tagged thinker: quarantine_misattributed (2025-10-10, 2026-04-21, 2026-06-23). Under the four-lane attribution policy these carry an unresolved channel-versus-speaker conflict and are excluded regardless of content. Two were bullish, one bearish, so the exclusion is not directionally convenient.
• 4 near-identical restatements by the same management-class speaker on China and lithography (2026-02-08 ×2, 2026-07-25, 2026-07-27, 2026-07-29 — five in total, one retained as representative). All say the same thing in the same words.
• 2 duplicate restatements within a single business_breakdowns document (2023-08-10) where the same claim about ASML's competitive advantage appears twice.
• 14 further claims where the search term is a generic noun or the claim is about a different company — Samsung's EUV capex, Intel's EUV under-investment, SMIC's 7nm yields, neon supply from Ukraine in 2022, mask-processing software, the layers of the compute stack above lithography, the philosophy of technological path-dependence, and a macro claim about avoiding European equities in which ASML is named only as the exception.

The moat lane — business_breakdowns, Tom Walsh (speaker), 2023-08-10, all independent:

> bullish · conviction 82 · entity: ASML

> "ASML has essentially 100% of most-advanced EUV lithography and 90%+ of next-gen; of all competitive advantages he's seen he can't think of one more significant — no rival within 10 years."

> bullish · conviction 78 · entity: ASML

> "No rival is close on EUV; Nikon and Canon abandoned it, and with 25+ years and €10bn+ R&D sunk plus bespoke components, any approximation is a decade away at best."

> bullish · conviction 63 · entity: ASML

> "As sole leading-edge lithography supplier ASML's revenues have become less cyclical; the story is now secular structural growth driven by leading-edge logic and memory demand, despite ongoing order oscillations."

That third claim is the analytically important one and the file supports it. Installed-base service revenue of EUR 8,193.0M grew 26.2% in a year when DUV unit shipments fell 25%. The damper is real and measurable.

The chokepoint lane:

> all_in · 2025-04-19 · bullish · conviction 85 · entity: ASML

> "ASML is the sole maker of ~$200M EUV lithography machines; that monopoly on 13.5nm-wavelength printing is the chokepoint gating advanced chip manufacturing."

> dwarkesh · 2026-02-05 · bullish · conviction 80 · entity: ASML, TSMC

> "ASML, not TSMC, is the true chokepoint; China is behind on chips only because it can't replicate/buy ASML lithography, not fabrication itself."

> geopolitical_cousins · 2026-07-29 · speaker Peter Zeihan (independent, guest) · neutral · conviction 80 · entity: ASML

> "Sub-7nm chips require ASML's EUV, with 100,000+ supply-chain steps and 400 single-product single-user firms across 60 countries—remove a few and it halts; rebuilding takes ~30 years."

> eurodollar_university · 2026-07-28 · neutral · conviction 58 · entity: ASML

> "European suppliers (ASML, Carl Zeiss, Trumpf, Siltronic, Aixtron, SUSS MicroTec) control irreplaceable semiconductor/precision-optics bottlenecks that took decades to develop and China can't easily replace."

The Zeihan claim cuts both ways and the dive treats it that way. A supply chain of 400 single-product, single-user firms across 60 countries is an unassailable moat and an unhedgeable fragility — the same fact stated twice. The oldest claim in the set makes the second half explicit: real_vision, 2023-03-23, conviction 85, "ASML is the world's best company and sole maker of leading-edge lithography machines — irreplaceable, but its distributed supplier constellation is a single point of failure."

The deepest single lane is Jordi Visser — 10 entity claims, skill weight 2.0, running 2025-09-21 to 2026-05-30. Where the doc_id shows a different channel from the speaker, the claim is attributed to Visser, with the channel noted separately, per house attribution policy:

> 2025-09-21 · bullish · conviction 75 · entity: ASML

> "ASML's EUV monopoly is critical for producing the advanced chips that serve as the brain for AI-native devices, robotaxis and humanoids; unchanged since 2021, backlog building."

> 2026-02-05 · bullish · conviction 68 · entity: Teradyne, Lattice Semiconductor, ASML

> "Analog/under-participated semis (Teradyne, Lattice) and ASML have a long road ahead despite ASML's run — rotating into names that haven't moved yet."

> 2026-05-30 · bullish · conviction 70 · entity: ASML, Siemens Energy (spoken by Visser; channel: anthony_pompliano)

> "AI trade is global, not US-only: data-center/optical/memory buildout lifts international semis and manufacturers (ASML, Soitec, Ajinomoto, Korea/Japan memory, Siemens Energy); strength in Korea/Japan confirms it."

Note the 2026-02-05 claim carefully — it contains its own qualifier, "despite ASML's run," and its actual recommendation is to rotate toward names that have not moved. That is the closest thing in this entire set to a caution on the entry, and it comes from the highest-weighted source in the lane.

The bear lane — entirely in the match: text band, and mostly low-weight:

> anthony_pompliano · 2026-04-18 · bearish · conviction 55 · text · speaker's own voice, LOW WEIGHT under the promoter-class rule

> "Massive competition and abundant private capital flooding into chips/hardware will push semiconductor prices lower over ~5 years — unless supply stays concentrated (three foundries, one lithography supplier)."

> Test attached to the claim: "Semiconductor prices are lower in ~5 years than today."

This is a bearish claim that exempts ASML in its own subordinate clause — "one lithography supplier" is precisely the condition under which the deflation thesis fails.

> elon_musk · 2026-02-08 · bullish (for Tesla) · conviction 65 · text · speaker_role: management, skill weight 0.5 — HALF-WEIGHTED and then some

> "China's advanced-chip bottleneck is sanctioned ASML EUV lithography, not TSMC; Elon says they need 3-4 years for domestic alternatives, giving Tesla a 2-4 year intelligence lead."

> doomberg · 2026-05-28 · bullish (for China) · conviction 72 · text

> "China will catch up and lap the US in semiconductors and AI; Huawei reaching 1.5nm without EUV proves it, and a chip glut is coming."

> geopolitical_cousins · 2025-05-06 · neutral · conviction 45 · text

> "Chinese breakthroughs (fast flash memory, EUV, sodium-ion batteries, Huawei AI chips) are early-stage, not production; each could be transformative but it's too early to call a tech-war win."

The bear case is real but it is weakly sourced and internally contradictory. The same speaker asserts across five separate dates that China is "closer than most realize" to solving lithography, at skill weight 0.5, with a management role tag; another source says China has already lapped the West; a third says the breakthroughs are journal-stage rather than production. They cannot all be right, and the strongest-weighted sources in the set — Walsh, Zeihan, dwarkesh — contradict all three. The honest summary is that the bear case exists, is not absurd, and has not produced a single entity-matched claim in four years.

The export-control lane is genuinely two-sided and worth quoting:

> dylan_patel · 2024-10-02 · neutral · conviction 70 · entity: ASML, Applied Materials

> "Chip export controls work but equipment controls are a failure — China was recently ~48% of ASML and ~45% of Applied Materials revenue, and the tools shipped build chips better than the West legally sells China."

That 48% figure was a quarterly peak; the FY2025 annual number in this file is 29.1%, down from 36.1% in FY2024. The controls have bitten since that claim was made.

What this lane establishes, and what it conspicuously does not.

It establishes the moat about as firmly as a knowledge base can: 17 sources, four years, 31 entity-matched claims, zero bearish, from technology analysts, macro strategists, geopolitical analysts and equity specialists who agree on almost nothing else. What it does not contain is a single claim about ASML's valuation, its price, or its entry. No source names a fair value. No source calls it cheap or expensive. The only price-adjacent remark in the entire set is Visser's "despite ASML's run." The panel underwrites the business and is silent on the trade — and this dive's hesitation is entirely about the trade. That asymmetry is why the conviction rating reads "High on the business, absent on the price" rather than simply High, and why a KB lane this strong does not by itself produce a Buy.

A weighting note. Six of the used claims come from Jensen Huang, a chief executive of a large customer and partner of ASML. Those are management-voice claims about a supplier relationship and are half-weighted accordingly — including the 2026-04-15 claim that "all upstream bottlenecks (logic, CoWoS, HBM, EUV) are solvable 2-3 year problems given a demand signal," which is, read plainly, a customer arguing that ASML's scarcity is temporary. It is quoted here because it cuts against the position, not for it.

8. Data integrity — what we rejected from the vendor file and why

Twelve defects and limitations. This section is the point of the exercise.

1. CURRENCY MIXING — the defect that matters most, and it is structural rather than a bug. quote, profile, pt and earn_cal are in USD; inc_a, inc_q, bal_a, cf_a, est and seg_prod/seg_geo are in EUR; km_ttm and ratios_ttm are computed on an EUR price of EUR 1,472.60. Implied rate 1.16249 USD/EUR, derived from quote.marketCap ($659.793B) ÷ km_ttm.marketCap (EUR 567.567B) and independently confirmed against priceToBookRatioTTM × bookValuePerShareTTM. A naive $1,711.89 ÷ EUR 27.55 gives 62.1x rather than the correct 53.4x — a 16.3% overstatement on every multiple. Every figure in this dive states its basis.

2. earn_cal is in USD and cannot be compared with est, which is in EUR. The 2026-07-15 entry reports epsActual 8.68 on revenueActual 10,665.68M; the Q2'26 income statement reports EUR 7.58 on EUR 9,326M. The implied conversion is ~1.144, a period-average rather than the 1.16249 spot. Any "beat versus estimate" computed across the two blocks is meaningless. Within earn_cal the comparison is internally consistent and is used only that way (Q3'26 consensus USD 11.74 / USD 12.960B).

3. seg_prod FY2025 is INCOMPLETE — the installed-base line is missing — and we reconstructed it against the filing. The seven disclosed FY2025 lines sum to EUR 24,474.3M against reported revenue of EUR 32,667.3M, leaving EUR 8,193.0M unaccounted. The FY2024 row contains a "Service And Field Options" line of EUR 6,494.2M; the FY2025 row does not. The reconstruction is confirmed twice by the 20-F: the residual of EUR 8,193.0M implies +26.16% on FY2024, and the 20-F states net service and field option sales increased 26.2%; the systems total of EUR 24,474.3M implies +12.43%, and the 20-F states net system sales increased 12.4%. Both reconcile to within 0.05 percentage points. We use EUR 8,193.0M.

4. seg_prod.reportedCurrency is MISLABELLED for every year before FY2025. FY2024, FY2023 and every earlier row are tagged USD. The FY2024 lines sum to EUR 28,262.9M, which is the FY2024 revenue in euros to the last decimal. The data are EUR; the label is wrong. Anyone converting those rows to euros would understate historic revenue by 14%.

5. seg_geo does not sum to revenue in either year. FY2025 sums to EUR 32,143.0M against EUR 32,667.3M — EUR 524.3M, 1.6% unallocated. FY2024 sums to EUR 26,940.8M against EUR 28,262.9M — EUR 1,322.1M, 4.7% unallocated. The "Rest of Europe and Middle East" line present through FY2022 disappears thereafter, and the Netherlands line of EUR 4.7M is implausible as a complete European figure. Geographic shares are quoted both as a percentage of total revenue and as a percentage of the disclosed geographic total, and the residual is stated.

6. Backlog, remaining performance obligations, unit shipments and customer concentration are ABSENT from the vendor file entirely — the filing supplies all four. RPO EUR 46.5B at 2025-12-31 (2024: EUR 43.3B), 65% convertible within twelve months (2024: 59%); backlog EUR 38.8B per the CEO letter; 4 EXE + 44 NXE recognised in FY2025 (2024: 2 + 42) and 279 DUV units (2024: 374); four customers each above 10% of net sales totalling EUR 20.0B or 61.2% (2024: 53.8%; 2023: 53.9%). These are the most important facts in the dive and none of them is in the payload.

7. FY2026 GUIDANCE is absent from the vendor file and the filings show three successive raises. 20-F 2026-02-25: EUR 34-39B at 51-53% gross margin. 6-K 2026-04-15: EUR 36-40B at 51-53%. 6-K 2026-07-15: EUR 43-45B at 54-56%. Consensus FY2026 revenue of EUR 43.232B sits at the bottom of the current range.

8. The 20-F contradicts the cash-flow statement on the buyback, and the filing wins. Vendor cf_a.commonStockRepurchased: EUR 5,715.0M (FY2025) and EUR 522.3M (FY2024). 20-F note 22: "we repurchased 8,323,320 shares (2024: 574,925 shares) for a total consideration of EUR 5,950.0 million (2024: EUR 500.0 million)." We use EUR 5,950.0M and EUR 500.0M. Settlement timing explains the difference.

9. est.ebitAvg and est.netIncomeAvg are CORRUPT in the out-years — REJECTED; epsAvg used throughout. FY2027: netIncomeAvg EUR 20.562B against ebitAvg EUR 18.537B — net income above operating income, impossible. FY2029: EUR 29.808B against EUR 25.290B. FY2030: EUR 37.839B against EUR 27.238B, a 39% inversion. Separately, FY2026 netIncomeAvg of EUR 12.778B is inconsistent with epsAvg of EUR 38.13 on ~385M shares (which implies EUR 14.7B). All ebitdaAvg, ebitAvg and netIncomeAvg rows are rejected. Every forward figure in this dive uses epsAvg and revenueAvg only.

10. ratios_ttm.priceToFairValueTTM (26.00365) is IDENTICAL to priceToBookRatioTTM (26.00365) to eight decimal places — REJECTED. The vendor's "fair value" is book value relabelled. It carries no valuation information and is not used.

11. Payables and coverage fields are unpopulated, not zero — REJECTED. payablesTurnoverTTM = 0, averagePayablesTTM = 0 and daysOfPayablesOutstandingTTM = 0, which makes cashConversionCycleTTM (346.06 days) identical to operatingCycleTTM (346.06) and therefore an overstatement of the true cycle by the full payables term. Likewise interestCoverageRatioTTM = 0, debtServiceCoverageRatioTTM = 0 and shortTermOperatingCashFlowCoverageRatioTTM = 0 reflect EUR 0 of interest expense in the quarterly income statements, so coverage is undefined rather than nil. grahamNumberTTM (EUR 187.54) and grahamNetNetTTM (−EUR 21.91) are not meaningful for this business and are not used.

12. profile.lastDividend (8.35043) reconciles to nothing and its currency is unstated — REJECTED. The 20-F declares a total 2025 dividend of EUR 7.50 per ordinary share (+17.2% on EUR 6.40), comprising three EUR 1.60 interims and a EUR 2.70 final proposal. ratios_ttm.dividendPerShareTTM of EUR 7.78 is a coherent trailing-twelve-month figure spanning two declaration years and is used for the yield calculation (0.53% on the euro price). 8.35043 is neither, in neither currency, and is discarded.

Two further limitations that are not vendor defects but constrain the dive:

A. The insider block is EMPTY — zero transactions. This is almost certainly structural rather than a data failure: as a foreign private issuer, ASML's officers and directors are exempt from Section 16 insider-reporting requirements. No insider read is possible for this name, in either direction. See Section 10.

B. All three 6-Ks are cover pages; the exhibits were not extracted. Exhibits 99.1 (press release), 99.2 (presentation), 99.3 (summary US GAAP consolidated financial statements) and 99.4 (statutory interim report for the six months ended 2026-06-28) are referenced but absent. Consequently: no Q2'26 balance sheet, no interim cash-flow statement, and no quarterly net system bookings. The cover pages do carry hard headline data — revenue, net income and the guidance range appear in the exhibit titles — and those are used. The quarterly granularity available for ASML is materially thinner than for the US-listed peers in this batch, and that is stated rather than papered over.

CORRECTION — the FY2025 debt line is NOT correctly reported, and the filing overrules it. The vendor's longTermDebt (EUR 2.708B), shortTermDebt (EUR 0) and capitalLeaseObligations (EUR 0) are not double-counted, but they are incomplete: the 20-F discloses EUR 3.8B of Eurobond principal across five notes maturing 2026 to 2032, plus EUR 693.0M outstanding under the EUR 1.5B ECP programme (carrying amount EUR 691.7M, 2.03% weighted-average rate, against nil in 2024). The current Eurobond maturity and the entire commercial-paper balance are both absent from the vendor file, understating total debt by roughly EUR 1.78B and net cash by the same less the omitted short-term investments. We use the filing throughout: approximately EUR 4.49B of debt and EUR 8.824B of net cash. The FY2024 row (EUR 1.079B short-term + EUR 3.677B long-term + EUR 237.4M leases = EUR 4.994B total) does reconcile internally, which is why the defect is confined to the structurally degraded FY2025 row. stockBasedCompensationToRevenueTTM of 0.41% is consistent with the cash-flow statements. tech.hi52 ($1,989.44) and tech.lo52 ($689.63) disagree slightly with quote.yearHigh ($1,999.96) and quote.yearLow ($684.24); we quote the quote values for the range and the tech values for the computed percentage from the high (−14.0%), and note the discrepancy.

Non-equity tripwire — checked and passed. ASML on Nasdaq is common equity, specifically ordinary shares in New York registry form: ISIN USN070592100, CUSIP N07059210 (the "N" prefix denoting Netherlands domicile), isAdr: False, isEtf: False, isFund: False, CIK 0000937966. It is not a preferred, a baby bond or a derivative: the price is $1,711.89 (nowhere near a $25 or $1,000 par), beta is 1.394 (not the sub-0.3 of a rate instrument), the dividend is variable and growing (EUR 6.40 to EUR 7.50, +17.2%) rather than fixed, average volume is 1.91M shares a day (deep), and the 52-week band of $684.24 to $1,999.96 is a 192% span, the opposite of a narrow par-anchored range. The 20-F separately describes cumulative preference shares issuable to a foundation as an anti-takeover structure — those are a different class, are excluded from the diluted share count, and are not the listed instrument. Check done, check passed.

9. Technicals

• Price $1,711.89, +4.22% on the day (previous close $1,642.52). Opened $1,697.03 — a 3.32% gap up — and traded $1,686.01 to $1,717.56, closing near the high.
• −14.0% from the 52-week high of $1,999.96, and +150.2% above the 52-week low of $684.24. Maximum drawdown from peak over the trailing year: −14.0% — the stock is essentially at its worst point since the peak.
• BELOW the 50-day moving average: −2.2% against $1,751.12. ABOVE the 200-day: +23.0% against $1,392.08. The 200-DMA is rising steeply; the 50-DMA has rolled over.
• RSI 41.3 — soft, not oversold. MACD −33.73 — negative, and this is the only meaningful semiconductor name in the batch with a negative MACD.
• Relative performance: 3-month +23.4% vs SPY +7.6% and QQQ +7.7%; 6-month +18.7% vs SPY +11.1%; 12-month +148.0% vs SPY +24.3% and QQQ +30.8%. A 124-point twelve-month excess return.
• Volume: 1,494,058 shares against an average of 1,909,658 — 78% of normal.

Today's move and what it does to the entry

ASML rose 4.22% on a day when LRCX rose 7.85%, MU 7.62%, KLAC 6.95%, AMAT 5.48%, INTC 10.92% and MRVL 12.81%. It was the smallest semiconductor move in the batch, and the most monopolistic asset in the complex participated least. That is not an anomaly; it is a correct read of what the day was about.

Today was a memory-led session. Micron rose 7.62% and the memory-adjacent equipment names — Lam and KLA, whose revenue mix is heaviest in etch, deposition and process control for DRAM and NAND capacity — rose most. ASML is not a memory-capacity play. Its revenue is levered to leading-edge logic and to DRAM node migration, which is a different thing: adding DRAM wafer capacity requires etch, deposition and metrology tools long before it requires another EUV scanner, whereas shrinking a DRAM node to 1B/1C requires EUV. The knowledge base contains exactly this distinction, from odd_lots on 2026-02-16: "Node migration (to 1B/1C) is the only near-term supply lever under wafer constraints, but it takes time, needs EUV, and competes with HBM for the same advanced capacity." A memory up-day is a second-order ASML event and a first-order Lam and KLA event, and today's relative moves priced it exactly that way.

The more important point is what today's move did not do. It did not repair the chart. ASML gapped up 3.3% and added a further 0.9%, but it closed $39.23 below its 50-day average and $288.07 below its 52-week high, with MACD still negative. The move happened inside a drawdown, not out of one. And it happened on 78% of average volume — 1.49M shares against a 1.91M average. A 4.2% advance that nobody had to fight for, on below-average participation, is a gap that was granted rather than earned. Compare the alternative reading: if this were genuine institutional accumulation into the highest-quality asset in the complex, volume would be above average, not a fifth below it.

So: are we asking the reader to buy into a one-day gap? No — and the honest answer is that we would not ask them to buy at all today. The stock is up 148% over twelve months. The base case is +10.1%. The chart is broken in the short term and intact in the long term. There is no print for 71 days and two export-control suspensions expire in 98 days. The sensible posture is to watch, with two concrete triggers: a reclaim and hold of the $1,751 50-day average would repair the short-term structure and argue for a first tranche; a Q3'26 print on 2026-10-14 that confirms the Q4 ramp would convert this to Buy — Tactical outright. A retest toward the $1,392 200-day average, roughly 19% lower, would be a genuine accumulation opportunity for a business of this quality and would sit close to our bear case of $1,216.

Cross-batch note. The wafer-fab-equipment complex — ASML, LRCX, KLAC, AMAT — is one trade with four expressions, and ASML is the highest-quality and the most expensive of them. It is also the only one with a monopoly rather than a duopoly or an oligopoly. If exposure to the complex is already held through Lam, KLA or Applied, ASML adds quality and geopolitical concentration rather than diversification — the export-control exposure is specific to the Dutch entity, and 29.1% of revenue ships to China.

10. Insiders — no data exists

The insider block in this file is EMPTY. There are zero transactions.

This is not a vendor failure to investigate. ASML is a foreign private issuer, and the officers and directors of foreign private issuers are exempt from the Section 16 beneficial-ownership reporting requirements that generate the Form 3, 4 and 5 filings from which insider data is compiled for US domestic registrants. There is nothing to report because nothing is required to be filed.

The consequence for this dive, stated plainly: no insider signal is available in either direction. We cannot say management is accumulating and we cannot say it is selling. Where other names in this batch offer a corroborating or contradicting insider read, ASML offers none, and the absence is a genuine informational disadvantage rather than a neutral result. It is one of the reasons the risk score is a 5 rather than a 4.

The nearest available substitute is corporate action rather than personal transaction, and it is constructive but weak: the company repurchased 8,323,320 shares for EUR 5,950.0 million in FY2025, cancelled 5,683,018, announced a further EUR 12 billion programme on 2026-01-28 running to 2028-12-31, and raised the declared dividend 17.2% to EUR 7.50. That is a board allocating capital as though it believes in the business. It is not the same evidence as a director buying stock with personal money, and it is weakened by the disclosure that the previous EUR 12 billion authorisation was only 63% executed (EUR 7.6B of EUR 12.0B).

11. Verdict, kill-criteria and flip conditions

Watch.

This is the best business in the batch and, on the numbers in front of us, an average entry into it.

Quality: the only true monopoly in the complex — 44 NXE and 4 EXE systems recognised in FY2025 against zero competing units anywhere; a 52.8% gross margin heading to a guided 54-56%; a 52.4% return on equity and 36.9% return on invested capital; EUR 8.824 billion of net cash on filing-corrected debt, with Moody's and Fitch both upgrading during 2024-2025; EUR 8.193 billion of installed-base service revenue growing 26.2% and now a quarter of the company; and EUR 46.5 billion of contracted remaining performance obligations with the twelve-month conversion share rising from 59% to 65%. Conviction: 37 usable knowledge-base claims from 17 independent sources across four years, zero bearish among 31 entity matches, with the bear case confined to the text lane and asserted principally by a half-weighted management voice.

What stops it being a Buy is three specific things, none of which is a criticism of the company.

First, the arithmetic. At 28.2x FY2027 consensus EPS, a constant forward multiple produces no return at all. Our +10.1% base requires the market to pay 31x FY27E at the end of the horizon, against consensus revenue growth decelerating from +32% to +8% between FY2026 and FY2030. The asymmetry is 29% down against 36% up — 1.24 to 1, which is not enough after a 148% twelve-month advance.

Second, the ramp. FY2026 guidance of EUR 43-45 billion against an H1 of EUR 18.093 billion requires an H2 of EUR 24.9-26.9 billion, and on the vendor's own Q3 consensus a Q4'26 of EUR 13.8-15.8 billion against the EUR 9.326 billion just reported. The order book supports it and the guidance ladder has moved up three times running. It is still the largest single-quarter step being asked of any company in this batch, and it does not resolve until 2026-10-14 at the earliest.

Third, the street's target is not what it appears. The $2,305.75 consensus is 37.9x FY27E — a demand that the market pay next year's earnings the multiple it pays this year's, with no de-rating whatsoever. The entire published band sits above spot, with a low of $2,000 that is four cents beneath the 52-week high. That is at least partly stale marks, and the +34.7% headline gap should be discounted accordingly.

What you would be waiting for. Not much, and not long. The Q3'26 print is 71 days out. The export-control cliff is 98 days out. A reclaim of the $1,751 fifty-day average would repair the short-term structure. The cost of patience here is one quarter of a business that is not going anywhere; the cost of impatience is buying a 38.6x current-year multiple into an unproven ramp.

Pre-registered KILL criteria — what would break the thesis:

• An FY2026 guidance CUT at the 2026-10-14 print. After raises on 2026-02-25, 2026-04-15 and 2026-07-15, a downward revision would reset the multiple and, more damagingly, the credibility of the ladder itself.
• Q3'26 revenue materially below the ~EUR 11.1 billion implied by consensus, which would make the Q4 step required to reach EUR 43 billion arithmetically implausible.
• Gross margin printing below 52%, indicating that EXE dilution is outrunning NXE mix and service margin — the 20-F already identifies EXE as margin-dilutive.
• Remaining performance obligations falling year on year at the next annual disclosure, or the twelve-month conversion share dropping back below 60%.
• Either export-control suspension lapsing on 2026-11-10 without replacement, given that China was 29.1% of FY2025 revenue.
• Credible evidence of a production-scale — not journal-stage — Chinese EUV tool. This is the one event that would invalidate the moat rather than the cycle, and it is the only claim in the knowledge base capable of doing so.
• Customer concentration rising further above 61.2% while a top-four customer defers deliveries.

Pre-registered FLIP TO BUY — TACTICAL:

• A Q3'26 print on 2026-10-14 that confirms the ramp — revenue at or above ~EUR 11.1 billion with the FY2026 guide reaffirmed at EUR 43-45 billion and gross margin at or above 54%.
• A reclaim and hold of the $1,751 fifty-day average with a positive MACD crossover, on above-average volume.
• A price below roughly $1,500 — 24.7x FY27E — at which point the base case offers better than 25% upside and the asymmetry inverts. A retest of the $1,392 200-day average would be an emphatic accumulation zone.
• A new long-range model above EUR 70 billion for 2030, replacing the obsolete EUR 44-60 billion frame and validating the FY2029-30 estimates that the terminal multiple depends on.
• Disclosed net system bookings that push remaining performance obligations above EUR 50 billion.

Where ASML fits in the Synthos Framework Portfolio. The semiconductor capital-equipment sleeve, as a watchlist name with a 1-2% target weight on trigger, not an initiation today. If a position in the complex is wanted immediately, the honest observation is that ASML is the highest-quality expression and the one with the least margin of safety — and that holding LRCX, KLAC or AMAT alongside it adds correlation, not diversification, since all four sell into the same capital budgets. ASML's differentiator within the sleeve is monopoly and installed-base annuity; its distinguishing risk is that 29.1% of revenue ships to China from a Dutch entity subject to Dutch, EU and US export controls with two suspensions expiring on the same day. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $1,711.89, with the flip conditions above pre-registered.

Single biggest risk: the second-half ramp. Everything in the FY2026 multiple depends on a second half 38-49% larger than the first, and on a fourth quarter roughly 48-69% larger than the quarter just reported. The EUR 46.5 billion of contracted obligations makes it credible; the fact that revenue recognition depends on "the customer facility readiness to take delivery of our goods or services, as well as applicable export control restrictions" — the 20-F's own language — makes it fragile in a way ASML does not fully control. A slip is not a demand event. But it would move earnings across a fiscal boundary at a 38.6x current-year multiple, and the multiple would not wait for the explanation.

Most fragile assumption in the price: that the market keeps paying a high-thirties multiple for a business whose consensus revenue growth decelerates from +32% to +8% over the next four years. The bull case, the street's target, and a good deal of the current price all require no de-rating at all. Our base case requires only mild compression and still yields just +10.1%. If growth decelerates on the consensus path and the multiple normalises toward the mid-twenties on forward earnings, this is dead money for two years even if every operational thing goes right. That is the specific, unglamorous risk in owning an outstanding company at a fair-to-full price.

Provenance & disclosures

• Traceability: 60 raw knowledge-base hits on the terms ASML, EUV and lithography; 37 used, 23 discarded (3 tagged quarantine_misattributed and excluded under the four-lane attribution policy; 4 near-identical restatements by a single management-class speaker; 2 duplicate claims within one business_breakdowns document; 14 match: text hits where the search term is a generic noun or the claim concerns Samsung, Intel, SMIC, Huawei, neon supply, mask-processing software, the compute stack above lithography, technological path-dependence, or European macro allocation). 37 hits were match: entity; 31 of those were retained, of which 22 are bullish and 9 neutral and NONE is bearish. Breadth 17 distinct sources: business_breakdowns (Tom Walsh, David Samra), jordi_visser, jensen_huang, dwarkesh, dylan_patel, chinatalk, real_vision, all_in, luke_gromen, macrovoices, andreas_steno, eurodollar_university, geopolitical_cousins (Peter Zeihan), odd_lots, doomberg, anthony_pompliano, elon_musk. Span 2022-10-26 to 2026-07-29. Where speaker and thinker differ the claim is attributed to the speaker and the channel noted separately — three Jordi Visser claims are carried on the anthony_pompliano channel and one Andreas Steno claim on the mikkel_rosenvold channel, and are attributed to Visser and Steno respectively. Half-weight applied to: six Jensen Huang claims (chief executive of a large ASML customer, management voice about a supplier relationship); one Elon Musk claim (speaker_role: management, skill weight 0.5); one Anthony Pompliano claim (own voice, promoter-class low weight). All quotations are verbatim from the stored claim text. No claim in the entire set addresses ASML's valuation or price.
• Data as-of: income statement through 2026-06-30 (Q2'26, reported 2026-07-15, twenty days before this dive) · balance sheet and cash flow 2025-12-31 (FY2025, per the 20-F filed 2026-02-25) — no interim balance sheet or interim cash-flow statement exists in this file · estimates 2026-08-04 · prices 2026-08-04 (quote.timestamp 1785873601 = 2026-08-04T20:00:01Z, the 16:00 ET close; $1,711.89, 50-DMA $1,751.12, 200-DMA $1,392.08, RSI 41.3, MACD −33.73) · knowledge-base claims 2026-08-04. All figures come from the Synthos vendor payload for ASML and from the SEC filings in filings/ASML/; no figure comes from memory, recall or external retrieval.
• CURRENCY BASIS — the most important note in this dive. ASML reports in EUR; the Nasdaq listing is quoted in USD. The vendor payload mixes both: quote, profile, pt and earn_cal are USD; inc_a, inc_q, bal_a, cf_a, est, seg_prod and seg_geo are EUR; and km_ttm/ratios_ttm are computed on an implied EUR price of EUR 1,472.60. The implied rate of 1.16249 USD per EUR is derived as quote.marketCap ($659,793,204,613) ÷ km_ttm.marketCap (EUR 567,566,533,547) and independently confirmed against priceToBookRatioTTM (26.00365) × bookValuePerShareTTM (EUR 56.63) = EUR 1,472.60. Every multiple in this dive is euro-price against euro-earnings. Fair values are computed in euros and converted to dollars at 1.16249. A naive $1,711.89 ÷ EUR 27.55 gives 62.1x rather than 53.4x, a 16.3% error, and it is the single most likely mistake anyone will make on this name.
• Filing-over-vendor corrections (detailed in Section 8): FY2025 buyback — vendor cash flow EUR 5,715.0M, 20-F note 22 EUR 5,950.0M for 8,323,320 shares; we use EUR 5,950.0M (FY2024: vendor EUR 522.3M, filing EUR 500.0M for 574,925 shares). FY2025 seg_prod missing the installed-base line entirely — reconstructed at EUR 8,193.0M from the reported total and confirmed twice against the 20-F's disclosed +26.2% service growth and +12.4% system growth. seg_prod.reportedCurrency mislabelled USD for FY2024 and all earlier years when the data are EUR. seg_geo unallocated residuals of EUR 524.3M (FY2025, 1.6%) and EUR 1,322.1M (FY2024, 4.7%). Vendor profile.lastDividend of 8.35043 rejected against the 20-F's declared EUR 7.50 total 2025 dividend. Backlog (EUR 38.8B), remaining performance obligations (EUR 46.5B, 65% within twelve months), unit shipments (4 EXE, 44 NXE, 279 DUV), customer concentration (four customers, 61.2% of net sales) and all FY2026 guidance (EUR 34-39B → EUR 36-40B → EUR 43-45B) exist only in the filings.
• Vendor-internal rejections: est.ebitdaAvg, est.ebitAvg and est.netIncomeAvg rejected across all years (FY2027, FY2029 and FY2030 show net income above operating income, which is impossible; FY2030 by 39%) — all forward work uses epsAvg and revenueAvg only; priceToFairValueTTM rejected as identical to priceToBookRatioTTM to eight decimals; payablesTurnoverTTM, averagePayablesTTM, daysOfPayablesOutstandingTTM and therefore cashConversionCycleTTM rejected as unpopulated; interestCoverageRatioTTM, debtServiceCoverageRatioTTM and shortTermOperatingCashFlowCoverageRatioTTM rejected as undefined (EUR 0 interest expense reported); grahamNumberTTM and grahamNetNetTTM not used; bal_a.netDebt of −EUR 10.203B corrected to net cash of EUR 8.824B, by including the EUR 405.7M of short-term investments the vendor omits AND by replacing the vendor's EUR 2.708B of total debt with the 20-F's disclosed EUR 3.8B of Eurobond principal plus EUR 691.7M of commercial paper — a filing-over-vendor correction of roughly EUR 1.78B on the debt line, with shortTermDebt of EUR 0 and capitalLeaseObligations of EUR 0 identified as the specific defective fields (Section 4); km_ttm.enterpriseValueTTM of EUR 562.879B implies EUR 4.688B of net cash against the filing-corrected EUR 8.824B, a EUR 4.14B unreconciled gap that cannot be resolved because no post-2025 balance sheet exists in this file — both figures are reported, the difference is 1.05% of market capitalisation, and no conclusion depends on it; tech 52-week extremes ($1,989.44 / $689.63) differ slightly from quote ($1,999.96 / $684.24), and we use quote for the range and tech for the −14.0% computed drawdown.
• Disclosure-cadence note: ASML is a foreign private issuer filing 20-F and 6-K, not 10-K and 10-Q. This folder holds four documents, not six: 20-F 2026-02-25, 6-K 2026-04-15 (Q1'26 results), 6-K 2026-04-23 (AGM results) and 6-K 2026-07-15 (Q2'26 results). All three 6-Ks are cover pages only — exhibits 99.1 through 99.4, which carry the press releases, presentations, summary US GAAP consolidated financial statements and the statutory interim report, were not extracted. There is therefore no Q1'26 or Q2'26 balance sheet, no interim cash-flow statement, and no quarterly net system bookings figure available to this dive. Quarterly granularity is materially thinner than for the US-listed peers in this batch and that is a real limitation on the analysis, stated rather than concealed.
• Insider note: the insider block is EMPTY. Foreign private issuers' officers and directors are exempt from Section 16 reporting, so no Form 4 data exists to compile. No insider signal is available in either direction, which is an informational disadvantage rather than a neutral finding, and it contributes to the risk score of 5.
• Basis note: consensus EPS tracks reported GAAP diluted EPS closely — FY2025 EUR 24.77 estimate against EUR 24.71 reported, FY2024 EUR 19.29 against EUR 19.24, FY2023 EUR 19.11 against EUR 19.54 — so the forward multiples in this dive are effectively on a GAAP basis and are not flattered by an adjusted-earnings convention. earn_cal, by contrast, is USD-denominated at an implied period-average rate of roughly 1.144 and must never be compared with the euro-denominated est block.
• Estimate coverage: deep in the near years, thin in the far ones — 22 analysts on FY2026 EPS, 24 on FY2027, 27 on FY2028, but only 12 on FY2029 and FY2030. FY2028 EPS dispersion is 50% (EUR 46.31 to EUR 80.13). The FY2027 row is the primary forward anchor for all three fair values; no conclusion rests on the FY2029 or FY2030 rows, which are shown only to illustrate the deceleration path underpinning the multiple-compression assumption in Section 6b.
• Peer note: the vendor peer set (AMAT, AMD, CSCO, IBM, KLAC, LRCX, MU, QCOM, SAP, TXN) contains no multiple data and is not comparable line for line. The analytically relevant comparison is the wafer-fab-equipment group — ASML, LRCX, KLAC and AMAT are one trade with four expressions — and today's closing moves (ASML +4.22%, AMAT +5.48%, KLAC +6.95%, LRCX +7.85%, MU +7.62%) are used only as batch context, not as valuation inputs.
• Fair-value caveat: the $1,216 / $1,884 / $2,328 anchors are 20x / 31x / 38.3x FY2027 consensus EPS of EUR 52.28, computed in euros (EUR 1,046 / EUR 1,620.68 / EUR 2,002) and converted at the implied rate of 1.16249, cross-checked against FY2028 consensus EPS of EUR 67.24 (15.5x / 24.1x / 29.8x). Stated multiple arithmetic, not a discounted cash flow. The base case assumes mild multiple compression on growth deceleration and is explicitly NOT reachable at a constant forward multiple — at an unchanged 28.2x FY27E the euro price is EUR 1,474, which is where it already trades. The bull case at 38.3x is essentially the street's own target (37.9x) and requires no de-rating at all.
• Timing: Q2'26 was reported 2026-07-15, twenty days before this dive, so the most recent quarter is fully in the income-statement data (though not in the balance sheet — see above). The next print is 2026-10-14, 71 days away, and a second dated catalyst falls on 2026-11-10, 98 days away, when the suspended US Affiliates Rule and China's suspended rare-earth export restrictions both expire. Both dates are explicit verdict inputs.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.