On June 12, 2026, ASML (ASML.AS, Euronext Amsterdam) closed at EUR 1,629.60 — up 65.2% year to date, with the close itself a 52-week high (Yahoo Finance). What the market is paying for the world’s only EUV lithography equipment maker is the net of two ledgers.
The first ledger is AI. Net system bookings reached a record EUR 13.2 billion in the fourth quarter of 2025, EUR 7.4 billion of it EUV; year-end backlog stood at EUR 38.8 billion, roughly 1.19 times fiscal 2025 revenue, with visibility covering all of 2026 and stretching into 2027. First-quarter 2026 revenue of EUR 8.8 billion beat market expectations, and the company raised full-year guidance from EUR 34–39 billion to EUR 36–40 billion. The second ledger is China. China’s share of system sales fell from about 41% for full-year 2024 (with quarterly peaks near 49%) to about 33% in 2025 and to 19% in the first quarter of 2026; the MATCH Act, introduced in the US Congress on April 2, would pull DUV immersion tools and services to China into the control regime as well.
Beyond the two ledgers sits a third variable: High-NA EUV, the next-generation flagship at roughly USD 380 million per tool. Intel has put its first unit through acceptance testing and ordered a second; Samsung received its first, with a second reported arriving in the first half of 2026; TSMC has said plainly that it will not adopt High-NA at the 2nm node. The second curve currently has a customer list of two — and the more eager of the two is itself in deep restructuring.
What this piece takes apart is the question hanging over this year’s entire European semiconductor rally, in its ASML form: of the 65% year-to-date gain, how much pays for cash flows already written into contracts, and how much is option premium on two propositions — that AI capex keeps filling the China gap, and that High-NA eventually goes mainstream.
The two ends of the 52-week range tell two different stories. The low of EUR 592.90 (closing basis) came in the first half of 2025, when export-control headlines sat on the stock; the high is the June 12 close of EUR 1,629.60, roughly 2.7 times the low (Yahoo Finance). Same company, same monopoly — within twelve months the market priced it first on geopolitical risk, then on AI orders.
On the timeline, the two ledgers collided head-on only once. On January 28, the company reported fiscal 2025 results and the record fourth-quarter bookings; on February 25, the 20-F went to the SEC, with fiscal 2026 guidance of EUR 34–39 billion. On April 2, the MATCH Act — the Multilateral Alignment of Technology Controls on Hardware Act — was introduced in the US Congress, proposing to restrict DUV immersion tools and services to China; per CNBC reporting, ASML fell about 3.4% in a single day on the news. Then came April 15: first-quarter revenue of EUR 8.8 billion, a 53.0% gross margin, net income of EUR 2.8 billion, and full-year guidance raised to EUR 36–40 billion. A week later, the annual general meeting approved a full-year dividend of EUR 7.50, up 17% year over year, and authorized a new buyback program. Since then, the second ledger has left no further mark on the tape.
Where the pricing lands: a market capitalization around EUR 475.8 billion (GuruFocus / StockAnalysis approximations), a trailing price-to-earnings ratio of roughly 55–59x (sources differ on basis; see the blind-spot section), a forward multiple of roughly 40–46x (GuruFocus, June 3, 2026), and a trailing multiple about 62% above its own historical average (CompaniesMarketCap). Guidance midpoint of about EUR 38 billion implies roughly 16% growth over fiscal 2025 — paired with a trailing multiple in the mid-50s, the market is plainly buying more than 2026. Our earlier European semiconductor research covered the power and analog layer represented by STMicroelectronics and concluded that the cycle bottom was real but the recovery already priced. ASML is the other pole: not a cyclical-repair story but a rerating of monopoly stacked on AI visibility. Asking where the bottom is would be the wrong question here. The question is what is packed into the multiple.
ASML is the world’s only maker of EUV lithography systems, and its customer list is the full roster of advanced logic and memory: TSMC, Samsung, Intel, SK hynix, Micron. On the DUV side it faces Canon and Nikon, but it dominates the high-end immersion segment as well. Fiscal 2025 R&D spending ran to about EUR 4.7 billion, roughly 14% of revenue — the monopoly is no static license; the position has to be repurchased every year out of the R&D budget. The workhorse platform keeps iterating too: the production-grade NXE:3800E runs 230 wafers per hour, up 44% from the NXE:3600D’s 160.
The monopoly expresses itself financially on two levels. The first is system sales: fourth-quarter 2025 net bookings of EUR 13.2 billion, of which EUR 7.4 billion was EUV and roughly EUR 5.8 billion non-EUV (a derived figure); year-end backlog of EUR 38.8 billion, about 1.19 times fiscal 2025 revenue — on the company’s framing, order visibility covers all of 2026 and extends into 2027, driven by AI-infrastructure-led expansion of advanced nodes. The second level is the annuity: installed base management (service) revenue once tools are on the line — EUR 2.5 billion in the first quarter of 2026 alone. As long as the installed base keeps running, that money depends on no new order.
The re-acceleration is written in the annual accounts: fiscal 2023 revenue of EUR 27.6 billion; fiscal 2024, EUR 28.3 billion, up just 2.5%; fiscal 2025, EUR 32.7 billion, up 15.6%. Gross margin moved from 51.3% to 52.8%, and net income from EUR 7.8 billion to EUR 7.6 billion to EUR 9.6 billion. The first quarter of 2026 extended the slope: revenue of EUR 8.8 billion, with system sales of EUR 6.3 billion (EUV accounting for EUR 4.1 billion) and service at EUR 2.5 billion, a 53.0% gross margin, and net income of EUR 2.8 billion. Bookings, backlog, raised guidance — the defining feature of the first ledger is that most of it is not expectation. It is signed contract.
China’s share of ASML’s system sales: about 41% for full-year 2024, about 33% in 2025, 19% in the first quarter of 2026, with management guiding around 20% for the full year. A market that once topped forty percent of the book — nearing half in peak quarters — has shrunk to less than a fifth in two years. Two numbers need separating here. Roughly EUR 8–10 billion a year is the peak annual size of the China business in its forty-percent era — it measures how big China once was, not how much has been lost. On management’s roughly 20% guidance, China is still contributing in 2026; what other regions actually have to replace is the stretch from about forty percent down to about twenty — on the order of half that peak volume.
So far, it has been replaced — and this is the most easily overlooked fact in the second ledger. Fiscal 2025 revenue grew 15.6% over the very stretch in which China’s share fell from 41% to 33%; in the first quarter of 2026, with China down to 19%, the company still beat expectations and raised guidance. What filled the gap is spelled out in the quarterly accounts: of EUR 6.3 billion in first-quarter system revenue, EUV was EUR 4.1 billion. What was lost was the DUV China could buy; what replaced it was the EUV China cannot — and gross margin rose from 51.3% to 53.0% along the way, so the substitution cost nothing in profitability.
But the phrase “monopoly pricing power fills the geopolitical gap” needs to be taken apart. Pricing power governs margin: customers have no second EUV supplier, so negotiation happens over delivery slots and configuration. Pricing power does not govern demand itself — a tool that export licensing will not release cannot be priced into revenue at any level of power. What actually filled the gap over the past several quarters was not pricing power; it was AI capex erupting in the same window. The first ledger funding the second rests on one premise: that advanced-node expansion continues at current intensity. That premise is underwritten by backlog through 2027. Beyond that, it is an assumption.
The MATCH Act threatens the part that has not yet been lost. The bill’s proposed scope — DUV immersion tools and services to China — is itself a map of what remains: the residual China business sits in DUV and in servicing the installed base. The service line deserves separate attention: installed base management is an annuity-type revenue stream running at EUR 2.5 billion a quarter globally, and if controls extend to maintaining already-installed tools, the damage lands not on one-time system sales but on standing cash flow. How deep that cut would go has no public quantification (see the blind-spot section); the roughly 3.4% single-day drop in April remains the only price the market has ever put on it. Nor is legislation the only threat to what remains: China’s DUV stockpiling rush of 2023–2024 left an inventory-digestion phase behind it, and the share could drift lower without any new bill — the roughly 20% guidance is not necessarily a floor.
High-NA (0.55 NA) EUV is ASML’s next-generation flagship. The first production-specification model, the EXE:5200B, runs about USD 380 million per tool (roughly EUR 350 million, per TrendForce and multiple media accounts); standard EUV, by comparison, runs about USD 180–200 million — one tool for the price of two. The technology itself is not the slow part: the first full-specification tool was delivered at the end of 2025, customers have cumulatively run more than 400,000 wafers, and the company plans to ship the first second-generation High-NA system within 2026.
The slow part is the customer list. Intel is the most committed buyer: its EXE:5200B passed acceptance testing in December 2025, supports 14A process development, and a second unit has been ordered. Samsung received its first tool at the end of 2025, with a second reported arriving in the first half of 2026, aimed at 2nm foundry work and next-generation memory (for the structure of memory capex, see our earlier research on Korea’s memory industry). SK hynix is reportedly watching, with no confirmed order. And the customer that matters most has answered in the negative: TSMC has said it will not adopt High-NA at the 2nm node and plans to introduce it from 1.4nm — its judgment being that 2nm can be reached with standard EUV and multi-patterning.
That is the heart of the economics dispute: a USD 380 million tool pays for itself only when single-exposure replaces enough process steps and the process-window gains are large enough. TSMC’s choice amounts to a public statement that at 2nm, the math does not clear. Near-term High-NA demand therefore concentrates in two customers — and the more committed of the two, Intel, is itself in deep restructuring through 2025–2026, with layoffs and repeatedly revised fab investment plans. If the 14A timeline slips, the single largest engine of early EXE orders decelerates with it.
Put back into this piece’s frame: High-NA today is not a second revenue curve. It is an option — exercisable when advanced nodes broadly discover, in the 1.4nm generation, that multi-patterning no longer gets them through. The option itself is real, and the physics roadmap leans ASML’s way. The question is only how much certainty a trailing multiple of 55–59x has already assigned to it.
This piece has leaned skeptical so far, so by our convention the bull case gets written at full strength.
First, backlog is not expectation; it is contract. An order book of EUR 38.8 billion — about 1.19 times annual revenue — fills 2026 and reaches into 2027. In a year when the market loses sleep over whether AI capex breaks, ASML is one of the few companies that had the answer signed in advance. One boundary on this plank: the contract protects against default, not against the calendar — industry practice in equipment is to reschedule orders rather than cancel them, a pushed-out delivery breaches nothing, and the protection a backlog offers can shrink without a single order being cancelled.
Second, the filling of the China gap is not a promise; it is something that has already happened once. Across the entire slide from 41% to 19%, revenue rose (up 15.6% in fiscal 2025), gross margin expanded, and guidance went up. Of the bear’s worst-case scenario — China going to zero while AI recedes — the first half has largely played out already, and the accounts show no crack.
Third, the service annuity only compounds. Every delivered tool joins the installed base, and the EUR 2.5 billion-a-quarter service line grows with that stock, independent of any new order.
Fourth, the High-NA option is carried at the customers’ expense. The R&D is absorbed by a budget of about EUR 4.7 billion a year, and the tools sell at USD 380 million apiece — holding the option costs ASML close to nothing net. TSMC’s “not at 2nm” is simultaneously “at 1.4nm”: deferral, not cancellation. The second-generation machine ships in 2026 regardless; the roadmap has not paused for customer hesitation.
Fifth, capital returns record management’s own judgment: a EUR 7.50 full-year dividend, up 17%, plus a new buyback authorization — this is not a company hoarding cash against a demand cliff.
Sixth, the MATCH Act is a bill, not a law. Export controls have tightened round after round in recent years, and ASML grew revenue from EUR 27.6 billion to EUR 32.7 billion against exactly that backdrop. The 52-week low of EUR 593 is itself a specimen of the last controls panic — a pricing the market later overturned on its own.
Each of these six planks holds on its own; the weakness they share is the multiple, which none of them resolves. Every one argues that the earnings base keeps growing underneath the price. None of it argues that 55–59x trailing — about 62% above the company’s own historical average — is itself cheap. The bear does not need to deny the monopoly. The bear only needs any one of the options — the continuation of AI capex, the mainstreaming of High-NA, the shelving of the legislation — to pay off one quarter late.
The first-quarter 2026 net bookings figure did not make it into this piece’s data collection — the largest blind spot here: how the order cadence runs after EUR 13.2 billion cannot be quantified in this piece and awaits the quarterly filings. The fourth quarter’s EUV versus non-EUV revenue split is missing, and within the bookings split the EUR 5.8 billion non-EUV figure is derived rather than reported. Fiscal 2025 operating profit in absolute terms was not extracted from the 20-F statements, so profitability is discussed here through gross margin and net income. Trailing price-to-earnings readings span roughly 54.5 to 58.6 across sources on differing bases (IFRS and US GAAP earnings per share differ slightly), so this piece gives a range and no precise figure. There is no official quarter-by-quarter company disclosure of cumulative High-NA orders; the Intel and Samsung unit counts come from TrendForce and other media reporting and may understate the total. This piece carries no AEX or Philadelphia Semiconductor Index comparison, so the 65.2% year-to-date gain lacks a relative coordinate. The final gap bears directly on the MATCH Act: there is no public quantification of the China exposure inside DUV service and installed-base revenue — the second ledger’s largest tail risk currently cannot be weighed with precision.
This report is an independent KSINQ market observation for informational purposes only. It is not investment advice. Data snapshot: June 13, 2026; price data as of the June 12, 2026 close.