{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreieitlz6arc3umtyhzt2sonflnhikqbvb2763mnuwthq4l56rn6lnm",
    "uri": "at://did:plc:ls6rbbwjyqeakittcsi3k6x3/app.bsky.feed.post/3mjlxs5w63762"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreifnnl6orrctlkppj5l53y6rjszzmrsrngfxi3vqnqagg6zv7dt72a"
    },
    "mimeType": "image/png",
    "size": 312776
  },
  "description": "When ASML raises its FY26 revenue midpoint by €1.5bn yet leaves the gross margin range untouched, what is the company quietly telling us about where the next leg of demand is actually coming from — and why is the market reading this as a soft beat?",
  "path": "/asml-1q26-earnings-review-buying-tomorrows-capacity-with-todays-margin/",
  "publishedAt": "2026-04-16T08:15:34.000Z",
  "site": "https://www.jasonandjarvis.org",
  "tags": [
    "Open this more visual friendly version in a new tab/点击跳转查看原文,左上角切换中文",
    "Subscribe now"
  ],
  "textContent": "Subscribe\n\n# ASML 1Q26: When \"AI-Driven\" Stops Being a Talking Point\n\nASML's earnings days carry a particular kind of tension. The company is both the master switch for global semiconductor capex and one of the few willing, every single quarter, to lay its full-year guidance on the table for the market to scrutinize. The 1Q26 scorecard: revenue of €8.77bn, gross margin of 53.0%, basic EPS of €7.15 — all three core metrics ahead of Bloomberg consensus, with EPS beating by +7.5%. More importantly, the company lifted its FY26 revenue guidance from €34bn–€39bn to €36bn–€40bn, a midpoint increase of €1.5bn (+4.1%), translating into YoY +20%.\n\nThis should have been a clean victory script. Yet the share-price reaction on the day was tepid. The reason hides in a single detail: **the gross margin range was kept unchanged at 51%–53%**.\n\nThat is where this report becomes worth your time. A guidance move that lifts revenue but refuses to lift gross margin — paired with a CEO whose conference-call refrain was \"We do not want EUV to be the bottleneck\" — frames the real narrative tension of 1Q26. **ASML is using an almost deliberate restraint to sketch a 2026 and 2027 that are stronger than the market expected, but harder to extrapolate in a straight line.**\n\n## Where, Exactly, Is the Upgrade Coming From?\n\nBy habit, the market asks first: is China sweeping the shelves again? The answer is no. CFO Roger Dassen reiterated in the Q&A that FY26 China share remains at the ~20% midpoint — perfectly aligned with the 19% China ship-to share in Q1, and the lowest level since 1Q23. **The guidance raise has nothing to do with China.**\n\nThe actual driver was buried in a fairly plain sentence during the Q&A:\n\n> \"Previously we were expecting [non-EUV] to be flat in comparison to last year. Right now, what we're looking at is in fact an increase of demand there as well.\"\n\nTranslated into investment language: non-China Immersion DUV has been upgraded from \"flat\" to \"growing.\" Bernstein's decomposition shows that even on the conservative assumption that China DUV revenue stays flat through 2026–27, non-China DUV can still grow +30% YoY, with overall DUV carrying +17% upside.\n\nASML DUV sales: China vs Non-China (Bernstein)\n\nsource: Bernstein ASML 1Q26 review\n\nThe shift shows up directly in the product mix. Q1'26 lithography systems shipped totaled 79 units — 16 EUV, 17 ArFi, 30 KrF. EUV beat consensus by +17%, KrF by +12.5%. Only ArFi units came in below consensus — but management explained the reason cleanly on the call: 2024 Immersion demand was meaningfully soft, and the supply ramp takes time to catch back up. \"That has now reversed itself.\"\n\nIn other words, Immersion supply is chasing pent-up demand; EUV, propelled by AI, set a fresh quarterly revenue record (>€4.1bn in EUV system sales).\n\n## Korea and Memory: A Customer Mix Redrawn by the \"Perfect Storm\"\n\nThe most structurally meaningful change in the report is the inversion of end-use and regional mix.\n\nOn the end-use axis, Memory's share of net system sales jumped from 30% in Q4'25 to 51%, while Logic dropped from 70% to 49% — a 21ppt quarter-over-quarter swing. Geographically, South Korea climbed from 22% in Q4'25 to 45%, an all-time high. Taiwan rose to 23%, China fell to 19%, and Japan was zero for the quarter.\n\nNet system sales by region (Bernstein)\n\nsource: Bernstein ASML 1Q26 review\n\nThis dovetails neatly with what sell-side preview notes had flagged: an SK Hynix EUV order of roughly €8bn, and Samsung ordering 20 EUV units for its P5 fab. But the more telling quote belongs to CEO Christophe Fouquet, relaying what customers told him on the call:\n\n> \"Many customers have confirmed that they are sold out for the remaining of the year, and that they expect the supply limitation to persist beyond 2026.\"\n\nHe went further, describing DRAM as \"a bit the perfect storm for ASML\" — memory price strength is fattening customer P&Ls and funding capex; 2025 is the breakout year for DRAM EUV adoption; and EUV displacing multi-patterning frees up scarce fab footprint. Three forces firing at once, all in ASML's favor.\n\nThe fair counter-argument: Q1'26 Memory weight is partly an order-fulfilment pulse, and the mix should normalize over the coming quarters; South Korea at 45% may not be repeatable either. But even taking those as a single-quarter peak, the conclusion that \"Memory customers have entered a multi-year EUV capex phase\" looks largely confirmed by this set of numbers.\n\n## The Restraint on Gross Margin, and ASML's Capacity Philosophy\n\nBack to the question this piece opened with: why raise revenue and leave gross margin alone?\n\nThe CFO's explanation in the Q&A was disarmingly direct. Of the 53% gross margin in Q1 (the upper end of the guidance range), **a portion came from one-off, very-high-margin components inside the Installed Base business — specifically the PEP-E qualification and upgrade work on NXE:3800E**. That piece is not repeatable. Q2 gross margin guidance therefore steps back from 53% to 51%–52%.\n\nAs for why FY26 gross margin was not lifted alongside the richer Immersion mix, the answer was more structural:\n\n> \"Every ramp, you always have a little bit of cost before you have the benefit of that.\"\n\nThe accelerated move-rate ramp requires hiring and training costs to land first. The mix tailwind from rising Immersion is offset by those front-loaded costs. Morgan Stanley said outright that this was the main reason for the muted share-price reaction on the day.\n\nUnderneath sits one of ASML's core capacity philosophies. The CEO repeated the line several times on the call:\n\n> \"We do not want EUV to be the bottleneck.\"\n\nTo make sure they never become that bottleneck, ASML is pushing on five fronts in parallel — availability, productivity, unit numbers, capacity, upgrades. Low-NA target capability is 90 units; Deep UV is 600. NXE:3800E throughput was lifted from 220 WpH to 230 WpH via the PEP-E package, and \"most of the time, by the way, with a software switch and some qualification\" the upgrade is delivered to all customers immediately.\n\nNXE:3800E PEP-E improvements\n\nsource: ASML 1Q26 earnings call deck\n\nFrom a valuation lens, this restraint pays off over the medium term. Bernstein notes that EUV ASP correlates very strongly with mixed throughput — and management confirmed in the Q&A that the ~4% productivity bump from NXE:3800F's 250 WpH to 260 WpH translates into a roughly proportional ASP uplift.\n\nEUV ASP–throughput correlation (Bernstein)\n\nsource: Bernstein ASML 1Q26 review\n\nASML, in other words, is trading this year's OpEx for next year's and the year-after's ASP — a choice that is friendly to long-duration shareholders, and rather unsexy to anyone hoping for an immediate EPS upgrade.\n\n## 2027 and 2028: From a Capacity Pledge to an Air Pocket\n\nThe most forward-looking piece of the report is not the FY26 guidance, but a single line on FY27: \"at least 80 units of Low-NA EUV.\"\n\nASML shipped 44 units of Low-NA EUV in 2025. The CEO leaned heavily into the 80-unit number:\n\n> \"What we did in 2025 is what we could do in basically 2027 with the at least 80 system. We more than double basically the total EUV capacity we can ship to our customer. This is done basically in about two years ... We do that, I would say, without blinking too much.\"\n\nThe \"doubling\" works because those 80 units in 2027 will leave the factory at the higher 230 WpH throughput — each tool delivering nearly twice the wafer output of a 2025 unit. That is a clear upward revision against Goldman Sachs' prior consensus of 72 units, and it implies further ASP improvement in 2027 as the D model approaches zero and the F model begins shipping in small volumes.\n\nEUV revenue and shipments (Bernstein)\n\nsource: Bernstein ASML 1Q26 review\n\nBut the 2028 story does not extrapolate cleanly. BofA flagged a specific air-pocket risk: as DRAM customers transition from current nodes to the 4F2 architecture, EUV shipments could see a temporary dip in 2028. BofA's model has Low-NA EUV easing to 75 units in 2028, and within EUV's end-use split, DRAM falls from a 2027 peak of 36 units back to 23 in 2028.\n\nThat air pocket is precisely why management deliberately framed the call around \"2026 solidified, 2027 under active discussion, 2028 still too far away.\" This isn't evasion — it is honesty. The company can credibly see customer capex shapes through 2026–2027; 2028 depends on the actual pace of node transitions like 4F2, and on whether High-NA can accelerate adoption in real product wafers across logic and DRAM.\n\nHigh-NA itself deserves a dedicated note. As of 1Q26, the platform has cumulatively exposed more than 500,000 wafers, with availability of 80%. The CEO relayed what Logic and DRAM customers were saying: \"going from 3 to 1 mask for EUV using High NA ... can reduce the number of process steps from 100 to 10.\" If this single-expose path is validated on real product wafers before 2027, the 2028 air pocket could plausibly be filled by an earlier-than-modeled High-NA ramp.\n\n## Export Controls: Uncertainty Built Into the Bandwidth\n\nThe unavoidable subject is China and the MATCH Act. The CEO handled it as a technical problem — explicitly noting that the FY26 €36bn–€40bn guidance bandwidth \"accommodates potential outcomes of ongoing discussions around export controls.\" The potential restriction is not deducted from the midpoint; it is absorbed into the width of the range.\n\nMorgan Stanley judges MATCH Act implementation probability as low, with the earliest impact materializing in late 2026. BofA treats the risk as \"already embedded at the midpoint,\" and views any post-meeting delay or relaxation following the Trump–Xi meeting in mid-May as outright upside. This is a textbook case of management refusing to forecast geopolitics, while quietly buying insurance through guidance width.\n\nFor the buy side, the framing is helpful: when actual policy outcomes arrive, the market won't have to relitigate \"should the guidance change\" — guidance has already absorbed the optionality.\n\n## An Underrated Detail: The Quiet Repricing of Installed Base\n\nA final detail worth real attention: the Installed Base Management business.\n\nQ1'26 IBM sales of €2.49bn (YoY +24.3%) beat consensus by +5.3% — what BNP Paribas called the most impressive line in the quarter. Behind this sits a long-underappreciated story. The CFO mentioned in the Q&A that 4–5 years ago, EUV service margins were negative — his exact phrase was \"bleeding money.\" Today, EUV service gross margin is \"no longer very much apart from ... the corporate gross margin\" — i.e., **it has quietly converged to corporate-level gross margin**.\n\nASML, in other words, has completed a silent repricing of its EUV service profitability. Layered on top of high-margin software-based upgrades like PEP-E, the IBM line is evolving from a \"wraparound service\" into a profit engine with its own narrative. As Low-NA EUV installed base scales from a few hundred tools today toward the 2027 run-rate of at least 80 units per year, the compounding effect on IBM gets meaningfully larger.\n\n## Back to the Question\n\nASML lifting the FY26 midpoint by €1.5bn while leaving gross margin untouched is, on its own, telling the market two things.\n\nFirst, **the company is choosing to spend this year's incremental profit to make sure it isn't the bottleneck next year or the year after**. Front-loaded hiring and training costs from the move-rate ramp are the price of delivering the at-least-80-unit 2027 plan and the €44–60bn 2030 revenue vision. To long-term shareholders the trade is unambiguously positive; to quarter-by-quarter traders it reads as \"upside swallowed.\"\n\nSecond, **the real demand strength is coming from AI-driven Memory customers entering an EUV capex super-cycle and from non-China Immersion staging a comeback — not from China stockpiling**. That structural shift makes ASML's demand quality healthier than in 2024: customers are more diversified, use cases are longer-dated, and order visibility is meaningfully higher.\n\nAs for the 2028 air pocket, the MATCH Act tail, and the pace of High-NA adoption — these are real questions, and all of them will get clearer answers from new data points over the next 12–18 months.\n\nBut what 1Q26 already delivers is a fairly crisp medium-term narrative: in a world where AI capex continues to accelerate, **the only company capable of defining the upper bound of leading-edge semiconductor capacity is choosing \"do not be the bottleneck\" as its single most important KPI.**\n\nASML 1Y forward P/E (Bernstein)\n\nsource: Bernstein ASML 1Q26 review\n\n### Earnings Call Recap\n\n\n   Open this more visual friendly version in a new tab/点击跳转查看原文,左上角切换中文\n\n\nI will now share a more detailed version of this earnings review behind the paywall\n\n### This post is for subscribers only\n\nBecome a member to get access to all content\n\nSubscribe now",
  "title": "ASML 1Q26 Earnings Review - Buying Tomorrow's Capacity With Today's Margin",
  "updatedAt": "2026-04-16T08:15:35.235Z"
}