Record €11.7B Quarter at 57.5% Margin, Q4 Guided to €14.2-15.2B, Price Increases Signed and 2028 Capacity Under Study Above +30%. The Semicap Tape Sold It Anyway. Maintaining Outperform.

Results vs. Consensus

Q3 2026 Scorecard (EUR)

Year-over-Year Comparison (Q3 2026 vs. Q3 2025)

Sequential Comparison (Q3 2026 vs. Q2 2026)

Revenue assessment. The quarter landed at €11.66B against a €11.0-12.0B guide, toward the high end, which is precisely the place it needed to land for the year to work. The composition is the more important read. Net system sales (+27.5% QoQ) grew faster than Installed Base (+19.1% QoQ) for the first time this year, because 121 litho units shipped against 91 in Q2: the physical output ramp, not upgrades, carried the sequential step. EUV system sales rose to €4.1B (from €3.8B) including one High-NA system, and non-EUV jumped to €4.3B (from €2.8B), edging above EUV in the quarter. End-market mix moved back toward logic, 54% against memory's 46%, from a near-even 51/49 in Q2.

Margin assessment. Gross margin of 57.5% is a record and 590bp above the year-ago quarter. Management attributes the beat to very high-margin upgrade components, and the arithmetic agrees: Installed Base is the only line that beat guide, and its upgrade component carries margins well above the corporate average. The quality concern is that this is the line management itself calls the swing factor for 2027. Operating margin of 44.4% expanded 730bp sequentially on 25.0% revenue growth and slightly lower operating expenses (€1,527.0M against €1,579.3M in Q2). The Q4 gross margin guide of ~57% steps down about 50bp despite roughly 26% more revenue, which management attributes to ramp costs for the 2027-2028 capacity expansion. That is the first time in this cycle that capacity investment has shown up as a margin headwind, and we treat it as the price of the next leg rather than a warning.

EPS assessment. Reported basic EPS of €12.95 is +135.9% YoY and +70.6% QoQ, and roughly €1.33 of it is the Mistral gain. Excluding that gain, and assuming it is untaxed as the tax-rate arithmetic implies, net income is about €4,450.7M and EPS about €11.62, still +112% YoY and +53% QoQ. Against the €10.76 consensus, reported EPS beat by 20.4% and ex-gain EPS by about 8.0%. The consensus almost certainly excluded the Mistral gain, so 8% is the honest beat, and income from operations of €5,172.8M, which excludes the gain entirely, is the cleanest single measure of the quarter. The share count fell 1.1% YoY and the Q3 buyback of €4.1B was nearly four times the Q2 pace, so the per-share number will keep compounding faster than net income into Q4.

Segment Performance

ASML does not report segment margins. The analysis below works from net sales composition, unit data, and the FY2026 growth rates management updated on the call. Several of those rates changed this quarter, and the direction of the changes matters as much as the totals.

Revenue Composition and FY2026 Growth Guidance

EUV and non-EUV quarterly figures are management's rounded call figures; the net system sales and total rows are the income-statement values.

End-Market Mix (net system sales)

EUV: the ramp is physical, and the price list changes from 2027

EUV system sales of roughly €4.1B rose about 8% sequentially and included one High-NA system, the same High-NA count as Q2. The full-year frame is unchanged: roughly 65 Low-NA systems in 2026, which the CFO put against 44 shipped last year, and EUV net system sales growth above 45%. What changed is the forward price list. Increases on what management calls the most critical products are now agreed, and the 2027 EUV mix drops the D models entirely, so both price and mix move in ASML's favour at the same time next year.

"For low NA EUV, anticipate output of roughly 65 systems for 2026, which is nearly 50% more than the 44 systems that we shipped last year." — Roger Dassen, CFO

Assessment: EUV is doing exactly what the July plan required, and the 2027 setup is better than it was in July because pricing is now contracted rather than discussed. 65 against 44 is +48%, so "nearly 50%" is the accurate framing (the video interview rounds it to "more than 50%"). The EUV line is no longer where the uncertainty sits.

Non-EUV: a big quarter inside a trimmed year

Non-EUV system sales jumped to roughly €4.3B from €2.8B, the largest sequential move of any line. Yet the full-year growth guide for non-EUV fell to ~20% from ~25% in July. Those two facts reconcile only if Q4 non-EUV steps back down from the Q3 level, or if the July figure was simply too generous. Management offered no explanation for the cut, and no analyst asked. The immersion picture improved in words (2026 shipments now similar to 2025, from a year that was once planned well below it), so the shortfall most likely sits in dry DUV or metrology timing, but that is our inference, not a disclosure.

"For DUV immersion systems, while our initial output target for these systems was well below the prior year, we now expect we will deliver a similar quantity in 2026 as we did in 2025." — Roger Dassen, CFO

Assessment: A quiet negative inside a positive guide. Five points of growth on a non-EUV base that cannot exceed 2025 net system sales of €24,474M is worth at most about €1.2B, and the Installed Base raise (about 20 points on 2025's €8,193M, or roughly €1.6B) more than covers it. But it is the one growth rate that went down, and it went down without comment.

Installed Base Management: the engine and the swing factor

Installed Base sales of €3,288.8M rose 67.6% YoY and beat the ~€2.9B guide by roughly €389M, after a roughly €260M beat in Q2. The full-year growth guide goes from above 30% to around 50%, and Q4 is guided to ~€3.7B, another 12.5% step. The mechanism is unchanged from July: customers who cannot get new tools fast enough buy productivity upgrades for the tools they own. New in Q3 is the CFO's framing of it as the 2027 swing factor for gross margin, with a two-sided argument: a still-large population of tools not yet upgraded, plus new upgrade products, against an obviously tough comparison.

"If you look at the number of upgrades that we have performed in the field, it is, you know, it is quite a few tools, but there are still quite a few tools out there that have not yet been upgraded." — Roger Dassen, CFO

Assessment: The highest-quality and highest-margin line in the business, and now the line most likely to make 2027 gross margin look disappointing against 2026. An Installed Base business running at ~€3.7B a quarter does not need to fall for margins to compress; it only needs to stop growing faster than systems. We model 2027 Installed Base growth well below 2026's ~50% and treat any repeat of this year's upgrade strength as upside.

Memory and logic: the mix rebalances

Memory's share of net system sales fell to 46% from 49%, and management for the first time quantified total logic growth for the year at ~10%, against advanced logic above 25%. The implication is that mainstream logic is shrinking this year. Memory growth is guided at ~75% (July: above 75%), mostly DRAM. On the forward view, management expects logic to outweigh memory in its 2027 order book, and said DRAM customers are currently the more aggressive builders.

"But I still believe that logic will outweigh memory in the in total book of business for next year." — Roger Dassen, CFO

Assessment: Helpful for the cycle debate. The bear case we carried in July was that memory, which has never been secular, was driving most of the growth. A 2027 book that leans back toward logic, at a time when DRAM customers are adding new EUV entrants, is a better-balanced demand base than the 49% memory share in Q2 suggested.

China: steady at ~20%, and back on the agenda

China remains guided at about 20% of 2026 net sales. Unlike July, when the topic did not arise at all, two analysts raised it. Management described intrinsic Chinese demand for mainstream logic and memory as strong, expects that to hold beyond this year, and dismissed the domestic-competition threat as a 25-year-old story with little factual progress. The export-control accommodation that management volunteered in April was again not restated.

"If you stick to fact, I think the fact this year are not different from the fact a year ago or even 5 years ago." — Christophe Fouquet, CEO

Assessment: Neutral. China is growing in line with the business, which is what it has done all year. The only change is that the forward China demand view is now explicitly constructive, which slightly raises the stakes on any future export-control tightening, a risk management again did not address.

Key KPIs

Implied DSO uses period-end net accounts receivable against quarterly net sales over 91 days. The Q3 buyback average price is our division of €4,071.0M by the ~2.8M shares disclosed.

Guidance & Outlook

Q4 2026 Guidance

FY2026 Guidance: What Moved

The guide now spells out the shape of the year rather than leaving investors to back-solve it. Nine-month net sales were €29,755.4M, so the €44-45B range implies Q4 of €14.24-15.24B, which is exactly the Q4 range given. The CFO framed the step-up as planned rather than stretched, and the video interview made the same point in plain terms.

"We said that we expect the second half to be stronger than the first half. You see that momentum really very, very strongly building up." — Roger Dassen, CFO

The CEO's framing was about duration rather than the quarter. The constraint, in management's telling, is no longer whether customers want the tools but how long the shortage lasts and how fast ASML can build.

"This extended visibility has led them to conclude that demand for their product will outpace supply through at least 2028." — Christophe Fouquet, CEO

Implied Q4 ramp: Q3 grew 25.0% sequentially, which was the first of the two roughly 25% steps the July guide required. Q4 at the €14.7B midpoint requires a further 26.0%. H2 at the midpoint is €26.36B, +45.7% on H1's €18.09B. The low end (€14.2B) still needs +21.8% sequentially. At ~57% gross margin and ~€1.6B of operating expenses, the midpoint implies roughly €6.8B of operating income and a ~46% operating margin in Q4.

Street at: Before the print, consensus sat at €14.19B for Q4 sales with a 55.4% gross margin, and at €43.6B for FY2026. The Q4 guide midpoint is 3.6% above that sales figure and its ~57% margin 160bp above; the low end of the Q4 range sits on the consensus. The FY2026 midpoint is 2.1% above. After the print, FY2026 consensus moved to €44.47B, essentially the new midpoint, and FY2027 to €57.21B of revenue and €55.93 of EPS, from €52.03 of EPS a week earlier.

Guidance style: This management team has now landed in the upper half of its own guide for three quarters running (Q1 at the top, Q2 above the high end, Q3 toward the high end) through the steepest ramp in its history. The Q4 guide carries a wide €1B range, which is honest for a quarter in which shipment timing on a handful of EUV tools moves the total by several hundred million euros. The new information is that the low end alone would be a record by 46%.

Key Topics & Management Commentary

Overall Management Tone: Management was as confident as in July and noticeably more specific: where July treated demand as settled, this call treated pricing, order acceptance and supply-chain readiness as settled too, and spent its energy on capacity beyond 2027. The one area of visible hedging was gross margin beyond 2026, where the CFO volunteered two headwinds (ramp costs and the upgrade comparison) before anyone asked. Tone was the most commercially assertive since our coverage began, with management openly declining to accept 2029-2030 orders until it had allocated capacity fairly.

1. The Back-Half Test, Leg One: Passed

In July we wrote that the whole FY2026 raise rested on two consecutive quarters of roughly 25% sequential growth, which this company had never delivered. Q3 delivered the first: €11.66B, +25.0% QoQ, toward the top of the €11.0-12.0B guide, on 121 litho units against 91. The margin beat was larger than the revenue beat, and every guided operating line landed at or below plan.

"Total net sales in the third quarter of 2026 were a record at €11.7 billion towards the high end of our guidance and reflecting strong execution of our capacity ramp." — Roger Dassen, CFO

The unit data is the part that matters most. In July the risk was manufacturing throughput, and 30 more litho units in one quarter (113 new against 86, 8 used against 5) is direct evidence of throughput, not mix or price.

Assessment: The single most important commitment from our July checklist is met. We said a Q3 print at the low end would put the year at risk; it printed toward the high end. Bear 4 (back-half execution) moves from EMERGING to CONTAINED, with Q4 still to deliver.

2. Q4 at €14.2-15.2B: The Record Quarter Is Now a Guide

The July guide implied a Q4 of about €14.4B at its midpoint, 48% above anything ASML had ever reported, and management did not walk through that arithmetic. This quarter the company guided Q4 directly at €14.2-15.2B with ~57% gross margin and ~€3.7B of Installed Base. The midpoint is about €0.3B above what we back-solved in July. Our July line in the sand was that a Q4 below €13B would damage the thesis; the low end of the guide is €1.2B above it.

"We expect Q4 total net sales to be between €14.2 billion and €15.2 billion with a gross margin of around 57%." — Roger Dassen, CFO

Assessment: Converting an implied number into a stated one is a meaningful signal from a company that guides conservatively. The residual risk is shipment timing at quarter end, which management addressed on the call by saying it allocates tools only where customer facilities are ready. A Q4 at the low end would still meet the full-year guide.

3. Pricing: From Free Option to Signed Contract

In July the CFO conceded unusual pricing leverage three times and dated it zero times. In October the conversations are finished. Price increases on what management calls the most critical products, the high-volume manufacturing tools, are agreed with key customers. Upgrade prices follow tool prices. Nothing lands in 2026, some of it shows up during 2027, and the full effect arrives from 2028.

"Conversations have been completed and, you know, have indeed led to pricing increases for what we call the most critical products for our customers. So that has been done." — Roger Dassen, CFO

Two qualifiers came with it. First, management insisted this is not an annual event, and framed it as a one-time reset of lithography's share of ecosystem value inside its value-based pricing model. Second, not all of it reaches gross margin: part of the increase funds the ramp, including help for suppliers that must expand capacity to match ASML's plan.

"So there is some portion of the increase, of the price increase that we need to cover for the extraordinary ramp that we are going through But I would say that is still a significant part of it is directly accretive to the gross margin." — Roger Dassen, CFO

Assessment: The largest change to the long-term model this quarter. We carried pricing as a free option outside our numbers in July. It is now a contracted lever with a dated profile, and we put a partial effect into FY2027 gross margin. We still decline to size the increase, because management declined to, and because some of it is earmarked for ramp cost. The "not every year" framing caps the upside but also protects the downside: management explicitly ties the model to never discussing price decreases in a downturn.

4. Capacity: Planned Output Will Not Meet 2027-2028 Demand

In July ASML planned +30% Low-NA EUV and immersion capacity for 2027 (about 85 Low-NA units) and was investigating another +30% for 2028. In October management said those plans will not fully meet customer demand in either year and is now studying a 2028 step above +30%. It gave no new unit numbers, and said the incremental opportunity is easier to capture in 2028 than in 2027.

"However, given the dynamics of this quarter, we now anticipate that our current planned levels of output in 2027 and 2028 will not fully meet the demand of our customers." — Roger Dassen, CFO

The levers are the same three as in July: using more of the existing footprint for output, cutting factory cycle time (including dropping qualification steps that customers now agree are unnecessary), and expanding the supply chain. Management said ZEISS is not a bottleneck even beyond +30% for 2028, and pointed to this year's roughly 50% EUV output increase as evidence that the supply chain can scale.

Assessment: Positive for ASML's revenue visibility and a sharper edge on our cycle-risk pillar. A supplier that says demand exceeds its planned capacity two years out is describing pricing power and order security. But the more capacity ASML adds into 2028, the more wafer capacity reaches a memory market whose shortage is the reason for the demand. We note the shift and keep Bear 5 at EMERGING rather than lowering it.

5. 2029-2030 Orders: Requested, Deliberately Not Yet Accepted

Customers are asking to place purchase orders for 2029 and even 2030. ASML has not accepted them. The stated reason is allocation: management wants a full picture of total 2029-2030 demand and of its own capacity before committing, so that early movers do not lock out long-standing customers. The CFO also described what an ASML purchase order costs a customer, which is the best answer yet to the double-ordering concern.

"1 serious commitment is a commitment to make a down payment, and as I qualified it before, it is a down payment that is enough to hurt." — Roger Dassen, CFO

Assessment: A monopolist rationing future capacity is not a normal position for a capital-equipment company, and it is the clearest evidence this quarter that the demand is real. The down-payment requirement turns order intake into cash before revenue, which is exactly what the Q3 balance sheet shows.

6. Down Payments Rebuilt the Balance Sheet in One Quarter

Free cash flow was €11,351M on operating cash flow of €11,797.6M. Non-current contract liabilities rose from €3,570.3M to €9,300.6M, and current liabilities rose a further €2,749.7M. That reverses the H1 picture we flagged in July, when H1 free cash flow was negative €1,291M against €5,674M of net income: nine-month free cash flow is now €10,060M against €10,634.7M of net income, about 95% conversion.

"Our free cash flow in Q3 was €11.4 billion, underpinning the strong order momentum of our customers and related down payments." — Roger Dassen, CFO

Receivables did not reverse. Net accounts receivable rose another €2.0B to €9,252.8M, and implied DSO edged up from 70.8 to 72.2 days. Inventories, by contrast, were essentially flat at €11,907.3M ahead of a 26% sequential step.

Assessment: Bear 6 (working capital and cash conversion) is resolved on the cash measure and still open on the receivables measure. We said DSO near 70 days without a collections offset would escalate the risk. The offset came, from down payments rather than collections, and at roughly €5.7B of new long-dated customer money it is larger than the entire H1 receivables build. We lower Bear 6 to CONTAINED and will keep watching DSO through the Q4 shipment wave.

7. Installed Base: The Beat Driver Is Also the 2027 Swing Factor

Installed Base beat its guide by almost €400M and drove the gross margin above guide. The CFO then did something unusual: he named it as the swing factor for 2027 gross margin before analysts did, setting a still-large pool of un-upgraded tools and new upgrade products against an exceptionally strong 2026 base.

"Installed base management sales for the quarter were €3.3 billion exceeding our guidance by almost €400 million primarily due to the strong demand for our productivity upgrade products in both EUV and DUV." — Roger Dassen, CFO

Assessment: We read the volunteered caution as expectation management for the January 2027 guide, not as a demand warning. It does, however, tell us how to model: upgrades are a growth business with a finite installed pool, not a permanent 50% grower.

8. Gross Margin: Record Now, Puts and Takes for 2027

Q3 printed 57.5%, the full year is guided to ~56% against 51-53% at the start of the year, and Q4 dips to ~57% on ramp costs. On 2027 the CFO declined a number (it comes in January) but listed the drivers: volume and fixed-cost coverage, a richer mix of immersion and Low-NA EUV, the price increase, and a better EUV mix as the D models leave and the first F models arrive. Against those: ramp costs and the upgrade comparison.

"And, of course, we are not going to give you a gross margin percentage for 2027. … So how much cost are we incurring to ramp not just, for 2027, but also for beyond?" — Roger Dassen, CFO

Assessment: Four structural tailwinds against two manageable headwinds. We model FY2027 gross margin at about 58%, up from our July range of 56-58%, with pricing contributing only partially in 2027. The Q4 dip is the honest version of the ramp story: margin leverage on volume is being partly reinvested in capacity, and we would rather see that than a company maximizing margin at the cost of the 2028 opportunity.

9. High-NA: Three Customers, Three Dates

High-NA moved further from qualification toward production. Intel has exposed more than one million wafers with performance at or above Low-NA on comparable layers. Samsung plans High-NA in high-volume manufacturing from 2028 and TSMC around 2030. Fleet availability is approaching 90%. ASML and TSMC also launched an industry initiative for 12-inch photomasks, which every major customer has joined, targeting a pilot line by 2031 and production in 2033.

"Additionally, Samsung and TSMC announced plans to introduce High-NA EUV into high volume manufacturing in 2028 and 2030, respectively." — Christophe Fouquet, CEO

The CEO's argument for the mask initiative is the most interesting point: nobody would invest in a new mask format for a technology they planned to use sparingly.

"Because if it were not the case, no 1 would really bother to go to 12 inch mask on High-NA." — Christophe Fouquet, CEO

Assessment: Bull 5 moves from "still immaterial" to "dated." TSMC's ~2030 date is close to the ~2029 deferral we carried in July, so this is not an acceleration at the largest customer. What is new is a second HVM customer with a firm year (Samsung, 2028) and a mask roadmap that only makes sense with rising layer counts. Revenue impact remains small through 2027.

10. New EUV Customers, Especially in DRAM

Management said it continues to win new Low-NA EUV customers, especially in DRAM, and that the trend extends beyond the prospective new entrant one analyst alluded to. It reported a similar pattern for High-NA. The CEO also confirmed that the DRAM EUV discussion excludes China.

"We continue to see, in fact, new customers on low-NA EUV, especially when it comes to DRAM." — Christophe Fouquet, CEO

Assessment: A broader EUV customer base is good for the monopoly and complicates the capacity allocation that management is trying to manage. It is also further evidence for our Bull 4 argument that DRAM lithography intensity is rising independently of volume, because new entrants are adopting EUV at the start of their roadmaps rather than at the end.

11. Product Roadmap: More Wafers per Tool

The capacity answer is not only more tools. The NXE:3800F Low-NA system enters in small quantities in early 2027 at 260 wafers per hour, against 230 for the current model per the company's video interview. Critical immersion tools shipping from 2027 move to 330 wafers per hour. The YieldStar 550B overlay metrology system shipped for the first time, with more than 60% higher productivity, and management expects YieldStar demand to more than double versus 2025.

"This latest generation system continues to support major adoption of YieldStar by customers, with demand that is expected to more than double for the coming years versus 2025." — Christophe Fouquet, CEO

Assessment: Throughput gains add wafer capacity faster than unit counts, the same "boxes versus wafers" point from July, and management confirmed that DUV pricing will scale with throughput the way EUV pricing does. That turns the 330-wafer immersion upgrade into a pricing event as well as a capacity one.

12. Below the Line: Mistral, Buybacks and Debt

Net income includes a €509.7M gain on equity investments, which management attributes to the revaluation of its initial Mistral AI investment. Equity investments rose from €1,324.9M to €2,239.4M, which also reflects €405.7M of equity investment purchases in the quarter; the release does not name the investee. Capital return accelerated sharply: €4,071.0M of buybacks (about 2.8M shares), €721.3M of dividends, and €1,000.4M of debt repayment, while cash and short-term investments still rose to €12,735M.

"Net income in Q3 was €5.0 billion, which includes around €500 million as a result of an increase in the valuation of our initial investment in Mistral AI, representing 42.5% of total net sales and resulting in earnings per share of €12.95" — Roger Dassen, CFO

Assessment: The Mistral gain is non-operating, likely non-recurring, and should be excluded from any run-rate. The buyback step-up is the more durable signal: with down payments funding working capital, ASML can return roughly a full quarter's net income in buybacks and still rebuild cash. The €12B 2026-2028 program now looks conservative.

Analyst Q&A Highlights

How Much Capacity Beyond +30%, and Whether 2027 Can Exceed the Mid-80s

The opening question asked management to size the extra capacity it now says it needs and to say whether 2027 could move above the roughly 85 Low-NA units discussed in July. Management declined to give numbers for either year, called it far too early, and said the incremental opportunity is easier to capture in 2028 than in 2027.

Q: "I was wondering if you could provide maybe just a little additional color on the capacity expansion that you need beyond the 30% you talked about. And just what do you think you can do in 2027 relative to that kind of mid-80s that we talked about last quarter?"
— Joe Quatrochi, Wells Fargo

A: "Now if we look at some of the opportunity we see, of course, it is a bit easier for us to enable those opportunity later. So mostly in 2028 rather than on the shorter term. So this is what we are focusing on. I will not give new numbers for 2027 or 2028 because it is far too early to give those number."
— Christophe Fouquet, CEO

Assessment: A non-answer on numbers, but a useful answer on timing. 2027 is roughly fixed near the July plan; the upside is a 2028 story. For modelling, hold 2027 Low-NA units near 85 and let the 2028 distribution skew higher.

Whether Pricing Covers Systems, Upgrades and the Existing Backlog

A follow-up asked whether the price increases apply across systems, services and spare parts, and whether ASML is repricing its backlog. Management refused to describe the mechanics but confirmed the scope: the most critical high-volume tools, with upgrade prices following tool prices, and the effect starting during 2027. The implied answer on backlog is no: orders already placed keep their prices, which is why the effect builds through 2027 into 2028.

Q: "On the value pricing, is this across the board? Is it for systems, services, spare parts, everything? And I am guessing based on your commentary, you are not repricing the backlog."
— Krish Sankar, TD Cowen

A: "But as I mentioned, we are increasing prices for what we call the most critical products. Those products that bring most value to our customers. So these are the high-range, high-volume manufacturing products."
— Roger Dassen, CFO

Assessment: The scope is narrower than "across the board" but covers the products that matter for revenue. The upgrade linkage is the underappreciated part: the highest-margin line in the business inherits the price increase.

Whether Customers Are Inflating Orders to Secure Slots

The double-ordering question, which every capital-equipment upcycle eventually raises, came early. Management said it validates customer demand against detailed fab construction and release plans, and that the demand it cites is what it believes is real. The same answer said ZEISS is not a bottleneck even beyond +30% for 2028.

Q: "In other words, if the demand is there, can you do 140 or 150 units in 2028? Along the same line, do you worry that customers are pulling in orders or giving you bigger forecast just to secure a slot?"
— Krish Sankar, TD Cowen

A: "So I think that, you know, we talk about demand, when we refer to demand in those discussion, we really refer to demand we believe is real and not inflated by any potential buying behavior from our customer."
— Christophe Fouquet, CEO

Assessment: Management cannot prove a negative, but the structure supports the claim: down payments that "hurt," cancellation costs beyond the down payment, and tool allocation conditional on facility readiness. The 140-150 unit framing for 2028 went unchallenged and unendorsed; we would not anchor to it.

Why Q4 Gross Margin Falls on Much Higher Revenue

A question on the Q4 margin guide drew the most candid answer on costs. The ~50bp step down despite roughly 26% more revenue is ramp cost for 2027-2028 capacity, which offsets the expected volume leverage. The follow-up on pricing confirmed that part of the increase funds the ramp, including support for suppliers.

Q: "Quickly, Roger, on the gross margin, your guidance on gross margin in the fourth quarter is down from the current quarter, even though your revenues are going up significantly?"
— Sandeep Deshpande, JPMorgan

A: "That is primarily related to the fact that we are seeing some ramp costs."
— Roger Dassen, CFO

Assessment: Credible and modest. Ramp cost is a choice to fund 2028 capacity now, and it sits alongside a price increase partly designed to pay for it. The risk is that ramp costs persist through 2027 while pricing builds slowly, which is why we model only part of the pricing effect in 2027.

Whether ASML Will Tighten Contract Terms in a Shortage

A question asked whether ASML would raise prepayments or demand firmer orders earlier to filter out speculative demand. The answer described the existing commitment as already severe and explained why 2029-2030 order requests are being held: ASML wants to see total demand before allocating, so that no long-standing customer is told it is sold out.

Q: "I was wondering if you were considering any change of the way you want to contract with your customers in those times? For example, you know, increasing the prepayments or also, you know, like, more the firm orders earlier."
— François Bouvignies, UBS

A: "So the only hesitation we have at this stage is that we really want to do our homework and be very diligent on the acceptance of the orders that customers are offering us to make sure that no single customer, you know, feels left out."
— Roger Dassen, CFO

Assessment: The most strategically interesting exchange of the call. ASML is behaving like a company allocating scarce supply rather than chasing orders. That posture supports pricing and limits the risk of a single customer pre-empting capacity, at the cost of some near-term order visibility for investors.

Down Payments as a Proxy for Order Intake

With bookings no longer published, a question asked whether the roughly €5B jump in non-current liabilities could be read as a large Q3 order intake. Management confirmed that the line includes down payments and called it a reasonable proxy, while noting that it lags order signing. The same exchange established that interest in 2029-2030 orders is broad rather than concentrated in one new customer.

Q: "So can we make a read through to a relative size of a big order intake in Q3?"
— Lee Simpson, Morgan Stanley

A: "So on the on the first 1, the non current liabilities, does that include down payments? Yes, it does. Right? So in that way, it is a interesting gauge of, you know, of how order intake is developing. I should say it is a bit of a lagging indicator."
— Roger Dassen, CFO

Assessment: The best disclosure of the call for anyone tracking demand. A €5.7B rise in non-current contract liabilities in one quarter, against a line that sat near €3.6B for the prior year, is the order-book signal investors lost when ASML stopped reporting bookings. Because it lags signing, Q3 understates Q3 orders rather than overstating them.

Whether Price Increases Become an Annual Event

A direct question asked whether ASML could raise prices every year if the shortage persists. Both executives said no: this was a one-off reset of lithography's share of ecosystem value, and the value-based model exists precisely so that ASML never has to negotiate price cuts in a downturn.

Q: "Do you think you will be able to negotiate price increase every single year?"
— Tammy Qiu, Berenberg

A: "Well, Tammy, that was not the plan, to be honest. … But we would not be looking at repeating this every year or so."
— Roger Dassen, CFO

Assessment: The right answer for a monopolist with a long memory. Opportunistic annual increases would invite customer resistance, regulatory interest and second-source investment. A one-time reset defended on value is much more durable, and it is what we model.

Logic Versus Memory in the Forward Book

A question asked whether the incremental demand had broadened from memory to logic. Management said memory is gaining share, mostly DRAM, but that logic remains very strong and is expected to outweigh memory in the 2027 order book. On 2029-2030 interest, management described equal interest from both and said DRAM customers are currently the more aggressive builders.

Q: "I wondered if it is fair to say that in the prior quarter, the incremental demand seemed to be very much related to memory. But now you seem to be seeing quite a strong balance across both logic and memory."
— Alexander Duval, Goldman Sachs

A: "So you are right. Memory is getting closer to logic, but logic remains very, very important."
— Roger Dassen, CFO

Assessment: Supportive of a more balanced cycle than the Q2 memory share implied. The disclosure that total logic grows only ~10% this year (mainstream logic is shrinking) is the counterweight: advanced logic and DRAM are carrying everything.

What They're NOT Saying

1. Bookings or backlog, again. The order book is described as building "quarter on quarter," and the only quantitative proxy is the €5.7B rise in non-current contract liabilities, which management itself calls a lagging indicator. The last published figures remain Q4 2025: €13.2B of net bookings and roughly €38.8B of backlog.
2. New unit numbers for 2027 or 2028. Planned output "will not fully meet" demand, but management would not say by how much or what the new 2028 target is. "Far too early" was the answer.
3. The size of the price increase. Agreed, dated and scoped, but never quantified. Management also would not say what share goes to supplier ramp support versus gross margin.
4. Why non-EUV growth was cut to ~20% from ~25%. No explanation was offered and none was requested, in a quarter when non-EUV system sales jumped more than 50% sequentially.
5. 2027 gross margin or revenue. Deferred to January 2027. The CFO listed drivers on both sides without weighting them.
6. The Mistral gain's tax treatment. Our arithmetic suggests the gain was untaxed, but management did not say so, and it did not identify the €405.7M of equity investment purchases in the quarter.
7. Receivables. DSO rose again to about 72 days. No question, no comment; the cash story drowned it out.
8. Export controls. China came up in Q&A after going undiscussed in July, but only as demand and competition. The export-control accommodation volunteered in April has not been restated in two quarters.
9. Who the new entrants are. Management confirmed new Low-NA and High-NA EUV customers, mostly in DRAM, and beyond the prospective new entrant one analyst alluded to, but named none.
10. 3D integration and the XT:260. Absent from the call for a third consecutive quarter. It survives only as a line in the investor key messages.

Market Reaction

• Pre-print setup (October 13 close): ADR $1,717.76; Amsterdam €1,509.60. Entering the print the ADR was +60.6% year to date and +74.7% over twelve months, against an S&P 500 up 14.0% year to date, and +9.1% over the trailing 30 days. It was 13.7% below its 52-week closing high of $1,989.44. The two sessions before the print were already weak: the ADR fell 3.5% from $1,780.34 on October 9, and Amsterdam fell 5.7%, including 4.20% on October 13 alone.
• Options-implied move: roughly 4.6% to 6.5% in either direction, depending on the listing and the reading, against a median absolute move of about 4.7% over the prior eight reports in Amsterdam.
• Amsterdam (the primary listing): opened flat at €1,510.00, traded between €1,494.00 and €1,564.00 (+3.6% at the high), and closed at €1,528.20, up 1.23%, on 1.3x average volume. The AEX rose 0.69%.
• US ADR: opened $1,676.80 (-2.4% gap), traded between $1,648.80 and $1,734.50, and closed at $1,650.00, down 3.94% and $1.20 above the day's low, on 2.1x average volume. The S&P 500 fell 1.11% and SOXX fell 4.50%.
• Print-day peers: AMAT -3.79%, LRCX -4.71%, KLAC -2.82%, TSM -3.40%, MU -4.10%. In Europe, ASM International rose 0.44% and BE Semiconductor rose 1.19%.
• The following two sessions: Amsterdam fell 4.19% on October 15 to €1,464.20 and rose 0.89% on October 16 to €1,477.20. The ADR rose 0.45% to $1,657.44, then fell 0.96% to $1,641.52.
• The week (October 9 to October 16 closes): ADR -7.80%; AMAT -19.28%, LRCX -17.22%, KLAC -15.25%, MU -14.59%, TSM -8.98%, SOXX -11.80%.

Same Verdict, Different Clocks

The two listings did not disagree about the print; they traded it at different times. Amsterdam reacted in the European morning, rising about 2% early and reaching +3.6% at the high, as European tech rebounded from a global chip sell-off the day before. US September CPI then came in hot at 14:30 Amsterdam time (3.7% year on year), three hours before the Amsterdam close, and Amsterdam gave back most of its gain to finish +1.23%. The ADR traded almost entirely after that: rate-hike odds for the October Fed meeting jumped, the 10-year yield spiked above 5.3% intraday, the bonds of a large AI-cloud operator sold off, a sell-side downgrade hit a large AI server maker, and the Philadelphia semiconductor index fell 4.3%. The ADR's -3.94% sits close to SOXX's -4.50%. Amsterdam then caught up on October 15 (-4.19%) while the ADR rose 0.45%. Measured over the same news, October 13 to October 15, the two listings converge: ADR -3.51%, Amsterdam -3.01%.

The Sector, Not the Print

The more important number is relative. From October 9 to October 16 the ADR fell 7.80% while AMAT fell 19.28%, LRCX 17.22% and KLAC 15.25%. SOXX fell in every session from October 12 to October 16, on four overlapping concerns, none of them about ASML: doubts about AI returns (a revenue scare at a leading model developer on October 8, a large chip designer's chief executive comparing AI investing to a casino on October 12, and a slowdown in another model developer's growth reported around its October 14 investor day); Chinese DRAM progress that hit memory makers and their tool suppliers from October 10; the largest US equipment maker declining to forecast wafer-fab equipment spending at its October 13 investor event; and a hot CPI and PPI pair that pushed up rates and widened credit spreads on AI-infrastructure debt. We found no new export-control, tariff or hyperscaler-capex news in the window.

Against that backdrop, ASML lost roughly half as much as its US peers in the same week it reported. The message of the print, that planned 2027-2028 supply will not meet demand and that customers are asking to order for 2029-2030, is the most likely reason. The market believed the print. It is also de-rating the whole AI capital-spending complex on rates and returns, and one company's order book does not exempt it from that.

Positioning Going In

Pre-print trading-desk commentary described ASML as a common funding short for other semiconductor longs, with much of the bull case already understood, and the premium to US equipment peers near its lowest in more than a decade. Options priced a move of roughly 4.6% to 6.5% depending on the listing and the reading. Both closes fell inside that range. The print did what a crowded short-funding leg needed it to do in relative terms, and nothing it could have said would have offset a 4.3% fall in the semiconductor index.

Street Perspective

Sell-side reaction was muted: a handful of small price-target changes in both directions and no rating changes that we could find. The debate has moved from whether demand exists to how long it lasts and what the market should pay for it.

Debate: Sold Out Through 2028, or Building Into a 2029 Peak?

Bull view: ASML is effectively sold out through 2028, the binding constraint is its own output, and the 30% capacity steps are minimums rather than targets. Customer down payments, contracted price increases from 2027 and requests for 2029-2030 slots make this the best-visibility cycle the company has had. Some on the Street already model more than 85 Low-NA units for 2027.

Bear view: Memory spending is cyclical, and a supplier adding more than 30% capacity in 2028 into a memory market that is short today is adding capacity just as demand may peak. Uncertainty rises sharply after 2028, AI returns are being questioned, higher rates make AI project financing harder, and the multiple could fall into the low 20s in a downturn. Orders are concentrated in a handful of leading-edge chipmakers.

Our take: The bulls are right about the next 24 months and the bears are right that the question has moved to 2029. What tips us to the bull side for a 12-month horizon is that the 2028 capacity still sits inside the existing footprint, so the cost of being wrong about 2029 is mostly lost upside rather than stranded assets. The bear case needs either a 2028 order cancellation or a January guide that frames 2027 as a peak; neither is in sight.

Debate: Are 2029-2030 Order Requests Real Demand or Reservation Behaviour?

Bull view: Customers asking to order three to four years out, at the price of a down payment that hurts and cancellation costs beyond it, is the strongest demand signal a capital-equipment company can receive. Interest is broad across logic and DRAM rather than tied to one new entrant, and confirmation of the first 2029 orders is the next catalyst.

Bear view: Interest is not an order. In a shortage, every customer has a reason to reserve slots it may not need, and ASML has not accepted a single 2029-2030 order. Some of what reads as demand may be insurance.

Our take: We side with the bulls on substance and the bears on accounting. The contract structure is the best evidence that the demand is real, and ASML's refusal to accept orders until it has allocated capacity is the behaviour of a company that expects more requests than it can fill. But until orders are accepted, it is an option on 2029, not a forecast. The non-current contract liabilities line, up €5.7B this quarter, is where acceptance will show first.

Debate: How Good Was the Quarter, Really?

Bull view: Record revenue and gross margin, an Installed Base beat of almost €400M, €11.4B of free cash flow, a Q4 guide above consensus on both sales and margin, and a full-year margin guide raised to ~56%. The guide now covers a Q4 that would be a record by half.

Bear view: Revenue was slightly below consensus. Roughly €0.5B of net income is a one-off investment gain. Only the bottom of the full-year range moved, lifting the midpoint about 1%. Q4 gross margin steps down on higher revenue, and the margin beat came from upgrades, the line management calls a swing factor. Much of this was already priced in.

Our take: The bears have the facts right and the weighting wrong. The revenue shortfall against consensus is within rounding of a guide the company landed toward the top of, and the Mistral gain is easy to strip out: ex-gain EPS still roughly doubled YoY. The two points that matter for the next year, Q4 guided directly and pricing contracted, are both better than consensus expected going in. The legitimate concern is the 2027 margin bridge if upgrades slow, which we address in our model by holding 2027 gross margin below the trajectory the Q3 print alone would suggest.

Debate: Is the De-Rating an Entry Point or the Start of a Cycle Multiple?

Bull view: The forward multiple has fallen from above 45x in June to around 30x on Street numbers, and ASML's premium to US equipment peers is near its lowest in more than a decade, despite the strongest competitive position and the best visibility in the group. Holders describe ASML itself, not demand, as the bottleneck.

Bear view: The de-rating reflects doubts about very high 2028 and 2029 expectations, not a temporary sentiment swing. If the semicap group is pricing a cycle peak, ASML's multiple converges toward cycle-trough levels regardless of its order book.

Our take: The de-rating is real and may not be finished, which is why we lowered our base-case multiple. But the bears' scenario requires 2028 estimates to fall, and this quarter's disclosures moved them up. A stock that trades at about 24x our FY2027E with contracted pricing ahead of it is pricing more cycle risk than the order book supports.

Model Update & Valuation Framework

Why the targets hold while the estimates rise. Our FY2027 EPS estimate rises about 9%, from ~€56 to ~€61, on a higher revenue base, a better gross margin and a larger buyback. We do not pass that through to the target. Instead we cut the base-case multiple from 32-36x to 29-33x, because the semicap group has de-rated sharply since July and we do not think one company's guide can hold its multiple against that. The result is an unchanged base target on a better earnings base, which is the more defensible combination if the sector de-rating persists.

Where the stock trades. At $1,641.52 (October 16 close) and EUR/USD of about 1.124, the ADR trades at about 35x our FY2026E ex-Mistral EPS of ~€41.4 (~$46.5) and about 24x our FY2027E EPS of ~€61 (~$68.7). At our July note it traded at about 28x the ~€56 FY2027E we then carried. The stock is down 9.6% since then in dollars and 4.7% in Amsterdam (€1,549.40 to €1,477.20); the euro's slide from about 1.146 to about 1.124 explains roughly two points of the gap. On the post-print FY2027 consensus EPS of €55.93 (~$62.9), the ADR trades at about 26x. Our ~€61 sits about 9% above consensus, almost entirely because we carry a higher 2027 gross margin with part of the price increase in it.

The 2030 framework, again. Our FY2027 model of ~€58B sits at 97% of the top of the €44-60B 2030 framework that ASML will reset at its June 10, 2027 Capital Markets Day. In July we argued that reset had to be upward. With pricing contracted from 2028, 2028 capacity under study beyond +30% and customers asking to order for 2029-2030, the open question is no longer direction but magnitude.

Risk-reward at $1,641.52. Base case $2,000-2,250 implies +22% to +37%. Bull case $2,600-2,900 implies +58% to +77%. Bear case $1,250-1,400 implies -24% to -15%. Weighting base at 55%, bull at 20% and bear at 25% gives an expected return of roughly +25%, against roughly +13% at our July note. Fundamentals improved and the price fell, so the asymmetry widened from both ends.

What would change our mind. Upward: a Q4 above €15B, a January 2027 guide that puts FY2027 gross margin above 58% before most of the pricing lands, or a new 2028 unit plan above ~110 Low-NA systems. Downward: a Q4 below €14B, a January guide that frames 2027 Installed Base as declining, any sign that customers are deferring 2028 slots, or a further leg of the semicap de-rating driven by hyperscaler capex cuts rather than by sentiment.

Thesis Scorecard Post-Earnings

Graded against the standing thesis as of our Q2 2026 note. Bull 1 (demand-framework progression) and Bear 2 (customer fab readiness) were retired in July and are not re-argued here.

Overall: The thesis is strengthened. Every commitment on our July checklist that this quarter could test was met or exceeded: Q3 landed toward the top of guide at a record margin, Q4 is guided above what we back-solved, R&D returned to plan, cash conversion recovered, 2028 firmed into orders, and pricing converted from a concession into a contract. Two of three July-era bear points (back-half execution, cash conversion) move to CONTAINED. What we add is a new risk the July note did not carry: that the market is pricing the semicap cycle, not ASML's order book, and that a sector de-rating can outrun fundamentals for longer than our 12-month horizon tolerates.

Action: Maintaining Outperform, conviction up to 8 of 10 from 7. Existing positions: hold full weight. New money: in July we said we would not chase at $1,815 and that the better entry would come on Q3 receivables or Q4 ramp anxiety. It has come instead on a sector sell-off, with the receivables question partly answered and the Q4 question converted into a guide. We would add at current levels, sized for a stock that can fall with its group regardless of its own numbers.