FY2026 Guide Blown Through to €43-45B With Margin to 54-56%, 2027 Nearly Sold Out at ~85 EUV Units, the 2030 Framework Reached Four Years Early. Amsterdam Still Closed Red. Maintaining Outperform.

Results vs. Consensus

Q2 2026 Scorecard (EUR)

Year-over-Year Comparison (Q2 2026 vs. Q2 2025)

Sequential Comparison (Q2 2026 vs. Q1 2026)

Revenue assessment. The headline beat is real but the composition matters more than the magnitude. Net system sales grew 17.3% YoY while Installed Base Management grew 31.8%, so the mix shifted toward the higher-margin, more annuity-like half of the business. That is the mechanical explanation for gross margin printing 200bp above the top of the guide, and management said so directly. The 91 litho units sold (86 new, 5 used) against 79 in Q1 and 76 a year ago show the output ramp is real in physical terms and not just a pricing or mix artifact. The end-market split was almost exactly even, logic at 51% and memory at 49% of net system sales, which is a historically high memory share and the clearest single indicator of what is driving this cycle.

Margin assessment. Gross margin of 54.0% is the highest in the five-quarter series (51.6% in Q3 2025 was the trough) and sits above the top of the FY guide that was in force this morning. Operating margin of 37.1% expanded 250bp YoY on 21.3% revenue growth against 7.8% combined OpEx growth. The forward setup is better still: the Q3 guide of 55-57% and the FY guide of 54-56% imply an H2 gross margin of roughly 56%, and management identified four separate drivers (richer immersion and Low-NA EUV mix, better-priced EUV as the D models run out, sustained high-margin Installed Base, and fixed-cost coverage on much higher volume). None of those four is a one-off. One note on the six-month figures: the six-month gross margin prints 53.5%, which sits between the two constituent quarters (Q1 at 53.0% and Q2 at 54.0%), exactly where a revenue-weighted average of the two should land. There is no filing quirk here.

EPS assessment. EPS of €7.59 grew 28.6% YoY, ahead of the 27.4% net income growth, with the 1.0% reduction in share count from the buyback accounting for the gap. The quality is clean: the beat was operational, not tax-driven (the 17.5% rate was only 60bp below the year-ago quarter and is in line with the ~17% full-year guide) and not below-the-line (equity method income of €41.1M actually fell from €88.8M a year ago, which was a modest headwind rather than a help). We could not locate a vendor-attributed euro EPS consensus; the two available implied readings put the beat somewhere in the +9% to +11% range, and we decline to quote a false-precision figure. Investors should also disregard the widely-circulated dollar-denominated YoY EPS growth figures, because the three US vendors disagree on the year-ago base by a range of $4.55 to $6.84 against an actual of €5.90.

Segment Performance

ASML does not report segment operating margins, so the analysis below works from net sales composition, the unit data, and the FY2026 growth rates management gave on the call. The FY2026 growth guidance in this table is new or updated as of today and is not in the press release.

Revenue Composition and FY2026 Growth Guidance

End-Market Mix (net system sales)

Memory: the quarter's engine, and the quarter's biggest question

Memory at 49% of net system sales is an unusually high share for ASML and the single most important number in the composition. The FY2026 guide of over 75% growth in memory-related net system sales is not a normal cycle figure. What makes it analytically interesting is that management attributes it to two independent effects stacking, rather than to one. The first is volume: customers are adding fab capacity because both DDR and HBM are supply-short and priced accordingly. The second is lithography intensity: as DRAM migrates to 1b and 1c, EUV layer counts rise, and, critically, customers are replacing multi-patterning with single-expose EUV because it is cheaper. The second effect is the durable one. It is a permanent shift of spend from non-litho process steps into litho, and it does not reverse when DRAM pricing normalizes.

"The 1c nodes, for example, which is going to be enormous node or even 1b, are using more EUV layers. This is really this combination which create a bit the perfect storm for ASML on DRAM this year and most probably the next few years to come."
— Christophe Fouquet, CEO

Assessment. "Perfect storm" is management's own phrase and it cuts both ways: a storm is by definition not the weather. Management declined to split the HBM and DDR contributions, which is the split that would tell you how much of the 75% is structural AI demand versus a DDR pricing spike that mean-reverts. We take the litho-intensity half of the argument as durable and the volume half as cyclical, and we think the honest read is that a meaningful slice of that 75% is cycle rather than secular. That is the strongest bear argument in the name right now, and notably nobody on the call made it.

Logic: less spectacular, more reliable

Advanced logic foundry net system sales growing over 25% is the quieter half of the story and the more dependable one. The disclosure worth noticing is the breadth: management described investment not only at 3nm for the current AI accelerator generation, but also at 5nm and 4nm to support what it called the diverse set of chips required by AI products, alongside a rapid 2nm ramp for HPC and mobile and the beginning of 1.4nm development spend. That is four or five nodes being funded simultaneously, which is not what a narrow AI-accelerator boom looks like. It is what a broad-based build looks like.

"In logic, there is a continued investment, not only to enable the expansion of 3 nm capacity in support of the latest generation of AI accelerators, but also at both the 5 nm and the 4 nm nodes to support the diverse set of chips required by AI product. At the same time, the 2 nm node continues to ramp rapidly to support next generation HPC and mobile applications. Our customers are already planning investment to support the development of the 1.4 nm nodes."
— Christophe Fouquet, CEO

Assessment. The multi-node breadth is the best evidence in the release that this is not a single-product cycle. A cycle driven purely by one accelerator generation would show up as concentrated 3nm spend. Spend across 5nm, 4nm, 3nm, 2nm and early 1.4nm implies the AI build is pulling through a whole chip ecosystem, not just the GPUs. This is the part of the guide we have the most confidence in.

Installed Base Management: the flywheel that beat the quarter

Installed Base grew 31.8% YoY to €2,761.7M and came in roughly €260M above guide, and it was explicitly the reason both revenue and gross margin beat. The mechanism is worth stating plainly because it is the most underrated part of the ASML model right now: customers want more capacity faster than clean rooms can be built, so they buy productivity upgrades on tools they already own. Every tool ASML has ever shipped becomes an upgrade opportunity, and management said the upgrade portfolio now covers every version of the EUV Low-NA and immersion fleet. Q3 is guided to roughly €2.9B, another step up.

"The key element today is that customer wants to get more capacity on their existing fab as quickly as possible. This creates really strong condition for system upgrade… The upgrade product we are developing are covering all the different version of our EUV Low-NA or immersion system. We are really capable, basically, to provide product to our customer that can be implemented to every single version of our EUV or immersion tool."
— Christophe Fouquet, CEO

Assessment. This is the highest-quality growth in the business and the market underweights it. It carries above-average gross margin, it requires no clean room, it compounds mechanically with the installed base, and demand for it is strongest precisely when customers are capacity-constrained. It is also somewhat counter-cyclical in a useful way: if new fab construction slows, the incentive to upgrade existing fleet goes up, not down. Management is accelerating new upgrade products into 2027 and 2028 at customer request.

China: steady at ~20%, and conspicuously undiscussed

China remains roughly 20% of total net sales for the full year, growing in line with the overall business, driven mainly by mainstream logic serving domestic demand. That percentage has not moved. What is notable is not the number but the silence around it: export controls did not come up on the call at all, from management or from any of the ten analysts who asked questions.

Assessment. Neutral on the number, mildly negative on the silence. Last quarter management went out of its way to say the FY guide could absorb potential export-control outcomes; this quarter the topic simply did not arise, and the guide was raised without any restatement of that accommodation. We do not read that as a signal of deterioration, but a raised guide with no restated policy cushion is a slightly thinner disclosure than the one we got in April.

Key KPIs

Guidance & Outlook

The FY2026 Raise

Q3 2026 Guidance

Implied quarter-over-quarter ramp: at the midpoint, Q3 must grow 23.3% sequentially and Q4 a further 25.3%. Two consecutive quarters of roughly 25% sequential growth would be without precedent in ASML's recent history. The company has never printed a quarter above €9.72B.

Street at: consensus sat at €39.4B for FY2026 and roughly €10.1-10.4B for Q3 before the print, so the guide is about 12% above the Street on both. One sell-side house had modelled Q4 at roughly €11.6B against the €14.4B now implied, a gap of about 24%.

Guidance style: this is a company that has guided conservatively and beaten consistently through this recovery, which is the strongest argument that the ramp is real. Q2 landed above the high end of its own guide; Q1 landed at the top of its range. But the pattern of beating a €9B guide by €326M is a different exercise from delivering a €14.4B quarter that requires the physical output rate to step up by half. Past conservatism is evidence about intent, not about manufacturing throughput.

Key Topics & Management Commentary

Overall Management Tone: Management was the most confident it has been in the eight quarters we have covered, and notably it was confident in a specific direction: about supply rather than demand. Where prior calls defended the demand framework, this one treated demand as settled and spent its energy on how much output can be forced through an existing footprint. The one place management was consistently less convincing was pricing, where it acknowledged unusual leverage three separate times and declined three separate times to say when any of it lands in revenue. Nobody on the call asked about cycle risk, and management volunteered none.

1. The Guide Raise: €36-40B to €43-45B in One Step

Ninety days ago ASML raised and narrowed FY2026 to €36-40B and we called it a textbook framework upgrade. Today the company moved the range to €43-45B. The midpoint jumped €6B, or 15.8%, in a single quarter, and the low end of the new range sits €3B above the high end of the old one. Growth versus FY2025 goes from +16% at the old midpoint to +34.6% at the new one. The gross margin guide moved up 300bp at both ends simultaneously, which is the rarer half of the raise: revenue raises are common, but a 300bp margin raise alongside says the incremental volume is arriving at better-than-corporate economics.

"Given the business dynamics discussed above, we now expect total net sales for 2026 to be between €43 billion and €45 billion, with a gross margin between 54% and 56%."
— Christophe Fouquet, CEO

The stated mechanism is a combination of customers revising capex plans upward mid-year and ASML being able to physically respond. Both halves are necessary. The demand revision alone would have produced backlog, not revenue.

"The combination of continued strong momentum in customer demand and our ability to respond to that by driving higher output through strengths in our supply chain, our manufacturing, and our installed teams in the field are the primary drivers of our improved guidance. Strong end market demand this year has motivated our customers to aggressively add capacity on their leading-edge nodes. A number of our customers have revised their capital expenditure plans upward for the year, and our ability to increase output has allowed us to meet their request for additional lithography system."
— Christophe Fouquet, CEO

Assessment. This is the largest single-quarter guidance raise in ASML's modern history and it validates the Outperform we have carried since Q3 2025 more completely than we expected. But it also changes what the thesis is about. Our rating has been carried on the proposition that the demand framework was being restored quarter by quarter. That proposition is now fully resolved in our favour and has no further room to run. From here the stock is a bet on manufacturing execution and cycle duration, which are different questions requiring different evidence.

2. The 2030 Framework Is Obsolete, and the Reset Is Eleven Months Away

This is the disclosure with the longest half-life and it received almost no attention on the call. ASML's standing long-term framework, reiterated as recently as our Q1 review, put 2030 revenue at €44-60B. The FY2026 guide midpoint is €44.0B. The company will reach the low end of its own 2030 revenue framework in 2026, four years early, and the high end no longer looks like a stretch case for the decade so much as a plausible 2028 outcome. The framework is not merely conservative now; it is not information. And management has deferred the reset to a Capital Markets Day on June 10, 2027.

"At our next Capital Markets Day, which will be held on June 10, 2027, we will update our longer-term views to reflect the market and technology dynamics since our last Capital Markets Day."
— Christophe Fouquet, CEO

Our Q1 review listed a "late-2026 Capital Markets Day" as the next major framework catalyst and flagged a potential 2030 upgrade as the principal driver of our bull case. That date has now slipped by roughly two quarters. Investors who were underwriting a long-term reset this year will wait until next summer for it.

Assessment. Two-sided, and we think net positive. The negative is real: the single cleanest re-rating catalyst in the name has moved eleven months out, and until then the only public long-term anchor is a framework the company will surpass this year. The positive is larger. A 2030 framework whose floor is hit in 2026 is not a ceiling that constrains the stock; it is a mathematical guarantee that the June 2027 reset must be upward, and probably by a lot. Between now and then the stock has to be carried by quarterly execution rather than by the long-term story, which is precisely why the H2 ramp matters as much as it does.

3. Capacity: +30% for 2027, Nearly Sold Out, and Another +30% Under Investigation for 2028

The forward disclosure that will drive 2027 estimates. ASML is adding 30% to its 2026 Low-NA EUV capacity of roughly 65 units, taking 2027 to roughly 85, and is investigating a further 30% for 2028, which analysts on the call put at roughly 110 without correction from management. The immersion DUV line follows the same pattern: roughly 130 units in 2026, +30% for 2027, another 30% under investigation for 2028. The critical qualifier is that 2027 is already close to being fully covered by orders, and 2028 already has a significant order book two years out.

"For 2027, we are now close to being fully covered with orders for Low-NA EUV, and we are planning to increase our Low-NA EUV capacity by around 30%. Looking ahead to 2028, we have already received a significant number of Low-NA EUV orders. Strong demand forecasts from our customers have led us to investigate a further 30% capacity increase for that year."
— Roger Dassen, CFO

Two things make this materially better than a normal capacity announcement. First, it is demand-backed rather than speculative for 2027. Second, and less appreciated, all of it fits inside the existing footprint. The new campus breaks ground this year but is explicitly a post-2028 asset and contributes nothing to these numbers. The levers are freeing up cabins currently occupied by prototypes and R&D tools, cutting cycle time, and lifting supply-chain move rates.

"I think all the capacity increase, either planned or investigated, we're talking about are based on our existing footprint. We have been working extremely hard, as you may have noticed, in the last six, nine months to increase our output… The number we are mentioning, we can achieve basically by optimizing the existing clean room space in the right way."
— Christophe Fouquet, CEO

Assessment. The existing-footprint point is the strongest single rebuttal to the overbuild thesis and it deserves more weight than the market gave it. Capacity that requires no new clean room is cheap to add, fast to add, and, most importantly, cheap to stop. ASML is not committing a decade of capital to a cycle it cannot forecast; it is raising utilisation of assets it already owns. If memory rolls over in 2028, the incremental cost of having been wrong is small and largely reversible. This is close to the ideal shape for capacity expansion into an uncertain cycle, and it is why we are more relaxed about the overbuild risk than the tape appears to be.

4. Boxes Versus Wafers: 30% More Tools, 45% More Capacity

The most useful piece of arithmetic on the call, and one that will cause estimate confusion if missed. The 30% capacity increase for 2027 counts tools, not output. Because the 2027 shipment mix is Es and Fs where 2026 is a mix of Bs and Es, the wafer capacity ASML delivers rises roughly 45%, not 30%. Layer the upgrade business on top and the effective capacity ASML adds to the industry next year is higher still.

"We're looking at 30%, which is boxes, if you like, 30% more tools. You should also recognize that the tool mix that we're going to ship next year is a different tool mix from the tool mix that we shipped this year… If you recognize the difference in output, then in essence what you're looking at is not 30% improvement of wafer capacity that we're adding, but approximately 45%. In addition to that, as you also know, we're offering a whole slew of upgrade packages to the install base to customers, which gives them another significant uptick."
— Roger Dassen, CFO

Assessment. This matters in two opposite directions and both are worth holding. For ASML's revenue, higher-productivity tools carry higher ASPs, so 30% more units translates to more than 30% more system revenue: that is a straightforward positive we have built into our FY2027 estimate. For the industry, ASML is adding roughly 45% more wafer capacity to a memory market that is currently short. Anyone modelling DRAM supply off ASML's unit count is understating incoming capacity by half. The tool that makes ASML's 2027 look good is the same tool that shortens the memory cycle.

5. The D-Model Sellout and the 2027 Margin Setup

A supply constraint that turns into a margin tailwind. ASML is effectively sold out of D-model EUV tools this year, constrained by a specific ZEISS optical component rather than by anything ASML controls. One or two may slip into next year, but from 2027 the EUV mix is E and F only, weighted toward Es. Those tools carry both higher ASPs and better gross margin profiles.

"At a certain point in time, we're simply sold out when it comes to D models, right? The D uses a specific QB from ZEISS, and at a certain point in time, we're just done… Then, from that moment onwards, there is no D POB left anymore, therefore we'll build E and Fs next year. Of course, that mix will come with a better ASP than the mix that we have this year, because it also will come with higher productivity as we just laid out, and also comes with a better gross margin profile."
— Roger Dassen, CFO

Assessment. The 2027 gross margin setup is better than the company will say out loud. Management declined to guide it but confirmed the EUV mix effect will be more positive next year than this year, and more positive than even the improved H2 2026 exit. Stack that on the four H2 drivers already in play (mix, better-priced EUV, high-margin Installed Base, volume leverage on fixed costs) and 2027 gross margin above the 54-56% FY2026 range is close to arithmetic rather than aspiration. We model 56-58%.

6. Pricing Power: Acknowledged Three Times, Dated Zero Times

The most interesting evasion of the call. Three separate analysts came at ASML's pricing from three different angles, and each time the CFO conceded that the current environment gives ASML unusual leverage and then declined to say when it converts into revenue. Long order lead times are the stated reason, and it is a legitimate one: tools shipping in 2027 were priced when they were ordered.

"I would agree with you that the current environment provides more flexibility for pricing than what you would have had in different days. Of course, you will also appreciate, given the long order lead times that we have, that that doesn't translate into pricing effects tomorrow. Clearly, the environment that we live in today, with the value that our products bring to customer, is substantial. Of course, gives us flexibility on pricing, more so than what you would have seen in the past. Of course, we're executing on that as well."
— Roger Dassen, CFO

The phrase "we're executing on that as well" is doing a lot of work and went unexamined. Separately, ASML has reportedly opened discussions on roughly 10% DUV price increases, with some Chinese chipmakers having already accepted, and has raised EUV increases directly with its largest customer, which is resisting.

Assessment. This is the single largest item in the ASML story that is in nobody's numbers, including ours. A monopolist with a two-year order book, customers who are capacity-desperate, and an explicit value-based pricing philosophy is describing pricing leverage it has not yet used. We deliberately model no pricing uplift beyond mix, which means our FY2027 estimate is conservative by construction. If ASML captures even a few points of like-for-like price, the earnings effect is close to fully incremental. Treat it as a free option, not as a forecast.

7. High-NA: Intel Reaches High-Volume Manufacturing, and It Is Still Four to Five Tools

ASML issued a second press release this morning announcing that Intel Foundry has entered high-volume manufacturing on Intel 18A using High-NA EUV, producing a subset of Core Ultra Series 3 processors. This is the first high-volume logic product in the industry built with High-NA, and it is a genuine milestone after two years in which High-NA was an R&D story with a large price tag.

"We are also very pleased to announce in a press release earlier today that Intel Foundry is using ASML High-NA EUV technology on the Intel 18A process node to produce a subset of its Intel Core Ultra Series 3 processors. These milestones marks an important step in demonstrating High-NA EUV readiness in a production environment."
— Christophe Fouquet, CEO

Three qualifiers matter and none was raised on the call. The layers are dual-qualified for both 0.33 NA and 0.55 NA scanners at yields matched to the existing platform, so this is an option Intel has taken rather than a dependency it has accepted. High-NA revenue recognition for 2026 remains four to five tools, unchanged, against roughly 65 Low-NA units. And ASML's largest customer has publicly called High-NA systems, at roughly double the price of current-generation EUV, too expensive, and has deferred adoption to roughly 2029.

Assessment. Directionally important, financially immaterial for at least two more years. The milestone does what it needs to do, which is prove the platform works in production, and Intel's own framing describes it as an option developed alongside the existing fleet rather than a replacement for it. That is closer to the sceptics' cost logic than a refutation of it. We treat High-NA as thesis-relevant for 2028 and beyond and as a rounding error in the numbers between here and there. The more interesting possibility, which management would not endorse, is that DRAM adopts High-NA before logic does.

8. Operating Leverage Is Now the Second Engine

Revenue grew 21.3% YoY in the quarter while SG&A grew 1.3% and headcount grew 1.7%. On the full-year guide, revenue grows roughly 35% against H1 operating expense growth of about 5.5%. That gap is where the margin story now comes from, and management confirmed it is deliberate: the strategy has shifted from adding R&D headcount to extracting more from the team already in place.

"If you contrast the way we've done it in the past, in the past we increased the headcount of R&D quite substantially. I would say that today we believe that with the team that we have today, we can really entertain a very aggressive roadmap going forward. All in all, I think you will continue to see us manage both R&D and SG&A quite nicely. As a result of that, the operating leverage that you imply, I think the operating leverage will indeed become better in the quarters and the years to come."
— Roger Dassen, CFO

The realised incremental operating margin was 48.4% YoY in the quarter and 45.6% for H1. On the full-year guide it rises to roughly 58%, which is what makes the 60% framing put to management on the call a forward statement rather than a trailing one.

Assessment. Confirmed and underrated. A business adding 35% revenue on 6% opex growth is converting a demand cycle into permanent margin structure, and the commitment to hold R&D headcount flat while running an aggressive roadmap is the specific choice that makes it work. The risk, which nobody raised, is that a flat R&D headcount through a period requiring High-NA maturation, Hyper-NA development and a 1,000-watt source is a bet that will not be testable for several years.

9. The Transformation Charge Nobody Asked About

R&D came in at €1,276.6M against roughly €1.2B guided, and the sole explanation offered was one clause in the CFO's prepared remarks. It was never sized, never broken out, and never asked about across ten analyst questions.

"Our operating expenses were higher than guided due to the recognition of estimated costs related to the technology and IT transformation in Q2, primarily in R&D."
— Roger Dassen, CFO

The gap to guide is roughly €77M. The Q3 R&D guide of roughly €1.2B implies the charge does not repeat, which in turn implies it was a discrete estimate rather than an ongoing programme cost. This appears to be the continuation of the restructuring disclosed in prior quarters, which the CFO alluded to later as "the reorganization that we talked about before."

Assessment. Immaterial in size, mildly unsatisfying in disclosure. Roughly €77M against €3.46B of operating income does not change any investment case. But an unquantified charge that pushes a guided line above its range, on a call where the company is asking investors to underwrite a 25% sequential ramp twice over, is exactly the sort of thing that should have drawn one question and drew none. We flag it as a disclosure-quality observation rather than a financial one.

10. Working Capital: H1 Free Cash Flow Went Negative

The quarter's least-discussed number. Q2 free cash flow was €1,317M, converting at 45% of net income. That is the good half. The full picture is that H1 2026 free cash flow was negative €1,291M against H1 net income of €5,674M, and H1 operating cash flow was negative €482.5M against positive €689.1M in H1 2025. Trade receivables rose from €3,023.0M at year-end to €7,252.6M, and receivables including finance receivables built roughly €4.16B across the two quarters. Implied DSO went from 29.2 days at Q4 2025 to 45.2 at Q1 to 70.8 in the quarter just reported, against 59.1 in the year-ago quarter. Cash and equivalents fell from €12,916.0M to €6,671.9M over the half.

There is a benign explanation and we think it is mostly right. Q4 2025 posted an extraordinary €10,940M of free cash flow on a customer-prepayment surge, and H1 2026 looks like the unwind of that. Current liabilities are already rebuilding, from €20,288M at Q1 to €22,155M at Q2. Shipment linearity in a quarter where units jumped from 79 to 91 would also push receivables up mechanically. Inventories were flat, which is the reassuring part: this is not a build of unsold goods.

Assessment. A watch item, not an alarm, but a real one. Nothing here suggests revenue quality has deteriorated, and the seasonal pattern in ASML's collections is well established. What concerns us is the interaction with the H2 ramp: if Q3 and Q4 deliver €25.9B of revenue, working capital will absorb cash before it releases it, and the year could end with a much weaker cash story than the earnings story implies. Not one analyst asked about this, and we can find no sell-side commentary on it at all. We will be watching Q3 receivables more closely than Q3 revenue.

11. What Management Did Not Say About the Cycle

Across a sixty-minute call in which the company raised its full-year revenue guide by 16%, disclosed that memory net system sales will grow more than 75% this year, and announced two consecutive 30% capacity expansions, the words "cycle," "downturn" and "correction" did not appear. Management offered no framing for what happens if DRAM pricing normalises, and no analyst asked for one.

Assessment. We do not read this as evasion, because nobody put the question. But the absence is itself the most important thing about this call. ASML is being asked to underwrite a capacity plan through 2028 on the strength of a memory market that is currently in the most acute shortage of the last decade, and the entire discussion, from both sides of the line, treated that shortage as the baseline. The one structural defence, which management made almost in passing, is that the expansion sits inside the existing footprint and is therefore cheap to reverse. That defence is stronger than the silence around it suggests, and it is doing more work in our rating than anything management said explicitly.

Analyst Q&A Highlights

Value-Based Pricing, and the Question of When

The opening exchange, and the most revealing of the call. The question was framed around a competitor-adjacent observation: if the industry's largest customer has publicly called High-NA too expensive, perhaps the market is telling ASML that Low-NA plus multi-patterning delivers more value than assumed, and perhaps Low-NA pricing should reflect that. The CFO's answer conceded the broader point immediately and generously, agreeing the current environment offers more pricing flexibility than ASML has historically enjoyed. He then declined to date it. The follow-up pressed the timing directly and got nothing more specific than a restatement.

Q: "When could that happen, Roger? I mean, to your point, it's not tomorrow. After tomorrow? When is it?"
— François Bouvignies, UBS

A: "That really depends on the order situation that you have with customers. As I said, it's obviously linked to order lead time. Of course, that varies from one to the other. The dynamic is clearly there, François."
— Roger Dassen, CFO

Assessment: A dodge, but an informative one. "The dynamic is clearly there" is a company confirming leverage it has chosen not to quantify, and the reason given (long order lead times) is genuine rather than evasive. What the exchange reveals is that pricing is a 2027-2028 earnings event that is not in the guide and not in consensus. The honest read is that management is negotiating and does not want to negotiate in public, which is the correct posture and also means investors get no visibility until it shows up in reported ASPs.

Whether 85 Systems for 2027 Is a Ceiling or a Placeholder

The most important question for 2027 estimates. If 2027 is already close to fully booked and the capacity plan is set at roughly 85 Low-NA EUV units, is that the maximum the company and its supply chain can support, or simply where the current demand-supply balance happens to sit? The answer was explicitly the latter, framed as a snapshot rather than a constraint.

Q: "The implied kind of 85 tools for next year, is that the ceiling of what you think you can support and your supply chain can support next year? It sounds like maybe the existing footprint could support more. Is there a possibility that that could move higher as we move closer to 2027?"
— Joe Quatrochi, Wells Fargo

A: "Joe, it's the balance as we see it today. The balance between demand and supply as we see it today gets us to the 30%, right? That's the way we do it. If customers are going to come to ASML and say, 'ASML, we need considerably more,' then just as we've been doing it in the past couple of months, we need to look ourselves in the eye, we need to look at all the supply chain and just see what can further be done. At this stage, given the conversations we have, we think the 85 is a nice representation of the balance between what customers are asking of us and what at this stage we've been asking ourselves in the supply chain to do."
— Roger Dassen, CFO

Assessment: The 85 is a floor dressed as a plan, and the answer was constructed to leave that unmistakable. Management has raised output repeatedly in the last six to nine months in response to demand and described the process as one it would simply repeat if asked. For modelling purposes, treat 85 as the base case and the distribution as skewed upward, with the caveat that the binding constraint is now ASML's own manufacturing rather than customer appetite. That is a much better problem than the one this company had eighteen months ago.

Committing Capacity Ahead of Purchase Orders

A pointed question about the order of operations, and the answer contains the single largest piece of risk in the capacity plan. Investigating a 110-unit scenario for 2028 implies either that the orders exist or that ASML is spending ahead of them. Management was refreshingly direct that it is the latter.

Q: "For you to increase, are you waiting for purchase orders from customers, before you start adding more capacity?"
— Krish Sankar, TD Cowen

A: "Clearly, Krish, we're not waiting until we get the orders, right? We said we're investigating the 110 scenario for EUV in Low-NA by 2028. Of course, we don't have orders for 110 EUV Low-NA at this stage. We're not waiting. We're preempting, but we are doing this because the demand signals that we're getting from customers that have not yet translated into full POs, but the demand signals we're getting from customers are quite strong."
— Roger Dassen, CFO

Assessment: The most honest answer on the call, and the one that defines the bear case. "We're preempting" on demand signals that "have not yet translated into full POs" is the textbook description of how capital equipment companies get caught at cycle tops. The mitigant, which management made elsewhere and not here, is that preempting inside an existing footprint costs very little and unwinds easily. The distinction between 2027 (order-covered, low risk) and 2028 (signal-driven, real risk) is the most useful line an investor can draw in this story, and management drew it clearly rather than blurring it. Credit for that.

The D-Model Sellout and What It Implies for 2027 Margins

A well-constructed question that used a supply constraint to extract a margin signal management had not volunteered. If the D models are gone from the 2027 mix, and E and F tools carry higher productivity and higher ASPs, then 2027 gross margin should mechanically improve. Management confirmed the mechanism while formally declining to guide the outcome.

Q: "Roger, you mentioned that next year there won't be any D in the mix. I just want to clarify, is that true? If so, is it fair to assume that gross margin next year should grow versus this year because if your mix is shifting to more volumes of E and eventually F?"
— Krish Sankar, TD Cowen

A: "Krish, at a certain point in time, we're simply sold out when it comes to D models… Of course, that mix will come with a better ASP than the mix that we have this year, because it also will come with higher productivity as we just laid out, and also comes with a better gross margin profile. Of course, I'm not going to guide you the gross margin for next year, Krish, but it is true that the mix effect on EUV next year will be more positive than it is this year."
— Roger Dassen, CFO

Assessment: A non-guide that is a guide. Confirming that the EUV mix effect will be more positive next year than this year, when this year's exit run-rate is already 56%, narrows the 2027 gross margin range considerably without a formal number. Combined with volume leverage and a still-growing Installed Base, 2027 above the 54-56% FY2026 band is close to arithmetic. This exchange did more for 2027 estimates than any other on the call.

Reconciling the Full-Year Guide: ASPs and the Second-Half Margin Bridge

The only exchange that engaged seriously with the shape of the guide rather than its level. The questioner had built the units and the Installed Base guidance into a model and could not make the full-year number work without something else moving, and asked what. The answer was the most detailed of the call and named four distinct drivers, plus a mid-point-by-mid-point walk to an H2 gross margin of roughly 56%.

Q: "When I look at your guidance on units for immersion and EUV, as well as your guidance for IBM, I struggle a little bit to get even to the midpoint. Sorry, I'm struggling to not get above the upper end. I apologize. I'm trying to understand if your ASPs are going higher in Q3, Q4, because of your Holistic Lithography attach rate on software, or is there any mix impact that is particularly meaningful?"
— Didier Scemama, Bank of America

A: "In terms of the gross margin, if you do the analysis and if you take midpoint by midpoint, I think you're looking for the second half at a gross margin of 56% approximately… Why is that? Again, it's the mix because we have quite a bit more immersion and Low-NA EUV in there, one. Two, because we have better priced EUV in there for the second half. Three, because of the installed base business, which remains quite strong. Four, of course, we have volume effect, and the fact that we have so much more volume in the second half than in the first half obviously gives you a positive fixed cost coverage."
— Roger Dassen, CFO

Assessment: Note what the question actually says, because it is the most bullish datapoint in the transcript and it came from a sceptic's spreadsheet. The complaint was not that the guide looked too high; it was that the disclosed units and Installed Base guidance make it hard to stay below the top of the range. If the components imply more than the guide, the guide is conservative. The margin bridge was answered thoroughly and all four drivers are structural. The ASP half of the question was, characteristically, left alone.

Memory: Whether Customers Are Closing the Supply Gap or Chasing HBM

The right question about the memory cycle, and the one that went least satisfyingly answered. The questioner wanted the split between HBM-driven lithography intensity and straightforward volume addition, because the two have very different durations: intensity is structural, volume is cyclical. Management declined to provide the split, on the grounds that it does not matter to ASML.

Q: "About the 75% memory growth in 2026, can you share with us how much is coming from HBM-driven lithography intensity versus volume addition, i.e., what I'm trying to know is also if with these investments, the memory manufacturer, your clients, are kind of closing the gap between offer and demand or are just investing in new technologies like HBM?"
— Stéphane Houri, ODDO BHF

A: "Well, I think the demand is really a combination of a lot more volume, both for HBM and DDR… It's hard to know the exact, I would say, ratio between the two. It doesn't matter that much in terms of technology for us because the DRAM, the RA itself is the same. HBM will require more wafer, there's a volume effect again… This is really this combination which create a bit the perfect storm for ASML on DRAM this year and most probably the next few years to come."
— Christophe Fouquet, CEO

Assessment: "It doesn't matter that much in terms of technology for us" is true and beside the point. It does not matter to ASML's tool design; it matters enormously to the durability of a 75% growth rate. The questioner asked whether customers are closing the supply gap, which is the question that determines whether 2028 demand exists, and it went unanswered. We take the litho-intensity component as durable and the volume component as cyclical, and the refusal to size them leaves investors unable to do that decomposition themselves. This is the most important unanswered question of the quarter.

Operating Leverage and the 60% Incremental Question

An attempt to convert the observed margin expansion into a forward target. The framing was precise: roughly 35% top-line growth against roughly 6% operating expense growth implies incremental operating margins around 60%, and the question was whether that is now the goal. Management would not accept the number but endorsed the direction unusually firmly.

Q: "We're clearly seeing in the numbers here with top line 35% growth and OpEx implicitly grown only about 6%. I'm curious, as you think about looking into 2027 and beyond, should we be thinking that this 60% incremental op margin is the target that you're going after?"
— CJ Muse, Cantor Fitzgerald

A: "Yes, CJ, we're not going to quantify that guidance. You are right that we have been managing our OpEx quite nicely, so SG&A and R&D… I would say that today we believe that with the team that we have today, we can really entertain a very aggressive roadmap going forward. All in all, I think you will continue to see us manage both R&D and SG&A quite nicely. As a result of that, the operating leverage that you imply, I think the operating leverage will indeed become better in the quarters and the years to come."
— Roger Dassen, CFO

Assessment: "The operating leverage will indeed become better in the quarters and the years to come" is about as close to a forward margin commitment as this management team gives. We verified the questioner's arithmetic and it holds: the full-year guide implies roughly 58% incremental operating margin, so the 60% framing is a fair characterisation of where the model is heading rather than a stretch. The reason it works is the deliberate choice to hold R&D headcount roughly flat while running the roadmap, which converts a cyclical revenue surge into structural margin.

What They're NOT Saying

1. Any bookings or backlog number, at all. ASML discontinued quarterly bookings disclosure from Q1 2026. "Extremely strong" order intake is the entire quantitative content. The last published figures were Q4 2025: €13.2B of net bookings and a year-end backlog of roughly €38.8B. In a quarter whose central claim is unprecedented forward visibility, the metric that would evidence it is no longer published.
2. The 2027 order coverage percentage, when asked directly. "Close to being fully covered" for 2027 and "a significant number of orders" for 2028 are the only calibration offered. The CFO explicitly refused: "We're not sharing order intakes. I'm not going to give you the coverage number." That refusal is the difference between a verifiable claim and an assertion.
3. Any framing whatsoever for cycle risk. Two consecutive 30% capacity expansions were announced into a memory market growing 75%, and neither management nor any of the ten analysts used the word "cycle." No downside scenario was offered, requested, or implied.
4. The size of the technology and IT transformation charge. It pushed R&D roughly €77M above guide and was disclosed in a single subordinate clause. Never quantified, never broken out, never asked about.
5. The HBM versus DDR split within the 75% memory growth. Declined on the grounds that it is technologically irrelevant to ASML, which sidesteps that it is the only way to judge how much of that growth survives a DRAM price normalisation.
6. When pricing leverage converts to revenue. Conceded three times, dated zero times. "The dynamic is clearly there" is the most specific formulation offered.
7. Any 2027 revenue or gross margin guidance. Deflected repeatedly, including a flat "I'm not going to guide in any way the business for next year." Reasonable this early, but it means the ~85-unit capacity plan is the only 2027 anchor investors have.
8. High-NA capacity for 2027 and 2028, when asked directly. The Low-NA and immersion capacity plans were given in detail; the equivalent High-NA question drew a discussion of optimisation principles and no number.
9. The negative H1 free cash flow and the receivables build. Zero airtime from management, zero questions from ten analysts, and no sell-side commentary we can find. H1 free cash flow of negative €1,291M against H1 net income of €5,674M passed without remark.
10. Export controls, and the export-control accommodation in the guide. Last quarter management volunteered that the FY range could absorb potential outcomes. This quarter the guide was raised by €6B with no restatement of that cushion, and the topic did not come up.
11. Any walk-through of the implied Q4. The full-year guide requires a quarter roughly 48% above the all-time record. Management never bridged to it, and only one analyst approached the arithmetic, from the direction of it being too conservative rather than too aggressive.
12. 3D integration and the XT:260 platform. Flagged as a signpost in our Q1 review and absent for a second consecutive quarter.

Market Reaction

• Pre-print setup (July 14 close): ADR $1,775.64; Amsterdam €1,555.80. Entering the print the ADR was +66.0% year to date (from $1,069.86) and +115.7% over twelve months (from $823.02), against an S&P 500 up 10.2% year to date. But it was also down 6.2% over the trailing 30 days (from $1,892.66 on June 15) and 10.7% below its 52-week closing high of $1,989.44. The 52-week closing range was $689.63 to $1,989.44.
• Options-implied move: 8.36% in either direction, more than double the four-quarter average realised move of roughly 4%. One outlier source put the expected move at 6.22% against a 7.00% historical average; the 8.36% figure carries two independent citations.
• Amsterdam (the primary listing): spiked to an intraday high of €1,678.20, +7.87%, essentially hitting the implied move, then gave back 7.67% from that high to close at €1,549.40, down 0.41%. Day range €1,536.20 to €1,678.20.
• US ADR: opened $1,825.06 (+2.8% gap), traded as low as $1,735.65 (-2.25%, genuinely negative), high $1,830.00 (+3.06%), closed $1,815.27, +2.23% (+$39.63). Volume 4.2M against a 2.1M 30-day average, 2.0x normal.
• Sector: the semicap complex fell while ASML rose. AMAT -2.73%, LRCX -3.08%, KLAC -2.55%, Besi -3.02%, ASM International -0.45%, TSMC -0.22%, SOXX -2.23%. The S&P 500 was +0.4%.
• Sell-side: no price target raises and no rating changes in either direction on the print. Two verifiable notes, both maintains at unchanged targets.

The ADR's +2.2% is a timing artifact and should not be read as the market's verdict. Amsterdam closes at 11:30 AM ET. The euphoric spike happened in the European morning immediately after the 07:00 CET release, hours before the US open, which is why the ADR never traded near it. The ADR opened after the European fade was largely complete, sank into negative territory, then rallied roughly 4.6% off its low through the US afternoon once Europe had gone home. Currency does not explain the 2.64-point divergence: EUR/USD was up 0.10% on the day, and the two listings tracked to within 0.06% on July 14. What the two tapes actually say is that European investors sold this print and American investors bought it.

The fade was not about the numbers; it was about who already owned them. Every price target circulating in today's coverage was set before the quarter: one house moved to $2,200 on June 22, another to €1,560 on July 2, another to €2,300 on July 6, another to €1,830 on July 8, another to $2,000 on July 14, the day before the print. The Street had spent a month front-running the beat-and-raise, and when it arrived there was nothing left to do. The most telling detail is that the most bullish sell-side model going into the quarter had already assumed 91 EUV systems for 2027 and 113 for 2028, both above the ~85 and ~110 ASML actually announced. A guide that beat published consensus by 11.7% still landed below the whisper. That is what a crowded long looks like: 33 of 44 ratings at the highest conviction tier, an average target implying only 10% upside, and no marginal buyer left to convert.

The sector move argues against reading the fade as an ASML-specific verdict. ASML outperformed SOXX by roughly 4.5 points on the day it reported. If the market had genuinely rejected this print, ASML would not have been the best-performing large-cap in its complex. The more coherent reading is that an AI-capex de-rate was working through semis generally and ASML resisted it, which is close to the opposite of the sell-the-news story. We cannot establish the causal direction, and we note that the two readings have very different implications: one says ASML is fine and the sector is wobbling, the other says the sector is wobbling because ASML told it something about capacity. The absence of any sympathy rally in AMAT, LRCX and KLAC on a 16% guide raise from the industry's most important supplier is, at minimum, unusual and worth sitting with.

Street Perspective

Debate: Is the Implied Q4 Credible?

Bull view: ASML has beaten its own guide every quarter through this recovery, including this one, and the H2 shipment schedule is not a forecast so much as a build plan for tools that are already ordered. The units exist, the customers exist, and the CFO walked the second-half margin bridge in enough detail to suggest the revenue bridge is equally worked. The one analyst who modelled the components bottom-up complained the guide looked too low, not too high.

Bear view: A €14.4B quarter is 48% above anything this company has ever printed, and it must arrive within six months. Two consecutive quarters of roughly 25% sequential growth have no precedent here. Guide-beating on a €9B base is a different exercise from stepping physical output up by half, and the entire full-year raise rests on it. Miss Q4 by 10% and the year lands at €42.5B, below the low end of a guide issued in July.

Our take: The bulls have the better of it, narrowly, and for one specific reason: the constraint is manufacturing rather than demand, and manufacturing is the thing ASML has been beating for the last three quarters. The bottom-up complaint from the sell-side is genuinely reassuring, because it means the disclosed units and Installed Base guidance already imply the number. But we would not describe this as a high-probability guide. It is a plan with real execution risk that the market is currently pricing as though it were an entitlement, and the single most likely path to a sharp de-rating in this name over the next two quarters runs directly through a Q4 that prints €13B instead of €14.4B.

Debate: AI Supercycle or Classic Top-of-Cycle Overbuild?

Bull view: Customers are signing multi-year long-term agreements with their customers, which gives them visibility they have never had and ASML visibility it has never had. 2027 is close to fully booked. Lithography intensity is rising independently of volume as multi-patterning converts to single-expose EUV, which is a permanent transfer of spend into ASML's line. And the expansion sits inside the existing footprint, so it is cheap to add and cheap to stop.

Bear view: A supplier announcing two consecutive 30% capacity expansions into a customer base growing 75% is the most reliable top-of-cycle signal in the history of semiconductors. The 2028 leg has no orders behind it, only "demand signals," by the CFO's own admission. And ASML is adding roughly 45% more wafer capacity to a DRAM market that is short today, which is precisely the mechanism by which shortages end. Memory is now driving the majority of ASML's growth, and memory has never once been secular.

Our take: Both sides are describing real things, and the resolution is in the distinction management drew and the market ignored. 2027 is order-covered and low-risk; 2028 is signal-driven and genuinely speculative. The bear case is correct about 2028 and largely irrelevant to the next twelve months. What tips us is the existing-footprint point: this is not a capital commitment that has to be defended through a downturn, it is a utilisation increase on assets already owned. ASML has engineered the cheapest possible way to be wrong about 2028, which is exactly what a company should do when it cannot forecast the cycle and knows it. The bears are right that the cycle will turn. They have not established that ASML will be hurt much when it does.

Debate: Does a 25% Earnings Revision on a 2% Stock Move Mean Cheap, or Saturated?

Bull view: Estimates moved roughly 25% higher today and the stock moved 2%. The forward multiple compressed from about 49x to about 40x on our numbers, and one widely-followed independent valuation shop that has been publicly bearish on ASML's multiple puts fair value at 35-40x forward, a band the stock has now entered because of this print. On pre-print forward multiples ASML screened cheaper than Lam and KLA and roughly in line with Applied, despite the best competitive position in the industry and having underperformed both year to date. The "it's priced in" argument was neutralised by the guide it was arguing against.

Bear view: The multiple compressed because the denominator moved, not because the market re-rated the business, and a 40x multiple on the peak year of a memory cycle is not obviously cheap. Positioning is extreme: 33 of 44 analysts at the top rating, an average target 10% above spot, and a stock that has doubled in twelve months. When a name gaps up 7.9% on a genuinely excellent print and closes red, the marginal buyer has been found and paid.

Our take: The bulls are right on the arithmetic and the bears are right on the tape, and the arithmetic wins over twelve months. Getting a business 18% cheaper on the day its guidance goes up 16% is the definition of a favourable revision, and revision slope has historically dominated positioning over a one-year horizon. But we hold the bears' point rather than dismissing it, because it is why our conviction steps down rather than up: a stock where the entire sell-side is already maximally positive and the average target implies 10% upside has to earn its return from estimate revisions alone, with no multiple help and no marginal convert left to buy it. That is a narrower path than we had at Q1, when the framework itself was still being disputed.

Model Update & Valuation Framework

The multiple compression is the core of the case. Entering the print at $1,775.64 on pre-revision FY2026E EPS of roughly €31.8, ASML traded at approximately 49x. Exiting at $1,815.27 on post-guide FY2026E EPS of roughly €39.7, it trades at approximately 40x. The stock rose 2.2% and got 18% cheaper on the day. Third-party data puts the post-print forward multiple at 37.9x, corroborating the direction. On FY2027E EPS of roughly €56, the stock trades at roughly 29x. All dollar conversions use EUR/USD of approximately 1.142, the rate implied by contemporaneous conversions of the reported actuals; a stronger euro is a modest tailwind to ADR earnings and is not modelled.

Risk-reward at $1,815.27. Base case $2,000-2,250 implies +10% to +24%. Bull case $2,600-2,900 implies +43% to +60%. Bear case $1,250-1,450 implies -31% to -20%. Probability-weighting our base at 55%, bull at 20% and bear at 25% gives an expected return of roughly +13%, against a mid-single-digit to high-single-digit expectation for the index. That is an Outperform, but it is a thinner one than the 2.5:1 asymmetry we published at Q1, and on a beta of roughly 1.4 the margin is not comfortable. The honest summary is that the fundamentals improved far more than the risk-reward did, because the price moved first.

What would change our mind. Upward: any pricing capture disclosed in reported ASPs, a Q3 print above €12B, or 2027 confirmed as fully booked with the unit plan raised above 85. Downward: a Q4 that prints below €13B, receivables continuing to build through Q3 without a collections offset, or any indication that the 2028 investigation is being scaled back.

Thesis Scorecard Post-Earnings

This quarter is graded against the standing thesis as last argued in our Q1 2026 review, not against a fresh set of pillars. ASML has not previously carried a formal thesis-of-record file; the pillars below are the ones we have been arguing publicly since the Q3 2025 upgrade, formalised here and carried forward from this point.

Overall: the thesis is strengthened on fundamentals and narrowed on risk-reward, and its character has changed. Five of six bull pillars confirmed, one new bull pillar established, two of three standing bear points weakened or retired, and three new bear points introduced. That looks like an unambiguous win, and on the business it is. But the pillar that carried this rating since Q3 2025, the progressive restoration of the demand framework, has now fully resolved and is retired: there is no more framework to restore. What replaces it is a narrower, more operational proposition. ASML must physically build roughly €25.9B of revenue in six months, and then a June 2027 Capital Markets Day must reset a long-term framework whose floor the company will clear this year. Both are likelier than not. Neither is the near-certainty that the framework trade was.

Action: Maintaining Outperform. Conviction 7 of 10, down from an implied 8 at Q1. Existing positions: hold full weight; the earnings revision outruns the price move and we see no reason to reduce into a 25% estimate raise. New initiations: we would not chase at $1,815.27 into a crowded long where the average Street target implies 10% upside. The better entry is any wobble on Q3 receivables or on Q4 ramp anxiety, which we think is more likely than not to arrive at some point between here and the January print. Position sizing should reflect that this is now a manufacturing-execution bet with a dated re-rating catalyst eleven months out, not the framework-restoration trade it was for the last three quarters.