ASML Holding NV (ASML.US) reported third-quarter revenue and gross margin that both hit record highs, driven by continued momentum in AI-related logic and memory chip capacity investment, which in turn strengthened demand for semiconductor manufacturing equipment.
Third-quarter total net sales reached 11.7 billion euros, near the upper end of the company's guidance range. Net system sales were 8.4 billion euros, with EUV systems contributing 4.1 billion euros and non-EUV systems contributing 4.3 billion euros. The EUV system sales included one High-NA EUV system. Logic accounted for 54% of net system sales, while memory accounted for 46%. Sales from installed base management, including services and upgrades, totaled 3.3 billion euros, exceeding guidance by nearly 400 million euros. Chief Financial Officer Roger J. Dassen said the beat was mainly driven by demand for productivity upgrades in EUV and deep ultraviolet (DUV) systems. Gross margin was 57.5%, above guidance, benefiting from a high-margin component within installed base management. Net income was 5 billion euros, including about 500 million euros from an increase in the valuation of ASML's initial investment in Mistral AI. Earnings per share were 12.95 euros.
Looking ahead to the fourth quarter, ASML expects sales between 14.2 billion and 15.2 billion euros, with a gross margin of about 57%. Installed base management sales are expected to be about 3.7 billion euros. For the full year, the company expects total net sales of 44 billion to 45 billion euros, with a gross margin of about 56%. This outlook implies sales growth of about 36% year over year, mainly driven by an expected increase of more than 45% in EUV sales, about 20% growth in non-EUV related system sales, and about 50% growth in installed base management sales.
As of the end of the third quarter, ASML held 12.7 billion euros in cash and equivalents and short-term investments. Free cash flow was 11.4 billion euros, which Dassen said reflected customer order momentum and related prepayments. During the quarter, the company repurchased about 4.1 billion euros of shares under its 2026-2028 buyback program and paid an interim dividend of 1.88 euros per ordinary share. The second interim dividend of 1.88 euros per share is scheduled to be paid on November 4.
Capacity Expansion Amid Strong Demand
Chief Executive Officer Christophe D. Fouquet said the pace of AI infrastructure buildout remains strong, while supply shortages in advanced logic and memory nodes have further intensified and the shortage window has been pushed further out. He said customers expect demand for their products to exceed supply at least through 2028, and are therefore accelerating capacity planning and considering expansion beyond that period. ASML said it is increasingly receiving orders for 2028, and customers have also expressed intent to place orders for 2029 and 2030. Management said it has not yet accepted these longer-dated orders because the company is assessing total demand and its own expected capacity for those years before allocating output among customers.
The company expects to produce about 65 low-NA EUV systems in 2026, an increase of nearly 50% from the 44 units shipped in 2025. It now also expects DUV immersion system deliveries in 2026 to be similar to last year, despite an initially lower production target. ASML had previously planned to increase capacity for both Low-NA EUV and DUV immersion systems by 30% in 2027 and has been evaluating another 30% increase in 2028. Dassen said the current plans for 2027 and 2028 cannot fully meet customer demand, and the company is now exploring further expansion beyond the 2028 growth level previously considered. Fouquet said the company does not currently view Carl Zeiss or other suppliers as a bottleneck to planned capacity expansion, but ASML is still working closely with its supply chain. Management identified improving factory floor space utilization and shortening cycle times as important levers for expanding output.
Productivity Upgrades and Technology Roadmap
ASML plans to begin small-volume introduction of its next-generation Low-NA EUV system, the NXE:3800F, in early 2027. The system specifies throughput of 260 wafers per hour, compared with 230 wafers per hour for the NXE:3800E. The company also plans to introduce a next-generation critical DUV immersion system with throughput of up to 330 wafers per hour. Fouquet said higher-throughput tools give customers more wafer capacity and create additional value, and ASML expects to capture that value through pricing. Dassen said completed pricing discussions with customers have led to higher prices for the company's most critical high-volume products. These changes are not expected to materially affect 2026 results, will begin to contribute gradually from 2027, and are expected to be fully reflected from 2028 onward. Dassen said a large portion of the price increase should help improve gross margin, although some of the benefit will be used to cover costs related to ASML's capacity ramp and supply chain expansion.
High-NA EUV Progress
Management emphasized growing momentum for ASML's next-generation lithography technology, High-NA EUV. Fouquet noted that Intel Corp (INTC.US) has exposed more than 1 million wafers using High-NA EUV systems, with performance meeting or exceeding comparable layers patterned with Low-NA EUV. According to ASML, Samsung and Taiwan Semiconductor Manufacturing Co Ltd (TSM.US) have each announced plans to introduce High-NA EUV into mass production in 2028 and 2030, respectively. With availability of the existing High-NA fleet approaching 90%, Fouquet said adoption should increase over time "node by node." ASML and TSMC also launched an industry initiative to develop larger EUV photomasks. The initiative aims to build a 12-inch mask pilot line by 2031 and reach mass production by 2033. Fouquet said broad customer support for the initiative shows the industry expects High-NA EUV usage to continue increasing over time.
The company also shipped its first YieldStar 550 metrology system during the quarter. ASML said this overlay measurement system is designed to support tighter process control requirements for 14-angstrom logic nodes and 1D DRAM nodes, while delivering more than 60% higher productivity than the previous YieldStar generation.
Conference Call Participants
Head of Investor Relations - Jim Kavanagh
Chief Executive Officer - Christophe D. Fouquet
Chief Financial Officer - Roger J. Dassen
Operator: Hello everyone, and thank you for waiting. Welcome to the ASML 2026 third-quarter financial results conference call for October 14, 2026. After the speakers' introductions, we will conduct a question-and-answer session. If you would like to ask a question during the call, please press 11 on your telephone. You will then hear an automated message advising your hand is raised. To withdraw your question, please press 11 again. Please note that today's call is being recorded. I would now like to turn the call over to Mr. Jim Kavanagh. Please go ahead.
Jim Kavanagh: Thank you, operator. Welcome everyone. I am Jim Kavanagh, ASML's Head of Investor Relations. Joining me on the call today are ASML CEO Christophe D. Fouquet and our CFO Roger J. Dassen. The subject of today's call is ASML's 2026 third-quarter results. The call will last 60 minutes, and questions will be taken in the order received. This call is also being webcast live at www.asml.com. A transcript of management's opening remarks and a replay of the call will be available on our website shortly after the call ends. Before we begin, I would like to remind listeners that management's comments during this call will include forward-looking statements. These forward-looking statements involve significant risks and uncertainties. For a discussion of risk factors, I refer you to the safe harbor statement in today's press release and the presentation on our website at www.asml.com, as well as ASML's Annual Report on Form 20-F and other filings with the U.S. Securities and Exchange Commission. I would now like to turn the call over to Christophe D. Fouquet for a brief introduction.
Christophe D. Fouquet: Thank you, Jim. Welcome everyone, and thank you for joining our 2026 third-quarter results call. Before we begin the Q&A, Roger and I would like to provide an overview and some comments on our 2026 third-quarter results, as well as some additional perspective on the current business environment and future business outlook. Roger?
Roger J. Dassen: Thank you, Christophe, and welcome everyone. I will start by reviewing our 2026 third-quarter financial results and then provide guidance for the fourth quarter and full-year 2026. Let me begin with our third-quarter results. Total net sales in the 2026 third quarter reached a record 11.7 billion euros, at the high end of our guidance range, reflecting strong execution on our capacity ramp. Net system sales were 8.4 billion euros, including 4.1 billion euros of EUV system sales, including one High-NA EUV system sale, and 4.3 billion euros of non-EUV system sales. Within net system sales, logic accounted for 54% and memory accounted for 46%. Installed base management sales in the quarter were 3.3 billion euros, nearly 400 million euros above our guidance, mainly due to strong demand for productivity upgrade products in EUV and DUV. Gross margin in the quarter was above guidance at 57.5%, mainly driven by the contribution of a very high-margin component in our installed base management business. Our operating expenses were in line with guidance, with R&D expenses of about 1.2 billion euros and SG&A expenses of about 400 million euros. The effective tax rate in the third quarter was 16.1%, and the expected annualized effective tax rate for full-year 2026 is about 17%. Net income in the third quarter was 5 billion euros, including about 500 million euros from an increase in the valuation of our initial investment in Mistral AI, representing 42.5% of total net sales, with earnings per share of 12.95 euros.
Turning to the balance sheet. At the end of the third quarter, our cash, cash equivalents and short-term investments were 12.7 billion euros. Free cash flow in the third quarter was 11.4 billion euros, supported by strong customer order momentum and related prepayments. Turning to shareholder cash returns. In the third quarter, ASML paid its first interim dividend for 2026 of 1.88 euros per ordinary share. In addition, we repurchased about 4.1 billion euros of shares in the third quarter under the current 2026-2028 share buyback program. The second quarterly interim dividend for 2026 will be 1.88 euros per ordinary share, payable on November 4, 2026.
Next, I would like to discuss our expectations for the fourth quarter and full-year 2026. We expect fourth-quarter total net sales between 14.2 billion and 15.2 billion euros, with a gross margin of about 57%. We expect fourth-quarter installed base management sales of about 3.7 billion euros. We expect fourth-quarter R&D expenses of about 1.2 billion euros and SG&A expenses of about 400 million euros. Regarding full-year expectations, we expect total net sales for full-year 2026 between 44 billion and 45 billion euros, with a gross margin of about 56%. Total net sales are expected to grow about 36% year over year, reflecting our expectation of more than 45% growth in EUV, about 20% growth in non-EUV related system sales, and about 50% growth in installed base management sales.
Turning to end-use. This year, we expect logic net system sales to grow about 10% and memory net system sales to grow about 75%. Memory growth is driven by both customer capacity expansion and increased EUV adoption at advanced DRAM nodes. As previously stated, we continue to expect China to account for about 20% of our total net sales this year. In summary, the third quarter set records for both total net sales and gross margin, and we expect our business to continue accelerating in the fourth quarter as we execute against very strong customer demand for our products and services. I would now like to turn the call back over to Christophe.
Christophe D. Fouquet: Thank you, Roger. If we step back and look at the market environment, we see the pace of AI buildout continuing to be very strong. Supply shortages in advanced logic and memory nodes continue to intensify in both scale and duration. Against this backdrop, we had very constructive discussions with customers throughout the quarter, and the demand signals we are receiving continue to strengthen. Our leading-edge high-volume customers are sharing very good visibility, driven by the growing number of long-term commitments they have with their customers. This extended visibility has led them to conclude that demand for their products will exceed supply at least through 2028. As a result, customers are accelerating capacity build plans and preparing for further expansion beyond 2028 to better meet expected demand levels. For us, their drive to add capacity translates not only into greater near-term demand across our entire lithography portfolio, but also improved long-term visibility and demand, especially for EUV products. As customers turn their attention to demand beyond 2027, our order momentum continues to strengthen. We are increasingly receiving orders for 2028, and customers have also expressed interest in placing orders for 2029 and even 2030. As our customers prepare to meet growing demand, we in turn continue to focus on increasing capacity to support their needs. Roger will now share more details on our latest plans to provide customers with additional capacity. Roger?
Roger J. Dassen: Thank you, Christophe. We have been very successful in our capacity ramp this year. For Low-NA EUV, we expect to produce about 65 systems in 2026, an increase of nearly 50% from the 44 systems shipped last year. For DUV immersion systems, although we initially targeted output well below the prior year, we now expect 2026 deliveries to be similar to 2025. Looking ahead, we shared last quarter our intention to increase capacity for Low-NA EUV and DUV immersion systems by 30% in 2027 and to study another 30% increase in 2028. However, given the dynamics this quarter, we now expect that the current planned output levels for 2027 and 2028 will not fully meet customer demand. We are therefore stepping up efforts to produce more systems, especially beyond 2027. To achieve this, we are now studying expanding 2028 capacity even above the 30% year-over-year growth target we were evaluating last quarter, covering both Low-NA EUV and immersion systems.
In addition to shipping new systems, we are supporting customers' need for more wafer capacity through productivity upgrades on the EUV and DUV installed base, as well as through new product launches. The ability to increase capacity through upgrades within the same footprint is very attractive to customers because the effect is an immediate increase in output capability. We will continue to focus on this important value driver, expanding the upgrade portfolio for customers in the coming years and offering new productivity options. New product launches are another lever we use to support increased wafer capacity. We plan to begin small-volume introduction of our next-generation Low-NA EUV system, the NXE:3800F, in early 2027. The system is specified at 260 wafers per hour, compared with the capability of 230 wafers per hour for the NXE:3800E. We also plan to introduce a next-generation critical DUV immersion system running at 330 wafers per hour. These new products will ramp over the coming years, providing the industry with more wafer capacity. I will now turn the call back to Christophe for a technology update.
Christophe D. Fouquet: Thank you, Roger. This quarter, we saw meaningful progress in High-NA EUV. Several industry announcements provided evidence of the technology moving from development to production. Intel announced it has exposed more than 1 million wafers using its High-NA EUV systems, with performance meeting or exceeding comparable layers patterned with Low-NA EUV. In addition, Samsung and TSMC each announced plans to introduce High-NA EUV into high-volume manufacturing in 2028 and 2030, respectively. With availability of the existing High-NA EUV fleet approaching 90%, we now expect adoption at these nodes to increase. To further ensure chipmakers can fully benefit from High-NA EUV, we also launched an industry-wide initiative with TSMC to develop larger EUV photomasks. This collaboration aims to establish a 12-inch mask pilot line by 2031 and move into production by 2033. Broad support for this transition reflects customer interest in maximizing the value of High-NA EUV and investing in next-generation technology.
In summary, a combination of very strong demand fundamentals, improved visibility and rising lithography intensity has strengthened our confidence in the business trajectory. The focus remains on supporting our customers as they continue to expand capacity across our multi-year period to advance their logic and memory nodes, while also delivering enhanced productivity by executing our technology roadmap. In the third quarter, we also shipped our first YieldStar 550B metrology system. This overlay measurement system is designed with higher accuracy and precision to support tighter process control requirements for logic 14-angstrom nodes and DRAM 1d nodes. The system is designed to deliver more than 60% higher productivity than the previous YieldStar generation, enabling customers to measure more wafers and thereby accelerate yield improvement in their processes. This latest-generation system continues to support broad customer adoption of YieldStar, with demand expected to more than double over the coming years compared with 2025. Finally, the transition of High-NA EUV to high-volume manufacturing creates a very attractive opportunity for customers to enhance the economics of their future roadmaps. We remain fully committed to helping them successfully insert it. With that, we are happy to take your questions.
Jim Kavanagh: Thank you, Roger and Christophe. The operator will soon guide you through the Q&A protocol. Before that, I would like to ask everyone to ask only one question, with a brief follow-up if necessary. This will allow us to take as many callers as possible. Operator, please provide the final instructions and then connect the first question.
Q&A Session
Operator: Thank you. As a reminder, to ask a question, please press 11 on your telephone and wait for your name to be announced. To withdraw your question, please press 1 again. We will now take the first question. Today's first question comes from Joe Quatrochi of Wells Fargo. Please go ahead.
Joe Quatrochi: Yes. Thank you for taking the question. I was wondering if you could provide some additional color on the capacity expansion beyond the 30% that you need. And relative to the mid-80s number we discussed last quarter, what do you think you can do in 2027?
Christophe D. Fouquet: Well, I think we are basically working very hard to increase capacity. As Roger explained, we did a lot of work this quarter, and that is now real. The same will be true next year, and the same in 2028. The reason, as we explained, is that we continue to see customers asking for more tools. Now, if we look at some of the opportunities we see, it is of course easier for us to realize those later. So mainly in 2028 rather than in the near term. So that is what we are looking at. I will not give new numbers for 2027 or 2028 because it is too early to give those numbers now. We will continue to do everything we can to secure additional capacity.
Joe Quatrochi: Okay. And then as a follow-up, you talked in your prepared remarks or transcript about value-based pricing having a bigger impact in 2027 and 2028? Could you share some additional details on that and how to think about the gross margin trajectory from here?
Roger J. Dassen: Yes. So that is two questions, right? First, on pricing. As we mentioned in July, at that time we were in conversations with customers about price increases. The conversations have been completed, and they did result in price increases for what we call our most critical products for customers. So that is done. As we also said in our prepared remarks, this will not have an impact on 2026. You will see some impact in 2027, but I would say the full impact really begins in 2028. So from 2028 onward, you will see a sustained impact, but during 2027 you will see it begin to show up in a positive way in revenue, average selling price and gross margin. Your second question on gross margin is obviously a broader question. But let me address it. So what are the favorable and unfavorable factors for gross margin? As you can see, gross margin this quarter was very, very strong. Stronger than we expected. I would say mainly due to the extra 400 million euros in our upgrade business. But we should also not forget that if you look at the full-year 56%, it is not just the upgrade business. Right? It is also higher volume. It is also because we have a very strong Low-NA EUV business in there. A very strong immersion business. So those are all helpful. And then, obviously, the question is how this develops in the years beyond this year. Right? So what is the impact for 2027 and beyond? Of course, we will not give a gross margin percentage for 2027. That will come in January 2027. But if you look at the key drivers of gross margin this year, you have to ask yourself how many of them can continue into 2027.
First, volume, plus the 30% growth in 2027 that Christophe just told you about, should of course help. Right? As long as we can control the cost of capacity additions, volume should give you better fixed-cost coverage. So that should be positive in that respect. The second factor is product mix. Again, the two key drivers of volume are the additional 30% in immersion and Low-NA EUV. From a gross margin perspective, both are clearly very positive. Of course, we will continue to add quite a lot of dry business this year, and the same next year. I would say the strength of EUV demand and immersion demand in the mix should also help. Then, as you mentioned, we have price increases, which, as I mentioned, will have some impact in 2027, with a bigger impact in 2028, but it will have some impact in 2027, and that will be helpful. Finally, on the positive side, I might point to the product mix effect within EUV. Right? Next year's product mix effect will be better than this year's because this year we still have quite a lot of DE models in our mix. Next year that will no longer be materially the case, and we will also have some APPS in there. That will help within total EUV. So those are the key factors.
Then what do we need to be vigilant about? First, as I mentioned, is ramp costs. Right? So how much cost are we incurring as we ramp for 2027 and beyond? So that is one thing we need to manage well. Then there is the swing factor of the upgrade business. Because as we mentioned, the upgrade business this year has been very, very strong. That does raise the question of how sustainable the upgrade business is. If you dig into the sustainability of the upgrade business, there are several elements to keep in mind. First, as of today, we still see a sizable addressable market in the upgrade business. If you look at the number of upgrades we have executed in the field, that is quite a lot of tools, but there are still quite a lot of tools that have not yet been upgraded. So that gives you good potential for future quarters and the year beyond. Second, as we mentioned in the video, we are also actively driving upgrade products. So we will bring new upgrade products to market, which of course will open up a completely new addressable market for us in the installed base. So that is helpful. And then, of course, there is the question of how the market will continue in a demand environment like the current one. As we just said, there is significant demand for these upgrades. So I would say upgrades are an important swing factor in how we see gross margin developing in 2027 and beyond. This year has been very strong. But I think, at least on our side, we are doing everything we can to create as many opportunities as possible for the upgrade business going forward.
Joe Quatrochi: Thank you for all the detail. Thank you.
Roger J. Dassen: You're welcome.
Operator: Thank you. Today's next question comes from Krish Sankar of TD Cowen. Please go ahead.
Krish Sankar: Yes. Hi. Thank you for taking my question. Christophe or Roger, my first question, I understand you do not want to give exact capacity numbers. I just want to know, in a strong demand environment, is there a natural ceiling on your EUV output, whether due to VDL or Carl Zeiss capacity? In other words, if demand is there, could you do 140 or 150 units in 2028? Also, do you worry that customers are pulling in orders or giving larger forecasts just to secure a slot? I have a brief follow-up after that.
Christophe D. Fouquet: Yes. So maybe let me try to answer both questions. First, on what we are doing to increase capacity. I think I mentioned this earlier. This is a task we started a few months ago. I think there are really three elements. The first is optimizing our existing production sites so that we can basically build our machines. As you know, we have a fairly large footprint, and we are starting to make sure that most of that footprint can be used for output. This is true for Low-NA EUV. By the way, it is also true for DUV, including immersion. This has already created some opportunities that we can leverage and build on this year, next year and in 2028. The second, I think a very big opportunity and still a major parameter, is cycle time. That means the time we spend on a given machine in the factory. As we mature, for example on the NXT:1380E, we are able to shorten cycle times for certain given sequences, but we are also able to further optimize the sequences. That means some of the things we initially needed to do to validate a tool, over time, with customer agreement, we believe are no longer needed. I would say we can push this approach quite far. This is one of the truly key levers we see, basically, to move forward. As I mentioned earlier, the opportunities continue to materialize over time. That gives us more hope and more opportunity. And of course, the supply chain, indirectly, as in your question or comment, is something we must supply. That is something we take into account, starting with Zeiss. We do not believe Zeiss is a bottleneck today. For the capacity expansion we are discussing. Even beyond the 30% in 2028 that we mentioned. We have done a lot of work in recent years with the entire supply chain to make sure the supply chain does not become our problem. So of course we continue to watch this very carefully. We understand that every time we add capacity, of course, we create more challenges there. So we do that. The last thing I would say is that we have time. I mean, if you look at the progress we have made this year, I think it is very significant. We feel quite good about next year. And the time we have, again, for next year, for 2028, remains important for us to continue to be very creative in all those areas. So that is the answer to your first question.
On the second, I think I mentioned in my introduction that we had very good conversations with customers. The conversations were not only about how many tools they need. They were also about exactly when those tools will land in their fabs. And when it comes to what we call demand, the increased demand we refer to, I think we are very capable of validating demand through the more detailed plans customers share with us about fab construction, releases, and so on. So I think when we talk about demand, when we refer to demand in those discussions, we truly mean demand we believe is real, rather than demand exaggerated by any potential purchasing behavior by customers.
Krish Sankar: Got it. That is very helpful, Christophe. Then a brief follow-up for Roger. On value-based pricing, is this across the board? Is it for systems, services, spare parts, everything? I guess based on your comments, you are not repricing the backlog. Can you clarify? Thank you.
Roger J. Dassen: Yes. Krish, I will not share exactly how we do it. But as I mentioned, we are raising prices for what we call our most critical products. The products that bring the most value to customers. So these are high-end, high-volume manufacturing products. These are the products that provide the most value to customers in the current environment, and these are the ones for which we are raising prices. You will also understand that upgrade prices are closely linked to equipment prices. So that, that, they will also follow, upgrade prices will also follow that pattern. That is how we do it, exactly how we do it, when we do it, and so on. I do not want to go into any detail. But the bottom line is that you will begin to see the effect during 2027. But the full effect you will begin to see from 2028 and beyond.
Krish Sankar: Thank you very much, Roger.
Roger J. Dassen: You're welcome.
Operator: Thank you. Today's next question comes from Sandeep Despande of JPMorgan. Please go ahead.
Analyst: Thank you for letting me ask a question. My question goes back to the indication that you are considering adding more than 30% capacity in 2028. Increasing capacity involves multiple components, and some of ASML's suppliers will need to add physical capacity to be able to provide you with that kind of huge output growth, especially considering there is already 30% growth in 2027. Have those discussions already started with customers? Will ASML need to make any substantial payments to those suppliers to increase capacity? Are there any particular obstacles? And given that you yourself do not have additional physical capacity before 2029, how is the whole supply chain being managed to be able to produce more capacity in 2028? I have a brief follow-up after that. Thank you.
Christophe D. Fouquet: So I will try to add a little to what I said earlier. I think the only real challenge we have with the supply chain is making sure they can ramp. I think on their side, there is no issue with the intensity of demand, with the intensity of the ramp. I think the main challenge with all our suppliers, because in this case it is with all suppliers, is making sure they can grow 30% per year. Now Roger mentioned earlier in his introduction. We grew EUV 50% this year, which basically shows that these are things we can do together with suppliers. As you can imagine, this is a discussion we have with them regularly. We mentioned 30% growth in 2027 and more than 30% growth in 2028 as a result of our confidence that the entire supply chain can of course follow our own demand, whether EUV or DUV. I would say this exercise can of course be extended as needed. Because we are also continuing to add suppliers. So we continue to make sure we have more and more people who can produce some of our parts. Especially non-critical parts. When it comes to very critical parts, you know, Zeiss again comes to mind, and we are absolutely strategically aligned with the industry's trajectory, and therefore also with the demand trajectory. So these are very, very detailed and intensive discussions. But again, every number we share with you is the result of us validating that the supply chain can of course follow those numbers.
Analyst: Thank you, Christophe. A quick one, Roger, on gross margin, your fourth-quarter gross margin guidance is down versus this quarter, despite significantly higher revenue? And then, again on your earlier answer about value-based pricing, should we assume that most of the value-based pricing-related growth will go directly to the gross margin line? Or do these increases have a margin number?
Roger J. Dassen: On the first one, fourth-quarter gross margin, slightly lower than the third quarter. That is mainly related to some ramp costs we see. Right? So we are ramping for the growth Christophe just talked about. That has a small impact on the gross margin expectation. That is why you would expect the fixed-cost coverage benefit from higher volume to be offset, and more than offset, by the additional costs we are now incurring in the fourth-quarter ramp. That is the main reason there. As for the price increases and the extent to which they fully add to gross margin, of course, the same applies. Right? I mean, one of our arguments in the customer conversations is that we are incurring costs to ramp. So ramp costs give you some additional cost. So it will certainly add to gross margin. But again, you will see us incur some ramp costs in 2027 to reach the higher numbers Christophe just talked about. This is both for ourselves and for the supply chain, for the reasons you just mentioned. Right? We need to make sure the supply chain is fully aligned with our objectives. To do that also means some help in places to achieve that. So part of the price increase needs to cover the extraordinary ramp we are going through. But I would say a large part of it still directly adds to gross margin.
Analyst: Thank you, Roger.
Operator: Thank you. Today's next question comes from François Bourvignet of UBS. Please go ahead.
Francois Bourvignet: Thank you very much. Just want to follow up on this shortage situation. Obviously, we are in a very unprecedented period, with your customers facing capacity constraints. But today, you also claim that on your side, demand will also exceed supply. So I mean, normally the behavior is for people to order as much as possible. Especially if you start to see some ramp to make sure they have tools. We have seen some long-term agreements in other industries, some prepayments, to try to get a more realistic sense of demand. I was wondering whether you are considering changing the way you contract with customers in these times? For example, increasing prepayments, or recognizing more firm orders earlier. To make sure your customers are committed.
Roger J. Dassen: Yes. If we talk about this, we have said before, if you place an order with ASML, it is not a frivolous piece of paper that you sign your name under. It comes with two serious commitments. One serious commitment is the commitment of a prepayment, as I qualified earlier, a prepayment sufficient to hurt. So that is one element. Second, the purchase order is a real commitment. On both sides of the equation, that also means it is deep. If you want to get out of that commitment, it will not be easy. Right? There is a cost, not just the prepayment. And more than that. So therefore, before a customer signs a purchase order with ASML, they have already made up their mind. They do not do it lightly. On whether you would accept orders earlier? Well, as Christophe mentioned, we are receiving requests from customers asking us to accept their 2029 and 2030 orders and purchase orders. The reason Christophe phrased it that way, saying customers are asking us, and therefore we have not yet accepted those orders, is solely that we want to make sure we have a clear picture of the overall demand for 2029 and 2030, what customers are currently signaling. And compare that with the capacity expansion we see at that time. We want to do that because we do not want to accept orders now from customers A, B and C and then tell customer D, who has not yet placed orders for 2029 and 2030. Sorry. We are sold out. I mean, that would be completely inappropriate for customers who have been our business partners for a long time. So we are really using this time now to compare. To get a good understanding of what the total demand customers are signaling for the 2029, 2030 timeframe is. What does that look like? What is a reasonable expectation of our capacity for the 2029, 2030 timeframe? Once we have that picture, we are ready to accept orders. And also accept our side of the commitment to execute those orders. So our only hesitation at this stage is that we really want to do our homework, to be very diligent in accepting the orders customers are offering us, to make sure no customer feels left out.
Francois Bourvignet: Thank you. That was a very detailed and good answer. The second question is about the High-NA news flow. I mean, Christophe, you mentioned high-volume manufacturing in memory in 2028. And also TSMC in 2030. If I look back at Low-NA in the past, when we started high-volume manufacturing, I mean, we saw significant growth between shipments in those years. So I was wondering whether we should think about High-NA shipments increasing significantly on that basis in 2028, and possibly reaching double-digit High-NA shipments as we move forward.
Christophe D. Fouquet: Well, I think we are still looking at a very early ramp. I think you will indeed see that every year our customers will need more tools. As they validate in manufacturing that they like High-NA, and they likely plan to use more layers over time. I think this announcement, and TSMC's push first on 12-inch critical quality, is very interesting there because the first signal it gives is that, in customers' minds, High-NA will be used more and more. Because if not, nobody would really bother to do 12-inch masks on High-NA. So the push for 12-inch masks and broad customer adoption and consumption is just a sign that everyone has in mind developing basically more and more High-NA adoption over time. So with that, you know, it is always hard to say exactly how this will basically play out. We of course also have a very large ramp, so there is a lot of focus on building capacity, which sometimes has a positive or negative impact on the High-NA rollout, depending on whether customers see High-NA as additional capacity or perhaps as a slight slowdown. But, the number will certainly grow year by year. And, I think the message all our customers conveyed this quarter by participating in the 12-inch alliance is that they believe this truly becomes a very important technology over time, and the number of participants will certainly grow over time.
Francois Bourvignet: Great. Thank you very much.
Christophe D. Fouquet: You're welcome.
Operator: Thank you. Today's next question comes from Lee Simpson of Morgan Stanley. Please go ahead.
Lee Simpson: Hi, good morning. Thank you for letting me ask a question. I just want to follow up on the last set of answers. I find it interesting that if you look at non-current liabilities, there was a fairly large jump quarter over quarter. It is on the order of about 5 billion euros. And, typically most of that is prepayments. So can we infer from that the relative scale of large order intake in the third quarter? That is the first question. I think maybe at the same time, Christophe, in your answer to the previous question, it seemed to imply that maybe it is new customers that are bringing you to the point of considering the 2029 and 2030 framework. And as you said, you still want to keep the marginal buyer in the mix. Can we say that most of that comes from new customers? Thank you.
Roger J. Dassen: On the first one, non-current liabilities, does that include prepayments? Yes, it includes them. Right? So in that sense, it is an interesting indicator of how order intake is developing. I should say it is a somewhat lagging indicator. Because typically, prepayments and receiving the money lag slightly behind the actual signing of purchase orders. So it is a somewhat lagging indicator, but in the absence of any purchase order numbers from ASML, it is a decent proxy. Yes. Yes.
Christophe D. Fouquet: I am not sure I fully understood the second question.
Roger J. Dassen: So I think the second question, I think it is about the customers who expressed interest in doing so for 2029, 2030, and I think your question, Lee, is whether that is an entirely new customer? Possibly starting with T, which may be locking existing customers out in that way. But I had better phrase it better. I think the answer is that the interest expressed is much broader. It is much broader. This is not about one specific customer. There is broader customer interest in placing purchase orders in that timeframe.
Christophe D. Fouquet: Yes. Maybe to add, because you indirectly mentioned a new customer. I think what is interesting is that on Low-NA EUV, we talked about increased EUV adoption. We continue to see, in fact, new customers on Low-NA EUV, especially when it comes to DRAM. I would say this is a bit beyond the T customer. So this is another interesting dynamic that we really have to take into account for 2028, 2029, 2030. Of course, this is another challenge when we have capacity discussions. But I think it is good news when we talk about validation of how EUV will be used in the future. The last thing I would say is that we see a similar trend in High-NA. So I think that is also a very good set of news we are getting from potential new customers.
Lee Simpson: Great. Thank you.
Christophe D. Fouquet: You're welcome.
Operator: Thank you. Today's next question comes from Alexander Duval of Goldman Sachs. Please go ahead.
Alexander Duval: Yes. Thank you very much for taking the question. Just want to turn slightly to immersion DUV. You talked today about reaching 330 wafers per hour productivity. So I was wondering what that might mean directionally for pricing in 2028, and whether, as we think about the innovation you are delivering in immersion, we should think about the average selling price uplift that we have historically seen in EUV tools. And then secondly, as a follow-up, on logic versus memory. I was wondering whether it is fair to say that last quarter incremental demand seemed very much related to memory. But now you seem to be seeing a fairly strong balance between logic and memory. So I am curious how much you agree with that statement, and what the drivers are, whether it is drivers like Gen AI, or progress with existing customers or new customers. Thank you very much.
Christophe D. Fouquet: Yes. I can start. So I appreciate the question about 330 wafers per hour, because this is one of the very near-term actions we have taken to add capacity. We will bring the different immersion products up to 330 wafers per hour. You know we have slightly different versions of immersion tools depending on customer overlay requirements. But all versions will reach 330 wafers per hour early next year. When it comes to value, and therefore pricing, as we have discussed many times for EUV, price basically moves with productivity. I think what applies to EUV will also apply to DUV. So this is an opportunity for value, margin, pricing, and of course also very helpful on capacity. Because you see, of course, delivering the same capacity with potentially fewer tools. It does not mean we need fewer tools. It means we get more capacity for customers, of course. So this is another creative move we are making in the portfolio to capture more capacity. We expect, again, that value will come to us in the same way as with this new product. That is on DUV. On the second question, I think Roger gave some numbers. If you look at growth compared with last year, memory growth is very large. We are seeing a trend of memory becoming increasingly important, 75% growth, which is of course significant. So you are starting to see the results of this huge memory investment, which, by the way, is mainly DRAM, to be clear. We continue to see this moving forward. We have said in the past that the ratio between logic and memory usually shifts toward memory. Still, I would say logic is also very strong. So it is difficult to predict exactly where this will land, because I want to emphasize that we also see very strong demand in logic, because, of course, as you know, the two are very tightly coupled in delivering AI capacity to the market.
Roger J. Dassen: Yes. One more data point for you. Indeed, as I mentioned, logic grew about 10%, but the advanced logic portion within that is much higher, right? So it grew more than 25% this year and is expected to grow more than 25%. So you are right. Memory is getting close to logic, but logic is still very, very important. And, when you look at 2027, memory is getting close. But I still believe logic will exceed memory in total business next year.
Alexander Duval: Thank you.
Operator: Thank you. Today's next question comes from Chris Caso of Wolfe Research. Please go ahead.
Chris Caso: Yes. Thank you. Good morning. My question is about your customers' cleanroom constraints. And, maybe you could discuss that in the context of your decision to expand capacity into 2028. Do you expect cleanrooms will remain a constraint on your customers adding tools and adding capacity? And then what does that mean as you potentially move into 2029?
Christophe D. Fouquet: Well, you know, I think there are still some, I would call them tactical constraints. Maybe a system moves from one quarter-end to another quarter-start, and to be honest we do not see much of that. But that is still possible, because our customers have been ramping their cleanroom facilities extremely fast. If we look longer term, as I said earlier, apart from those small movements or those tactical movements, we have confirmed that the numbers customers give us are indeed consistent with their plans. In fact, in many cases, when supply is tight, before fully allocating tools to a customer, we make sure the space will be there and the facilities will be there, so that we do not ship a tool and then wait weeks for the customer to be able to use it, because of course other customers would be very disappointed about that. So we do a lot of work to make sure that across all of ASML's plans, there will not be more than a few days or weeks of delay in our product shipments, I would say. So the short answer is, I think, no. We do not believe our customers' cleanrooms are a major factor in our demand.
Roger J. Dassen: To your point, Christophe, it is very clear that everything being built today is on such a massive scale. On the one hand, this opens up enormous opportunities. But it also means our customers have to manage the execution of all that construction in 2027-2028 very diligently. Right? So they. They are looking at a huge expansion of their footprint, but of course that requires very diligent execution. That is obviously something we need to monitor very, very carefully. So in that sense, conceptually, there is a lot of opportunity rather than constraint, but in reality, we need to make sure they execute well so that it does not become a constraint.
Chris Caso: Thank you. As a follow-up question, on China, you talked about it representing 20% of system revenue this year. What should we expect from China going forward? Do you expect it to stay at these levels next year?
Roger J. Dassen: I think underlying demand in China is strong. Right? If you look at what China needs in mainstream logic, in memory, domestic consumption demand in China is very strong. If you look at the fabs being built in China, that is. There is a lot coming online there. Therefore, we expect China demand to be strong, 20% this year. I would say the same for 2028 and beyond based on underlying drivers, and the situation in China still appears very strong.
Chris Caso: Thank you.
Operator: Thank you. Today's next question comes from Tammy Qiu of Berenberg. Please go ahead.
Tammy Qiu: Hi, thank you for taking my question. First, on your value-based pricing. This year, you. You are raising prices. Looking ahead, assuming industry tightness persists. Do you think you will be able to negotiate price increases every year?
Roger J. Dassen: Well, Tammy, that is not the plan, to be honest. [smiles] Yes. We are smiling. Our customers are listening. Yes. That is right. Yes. That is right. That is not the plan. So, we. We have already had these conversations with customers, given everything the ecosystem is going through, and the efforts all of us are making to meet the ramp requirements. We believe we are entitled to a higher percentage, and we were able to agree with customers. We would consider this possible within our value-based pricing model. But we would not consider repeating this every year. That is not the plan. No.
Christophe D. Fouquet: Let me add, Roger. This remains, I think, mainly, you know, this is mainly the first time we have had this kind of discussion with customers. If you go back many, many years, this is something we have never done. Roger mentioned in his introduction that the discussions were constructive. I think they were not just constructive, meaning the overall value in the ecosystem may be shifting. This is triggered in part by AI, I would say, you know, we will see what the future brings, but at least we were able to have very good, very mature discussions, which I think is very important. So this is something we have never done before, and we do not want to do it every year, because that is not the purpose. But I think, in principle, at least we have the ability to occasionally discuss again what lithography is truly worth in the overall ecosystem, which is very good.
Roger J. Dassen: Absolutely. And, Tammy, as you know, one of our key objectives in the pricing conversations is to ensure we never deviate from the concept of value-based pricing. Because we do not want to be in a position where, if the market declines, we would effectively have to discuss price reductions with customers. I mean. And that is well understood in our discussions with customers. So in Christophe's words, these are conversations about fairness. But, you know, we are able to capture the value of next-generation tools, and therefore we are able to continue to raise prices in proportion to the value it brings. That is the principle we want to maintain, even if the market takes a slight step back. So that is why we are not opportunistic in pricing, we are not opportunistically seizing the moment, we execute in the way Christophe just described.
Tammy Qiu: Okay. Thank you. That is really clear. Second, on your China exposure, so I understand China may never be able to compete with you from a technology standpoint. But, let's say around 2029, when you develop lower-end versions of immersion tools, do you think that could affect your potential to price at a premium relative to local Chinese players? Or your potential margins in China may need to be lower?
Christophe D. Fouquet: Well, I think there is a lot of speculation in the question. But, where we are today, I think, with our Chinese customers, they all want to work with us. I think that is very clear. I will not give any market share numbers, but you can imagine it is very high, especially in immersion. So, you know, as long as we have good market access, as long as they can work with us. I think the likelihood of major competition emerging in China remains fairly low. That is the first point I would make. Second, of course, you know, if they cannot get some of our tools, then you get more momentum. But that still has a very, very long way to go. I think we have been having discussions for 25 years about China having tools to compete with us, you know, every other year, we get this message that it is coming. If you stick to the facts, I think the facts this year are no different from the facts a year ago or even five years ago. So I think progress is not very strong. The momentum is not strong at all. I think as long as. Again, as long as we maintain market access in a reasonable way, we should be fine.
Tammy Qiu: Okay. Thank you.
Christophe D. Fouquet: You're welcome.
Jim Kavanagh: Okay. I just want to remind everyone that we have time for one last question. So Sharon, could you please ask the last question? And then if there are follow-ups, please contact ASML Investor Relations. We will do our best to follow up. So, operator, could you please connect the last question?
Operator: Thank you. Today's last question comes from Stéphane Houri of ODDO BHF. Please go ahead.
Stéphane Houri: Yes. Hello. Thank you for taking my question. Actually, my question was originally about China, but just to understand whether your discussions about 2029 and 2030 are more because you are currently very optimistic about memory. Whether it is very balanced between memory and logic, or whether it is still memory makers giving you some visibility on 2029, sorry, 2030. Thank you.
Christophe D. Fouquet: Well, I think you need memory and logic to get some AI chips. I think we had a lot of discussions with both sets of customers, because they get the same information from their own customers. I think today, DRAM is a bit more aggressive in building these capacities. But, yes, we have discussions with both. So I think we see equal interest in the memory and logic demand trajectories, because you basically need both to bring products out.
Stéphane Houri: And in DRAM, is that from the U.S., Korea, and even China at the same time, or is it more concentrated geographically?
Christophe D. Fouquet: No. I think, in this case, we are mainly talking about EUV. So that does not include China.
Stéphane Houri: Okay. Yes. Okay. Thank you.
Jim Kavanagh: Okay. Thank you everyone. On behalf of ASML, I would like to thank you all for joining today's call. Operator, if you could formally close the call, it would be appreciated. Thank you.
Operator: Thank you. This concludes the ASML 2026 third-quarter financial results conference call. Thank you for your participation. You may now disconnect.
Comments