ASML - Earnings Call - Q3 2025 Investor Call
October 15, 2025
Transcript
Operator (participant)
Good day, and thank you for standing by. Welcome to the ASML 2025 third quarter financial results conference call on October 15, 2025. At this time, all participants are in a listen-only mode. After the speaker's introduction, there will be a question and answer session. To ask a question during the session, you will need to press star one and one on your telephone. You will then hear an automated message advising your hand is raised. To withdraw your question, please press star one and one again. Please be advised that today's conference is being recorded. I would now like to turn the conference call over to Mr. Jim Kavanagh. Please go ahead.
Jim Kavanagh (Head of Investor Relations)
Thank you, operator. Welcome, everyone. This is Jim Kavanagh, Head of Investor Relations at ASML. Joining me today at the call are ASML CEO, Christophe Fouquet, and CFO, Roger Dassen. The subject of today's call is ASML's 2025 third quarter results. The length of this call will be 60 minutes, and questions will be taken in the order that they were received. This call is also being broadcast live over the internet at www.asml.com. A transcript of management's opening remarks and a replay of the call will be available on our website shortly following the conclusion of this call. Before we begin, I would like to caution listeners that comments made by management during this conference call will include forward-looking statements within the meaning of the federal securities laws. These forward-looking statements involve material risks and uncertainties.
For a discussion of risk factors, I encourage you to review the safe harbor statement contained in today's press release and presentation found on our website at www.asml.com and in ASML's annual report on Form 20-F and other documents as filed with the Securities and Exchange Commission. With that, I would like to turn the call over to Christophe for a brief introduction.
Christophe Fouquet (CEO)
Thank you, Jim. Welcome, everyone, and thank you for joining us for our third quarter 2025 results conference call. Let me start by saying how pleased I am with the recent announcement of the reappointment of Roger Dassen and Frédéric Schneider-Maunoury to the board of management. We also announced last week the appointment of Marco Pieters as our Chief Technology Officer. This appointment is part of our robust succession planning process, and with over 25 years of experience at ASML, Marco brings a proven track record in technology leadership. Marco will take on the responsibility of driving our technology roadmap forward in support of our customers, and I look forward to our continued collaboration. In addition, ASML's supervisory board announced that it intends to appoint Marco to the board of management as of the company's next annual general meeting to be held on April 22nd, 2026.
Before we begin the Q&A session, Roger and I would like to provide an overview and some commentary on the third quarter results, as well as provide some additional comments on the current business environment and on our future business outlook. Roger?
Roger Dassen (CFO)
Thank you, Christophe, and welcome, everyone. Let me start with our third quarter accomplishments. In the third quarter of 2025, total net sales were EUR 7.5 billion, which is within our guidance. Net system sales were at EUR 5.6 billion, which includes EUR 2.1 billion from EUV system sales, including one High NA system, and EUR 3.4 billion from non-EUV system sales. Net system sales were driven by logic at 65%, with the remaining 35% coming from memory. Installed base management sales for the quarter came in as guided at EUR 2 billion. Gross margin for the quarter was also within guidance at 51.6%. For operating expenses, R&D expenses came in a bit below guidance at EUR 1.1 billion due to the timing of spending, and FG&A expenses basically came in as guided at EUR 303 million. The effective tax rate for Q3 was 17.8%.
For the full year 2025, we continue to expect an annualized effective tax rate of around 17%. Net income in Q3 was EUR 2.1 billion, representing 28.3% of total net sales and resulting in an EPS of EUR 5.49. Turning to the balance sheet, we ended the third quarter with cash, cash equivalents, and short-term investments at a level of EUR 5.1 billion. Moving to the order book, Q3 net system bookings came in at EUR 5.4 billion, split between EUR 3.6 billion of EUV systems and EUR 1.8 billion of non-EUV systems. Net system bookings in the quarter were slightly weighted towards logic at 53%, while memory accounted for the remaining 47%. In Q3, ASML paid the first interim dividend over 2025 of EUR 1.60 per ordinary share. The second quarterly interim dividend over 2025 will also be EUR 1.60 per ordinary share and will be made payable on November 6th, 2025.
In Q3 2025, we purchased shares for a total amount of around EUR 148 million. As of September 28th, 2025, ASML has acquired 9 million shares under this program for a total consolidation of EUR 5.9 billion. ASML does not expect to complete the EUR 12 billion share buyback program in full within the 2022-2025 timeframe. We intend to announce a new share buyback program in January 2026. With that, I would like to turn to our expectations for the fourth quarter of 2025. We expect Q4 total net sales to be between EUR 9.2 billion and EUR 9.8 billion. We expect our Q4 installed base management sales to be around EUR 2.1 billion. As previously discussed, we expect Q4 to be a very strong quarter, as was the case at Q4 of last year. Gross margin for Q4 is expected to be between 51% and 53%.
The expected R&D expenses for Q4 are around EUR 1.2 billion, and FG&A is expected to be around EUR 320 million. For the full year, we continue to expect total net sales to be around EUR 32.5 billion, with a gross margin of around 52%. With that, I would like to turn the call back over to Christophe.
Christophe Fouquet (CEO)
Thank you, Roger. As Roger has highlighted, we finished the third quarter with good financial results. Looking now to the market, there's been a positive news flow across the industry in recent months that has helped to reduce the level of uncertainty that we were reporting last quarter. First, there were a number of announcements around the continued investment in AI infrastructure that supports demand in both leading edge logic and advanced DRAM. Second, the positive momentum around AI seems to extend to more customers in both logic and DRAM. Third, we see continued momentum around customers adopting more EUV layers in both logic and DRAM, migrating multi-patterning DUV to single exposure EUV, and continuing to support lithography intensity.
On the other hand, we expect to see China customer demand, and therefore our China total net sales in 2026, to decline significantly compared to our very strong business there in 2024 and 2025. We believe that the impact of these dynamics will only partially affect 2026. However, overall, we do not expect 2026 total net sales to be below 2025. In this environment, we also expect the 2026 EUV business to be up, driven by the dynamic in advanced DRAM and leading edge logic, and the DUV business to be down compared to 2025, driven by the dynamics with our Chinese customer. We will provide more details on our 2026 outlook in January.
Turning to technology, there's been a lot of good progress this quarter with the latest achievement on EUV presented at industry conferences, the release of new 3D packaging lithography system, and the announcement of our strategic engagement with Mistral AI. For EUV, we presented a number of papers at recent SPIE and SEMICON events that highlighted the progress we have made in helping drive down the cost of technology on our customers' most advanced processes. With regards to the maturity of High NA, we shared data showing that we have now run cumulatively over 300,000 wafers on the system at our customer. Also, our customers have shared very positive data showing that the maturation level of the platform is well ahead of where low NA EUV was at the same stage in its introduction.
Further, SK Hynix announced this quarter that they started to take delivery of their first High NA system, the EXE:5200, positioning High NA as a critical enabler for future advanced DRAM devices. We are also happy to report that this quarter we shipped ASML's first 3D integration product, the XT260. The XT260 is an i-line scanner designed for applications that include advanced packaging and offer up to four times the productivity compared to existing solutions. 3D integration is of increasing importance to the roadmaps of our customer and the semiconductor industry, and our customers have been sharing with us a need to innovate in order to meet their future requirements. The discussion with our customers on those requirements points to a good opportunity to transfer some of our holistic lithography technology to 3D integration to meet their future needs.
The XT260 is the first example of several opportunities we are evaluating. With the XT260, we are able, as said, to multiply the existing productivity by up to a factor of four using a unique optical design. As mentioned, we shipped our first system this quarter and expect to ship this tool to quite a few more customers in the coming quarters, reflecting the strong interest in this technology solution. With that, I ask Roger to provide some insight into our recent engagement with Mistral AI. Roger?
Roger Dassen (CFO)
Thanks, Christophe. In September, we announced that we closed a strategic partnership with Mistral AI, a pioneering company in generative artificial intelligence with a strong business-to-business focus and widely recognized for its leadership in large language models that assist in areas such as software coding development. ASML is normally associated with hardware. Software plays an increasing role in driving the precision and speed of our tools. Our partnership with Mistral AI allows us to embed AI across our entire holistic portfolio in order to increase the performance and productivity of our systems and the yield of our customers' processes. Also, we believe this collaboration will allow for faster innovation, resulting in improved time to market and lower development costs when delivering state-of-the-art solutions to our customers.
In addition to the collaboration agreement, ASML has invested EUR 1.3 billion in Mistral AI's Series C funding round as lead investor, resulting in ASML holding around an 11% share in Mistral AI and having a seat at their strategic committee. It allows us to become even more closely connected to the AI ecosystem.
Christophe Fouquet (CEO)
Thank you, Roger. Looking longer term, as we shared in our capital market day, we start to see that the end market dynamics are leading to a product mix shift towards more advanced logic and DRAM. Those applications require a more intensive use of advanced lithography systems. We expect that to continue. The combination of our strong productivity roadmap on low NA and the introduction of INA supports further costs of technology reduction and the conversion of more multi-patterning layers to a single EUV exposure, especially on DRAM advanced nodes. In line with our 2024 capital market day, we expect a 2030 revenue opportunity between EUR 44 billion and EUR 60 billion, with gross margin expected between 56% and 60%. With that, Roger and I will be happy to take your question.
Jim Kavanagh (Head of Investor Relations)
Thank you, Roger. Thank you, Christophe. Now, the operator will instruct you momentarily on the protocol for the Q&A session. Beforehand, I would like to ask that you kindly limit yourself to one question with one short follow-up if necessary. This will allow us to get through as many callers as possible. Now, operator, could we have your final instructions and then the first question, please?
Operator (participant)
Thank you. As a reminder, to ask a question, you will need to press star one and one on your telephone and wait for your name to be announced. To withdraw your question, please press star one and one again. We will now take the first question. Your first question today comes from the line of François Bouvignies from UBS. Please go ahead.
François Bouvignies (Head of Europe Tech Hardware and Semiconductor)
Thank you very much. My first question is on this, what you said, Christophe, that you saw positive news in the last months that helped you reduce the uncertainty. I was wondering, can you elaborate on this? Obviously, you get these $5 billion orders, which is helping, but I guess just one quarter of orders doesn't give you the full flexibility on 2026 completely. Do you have more visibility in terms of the full capacities they need for 2026 now? Do they give you more numbers around all of that or the layers? I'm just trying to understand what changed versus last month, or you just rely on positive news out there and you try to extrapolate, just trying to understand these dynamics. Thank you.
Christophe Fouquet (CEO)
Yeah, François. I think first, I refer to a lot of positive news on AI infrastructure. I think you all know that usually this doesn't translate immediately into orders for us. This takes quite some time. If you look at the sum of the announcement, I would say this creates a pretty positive backlog of opportunity for AI moving forward. The second one is also quite important. I think we mentioned the fact that we see now that more customers will benefit from the AI opportunity. I think it's important for many reasons. The first one being that, in order to respond to this huge demand of good news, this huge amount of good news on AI infrastructure, you need to make sure that the market capacity will be high enough.
I think that seeing more customers entering logic opportunity or DRAM opportunity is pretty good news for the long term. To know exactly how this will affect the next few years is still difficult to say. I think, as I said before, only part of that will be effective next year. For the rest, I think it's far too early to say on our side.
François Bouvignies (Head of Europe Tech Hardware and Semiconductor)
Makes sense. Thank you. The follow-up is on China. I mean, that has been a key driver of growth in the last few years, and you mentioned it will go down significantly. It's again a bit on the visibility side here. I guess you see the strong edge too. I would imagine three to six months lead times because it's DUV. I would expect you not to see much beyond Q1 2026. Is it some conservatism because you see the strong base and maybe a soft Q1 2026, or do you really have the full picture already on 2026, how it's looking?
Christophe Fouquet (CEO)
I think that we have about the same clarity that we had last year about this time. I think around that same time, we still try to provide you a bit of our view of the market. I think on China, we have been very consistent that we thought that, you know, the level of business in the last two, three years was very high and in no way normal. I think we have been experiencing a very high cycle in China, especially through the last couple of years. Again, our expectation and the visibility we have right now is that next year we go back to more reasonable business.
François Bouvignies (Head of Europe Tech Hardware and Semiconductor)
Thank you very much for your answers.
Christophe Fouquet (CEO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Krish Sankar from TD Cowen. Please go ahead.
Krish Sankar (Managing Director)
Yeah, I was kind of thinking my question. I have two of them. Roger, you kind of mentioned about the recent AI investment strength. Some of it will come in 2026. I'm kind of curious, given your long lead times, how to think about linearity of your revenues or orders in 2026? Any early thoughts on what it implies for 2027?
Roger Dassen (CFO)
Yeah, Krish, interesting question. I think we're, you know, we set what we set on 2026. I think it's way too early to make any comments on 2027. You will see that, you know, orders came in strongly in the last quarter. Actually, the quarter before that also came in strongly. I think linearity of orders is an anomaly. We said orders always come in lumpy. We've had a healthy run in the past two quarters, but I don't think you can talk about linearity. It's way too early to talk about what this means for 2027. Going back to Christophe's earlier answer, the news flow that you got in the past couple of months is a positive news flow and is a positive news flow, particularly in the medium term. To now translate that into concrete expectations for 2027 is really quite a bit too early.
Krish Sankar (Managing Director)
Fair enough. Fair enough. A follow-up for Christophe. You know, clearly, you're seeing strength in DRAM. I'm kind of looking longer term. There's a view that when you go into 4F sq from 6F sq for DRAM architecture, that's actually a negative for EUV. The EUV layer count comes down. Can you just help us understand that, Christophe? Thank you.
Christophe Fouquet (CEO)
Yeah, it's a good question. It's a question we get a lot. The short answer is no. If we look at the number of EUV layers going from 6F sq to 4F sq, we do not expect the number of layers to drop. In fact, as 4F sq roadmap continues after the transition, we, in fact, expect the number of EUV layers to continue to grow. I make that statement after many discussions with our customers. On top of that, what I'd like to add is, you know, 4F sq has a bit of a more complex structure. It's, in fact, heading overall more litho mask, more advanced litho mask. There is a benefit also to some extent to advanced DUV. In any case, if you still doubt about it, 4F sq is in no way bad news for ASML. We are looking forward to it.
Krish Sankar (Managing Director)
Got a sense of that, Christophe. Thank you.
Operator (participant)
Thank you. Your next question comes from the line of Joe Quatrochi from Wells Fargo. Please go ahead.
Joe Quatrochi (Executive Director and Equity Research Analyst)
Yeah, thanks for taking the question. I was wondering if you could talk a little bit more about just the updated commentary for 2026. Is that more that your more positive view, a bit more DRAM slanted or equally balanced with logic? Then same for, you know, more customers benefiting from AI infrastructure build-out. Is that more of a DRAM comment?
Roger Dassen (CFO)
Joe, I think it's related to both markets, actually. Last quarter, we talked about uncertainties. I think the uncertainties, and Christophe went into the positive flow that has happened thereafter. One element of the uncertainties that we called out at that point in time was also the uncertainty around tariffs. That was out there. I think there is more clarity on that front right now, but that unclarity also prevented customers from being very concrete as to what exactly they were going to do and where they were going to build their capacity. I think that has decreased, and I think that has given rise to the commentary that we now make on 2026.
I would say in general, when we talk about the positive news flow on the leading edge, leading to our expectation that EUV is going up next year, that is related both to the DRAM and to advanced logic.
Joe Quatrochi (Executive Director and Equity Research Analyst)
Thanks for that. As a follow-up, I was wondering if you could maybe just walk through some of the puts and takes in your gross margin guide. I think what you're guiding for for the December quarter is a bit better than what you were implying a quarter ago. Is that something related to tariffs or mix or something else?
Roger Dassen (CFO)
Obviously, volume is quite high, right? That is a positive moving from Q3 to Q4. Mix has many elements to it. It's, of course, we expect two [EXEs] in this quarter in Q4, so that is a negative. On the other hand, we will also have a good low NA in there. We're a bit positive on upgrade business, as you could see, right? That business is a little bit up. If you then take it all in all, we see a slight improvement at the midpoint in comparison to what we had in last quarter. Those are the dynamics. I would say the fairly small movements, but all in all, leading at the midpoint to a slightly better gross margin than maybe we thought we were going to be forward.
Joe Quatrochi (Executive Director and Equity Research Analyst)
Thank you.
Roger Dassen (CFO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Didier Scemama from Bank of America. Please go ahead.
Didier Scemama (Head of EMEA Tech Hardware and Semiconductor Research)
Oh, yes. Thank you for taking my question and good afternoon. My first question is about things that you talked about a little bit before, but I just wanted to press you a little bit more. We've all seen the press release of SK Hynix and Samsung, following the visit of Sam Altman to Korea, talking about a letter of intent of 900,000 with a start per month of HBM capacity, which is probably more than double the current HBM capacity. My question is this. My estimate would be that there are about 30 EUV tools in DRAM today, not all of them for HBM, but the majority, presumably. That would imply if we believe those numbers, and many people think those numbers are absolutely grotesque, but let's do it for the sake of the argument.
You would need something like 65 EUV tools just for HBM, and then comes on top whatever Samsung foundry intent and TSMC would need. I guess the question is, A, what do you think of those numbers? B, do you have enough capacity for EUV low NA, let's say, by 2030 to get to satisfy that potential demand?
Christophe Fouquet (CEO)
I think we won't get into the calculation because I think we said it already a couple of times. We are a bit careful with how the big announcement can translate into real capacity need on the ground. The one thing I'd still like to stress one more time is we see the broadening of the customer base, very important news in that matter because whatever you do is the first set of news. I think we can all agree that we need to make sure that the market will not be supply limited. This has always been a risk with a limited amount of customers supplying AI chips, both in logic and DRAM. The broadening of the install base is very good news there. On the last question, you know, we have said for a few quarters that we have been preparing for growth.
We were following those dynamics, and I think we know now that EUV most probably will be stronger next year. We've been preparing for that. We have, as you know, also worked on longer-term capacity. We continue to track basically the market carefully, having in mind that we want to be able to follow the demand. I would say we don't have any concern there at this point of time.
Didier Scemama (Head of EMEA Tech Hardware and Semiconductor Research)
Okay. Great. My follow-up for Roger, perhaps, just wanted to understand what I'm missing. If we look, if we assume, like, let's say you book $5 billion of orders in Q4, you know, give or take, that would imply you exit calendar year 2025 with a backlog just about $30 billion. Even if you strip out the High NA tools in there, expectations, I'm going to call it $1.5 billion, $2 billion, that makes me comfortably above, you know, let's say, modest growth for next year. Is there a large portion of the current backlog which has got maturity beyond 2026? Is that the reason why you're still hesitant at calling 2026 a, let's say, strong growth year or high single-digit growth year or even double-digit growth year? Thank you.
Roger Dassen (CFO)
Of course, that's the big question, right? The big question is what is in the backlog that pertains to beyond 2026. That is a reasonable number. As a result of that, your math doesn't work exactly right for 2026 because there is a pretty healthy number in there for beyond 2026. We also know that there's always a question about a bit of pull in here, a bit of push out there, and that makes it extremely hard to make at this stage any concrete projections on 2026. To your point, there is already a healthy order intake in the backlog that is beyond 2026. Also, as it relates to High NA, I think that the High NA contingent in there is actually pretty strong.
Didier Scemama (Head of EMEA Tech Hardware and Semiconductor Research)
Would you say how many you expect to recognize next year?
Roger Dassen (CFO)
High NA? No, that's a January topic, Didier.
Didier Scemama (Head of EMEA Tech Hardware and Semiconductor Research)
Thank you, gentlemen.
Roger Dassen (CFO)
You did.
Christophe Fouquet (CEO)
Impressive.
Didier Scemama (Head of EMEA Tech Hardware and Semiconductor Research)
Thank you, guys.
Roger Dassen (CFO)
You're welcome. Thanks.
Operator (participant)
Thank you. Your next question comes from the line of Andrew Gardiner from Citi. Please go ahead.
Andrew Gardiner (Head of European Technology Equity Research)
Good afternoon. Thank you for taking the question. I might try Didier's question in a slightly different angle and see if I have any more luck. You've highlighted the news flow, Christophe. You know, we can all see it. It feels like you wake up every day to another massive announcement from somewhere within the AI food chain. You sort of spend a lot of time talking about how that could create a theoretical backlog for you, not yet orders. I'm just wondering, you are a critical supplier into this market. You have the potential to be a backlog, to be a bottleneck, rather, for the market. Of course, you don't want that to be the case. You're preparing for growth, etc. Do you feel like there's sufficient understanding through the chain, whether it's of where you sit or perhaps your customers?
How can you, as a critical supplier, make sure that the broader market isn't supply limited come 2027, 2028, given the kind of announcements that we're seeing on a near-weekly basis?
Christophe Fouquet (CEO)
First, I think we wish we had a formula to translate all the announcements on what it means exactly for us in the next few years. I think no one has that. The experience also of 2022 has mostly taught us a lesson to be ready and to have maybe more flexibility because we know the market can swing. We've done a lot of work on that in the last few years. We have prepared a building, as we discussed before, which are usually the longer lead time items. For the rest, when it comes to defining exactly how many tools we want to produce, I think the lead time is a lot shorter, so we have more flexibility. I would say we have structurally maybe improved ourselves so that, because we cannot answer those questions after the announcement, we at least have the flexibility.
The other thing I would like to add is, of course, our customers, at the end of the day, tell us what they need. I think that's a constant dialogue, a dialogue we try to always reflect with you on a quarterly basis. I think we continue to do that. We are prepared. I think we stressed a few times we were preparing for growth. This was also in light of some of this activity we have seen. Come January, we will be most probably knowing even more about what's happened then. We will continue to monitor the market. I think we are, I would say, a lot more prepared than we were a few years ago.
Roger Dassen (CFO)
I think Christophe said it exactly right. We are very well prepared, particularly, you know, by having invested in the long lead time items. That still, by the way, means that when it comes to shorter lead time items, of course, you need to have a dialogue with your customer that gives you a timely heads up, right? Because, you know, of course, we need to kick our supply chain into gear. We need to hire the people, etc., etc. We have a lot more flexibility than we used to have. It is also critical for our customers to give us a timely heads up such that we can make sure that, you know, within the long-term infrastructure that we have, we're also able to get to higher output levels and, you know, get supply chain and people in place.
Andrew Gardiner (Head of European Technology Equity Research)
Thank you, Roger. I suppose without naming names, do you feel like your customers are giving you that heads up, right? Is there a sufficient acknowledgement through the chain?
Christophe Fouquet (CEO)
I think they do their very best. I will say it this way because they have the same challenge as we do. I think they do their very best, and I think we're very happy with the transparency and the honesty around those discussions because over time, that's very helpful, also helpful in trying to really avoid the bigger surprise. I feel that this discussion really has improved in the last few years, and I think this is very helpful for them, for us, but also for their customers.
Andrew Gardiner (Head of European Technology Equity Research)
That's great. Thank you very much, guys.
Christophe Fouquet (CEO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Tammy Qiu from Berenberg. Please go ahead.
Tammy Qiu (Head of Tech Equity Research)
Hi. Thank you for taking my question. The first one is the High NA kind of order pattern and demand ramp-up curve. From my understanding, there hasn't been High NA orders for about two-ish, two-plus quarters now. I understand that you're working on your backlog. Does that mean if orders are only coming in from end of next year, that you may actually have, let's say, somewhere like 2027 or 2028, a few quarters of or a year of zero High NA revenue? The ramp-up curve of revenue of High NA is going to be quite lumpy. Is that correct?
Christophe Fouquet (CEO)
I think you summarized to me the situation pretty well. We are indeed working on NA out of the pretty healthy backlog we have. I think I explained in the past that this backlog allows our customers to be covered for, of course, R&D, pretty much done with most of those shipments, but also with the system they want to use for qualification and the insertion. For the rest, the next wave of orders we expect to happen basically when the data coming out of the qualification can confirm that the maturity of the tool is there. I think when it comes to performance, most probably we have passed that milestone. As we discussed last quarter, we expect that to happen most probably towards the second half of next year and after that. This being said, we do more than just waiting for orders with our customers.
It's not like nothing is going to happen also in the next 18 months in terms of assessing the progress and therefore also assessing the likelihood of the type of insertion we're going to look at. As we did, we just talked about the way we were preparing for growth on EUV in the last few months. We did that without having necessarily the full clarity of the demand. We would be able to do that through the discussion with our customer. In terms of orders, you look, yes, towards the end of next year. In terms of shipments, you look at 2028 and beyond.
Tammy Qiu (Head of Tech Equity Research)
Okay, thank you. That's very clear.
Christophe Fouquet (CEO)
You're welcome.
Tammy Qiu (Head of Tech Equity Research)
Now, the second question is on China. You mentioned that China revenue would be down significantly in 2026, and that's because of demand. I'm wondering if that's because the customers are not sure about the demand, so they wouldn't want to commit. Therefore, you've been conservative, or the customer told you for some reason their end market demand has been weak, so they're buying less. We don't actually currently see the Chinese market further weaken from where we are today. I wonder what's the reason for that comment to be weaker significantly on you?
Roger Dassen (CFO)
I mean, I think the comment that Christophe made, and I think he also said it goes back to also the comment that we made a year ago, right? For quite a while, we have been eating into our backlog. That's really what happened. For quite a while, the China sales were very high because we've been eating into a substantial backlog because of underserving the Chinese market. That was the reason why, in fact, a year ago, we said we expect the China sales to be more commensurate with the China percentage in the backlog, which at that stage was around 20% of the backlog. That's what we said. Actually, as Christophe said, we were quite surprised that the China sales this year are as strong as they are.
That is still the underlying assumption, and our underlying perspective on the Chinese market is still the way it was a year ago. That has to do with the fact that the Chinese market is a very specific one, right? It's focused on mainstream logic, as we call it. Simply given the dynamics of that market, it is our assessment that the sales level that we currently see this year is very high in comparison to what we would think is a normalized level for the mainstream market. That is the reason why we've indicated this assumption of a significant decline. That is based on our understanding of the market. It's based on the dialogues that we have with our customers. Could that change? Absolutely. I think we've seen that this year that it could change in comparison to our perspective.
If you ask us for an honest assessment at this stage, how we think it's going to be in 2026, it is as we communicated.
Tammy Qiu (Head of Tech Equity Research)
Okay, thank you.
Roger Dassen (CFO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Stephane Houri from ODDO BHF. Please go ahead.
Stephane Houri (Head of Equity Research and Technology Analyst)
Yes. Good afternoon, everyone. Actually, I've got two questions. The first one is on the advanced packaging product that you're highlighting in your presentation, the XT260. Can you maybe say a little bit more about what it's doing, the price of the machine, type of clients, who you're competing with, and the market outlook if you have a few information? I have another one.
Christophe Fouquet (CEO)
Yeah. Let me give you a bit of context. I think we mentioned that product in our press release, also in the call, not necessarily because of the high price, high value of the product, but because this is the first product ASML is providing to its customer to support 3D integration. I think that's mostly where the important news is. I think we all know that when it comes to Moore's Law, our customers are asking us to drive transistor density, still doubling it basically a factor of two every two years. I think that we also know that over time, litho scaling slows down. That creates a need basically for more either stacking or packaging of transistors.
Our customers are really asking us to help there because what they want is also speed, is also over time accuracy, and I would say some of the technology we have been developing for our litho portfolio. This is a bit the starting point. I think we also mentioned most probably we'll be looking at some more product there. I think what's very interesting is the interest we see from our customers on this product. It's an iLine scanner to answer your technology. This is based basically on iLine technology, which, of course, we have had at ASML for many years. In this case, we have a new optical design that really enables us to provide a 4x improvement on productivity. Competition are basically the people who do i-line scanners. I think you know them.
What's very interesting is that if we look at next year, we have quite a few customers very eager to take this technology. Because we provide a nice technology, some good improvement, I will say that the business and the benefit we can get out of this product is quite a lot higher than we have done historically on i-line.
Stephane Houri (Head of Equity Research and Technology Analyst)
Okay. The follow-up would be about the gross margin in 2026. I know you're not giving guidance and you will say more in January, but given what you have described, i.e., less China, more EUV, but also at the same time, more EUV, High NA, can we expect some increase in the gross margin next year?
Roger Dassen (CFO)
Yes, Stephane, indeed. We will give more clarity on that, obviously, in January. I think you're right. Our product mix, obviously, is very important in what drives it. As you know, what we ship to China today to a very large extent is immersion. Immersion comes with a very good gross margin. Less China business would be diluted on that front. You're right, EUV comes in with very strong gross margins. Therefore, if we predict that EUV, particularly low NA, goes up, that would be a positive again, on the other hand. There is the question on the number of tools that we're going to recognize for High NA, which, of course, is still diluted to the corporate gross margin. I think that's an important one. The other one is our expectation on the installed base business.
We know that typically the upgrade business is pretty important when it comes to gross margins. It's all those moving parts exactly where they land that will eventually, you know, determine what our expectation is. Those are indeed the moving parts. It's the product mix and it's the expectation on the composition of the installed base business, which, you know, first and foremost is going to drive that perspective.
Stephane Houri (Head of Equity Research and Technology Analyst)
Okay, thank you very much.
Roger Dassen (CFO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Chris Caso from Wolfe Research. Please go ahead.
Chris Caso (Managing Director)
Yes. Thank you. Good morning. I guess the first question is on installed base. If you could update us on your thinking as we go into calendar 2026. Previously, you had talked about having to back out some of the upgrade revenue as you go into 2026. What's the current thinking on installed base as we go into next year?
Roger Dassen (CFO)
I think you feel well, that's first over the installed base for this year, right? I think you will see that our expectation for the installed base this year has actually gone up a bit if you look at what we guide for, you know, for Q4. Originally, we thought that the first half was going to be, you know, quite a bit stronger than the second half. Now that the second half is as strong as the first half, what you see there is that actually the service business is developing quite nicely. While we might have had a bit more upgrade business in the first half, the second half is really benefiting from the service business. As you know, the service business is very much tied to the development of the installed base in EUV, right?
With the increase in the installed base in EUV, of course, that also, you know, further drives up the service business. That's an important one that I think you can sort of back of the envelope calculate what the impact of that is going to be for next year. The question indeed is how sustainable is the upgrade business? Clearly, it didn't fall off a cliff in the second half of this year. We'll give you an update in January what our expectation is for 2026.
Chris Caso (Managing Director)
That's helpful. Thank you. Just as a follow-up and maybe summarizing some of the earlier comments about, you know, some of the more optimism, the optimism you had with regard to some of the AI developments. I mean, is it safe to say that the bookings, you know, the total bookings haven't really increased here? Is this more a function of, you know, your customers are telling you based on some of these developments, there's sort of an expectation for, you know, stronger bookings in coming quarters because of the capacity needed to provide that as compared to, you know, what's in the backlog right now? I guess this is an indicator of potentially stronger bookings if this comes to fruition.
Roger Dassen (CFO)
Yeah. I think, Chris, that is indeed the way to look at it. You know our view on bookings, right? The bookings, we always say, are not necessarily a good proxy of the business momentum. I think the bookings that we had in this quarter are actually pretty decent, as they were pretty decent in the previous quarter, without being downright spectacular, but pretty decent for sure. I think most of the things, most of the positive developments that Christophe talked about, and Christophe actually said it, only partially affect 2026. Most of this, the AI investments, the fact that multiple customers of us are benefiting from AI, whilst in the past, we always said it's only a limited number of customers that are benefiting from AI. The technology progress that Christophe referred to, the fact that we see more and more layer transitioning to EUV.
All that is good stuff. All of that, I would say, is good stuff, not necessarily being cashed in for 2026, but primarily beyond that. I think that's the way to look at it. All good stuff. Definitely, some of those having a partial impact on 2026. The great optimism is also, I would say, for what it does to the business in the longer term.
Chris Caso (Managing Director)
Helpful. Thank you.
Roger Dassen (CFO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Timm Schulze-Melander from Rothschild & Co. Redburn. Please go ahead.
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Yeah. Hi. Thanks for taking my questions. Maybe to begin with, for Christophe, on the High NA technology maturity, could you maybe just talk a little bit about on what measures High NA is kind of ahead of 0.33 at a similar point in time? Just thinking out to sort of 2028 and that journey, are you kind of halfway there, three-quarters of the way there, and then add a quick follow-up for Roger?
Christophe Fouquet (CEO)
I think it's a good question. Typically, when we look at the maturity of the system, I think there's two elements to it. The first one is when do we demonstrate the final specification of the tool at our customer? When we look at the EXE:5200, we expect that to happen in the next few months, most probably this year. That means that we validate the overall capability of the system. That's point number one. Point number two is, you know, the availability of the tool, you know, how much, what percentage of the time the tool can be used to run wafer. I think there a major difference between low NA and High NA is that when we looked at low NA maturity back at the time we were ramping the product, the availability was really dragged down for many years by the source performance.
The source was by far the biggest detractor of our maturity on the low NA tool. Now, I think you're fully aware that the source of High NA is exactly the same as the one on low NA. If we look at the availability number of the source itself, we are exactly matching the performance of low NA. What is basically separating us from today to final maturity is just the platform itself. Our experience is that most probably in 12 months, 18 months from now, we will be in a very good shape. There's no showstopper in the way we look at it today when it comes to maturity of the tool. There's a few more months we have to work with our customer to validate that.
When I look at the technology, when I look at the reason why we were struggling with low NA, those reasons are not with us with High NA. That's why I think our customers are also eager to report that maturity. That's also what they see. That's also their logic there.
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Great. Very, very helpful. Maybe, Roger, just in terms of profitability, High NA that you recognize revenue on, could you just maybe talk about, I know gross margin is dilutive, but you know, are we positive? What's the sort of runway there? Maybe just thinking about the operating expenses. Given the maturity of the High NA platform, what's the outlook for the sort of cadence of R&D expense, of the R&D burden for the business going forward? Thank you.
Roger Dassen (CFO)
Yes, indeed, High NA is dilutive. The key thing that will make it less dilutive, or the key thing that will drive up the gross margin is volume, right? You have a significant capability both in the factory and also in the field for High NA. That total cost base is only absorbed by a very limited number of tools. It is volume that is ultimately going to drive up the gross margin. As Christophe was saying, you're looking at taking stuff into high-volume manufacturing in the 2028, 2029 timeframe, right? In all likelihood, you're going to see meaningful numbers. At that stage, you will see the gross margin profile improving. I did say at the capital market today that even by 2030, I still expect High NA to be margin diluted, right?
It takes quite a long time before you get on the maturity curve and before you get significant volume for this to meaningfully contribute. At that timeframe, once you get meaningful numbers, at that stage, the dilutive nature of it will be limited. In terms of.
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Are we profitable?
Roger Dassen (CFO)
What was that?
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Was it profitable at the gross margin level in the quarter, or is it still loss-making, please?
Roger Dassen (CFO)
You mean High NA?
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Yes.
Roger Dassen (CFO)
High NA is very low margins. It's very low margins. You're talking very low positive margins. That's what you look at. In terms of operating expenses, R&D, I would say we still have a formidable roadmap ahead of us. In spite of the fact that we have High NA up and running, you hear us, if you look at the roadmap, if you look at the significant breakthroughs that we think we can still push in terms of low NA, in terms of productivity of those tools, in terms of imaging quality there, but also the progress that we continue to make on some of the DPV tools that we talked about, we still have a pretty formidable roadmap. I would say, however, that we are looking at further increasing the efficiency that we get out of the organization.
From that vantage point, I do believe that you will continue to see us manage both our SG&A and/or R&D quite nicely because we do feel that out of the formidable team that we have here at ASML in our R&D department, we can get even more value and efficiency there. I think you will see us navigate those numbers quite diligently and keep the increases quite controlled.
Timm Schulze-Melander (Head of Semiconductor and Technology Hardware Research)
Extremely helpful. Thank you.
Roger Dassen (CFO)
You're welcome.
Operator (participant)
Thank you. Your next question comes from the line of Alex Duval from Goldman Sachs. Please go ahead.
Alex Duval (Head of Europe Tech Hardware and Semiconductors Equity Research)
Yes. Hi, everyone. Thanks for the question. You talked today about lithography intensity inflecting positively. I just wondered if you could clarify the timeline you're thinking about here and to what extent this is a function of progress having been made in the gate all around transition versus other factors. Secondly, we've discussed on this call today about the degree of ambition around AI investment. We've also seen news items talking about AI chipmakers being more aggressive on the nodes they target for future chips. That seems to be somewhat different to how one had thought about where they would locate themselves relative to the bleeding edge. I wondered if you could talk about the implications for ASML of a faster cadence of these more powerful AI chips over time. Many thanks.
Christophe Fouquet (CEO)
Maybe on the first question, I think nothing really new there compared to the last few quarters. I think on the logic, we explained a few times, you know, the gate all around transition is happening without an increase on the number of EUV layers just because customers typically do first the transistor change before they start shrinking more aggressively again, which we still expect to happen at A14, A10. I think very consistent there in the previous view. In DRAM, we talked about 4F sq already today. Since the capital market day last year, we've got a lot of confirmation that indeed DRAM could get more aggressive in terms of EUV adoption for 6F sq, for 4F sq as well, as I mentioned. There, I would say the trajectory is very strong towards more use of advanced lithography moving forward. The second question is a good one.
I think we also mentioned last year in November that more AI applications will drive more advanced logic and more advanced DRAM. Most probably that part is still to be seen because we're looking at 12 months, which is a fraction of a node in terms of timing. What you see happening is what you describe is those applications, the value people can extract out of those applications can justify most probably moving to new nodes that are more expensive, as you have noticed, looking at the price of the wafer. Today can be justified by the value of AI. This has changed indeed the way people used to look at the industry. When we look at the industry driving mobile only, there was a lot of doubt on does the next advanced logic node make sense. I think that those doubts have gone away quite a bit.
The size and the speed of the ramp of 2 nm node on logic are the very first proof of that. We would expect that trend to continue. We have not seen yet a real acceleration per se. We mentioned again the larger customer base for those products. As you know, also a larger customer base also means that you get more competition and potentially also more motivation to move faster on those advanced nodes. I think that question is a very good one and one that we all have to look at in the next 12 months.
Alex Duval (Head of Europe Tech Hardware and Semiconductors Equity Research)
Super. Thank you so much.
Jim Kavanagh (Head of Investor Relations)
Okay. We have time for one last question. If you were unable to get through on this call and still have questions, please feel free to contact ASML Investor Relations with your question. Now, operator, can we have the last caller, please?
Operator (participant)
Thank you. Your last question comes from the line of Mehdi Hosseini from Susquehanna Financial Group. Please go ahead.
Mehdi Hosseini (Senior Equity Research Analyst)
Yes. Thanks for excusing me. Only good questions have been asked. I just have two follow-ups. One for Christophe. Obviously, we're all seeing the headlines with AI and everything. You have been highlighting how AI could drive incremental investment. The way I see it, you're also constrained with increased concentration of customers, especially on your logic. One customer is doing all the investment for the leading edge. That by itself drives more volatility in your booking, backlog, and even quarterly revenue. I'm not asking you to name that customer, but is that the right way of thinking about how AI is incremental, but it does limit your visibility? I don't know how to follow up.
Christophe Fouquet (CEO)
Yeah. I don't think we need to mention the customer. Your point is clear. I think we discussed that. I don't think there was any concern in terms of either visibility or in terms of, you know, sometimes pricing power. That's a question we got a lot if we had only one customer. I think the only concern really when you have only one customer is, are we going to be supply limited? Are we going to look at a market that is supply limited? Because if you have one customer, you know, the market will have the size of what that market, that customer can deliver. I think this was a bit our bigger concern. I think that's a concern that most probably goes beyond ASML, right?
That's what we discussed before because it's great to get all those good news about AI infrastructure investment, but at some point of time, you have to get the chips out. You know, as we mentioned a few times already today, I think that the news in the last few months of most probably more customers being able to play in AI, both for logic and DRAM, I think is a very interesting development for the entire market. That applies also, of course, for us.
Mehdi Hosseini (Senior Equity Research Analyst)
Sure. A follow-up for Roger. How should I think about working capital intensity over the past several quarters? Inventories have actually gone up, same with accounts receivable. As shipment strength improves, should I expect some improvement here with working capital and just overall cash room operation?
Roger Dassen (CFO)
Yeah. I think working capital is going up. Inventory is going up to a large extent because of High NA, right? With High NA, it takes quite some time before a system is being installed. As a result of that, obviously, that drives up the inventory level there. I think that's the major element. The other element is down payments, right? That's the other side of the equation. To the extent that down payments kick in, and that obviously is related to order intake, that's an important negative. I would say those are probably the two most important elements in there. When it comes to High NA, an important element in there, obviously, is reducing cycle time. That's a very important driver of driving down the working capital that is tied up in High NA. That's something that we're very clearly working on.
Of course, you need to have the volumes in order to make meaningful progress there. I think the working capital levels that you're currently looking at, I think, are sort of reasonable, I would say, for the business that we have today. To the extent that we're able to drive down cycle time, that I think is the single biggest driver of reducing working capital for us.
Mehdi Hosseini (Senior Equity Research Analyst)
All right. Thank you.
Jim Kavanagh (Head of Investor Relations)
Now, on behalf of ASML, I would like to thank you all for joining us today. Operator, if you could formally conclude the call, I would appreciate it. Thank you.
Operator (participant)
Thank you. This concludes the ASML 2025 third quarter financial results conference call. Thank you for participating. You may now disconnect.