ASML Holding N.V. (ASML) Earnings Call Transcript & Summary

Good day, and thank you for standing by. Welcome to the ASML 2025 First Quarter Financial Results Conference Call on April 16, 2025. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference call over to Mr. Jim Kavanagh. Please go ahead.

Thank you, operator. Welcome, everyone. This is Jim Kavanagh, Vice President of Investor Relations at ASML. Joining me today on the call are ASML's CEO, Christophe Fouquet; and our CFO, Roger Dassen. The subject of today's call is ASML's 2025 first quarter results. The length of this call will be 60 minutes, and questions will be taken in the order that they are 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 to Securities and Exchange Commission. With that, I would like to turn the call over to Christophe Fouquet for a brief introduction.

Thank you, Jim. Welcome, everyone, and thank you for joining us for our first quarter 2025 results conference call. Before we begin our Q&A session, Roger and I would like to provide an overview and some commentary on the first quarter results as well as provide some additional comments on the current business environment and on our future business outlook. Roger?

Thank you, Christophe, and welcome, everyone. Let me start with our first quarter accomplishments. In the first quarter of 2025, total net sales were EUR 7.7 billion, in line with our guidance. Net system sales were at EUR 5.7 billion, which includes EUR 3.2 billion from EUV sales and EUR 2.5 billion from non-EUV sales. Net system sales were driven by Logic at 58% and the remaining 42% coming from Memory. Installed Base Management sales for the quarter came in at EUR 2 billion. Gross margin for the quarter was above guidance at 54%, driven by achieving customer productivity milestones on already installed EUV systems as well as a favorable EUV product mix and a rich configuration resulting in higher ASPs. Operating expenses were in line with guidance, with R&D expenses at EUR 1.161 billion and SG&A expenses at EUR 281 million. The effective tax rate for Q1 was 16.7%. For 2025, we expect an annualized effective tax rate of around 17%. Net income in Q1 was EUR 2.4 billion, representing 30.4% of total net sales and resulting in an earnings per share of EUR 6. Turning to the balance sheets. We ended the first quarter with cash, cash equivalents and short-term investments at a level of EUR 9.1 billion. After the very strong free cash flow generation in Q4, we ended Q1 with a free cash flow of minus EUR 475 million due to a combination of customer payment and down-payment dynamics and continued investments in fixed assets for future capacity. Moving to the order book. Q1 net system bookings came in at EUR 3.9 billion, which is made up of EUR 1.2 billion of EUV and rounded EUR 2.8 billion of non-EUV. Net system bookings in the quarter were weighted towards Logic, 60% of the bookings, while Memory accounted for the remaining 40%. In Q1, ASML paid the third quarterly interim dividend over 2024 of EUR 1.52 per ordinary share. Recognizing the 3 interim dividends of EUR 1.52 per ordinary share each paid in 2024 and 2025 with a final dividend proposal to the Annual General Meeting of EUR 1.84 per ordinary share, this would result in a total dividend for the year 2024 of EUR 6.40 per ordinary share. In Q1, 2025, we purchased shares for a total amount of around EUR 2.7 billion. With that, I would like to turn the call back over to Christophe.

Thank you, Roger. As Roger has highlighted, we started 2025 with good first quarter financial results. Turning to the market and consistent with our view from last quarter, the growth in artificial intelligence remains the key driver for growth in our industry. If AI demand continues to be strong and customers are successful in bringing on additional capacity to support the demand, there is a potential opportunity towards the upper end of our range. On the other hand, there is still quite some uncertainty for a number of our customers that can lead to the lower end of our range. We continue to see revenue from Logic increasing in comparison to 2024, with the ramp of leading-edge nodes, and we expect Memory revenue to remain strong, similar to 2024. Installed Base Management revenue is expected to grow in comparison to 2024. This is driven by increasing service levels as our installed base grows and increasing contribution from EUV and an increase in revenue from our upgrade business. Regarding recently announced tariffs, discussions are just starting and are very dynamic. The end-state will be unknown for a while. And until then, the potential impact on our customers, suppliers and ASML will continue to be unclear and will continue to evolve. Roger will provide more details, but it is clear that uncertainty is increasing in the macro environment as reported by many experts and businesses. With that caveat, we continue to expect revenue of between EUR 30 billion and EUR 35 billion in 2025 and continue to expect 2026 to be a growth year. With that, I ask Roger to provide some insights about how we are looking at the recent tariff announcements. Roger?

Thanks, Christophe. As Christophe highlighted, we are currently facing an elevated level of uncertainty surrounding tariffs, which may have both direct and indirect implications for our business. The total direct impact results from tariffs related to a number of areas, including new system sales and upgrades to our U.S. customers, the import of material for our U.S. manufacturing facilities, the import of parts and tools for our U.S. steel operations and, finally, imports of parts from the U.S. into other countries to the extent tariffs applied to those parts. We are working with our customers and suppliers to try to achieve that any direct impact of tariffs on our results is limited. As Christophe said, the tariff discussion is still very dynamic. The potential indirect impact on end-market demand is even more complex and impossible to determine at this stage. With that, I would like to turn to our expectations for the second quarter of 2025. We expect Q2 total net sales to be between EUR 7.2 billion and EUR 7.7 billion. We expect our Q2 Installed Base Management sales to be around EUR 2 billion. Gross margin for Q2 is expected to be between 50% and 53%. The bandwidth for gross margin is larger than usual, given the uncertainty around the scope and size of the tariffs and the value chain absorption of tariffs for the quarter. The expected R&D expenses for Q2 are around EUR 1.1 billion, and SG&A is expected to be around EUR 300 million. The gross margin in the second half of the year is expected to be lower than the first half, primarily due to the expected margin-dilutive effect of the revenue recognition of High-NA systems in the second half of the year, lower upgrade revenue as well as any potential impact of tariffs. For the full year, we continue to expect a gross margin between 51% and 53%, of course, with the caveat of the certainties around tariffs that we discussed before. With that, again, I turn it back over to Christophe.

Thank you, Roger. Turning to technology. In EUV, we have achieved some important milestones on both the Low-NA and the High-NA platforms. These are critical steps in providing a comprehensive EUV product portfolio that offers the necessary flexibility to support our customers' road map requirements and optimize their cost of technology. Let me first update you on our Low-NA NXE:3800E. We started to upgrade our systems in the field to its final 220 wafers per hour configuration this quarter, and we continue to roll out on the installed base split through 2025. We now ship all our new NXE:3800E system at full specification. In addition, our NXE:3800E maturity is reaching the level needed to support high-volume manufacturing, and several Logic and Memory customers are ramping their most advanced nodes using this system. The gain in productivity supports the execution of our cost of technology reduction road map with our customer, enabling more opportunities for EUV single-exposed adoption. This is especially relevant to DRAM as discussed at our Capital Markets Day. Let me turn now to High-NA. At the SPIE conference in February, there were a number of good results presented by our customers who highlighted the achievements of some key performance and maturity milestone. They also stressed the benefit of the technology in terms of process simplification, cost and cycle time reduction. Process simplification, leading to a fewer process steps, shorter cycle time, lower cost and better yield are the historical value driver of single-exposed lithography versus multi-patterning. These benefits drove the industry transition to EUV Low-NA and will drive the transition to High-NA EUV over time. One paper showed that the High-NA system maturity is far ahead of what we experienced on Low-NA at the same stage of its introduction, supporting a much lower risk of insertion and adoption for our customers. Intel reported the exposure of more than 30,000 wafer in 1 quarter and a significant process improvement by reducing the number of process steps from 40 to less than 10 on a given layer. With that comes a significant cycle time improvement. Samsung reported a 60% improvement in cycle time in one of their use cases as well. We shipped our fifth and final EXE:5000 High-NA system in Q1 and now have system at 3 different customers, with the follow-on on High-NA system model, the EXE:5200, shipping from Q2 this year. As we have described before, there are 3 phases of technology insertion our customer will follow with High-NA. We are currently in Phase 1, where our customers take a system into their R&D facility and work with us to understand the value and capability of High-NA for their next nodes. In Phase 2, which we expect to take place in 2026, 2027, customer will start running the system on 1, 2 layers to test its readiness for volume manufacturing. And Phase 3, when customers design in High-NA on their most critical layers, in their most advanced nodes and run in volume manufacturing. Looking longer term, the semiconductor market remains strong with artificial intelligence, creating growth in recent quarters, and we see some of the future demand for AI solidifying, which is encouraging. Our conversations so far with customers confirm our expectation that both 2025 and 2026 will be growth year. At the same time, as Roger and I have already explained, there is an increased uncertainty across the global economy due to the ongoing discussion on tariffs. As discussed in our Capital Markets Day, we expect that the end market dynamics will lead to a product mix shift more towards advanced logic and DRAM. The combination of our NXE:3800E product progress, our strong productivity road map on Low-NA and the introduction of High-NA will support the cost of technology reduction and the conversion of more multi-patterning layers to a single EUV expose, leading to higher litho intensity. In line with our 2024 Capital Markets Day, we expect a 2030 revenue opportunity between EUR 44 billion and EUR 60 billion, with gross margin expected between 56% and 60%. Finally, as a reminder, we host our Annual General Meeting on Wednesday, April 23, and we hope to welcome our shareholders again there. With that, we will be happy to take your question.

Thank you, Roger, and thank you, Christophe. The operator will now 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 a short follow-up, if necessary. This will allow us to get to as many callers as possible. Now operator, could you -- could we have your final instructions and then your first question, please?

[Operator Instructions] Your first question comes from the line of Francois Bouvignies from UBS.

My first question would be on High-NA. So you have been doing a lot of work and very promising in terms of breakthrough and technology developments like you described, Christophe, so I know you are working on this common platform to help on the cost side. But obviously, like you described, it's a question of cost of ownership versus multi-patterning, so cost performance benefits where they're going to make the decision. And with that in mind, I was wondering if you would consider some flexibility around the pricing of High-NA to facilitate adoption, I mean, to help your customers decide more quicker, if that makes sense. So I was wondering if it's something you consider at all at the moment, the pricing of High-NA to help the adoption.

Well, Francois, thank you. It's a good question. I think you remind everyone that indeed, in general, single-exposed lithography will be better for customers than multi-patterning. I mentioned that this typically lead to process simplification, yield improvement, cost reduction. So when it comes to any new lithography system, I think, of course, we want to drive the adoption of those tools as quickly as possible basically to get to single-exposed. We are still working on that, as you know, for Low-NA, so that also typically takes some time. And of course, we're going to do the same for High-NA. Now usually, the main reason to not adopt very quickly the new system fully for single-exposed is not the tool price, to be honest. It's the tool maturity. And I think it's very important to get that maturity at the right level because, if not, you will set basically a cost of technology point that will be not very optimized. And this is why, not only I refer to maturity in my introduction, but you also heard me quoting a customer also making the case at SPIE that the maturity of High-NA was, by far, ahead of the maturity of Low-NA at a similar stage. So this is still the work we have to do. I mentioned the 3 phases. I think it's a major part of Phase 2. And I think if you look at the focus of ASML today with the customer on High-NA is really to get this maturity in place as soon as possible so that the adoption is there. Lowering the price at low maturity will just create too much headache for our customer because the tool will not be very reliable.

Makes sense. And maybe my quick follow-up would be on the comment on single-exposed EUV that you start seeing, I mean, the adoption on more single-exposed versus multi-patterning, I guess, on Low-NA that you are referring. Can you help us on the timing? You mentioned DRAM specifically that you see more adoption on single-exposed. I mean what timing should we look at this opportunity? If you start seeing it, I would imagine, it's a 2 years view you are talking about, or could it be quicker?

I think those things are happening, I will say, almost as we speak. So I think every new customer nodes where we bring basically a tool with a better cost of technology, such as the 3800E, is an opportunity for more adoption. So I think this is, I would say, a permanent job that's something we have been working with our customers for a bit. I think the place we are today with Low-NA EUV, the maturity of the tool, the step we do on productivity, remember, the 3800 is 30% faster than the 3600. I would say, really give us a chance to be optimistic on that work with our customers in the coming years. But the work already started, Francois.

Your next question comes from the line of Krish Sankar from TD Cowen.

First one I had for Roger. I'm just kind of curious, you talked about growth next year. What kind of bookings run rate should we expect in the current and next quarter to see that growth? And if calendar '25 ends up in the upper end of the range, would calendar '26 still be a growth year? And then I have a follow-up.

Yes. So Krish, we're not going to comment on the magnitude of the growth, as we already also said on the video. We still believe '26 to be a growth year based on our technology, based on the conversations we have with customers, based on the intrinsic market demand as we see it, with the caveat that we also talked about, which is the general macro climate. In terms of what bookings do we need to get there, I'm not going to go into any detail there. I think you can easily figure out what the backlog is today. I think you can easily figure out that there is a very significant part in the backlog that actually pertains to the period beyond 2025. So there is already a very good booking level, I would say, for next year. But of course, bookings still need to come in order to make the growth year, indeed, happen. So I think that's what we should see in the next quarters. But I'm not going to quantify, also in light of the comment that we made before on the nature of the bookings and the extent to which it does not really reflect the business momentum always accurately of our business.

Got it. Got it, Roger. And then as a quick follow-up, do you still expect China to land somewhere in the mid-20% of sales? And can you give any color on the backlog composition, how much is EUV, deep UV in China in that mix?

Sorry, you broke up in the first part of the question. I couldn't understand it, Krish. Maybe you can repeat it.

Yes. Sorry, Roger. I was trying to figure out, is China still going to be around 25% -- mid-20 percentage of sales this year? And any color on the composition of the backlog?

Understood. So yes, China -- we believe China will be a little bit over 25% of sales this year. As you know, in the previous call, we said it's low 20s. The way we look at it today, the demand within deep UV is shifting a little bit, and I think the right way to look at it today would be that it's a little over 25% of our total sales this year. And in terms of the composition of the backlog, if what you're referring to is the China part of the backlog, I guess, that's what your question is, I think that the China part of the backlog is still in this -- around the same number. So it's still in that 20% to 25% range approximately.

Our next question comes from the line of Joe Quatrochi from Wells Fargo.

Yes. One on the tariff front. Can you talk about just what your customer conversations have been like over the last couple of weeks? Is there any interest in altering delivery schedules? And just kind of how do we think about your [indiscernible] in the context of the ability to like pull shipments in to kind of elude some of the tariffs?

Well, maybe I can start, and then Roger will add. But I think the short summary of our customer conversations so far, I think that the announcement of tariffs have not changed the business conversation we have with our customers. So that's the first point. The second point, I think the level of uncertainty we shared in the introduction, which, of course, as the result of the many announcements we have seen in a very short time, I think that uncertainty is also with our customers and with our suppliers. So a lot of people are still trying to understand exactly what it means for them. And I think in some way, we start to try to do that together. But that's extremely preliminary. And I will say, again, once more that, so far, those discussions have not changed fundamentally the business planning or the business discussions we have been having with our customers.

Because fundamentally, I mean, it's even impossible for a number of customers to do what we're suggesting here, because, of course, to a very large extent, what is gating for customers is space, right? So they need to have fabs to put tools into. So -- and at least for a number of customers, that is the #1 gating item, so it would even be impossible to opportunistically just pull in the system. That's not the way for us.

Yes. No, that makes sense. I appreciate the color. And then you talked about the 3800 Low-NA becoming the main tool for your system shipments now. Can you remind us, did you -- how many of the 3600 are still left that you prebuilt, I think, last year in inventory? Is there anything that we should think about like in terms of the salability of those tools? Or can you upgrade some of those tools to be closer to the productivity of the 3800 to sell?

Yes. We have no concerns actually on the ability to sell the 3600s. So that's -- the 3600s, the number of prebuilds, you will see that come down this year, and there is no concern around the ability for us to sell the remaining 3600s that we even build -- that we built or are in the process of building. No concern there.

Your next question comes from the line of Didier Scemama from Bank of America.

I just wanted to make sure I get everything right. So I think if you look at the blended ASPs of your Low-NA EUV, I think it came in at about EUR 230 million. If I do the math correctly, also on your gross margin, it looks like your EUV gross margins were at about 55%. So my question is, is that correct? And I've got a quick follow-up, especially if you could talk about whether there are any one-off or exceptional because I think you talked about upgrades, whether that came into the ASP, the gross margin or in IBM or in all elements.

So you're right on the -- on your calculation of the ASP. It is indeed close to EUR 230 million. I think it's EUR 227 million, is the number on the ASP. That's -- as we said, it's a little on the high side given composition of 3800, 3600, also configuration, et cetera. But I do believe on a go-forward basis, I think a blended rate a little bit below that point. So a blended rate of around EUR 220 million is probably the right number to have in your models as a blended rate for Low-NA. When it comes to the gross margin, we don't disclose that on a product-by-product basis. But I think we said in the past that the gross margin on Low-NA is clearly now above the corporate gross margin. So that's clearly the case, but we're not separately disclosing that.

Got it. And on China, I just wanted to also get a bit of color. So you said that from the previous guide of the low 20% of group revenues, you think China is going to be a bit above 25%. So can you talk about what's changed there? And also if you could maybe, within that, give us a sense of what's your directional guidance would be on China IBM revenues because, obviously, there's been restriction on maintenance and services on these 14 fabs. So just wanted to understand what you've provisioned for that IBM revenues in '25.

Yes. So Didier, when it comes to a few percentage points changes, I think we should recognize that you're looking at a couple of hundred million, right? So we're not looking at huge shifts. We're looking at couple of hundred million shifts, which I think is normal business dynamics that you have a shift like that. So I think what you notice is that the demand in China is still strong. And if you then look at where is the demand strong, it is still strong particularly in the mainstream business. That's where the demand continues to be very strong. And that is on the back of the demand for mainstream chips in China, both for domestic consumption but also for what China exports to the rest of the world. So that demand is still resilient and, in fact, a bit better than what we anticipated 3 months ago or 6 months ago.

Your next question comes from the line of Alexander Duval from Goldman Sachs.

Firstly, there were some reports mid-quarter talking about China being close to develop its own alternative to EUV lithography tools. I wondered if you could just share your latest view on those and to what extent those really have sufficient scalability, reliability and so on. And secondly, you talked today about AI demand solidifying, if I heard well in your prepared remarks. We saw NVIDIA at its GTC event talk about increasing inferencing and reasoning models, driving more semis demand. I wondered if that could be something that could impact your long-term TAM expectation in a positive way.

Okay. I'll take the easy question on EUV China. I think there is nothing really new there. I think that we expect to continue to see news here and there on some progress with regards to EUV in China. And I think this is mostly driven by a strong wish, I think, from China to have this tool and to display some progress. If you look at the fundamental, even if you look at what has been shown, we have seen some picture, I think that I would consider that as research news more than product news, and therefore, of course, it's always possible to generate some EUV light. It may even be possible to have an EUV mirror here and there. But in no way this is enough proof that there is a serious product on the way. So I think we are still on the same view that it will take many, many years for China to be able to make an EUV machine. And again, you should expect some more news because I think that's just what you do when you want to show progress.

And on the -- Alexander, on the AI demand, I think the comment that Christophe made on that, I think, is a comment in general. So what we hear from customers, both on the Logic side and on the Memory side, is that they still see strong demand there. I think the point that you make is right. I think there has been a lot of emphasis in the past quarters on the training side of life. I think more and more, which I think is logical, that you also see more and more emphasis being put on the inferencing side of the equation. So I think you will see the inferencing part becoming a larger component of AI demand on a go-forward basis. So I think you will continue to see that develop.

Your next question comes from the line of Chris Caso from Wolfe Research.

I guess first question would be on gross margins, and you have provided a wider range for Q2, taking into consideration some of the tariff impact. But speaking for the full year, what are the expectations there? What sort of direct tariff impact are you expecting in the gross margins? And what goes into the thinking with regard to the gross margin for the full year?

Yes, Chris, as we also said on the video, it's very hard, given the dynamics around tariffs to put any meaningful number on there, right, because on the one hand, the question is what tariff will eventually -- what tariffs are we eventually going to look at, in general, from one region to the other. But then more specifically, when it comes to semiconductors, it's pretty clear that, that is still under review by the U.S. government. So it's impossible to see what, as I mentioned before, what the size of the tariffs is actually going to be. And then the second question, to the extent that, that is clear, the question is, how will that ultimately be absorbed in the entire value chain? And we've made it clear also on the video that we're working very closely with everyone involved to try and minimize the total exposure of the ecosystem to tariffs. So we try to see what we can do to minimize the overall impact. But we also believe that once that has been minimized, that the burden of that should not be with ASML and that the burden of the tariff and the lion's share of the tariff burden should be borne by the next element in the value chain. I think that's the way we approach it. But there is still so much uncertainty out there that to make any judgment what impact that is going to have on the full year is absolutely impossible. The reason that we put it into the quarter that we said in the quarter, we have a wider bandwidth is that, obviously, the time frame there is a bit shorter and that allows you to try and include the impact into the quarter. But for the full year, it's impossible to put a number on it.

Understood. That's helpful. If I could ask as my follow-up, a bit of a bigger-picture question. And there's been a school of thought that if we need to diversify the geographic location of fabs going forward, if we're going to produce more in the U.S., if produce more in China, then that's ultimately good for WFE spending. And it would seem that the current tariff situation would at least advance that narrative. I'm just interested in your latest thinking with regard to that and sort of geographic diversity and the effect on your business and perhaps some of the conversations you may have had with customers with that regard.

I think that notion, on the one hand, still holds. So I think the notion that having dispersed fabs across the globe in all likelihood will drive to, ultimately, to more capacity having to be installed in order to still be able to drive the same number of wafers. So there will be a heightened level of inefficiency in there. I think you actually see that now in the entire value chain. You see it in semiconductor manufacturing. You see it in discussions on data centers being spread across multiple continents more so than has been considered so far. So I think this whole notion of having a more dispersed nature within our ecosystem, you see that in multiple basis. Eventually, I think that will drive up the demand for semiconductor, for wafers. And then to the extent that those wafers are made in different places, I think that will drive that up. So there is a potentially positive element in that. But as Christophe very clearly said it in the video and also in the call at the beginning, the uncertainty that we're currently having on tariffs is a dimension that we also have to consider there. So I think it's a story of puts and takes.

Your next question comes from the line of Mehdi Hosseini from Susquehanna.

Yes. And I do have 2. Starting with EXE platform. Christophe, can you remind us what are the key milestones as we go from R&D EXE:5000 to production, which is EXE:5200? And those milestones, it could be in terms of the throughput or any other factors that you can share with us.

Yes. I'll try to do it again quickly. So I think we usually talk about 3 phases. And the first phase is -- of High-NA for our customer is to receive the first tool or use the tool we have in the lab in ASML to test the technology. It's a completely new technology, new imaging, new performance. So they use basically this time to really validating some way that the design of the technology deliver what they had in mind. And this is done with R&D tools. So this is why we had the EXE:5000. This was a tool that we really basically sought through in order to support this R&D work. And this is a bit where our customers are today. That's also why we are usually happy when they share results because they show basically that this phase is progressing well. The next phase is basically when they will really start to test the tool in what I would call early production. So they will select a limited amount of layer, a limited amount of product, and then they will really run the tool in production. But this is also where our EXE:5200 become important. So this is a tool with higher productivity, more maturity, basically a tool that enable customers to really start this second phase. And as you heard, we are starting to ship this tool now. And once this is done, typically, a customer will be convinced that the tool can do the job, the tool is mature. And then they will use it fully for any future node in high-volume manufacturing. And that, I think, we have said will come mostly '27, '28. So that's a bit the sequence. And it can seem a bit long. But of course, you need to do that across a couple of nodes. And so far, I think that's the progress we are witnessing with our customer.

I see. And given that your earlier comment that focus on key takeaways -- I'm sorry. Given your earlier comment that has to do with the key takeaways from SPIE conference in February, it seems like advanced logic would be the first adopter, and then there's a debate as to what happens to foundry versus DRAM as the second type of adopter for these High-NA technology.

Well, I think it's hard to say because, to be honest, we said it in the past, the timing is very close. And when we refer to the 3 customers, when we refer in the past the customer having access to our lab, this is really covering both Logic and Memory customer. And I would say both Logic and Memory customer have good reason to use High-NA as soon as the maturity of the tool and, therefore, the cost of technology on the tool will be there. So to be honest, yes, historically, we always expect Logic to be first. It could be the case, but it's a very close call. So I will not make a bet today. But both Memory and Logic customer are really working hard to qualify the tool.

Your next question comes from the line of Tammy Qiu from Berenberg.

So first one is on China. So my understanding is your customer list has been -- keep expanding in China. Can you share if your percentage of China revenue from the big 4 chipmakers has been decreasing over time or not really changed?

Percentage of the...

Big 4.

Of the...

Big 4.

Big 4 from China.

Okay. Over time, that has decreased. Over time, that has decreased. But when you -- well, big 4, you mean the -- sorry, you mean the domestic ones? Sorry, sorry, sorry.

Domestic ones, yes. Domestic ones.

The tail in China has definitely become longer, right? So the number of players have become longer. That's for sure. But still a significant part of the shipments into China go to the large players. The tail has definitely become longer, but the number of tools that the tail takes is obviously smaller. So there is still quite a big part of the China sales going to the large players.

Okay. I see. Okay, that's clear. And also, currently, your best deep UV tools with China is 1970, 1980i. So assuming you can only ship 1950 or below, can China still make 28-nanometer from a technical perspective using multi-patterning based on 1950 or 1940?

Yes. 28-nanometer for sure. I think that this has been done in the past by other customers. So I think, yes -- if you go back in history, when 28-nanometer was run by some of the non-Chinese customer, these were the tool basically we were looking at. So it's definitely possible, I would say. Even smaller technology can be run with those 2.

Your next question comes from the line of Timm Schulze-Melander from Redburn Atlantic.

Maybe just 2 quick ones. You commented on the gross margin guide for the coming quarter. Just in terms of the order backlog, could you give us some color what proportion is priced ex works? And maybe what proportion of the backlog is ASML responsible for, for custom fees, delivery duty paid, et cetera?

In general, well, it depends from one country to the other, right, exactly how it works. But in general, we are the importers into the country. But of course, that is actually a bit irrelevant because, at the end of the day, it depends on the contractual agreements that we have with the customers, how you pass on the -- how you pass on tariff increases there. And as I mentioned before, we believe that the cost [ and burden ] tariffs are going to give, we believe that we should not be the bearer of that. There should be a fair allocation of that burden within the value chain.

Very clear, very clear. And then just a follow-up. You talked about the EXE:5200. I think, Christophe, you just mentioned that ASML is starting to ship this tool now. Maybe just looking at 2025, will we expect any of those tools to rev rec this year? Or is the High-NA rev rec entirely 5000s?

So we also expect 5200 rev rec in the year because remember, we said we are expecting 5 rev rec this year. We had 2 rev recs last year. And in total, as you know, we have 5 tools -- 5 5000s that we ship to customers. So that tells you that in the 5 that we talked about before for rev rec this year, there's also 5200 in there.

Your next question comes from the line of C.J. Muse from Cantor Fitzgerald.

I guess there's obviously greater uncertainty around geopolitics. But in Q1, NVIDIA will surpass Apple as the #1 customer at TSMC. And it's a very important milestone for HPC surpassing mobility. So with that as the backdrop, would love to hear the visibility you have today, the conversations you're having with your customers today as it pertains to 2026, 2027.

Well, I think maybe I'll answer the first part on your comment. I think Roger said it already. I think mostly I said it as well. I think the demand on AI is still very strong, and I think this has been confirmed by some of our customers, I think, by some of our customers' customers because when it comes to AI, at least in the next couple of years, you are looking at major investment basically. And that's why your reference to HPC is the right one. So because you are looking at major investment, investment has been committed, investment that a lot of company believe they have to make in order to basically enter this AI race, I think the threshold to change this behavior is pretty high. And this is why -- this is what our customers are telling us. And that's also why we mentioned that, based on those conversations, we still see '25, '26 as growth years. That's largely driven by AI and by that dynamic. Now '27 start to be a bit further away, so you're asking us too much, I think, to be able to answer basically what AI may look like in '27. But if you look at the next couple of year, so far, the commitment to the AI investment and, therefore, the commitment also to deliver the chips for AI has been very solid.

Great. Very helpful. I guess as a follow-up, and I guess to follow up on a prior question. At Capital Markets Day, the theme of replacing double-pattern with a single EUV step was a major focus. But my impression then was that there was real work to be done in terms of the throughput on EUV to do that. But given your commentary today in the video and on this call, it certainly sounds like that progression is happening sooner. So can you kind of speak to when you see that occurring? I believe you're talking about it happening first in DRAM, but would love to hear kind of how you see that playing out this year, next and how we should be thinking about the overall implications to litho intensity.

Yes. I think I tried to explain that to Francois before. I think we will see that gradually. So I think, yes, it's a lot of work. But every step we make on productivity, every step we make also on the, I would say, maturity, the efficiency of the tool of our customer bring basically the cost of technology of Low-NA EUV down. And as I said a few times already today, single-exposed is so much better for our customer than multi-patterning when it comes to complexity, cycle time, et cetera, that the minute we are going to match basically the cost of multi-patterning with single-exposed, most probably the customer will move. So we'll see that gradually. That's also why at Capital Market Day, we show basically that we expect every node both for DRAM and logic to bring more EUV layers. I think it's a bit more maybe spectacular on DRAM just because the number of multi-patterning in DRAM today is still higher than it is for logic. So this is where we see on the short term the -- maybe the fastest progress. But this is something we will experience for the few years to come.

Your next question comes from the line of Stephane Houri from ODDO BHF.

Yes. I have a first question about the order volatility and maybe to understand better from you if you think that the uncertainty around the tariffs in Q1 already had an impact on your orders or not at all. And the second question is about the coverage of the middle of the range for 2025. Basically, last quarter, you said that your middle of the range was covered for EUV. So the follow-up question is, what about the EUV, which is supposed to grow pretty strongly, notably out of China?

So let's start with that question. So you're right. We said last time that for the midpoint of the range, we are fully covered with EUV. For the midpoint of the range, when it comes to deep UV, we're covered approximately 90%. So nearly there, which, given order lead times, I think it's a very good place to be in. Back to your first question, order volatility. I think we go back to what we said about that before. I think order volatility and the lumpiness of order intake is what we've seen, not just in this environment, but what we've seen for the past couple of years. And that is, I think, first and foremost, driven by the fact that a customer that puts in a very, very significant order in 1 quarter has to go through a lot of governance steps in order to get that order approved and is, therefore, unlikely to then, the quarter thereafter, come back. So to the extent that you have major order intake as we had it, for instance, in Q4, that has a bearing on the order intake in the quarter thereafter. I think that erratic pattern, I think, you've seen in the past couple of years. So that's why we said it is not necessarily a good proxy for the business momentum. And so therefore, I think that's what you're looking at here, much rather than customers taking already a view on what tariffs mean for their business. I don't think you can read that from this order book.

Your next question comes from the line of Andrew Gardiner from Citi.

Another one on tariffs. I wanted to take the conversation in a slightly different direction. On one hand, the U.S. government is trying to encourage reshoring of semiconductor manufacturing in the U.S. And yet in achieving some of those goals, particularly with TSMC and a quite high-profile announcement earlier this year, they're now turning around and saying, well, yes, we'd like you to do that, but we're also going to slap tariffs on some of the tools that you need in order to make that happen. It does strike me as at least a little bit inconsistent. I was wondering about your position on that in Washington and perhaps, even more importantly, the position of your customers. I can imagine that they're not too pleased by it. I was just wondering whether -- how can the industry sort of push back or explain that there is no U.S. litho industry. That ship sailed over 40 years ago. And really, we need these tools, and we need some relief under this. It just doesn't make sense. Are those arguments being made? Is it too sensitive a topic to be made? Where are we on that?

Well, I think, Andrew, we understand your comments. It's not the first time we hear it. And that might be the reason why, as you know, since a couple of days, this industry or a significant part of the products within this industry are now not put on -- are now exempt from tariffs. But at the same time, U.S. administration has also said, we're reviewing the entire ecosystem to figure out how to deal with this. I think it is this complexity that is being recognized. I think it's being recognized by players such as ourselves. I think it's being recognized by customers, and I think it's being recognized by the U.S. administration. And that's why I think they've said they need a bit more time to understand how to do this and how they can achieve the objective with more onshoring of chip manufacturing in the United States and how they can reconcile that with this. I think that's the background of the delay that we now see and of the anticipated and announced review of this ecosystem.

Okay. Understood. And just a quick follow-up on the High-NA commentary. Christophe, you've talked about the different phases. Clearly, you're booked. You've had sort of long-term backlog to achieve Phase 1. Would you say that Phase 2 is also represented in the backlog, and therefore, really incremental orders that come are only necessary for the Phase 3, sort of '27, '28 production ramp?

Yes. I think we discussed that last quarter. I think we mentioned several times that we have double-digit booking for High-NA. So that's definitely enough to cover basically both Phase 1 and Phase 2. And the order for Phase 3, I will say, come -- will come when the level of confidence has achieved a place where the tool will be, for sure, used in HVM. So that, by the way, can also happen during Phase 1, but mostly, we still have a few months to go before we see that happening.

Okay. So 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, may we have the last caller, please?

Your final question for today comes from the line of Sandeep Deshpande from JPMorgan.

My question is, at this time, in April 2024, how was your order book and backlog looking for 2025? And how does the order book and backlog for 2026 look at this time of the year versus 2024?

Goodness me, Sandeep. I'm not into the history lesson here. I think I made a comment at the -- at sort of one of the earlier questions. I think it's not too difficult, I think, to figure out where our order book is for this -- our total order book is. And I also believe it's not too difficult to figure out what the component is at the midpoint of demand beyond 2026. I think you're able to figure that out. And then you will see that there is -- there are quite some orders already in there for the period beyond 2025. So I think there is a really good start for the year '26, but we refrain from making any further comments on what we think 2026 looks like. And we also refrain from saying, and therefore, this is exactly the number that you need for next year. I think we've indicated how we look at bookings. And therefore, I think it would not be appropriate for us to give further commentary on what bookings you still need in order to get to whatever midpoint is for 2026.

And my quick follow-up to Christophe on High-NA. You talked about the Phase 3. Do you expect to start getting these Phase 3 High-NA orders by the end of this year or the second half of this year?

Not too many. We could have a few, but I think that's still too early because, remember, we're still mostly in Phase 1. We're still installing some of the tools. So no, this will mostly take some more time.

Okay. 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 much appreciate it. Thank you.

Thank you. This concludes the ASML 2025 First Quarter Financial Results Conference Call. Thank you for participating. You may now disconnect.