Good afternoon. This is the Chorus Call Conference operator. Welcome, and thank you for joining the ASM Second Quarter 2024 Earnings Call. As a reminder, all participants are in listen-only mode. After the presentation, there will be an opportunity to ask questions. Should anyone need assistance during the conference call, they may signal an operator by pressing star and zero on their telephone. At this time, I would like to turn the conference over to Mr. Victor Bareño, Head of Investor Relations. Please go ahead, sir.
Thank you, operator. Good afternoon, and welcome everyone to our Q2 earnings call. I'm joined here today by our CEO, Hichem M'Saad, and our CFO, Paul Verhagen. ASM issued its second quarter 2024 results yesterday at 6 P.M., Central European Time. For those of you who have not yet seen the press release, it is accessible on our website, along with our latest investor presentation. As always, we remind you that this conference call may contain information relating to ASM future business and results, in addition to historical information. For more information on the risk factors related to such forward-looking statements, please refer to our company's press releases and financial reports, which are available on our website. Please note that the profitability measures mentioned in this call will be primarily based on adjusted non-IFRS figures.
For the reported figures, as well as the reconciliation between IFRS and adjusted results, please refer to the quarterly results press release. With that, I'll now hand the call over to Hichem M'Saad , CEO of ASM.
Thank you, Victor, and thanks to everyone for attending our second quarter 2024 results conference call. It's a real pleasure to talk to you today, and as most of you know, I took over as CEO last May. I already met many of you in our previous investor days and at other occasions, and I look forward to meeting many more of you in our upcoming investor conferences and roadshows. For today's call, Paul will first review our second quarter financial results. I will then continue with the discussion of the market trends and the outlook, followed by the Q&A, as usual. Over to you, Paul.
Thank you, Hichem, and once more, welcome, everyone. Before I begin, I'd like to make an organizational announcement. Following the appointment of Hichem as our CEO, we decided to redistribute some of Hichem's day-to-day responsibility at the executive committee level. As part of that, we expanded the executive committee with two new members, both with impressive track records and experience in our industry. Paul Ma, since 2018 with ASM, and responsible for the thermal products business unit, and Steve Reiter, already with ASM since 2015, and in charge of plasma products and epitaxy. Let's now review the financial results. In the second quarter of 2024, revenue came in at EUR 706 million. At constant currencies, up 6% year-on-year.
Revenue was slightly above the top end of our guidance of EUR 660 million-EUR 700 million. Equipment revenue increased 4% year-on-year at constant currencies, and was mostly driven by robust ALD shipments, which accounted for more than half of equipment sales. Spares & Services revenue had a solid performance with 20% year-on-year growth at constant currencies. In terms of customer segments, revenue was led by foundry, followed by memory, and then the power analog wafer segment. Combined logic foundry sales continued to represent the larger part of our sales and were roughly stable, both versus Q1 and compared to Q2 last year. Compared to the first quarter, sales related to the gate-all-around 2 nm technology increased, while mature logic foundry sales were lower. Memory sales increased strongly both year-on-year and versus Q1.
DRAM sales accounted for the larger part of the memory and reached a quarterly record high. This was mostly driven by ALD and high-k tools for high bandwidth memory. 3D NAND revenue also increased markedly, but that was from a relatively lower base. Sales in the power analog wafer segment, excluding silicon carbide, were somewhat higher than in Q1, but still meaningfully lower than the average trend last year. Gross margin in the second quarter came in at 49.8%, up from 49% in the same quarter last year, but down from the exceptionally high level of 52.9% in the first quarter, mostly explained by mix. Part of the mix development is the contribution of China sales, which were still strong in the second quarter of 2024, albeit below the record high level in the first quarter.
SG&A expenses increased by 19% compared to the second quarter of last year, and by 20% quarter-on-quarter. SG&A included one-off tax expense of EUR 8.4 million, resulting from the fresh vesting of previously granted performance shares. Excluding this one-off item, SG&A expense increased 7% year-on-year, and for the full year 2024, we continue to expect a moderate increase in SG&A. Net R&D in the second quarter increased 11% year-on-year. For the second half, we expect a higher increase in net R&D due to higher amortization of development expenses. As explained in the previous quarter, this is due to a number of newly developed products and applications for which amortization is expected to start over the course of the year. For the full year 2024, we expect net R&D to increase between 10% and 20%.
Operating margin decreased to 25.8%, down from 26.9% in the same quarter last year, last year. Not taking into account this one-off tax charge that I just mentioned, EBIT margin would have been 27%. Below the operating line, financial results include the currency translation gain of EUR 16 million. This compares to a translation gain of EUR 8 million in the second quarter of last year and EUR 23 million in the first quarter of 2024. As a reminder, these currency gains and losses are explained by the fact that we hold the largest part of our cash in U.S. dollars, and the translation differences are included in our financial results.
Then over to ASMPT, briefly, our share in income from investments reflecting our stake in ASMPT, amounted to EUR 4 million in the second quarter, down from EUR 5 million in the first quarter and down from EUR 9 million in the second quarter of last year. Now, back to ASM. Our order intake in the second quarter was at a strong level of EUR 755 million, an increase of 56% at constant currencies compared to the same quarter last year and up 8% from the first quarter. Orders were led by memory, followed by logic and foundry. Memory orders increased strongly to a quarterly record high and consisted primarily of orders for HBM-related DRAM applications. The underlying trend is robust, especially for HBM, and we expect continued healthy memory demands.
It should be noted that DRAM orders were particularly high for this quarter, and order intake can be a bit lumpy from quarter- to- quarter. Total logic foundry orders decreased compared to the first quarter. gate-all-around orders were solid and increased somewhat compared to Q1, mature logic foundry orders decreased year-on-year and compared to the previous quarter. In silicon-based power analog, bookings were at a decent level despite continued market weakness. silicon carbide epi orders were solid, the highest level in the past six quarters, and included multiple tool orders for a new 200 mm fab project in China. Next, the balance sheet. We ended the quarter with EUR 637 million in cash, down compared to EUR 712 million at the end of Q1.
In the second quarter, we generated a solid free cash flow of EUR 103 million, thanks to continued solid profitability and with only a modest outflow for working capital. CapEx amounted to EUR 36 million in the second quarter, and as a reminder, our 2024 target for annual CapEx continues to be between EUR 100 million-EUR 180 million. In Q2, we used EUR 135 million in cash for the payment of dividends of EUR 2.75 per share. In addition, we used EUR 59 million for share buybacks as part of the EUR 150 million buyback program that was started on May 15. And at the end of June, we had completed 39% of the program and 78% at the end of last week. And with that, I hand the call back over to Hichem.
Thank you, Paul. Let me first start with a recap of our strategic priorities. After I took over as CEO in May, a number of investors asked if I was going to make any changes to the strategy. My answer is clear: We will continue with our growth through innovation strategy. I joined ASM in 2015, and in the past five years, I was in charge of the entire product portfolio, as well as of research and development. In 2022, I joined ASM's management board, which means I have been privy to many of the decisions that were made over the past few years, including company strategy, investment in infrastructure, and so on. I feel very confident that ASM's growth through innovation strategy has been working, evidenced by our strengthened position with key customers and our revenue growth that has been clearly ahead of the industry.
It's an exciting time to be in our industry, and I feel upbeat about where ASM stands in our market today. The semiconductor market is on a solid growth trajectory, driven by structural trends such as digitalization and electrification. Third-party research firms forecast the semiconductor market to reach a size of EUR 4 trillion by the end of the decade, and AI is expected to account for a meaningful part of the expected growth. In addition, AI will require more powerful and power-efficient computing solutions. To enable the next generation AI chips, such as AI accelerators and high bandwidth memory, our customers are working on 3D device technologies and new materials, all of which will require more ALD and epi steps. In Logic Foundry, the transition from FinFET to gate-all-around requires new ALD depositions for gate stack features, a trend that will increase with subsequent generations of gate-all-around
followed by CFET by the end of the decade. We see a similar trend toward more ALD for memory products. DRAM is adopting more advanced logic processes as performance gets enhanced. We are confident these trends will keep the ALD market on a strong growth path towards our earlier communicated forecast of $4.2 billion-$5 billion by 2027. We target to keep our leading share in logic foundry ALD, and to step by step expand our position in the memory market. Another strategic priority is to expand our position in silicon epi. We expect epi intensity to increase in the transition to gate-all-around. Our innovative technology has been accepted by more and more customers. In our other product lines, we are making selective investment to drive additional growth.
A great example is the introduction of our new Sonora system in vertical furnaces, for which we want several new customers in the power, analog, and wafer markets. silicon carbide epi is another attractive product line where we are well-placed to expand our position as customers transition to 200 mm wafer size. To stay ahead, we are investing in our people. We are investing in R&D, and we are also investing in our infrastructure. A year ago, we announced the new site in Korea to double our footprint in R&D and also increase our manufacturing in Korea. In Arizona, last December, we announced a major expansion for both epi and ALD R&D. Lastly, we continue our focus on sustainability. A highlight in the first half of this year was the achievement of 100% renewable electricity for all of our global operations.
Let's now move to the current market trends. Conditions in the semiconductor market continues to be mixed, with a patchy recovery in smartphones and PCs and ongoing weakness in automotive and industrial. Accelerating growth in AI is, however, driving increased wafer fab equipment investment in high-bandwidth memory and in leading-edge logic and foundry. Momentum in the leading-edge logic foundry segment is picking up pace, supported by the transition to 2 nm gate-all-around. We had, again, solid gate-all-around orders in the second quarter, up somewhat from the level in the first quarter. Our customers are reporting good progress and yield improvement in their pilot line activities. The start of high volume manufacturing is still targeted in 2025 across our leading customers, although we already have some orders from customers who want to start initial volume manufacturing for part of their products by the end of 2024.
2 nm technology node offers significant improvement in device performance and in power efficiency. Our customers have been talking about strong demand for 2 nm from their own customers, including from AI chip makers. Based on customer traction, the gate-all-around transition will be a strong driver for ASM. As previously communicated, our served available market in logic foundry will increase by some $400 million for each additional 100,000 monthly wafer capacity, as many new ALD layers are required in gate-all-around devices. In silicon epi, we have been selected for various applications by all three leading customers, compared to one leading customer we had in the previous FinFET nodes. A significant portion of gate-all-around orders in the second quarter consisted of Intrepid ES epi tools for gate-all-around, reflecting our expanded customer base.
In R&D, we are strongly engaged with the next generation version of gate-all-around for both ALD and epi. In this next generation gate-all-around, we are working on advanced applications, such as for widening adoption of Backside Power Distribution on metal ALD and on selective ALD. As Paul mentioned, memory was the largest segment of our bookings in Q2. All customers are ramping up manufacturing of high-bandwidth memory to support the surging demands of complex AI computing. The high-speed DRAM chips used in these HBM stacks require high-k metal g ate ALD technology, in which we have the leading position. In the coming years, the DRAM industry will move to 4F², which we expect to result in additional ALD layers in the gate stack and increasing epi requirement.
While revenue was up in Q2 from a low base, demand in 3D NAND continued to be relatively soft in the second quarter. Following the cuts in investment and production last year, supply-demand conditions are gradually improving. Except for technology investment, spending levels continue to be low. Next, let's talk about China. Our total sales in China were still at a good level, but lower than in Q1. If you look at the different segments within China, the drop in sales was mostly explained by the memory logic foundry segment. For China as a whole, we think that Q1 has been the peak for now. Several customers seem to be moving into a phase of digestion. We expect both bookings and sales in the memory logic foundry segment in China to further decrease in the second half.
The power analog wafer segment, excluding silicon carbide, already started to slow down from the end of last year, not only in China, but also globally. The general trend in this segment continued to be soft in Q2, and is likely to remain so for the rest of the year. Despite the weaker market conditions, we actually booked very decent orders in this segment in Q2, thanks to multiple tool orders from new silicon power customers in China. Finally, silicon carbide epi. Condition in this market continued to weaken in the second quarter, and we now expect the silicon carbide epi market to be down for the full year. While we've reduced our expectation for 2024, we believe that our silicon carbide sales are still on track for double-digit growth this year as we continue customer pull. In addition, long-term prospects for this business remain solid.
We believe the slowdown in the EV market is only temporary and will pick up again as EV penetration rates are expected to increase in the coming years. We continue to have good traction in R&D engagement and new customer evaluations. We reported on another new tool selection in the Q2 press release for a major 200 mm silicon carbide fab project in China. Moving on to the outlook. As reported in our Q2 earnings press release, we expect sales in the third quarter to further increase to a range of EUR 740 million-EUR 780 million. We further expect sales in the second half to increase by approximately 15% compared to the first half. This means that we are on track for another year of healthy growth.
We expect a strong increase in second-half sales from the leading-edge logic foundry segment, predominantly driven by higher gate-all-around related sales. We also project memory sales to increase strongly, both in the second half and for the full year. As mentioned, China sales are expected to be somewhat lower. Looking at the mid-term, we are confident about the targets, the targets we provided in our investor day last year for both 2025 and 2027. While it's too early to provide more specific guidance, we expect 2025 to be a good year for ASM. The growth in wafer fab equipment, equipment in 2025 is expected to be stronger than the slight increase that's forecast for 2024, driven by a further recovery in the memory market and increased, and increased investments in 2 nm. This concludes our prepared remarks. Let's now move to Q&A.
We'd like to ask you to please limit your questions to not more than two at a time, so that everyone will have a chance to ask a question.
Okay, operator, we are ready for the first question, please.
Thank you. This is the Chorus Call conference operator. We will now begin the question and answer session. Anyone who wishes to ask a question may press star and one on their touchtone telephone. To remove yourself from the question queue, please press star and two. Please pick up the receiver when asking questions. Anyone who has a question may press star and one at this time. We will pause for a moment as participants are joining the queue. The first question is from Sandeep Deshpande with JP Morgan. Please go ahead.
Yeah, hi. Thank you for my letting me on with the question. Hichem, my question is, you're seeing strength in the memory market in terms of orders at this point. Is this strength, based on what you're hearing from your customers, likely to continue into the future quarters of the year? Or is this going to get replaced by GAA, which will be an important driver for you into 2025? And then I have a quick follow-up. You have this holding in ASM Pacific. I mean, can you clarify, I mean, having now become CEO, where this holding stands, in your view, is this a core holding or is this a financial holding? Thank you.
So to answer your first question, Sandeep, thank you very much for your question, and it's really to have met you a couple of months ago. As for, we see that the DRAM is really driven right now by high bandwidth. And the high bandwidth DRAM actually goes hand in hand with the... So we don't see a letup in the DRAM orders in the future. But at the same time, we can expect to see some lumpiness from time to time. But I think the DRAM, especially the high bandwidth, is going to grow in the future as the gate-all-around technology node is going.
As far as your question about ASM PT, I really have no comment right now. I'm just started to become a CEO for ASM for a couple of months right now. I didn't really have the time to review it any further. It has been a financial holding for us right now, and I don't see any change for this point in time.
Thank you.
The next question is from Andrew Gardiner with Citi. Please go ahead.
Good afternoon. Thank you for taking the question. I just had one sort of high-level question to some of the points you were just making at the end there, Hichem, regarding the outlook. If I look back over the recent quarters, your quarterly revenue guidance has proven pretty conservative. You've regularly been generating quarterly revenue towards the high end, if not above, the guidance ranges that you've provided quarter after quarter. Again, with yesterday's release, you've sort of edged up the 2024 revenue targets again, and now you're, you know, with that, you're within touching distance of a few percentage of the lower end of your 2025 targeted range.
And then if I come at it the other way, sort of a top-down basis, your WFE assumptions are looking rather conservative at this point, relative to, I think, anything anyone in the financial community looks at, or indeed, the industry analyst community looks at. So, you know, I hear your point that you're confident there, but, isn't it gonna be time soon to raise that 2025 target? You know, isn't it simply looking too conservative, given the way the industry's going and the market share gains that you guys in particular are making?
I think for this question, I'm gonna let Paul really give you the answer.
Yeah. Andrew, thanks, thanks for the question. Yeah, as Hichem already said, prospects for next year, we believe look, look good. Of course, with the trends that we see happening in, in Gate- All- Around, which of course will ramp up in the course of next year with three customers, which compared to the past, typically advanced node was with one customer, so that in itself already is, is great. Of course, the benefits, we see continued, strong demand in, in memory, particular DRAM, with some lumpiness from quarter- to- quarter. As Hichem already indicated, that Q2 was very high, but also pretty lumpy. So although we expect continued, good demand, it might be somewhat lower than Q2, but still good.
So, also in other markets, power wafer and analog is lower than last year, but we saw good demand and hopefully might, let's say, pick up somewhere in the second half. That's still too early to tell, to be honest. At the same time, we see a drop in China, which is still a little bit hard to tell how much that will be, but there will be a level of normalization and digestion. We see that already in mature logic foundry in our order book. So why do I say all this? There's still quite some variables in play, and we believe it's too early to actually touch our guidance. I think we try to be realistic. You call it conservative. We believe we try to be realistic.
You're right, that we many times at the high end of our, of our guidance, and sometimes marginally exceed it. But for now, we believe with everything that we know, with all the moving parts, that EUR 3 billion-EUR 3.6 billion is still the right guidance for next year. And maybe, just I'm 100% sure you know, we have increased it, of course, because of LPE, because our initial guidance that we disclosed in 2021 was EUR 2.8-EUR 3.4. So we increased it to EUR 3.0-EUR 3.6, and again, given all the moving parts, we believe for now that is still the right guidance for 2025.
Understood. If I could follow up quickly. I mean, you mentioned some of the moving pieces there. In terms of the potential headwind, you know, would you characterize China as the biggest risk there, sort of, where you may not have the visibility you have elsewhere with, you know, sort of advanced memory and advanced logic?
Definitely China will be one of them. Given, as you know, what we have said, exceptional levels of investments, we believe in 2023, and even in the first half of 2024, because already in the second half of 2024, we see a decline, and that might further continue into 2025. So yes, China is for sure a headwind. At the same time, as you know, our strategy is built around growth and leading-edge logic foundry, gate-all-around, and high performance and memory, which, yeah, shows very good and healthy drivers for continued growth. So they will more than compensate, of course, a drop in, a drop in China. Silicon carbide is still somewhat uncertain.
As Hichem said, we expect actually the market to be down this year, although we still expect double-digit growth. But the question, of course, is what will happen next year with the EV markets? It might come back. It might come back towards the latter end. It's still a little bit early to tell. That could be another, maybe small headwind compared to this year, but I don't know yet. Most likely case there will be growth, but again, too early to tell. But for China, based on everything we see and hear today, yes, China is for sure, almost sure, I should say, down compared to this year. Yeah.
Thanks very much, Paul.
The next question is from Janardan Menon with Jefferies. Please go ahead.
Yeah, hi, hi. Paul, just to follow on that question and the answer you gave. I mean, your initial guidance for 2025 was in 2021, and you added the LP numbers to that. But in 2021, China was extremely low. And I would assume that even with the decline that you're assuming in 2025, or with a moderate decline, it would still be much higher than the assumption you made for China in 2021.
So I'm just wondering, you know, I'm just going back to the same question, which is given, you know, the whole AI wave with HBM coming up so strongly, which is probably unexpected, even when you gave the 2023 Capital Markets Day, should there not be some upside, on the 2025 guidance, if things stay at this level into next year?
Thanks for the question, Janardan. Of course, we've given a pretty wide range for 2025, EUR 3 billion-EUR 3.6 billion, and of course, pluses and minuses, we hope that they can be incorporated in this range. So based on everything that you say, you might maybe expect us more to be towards the high end of the range compared to the low end of the range. That could be. So there, of course, is still a difference, because, again, it's EUR 600 million difference. That's not a small number for us. So, if, if, let's say, the, the...
Which we expect gate-all-around build-up of capacity will continue as we see it today, if memory continues to grow as we think it will be today, and even with the decline of a larger China, indeed, I think you're right. I think that's a fair assumption. Yeah, we still believe that EUR 3 billion-EUR 3.6 billion is, for now, the right range. But yeah, of course, the more tailwind we will have in terms of markets and growth that we might not have fully taken into account in 2021, yeah, the more likely it might be that we go, let's say, more towards the higher end of the range. But even that, I don't want to guide. This is no guidance. Our official guidance is EUR 3 billion-EUR 3.6 billion. A lot of things can still happen.
It's still early days, but we're confident about our position in the market, and as Hichem already indicated, confident about next year based on everything we see here today.
Understood. Thanks for that. And my second question is, just on the gate-all-around, 2 nm gate-all-around orders, you've said that you got a higher order from that in Q2. But in previous quarters, you said that you expected pilot orders in Q1 and Q2, and then the high volume orders from Q3, Q4. I just want to clarify, did you get any high volume orders in Q2, or was it still pilot orders? And you also said that, you know, some of the orders that came were more on the epitaxy side. So is it that we still expect quite a bit of the ALD high volume orders through Q3 and Q4, especially from the foundry side? Thanks.
Yeah, it's a little bit hard to tell, Janardan, simply because we... orders don't come in earmarked pilot or HVM. I mean, maybe to put things in context, we have seen already. We had last year, Q3, you might recall, we had higher orders than anticipated, and it was to a reasonable extent driven by gate-all-around, which were partially pulled in from Q4. Already four quarters in a row now, solid orders for gate-all-around, both ALD and epi. In this quarter, we talked about it. We believe there will also be some HVM orders.
And going forward, it will be more and more for HVM, because if you look at the four quarters of orders we've received, that's if you accumulate that, we don't disclose the number, but that's a decent number. So we think that most, if not all, of pilots has been received. But we never, again, we never know that precise, but that would be, I think, a fair assumption.
Yeah, let me add something here and provide you with a broader perspective. I think for as opposed to FinFET, Gate All Around, you know, has three customers, and in FinFET was really for most of the time, it was just one customer. So it's, you know, the order has taken a long time because it really depends on the customer roadmap and where they're gonna be in HVM. And we have seen the past few quarters, there's you know, plenty of N2 or Gate All Around orders. So we see, I mean, for this quarter, Q2, there's actually a mix like what Paul has mentioned, mix of pilots and HVM order for N2. So we feel confident that this is really what's happening.
Most of the orders are actually customers are getting ready for HVM in 2025, and we expect right now that 2025 is really gonna be the start of the gate-all-around HVM technology node.
Understood. Thank you very much.
The next question is from Alexander Duval with Goldman Sachs. Please go ahead.
Yes, many thanks for the question. I saw in your presentation that you'd upgraded your expectations for the number of logic devices to be driven by AI. I wondered if you could talk a little more about what specifically drove that, and longer term, to what extent you see that impacting other areas other than logic, that could benefit ASM?
So I think for AI, AI is driving two type of chips. It's driving both logic and also driving the memory. So for logic, you need to have really high performance, mainly for high-performance chips, which is a 2 nm technology node. But also with a 2 nm technology node, you have also higher efficiency and lower power consumption. So, 30% actually lower power consumption than FinFET. So that's a huge and as you know, you know, semiconductors are eating more and more energy. So that's on the logic side. But also for each AI chips, you have many, many HBM or high bandwidth memory DRAM. And, those high bandwidth memory DRAM, they are high performance. And when you say high performance, it means that they have more and more logic cores.
So the logic part of it, really, high performance needs lots of ALD and epi layers. So in DRAM, high-bandwidth memory, DRAM, we have lots of ALD, and you have lots of epi. And for the gate-all-around, before—because of just this, the gate-all-around structure by itself, that needs more ALD layer for dipole layers, for phase shifting, and so on and so forth. I mean, even in the back end of line, you have actually middle end of line, you have metal ALD and so on and so forth. So we see...
Also at the same time, for the logic part, for the gate-all-around, epi intensity is higher because right now, what happens is that actually the channel width is defined by epi for the first time. Before in FinFET, the channel is being defined by etch and litho. Now, actually, for gate-all-around, epi defines the channel, and you have like 5-6 layer of silicon, silicon germanium superlattice. So we see, you know, for both memory and for both logic, which constitute an AI chip, both ALD and epi intensity increase, and that's where we play right now.
Thank you very much.
The next question is from François Bouvignies with UBS. Please go ahead.
Thank you, very much. My first question would be on this, you know, SEMICON West. I mean, you had a few slides and discussion, and you described the first generation and second generation of GAA, with epitaxy layers getting another step up, in the second generation. And I was wondering, about your market share in epitaxy, specifically as you move forward. Because when we look at your forecast, I mean, and your epitaxy roadmap, it looks like, like you say, you're gonna move your market share from 15% to 30% in a matter of two years, which is, I think, unprecedented, to gain so much market share in a, in such short period of time for the industry.
So given that you, you know, compete with one very important players, AMAT, I was wondering how you think about the market share, your market share in epitaxy, specifically as you go to Gen 2, especially in a context where AMAT is, is I guess looking at this carefully. And at the SEMICON West, they talk a lot about these integrated systems where, you know, they can bring, you know, different proposition. And I was wondering what you, you think about this value proposition versus yours, I guess, because you don't have that. I'm sorry, it's a very long question. Thank you.
Yeah. Thank you very much for the question, and it's nice that you saw my presentation in SEMICON. I think that's for epi. As I mentioned right now, nanometer is the first technology node for gate around. The second generation is gonna be 1.6 nm, 1.4 nm, and the third generation would be 1.0 nm, which most likely gonna be CFET. So as you go to next generation, gate around, two things will happen. Shrink, shrink is gonna happen.
So, from that point of view, you know, you need more epi is happening there because, especially in the contact area, you know, when you have a very small, a very small width, having a better contact would be the best. And epi, you know, has a very intimate contact, you know, instead of having contact with CVD layer and so on, actually, epi will have more and more intensity. Just, I think from the contact point of view, we see that. And also, you know, you're gonna have, for some customer, you're gonna have, like, more superlattice layer. Instead of having four layers, you might go to six and so on.
And you know, when you end up with the CFET, actually, that's CFET is actually double gate-all-around. It's like you have NMOS on top of PMOS, so actually you increase the number of epi layer from that point of view. I think that for epi, you know, for epi, what we have done in epi the past few years is really bringing innovation to this technology, which haven't seen innovation for many, many years. And I think with the innovation that we have brought in the reactive technology, customers have seen the benefit of it, and that's how, you know, we've been able to penetrate and provide value to our customer.
As I mentioned before, you know, I mean, we as ASM, you know, we're built on a premise of breakthrough technology, innovation, and relentless research, and really try to provide something that's unique value to the customer. So we see that we have good technology being adapted by all three logic foundry players right now, and we're continuing innovation. I mean, we're really not sleeping on our laurels, and we have very much respect for competition, but we need to be close to our customer. We need to continue innovation, and that's what we see, and we think that our position is good, and as long as we're providing value to our customer.
Great. Thank you. And if I may, ask the second one, it will be shorter. About China, I mean, do you sell, is, memory a big driver of China demand as well? I was wondering if you were selling a lot to China, memory.
Yeah, maybe in China, we serve multiple segments. As you know, memory is one of them, but actually not a large one, to be honest. So we've said that multiple times. Of course, we see some growth in China, but also in memory. But our key play in the last quarter has been where we've seen what we then call exceptional growth, was in mature logic foundry, which is now actually gradually going to reduce and normalize some more, we believe. Then of course, last year we've seen very good growth in the power, wafer, analog, but this year will be less. But actually, the last two years we've seen very strong growth in power, wafer, analog.
Of course, we have consolidated our silicon carbide business, which we did not have, of course, before 2023, which was also mostly China. So adding that together, you get what we have today. But the biggest play that we used to have in the last, let's say, five, six quarters was in mature logic foundry. Going forward, I think the focus will be maybe even more on the power, wafer, analog and on silicon carbide, and maybe to a lesser extent on mature logic foundry, given again, the exceptional levels of investments we've seen, and of course, also some memory there, but they were relatively small.
Thank you, François.
The next question is from Stephane Houri with Oddo BHF. Please go ahead.
Yes, hello, thank you for taking the question. I have two, in fact. The first one is about the gross margin, because there is indeed some volatility around the gross margin, and we were expecting a kind of normalization, but maybe more happening in the second half, with China going down. Can you maybe explain what changed in the mix during this quarter? And how low can the gross margin go if China goes down in the mix? And I have a follow-up.
Yeah, good, good question. Thanks. Yeah, I mean, as we've said, and you know, of course, multiple times, is that, that, let's say more China, let's say, on average, means higher margin, less China typically means a little bit lower margin, simply because pricing is somewhat different in China compared to other markets. And then I think the most important one is lower volume per customer, and the second most important one is that, that, yeah, more mature products with sometimes higher throughput compared to, let's say, the initial launch of a product, so you can command, yeah, more value from your customer. It's hard to say. It depends, of course, ultimately on the mix in China, it depends on the mix outside of China.
We have quite a few products, each with their own margin profile, different layers with their margin profile, different customers. So yeah, the best guidance, again, I know that's not a very satisfying answer, but is this 46%-50%? I mean, if China would normalize, yeah, I think it will be very hard for us to get north of 50%, to be very honest. I've said it more times, it's something, yeah, you guys might not want to hear, but I think that is, at least for... as we see it today, the most likely scenario, that with a more normalized China, it will be hard to get it structural above 50%. And I don't think that we're there to achieve that.
Could it be, sometimes a quarter, if the mix really is really very positive, that all the points fall to the right side? Yeah, it could be, but definitely not structural.
Okay, thank you for that. And then the follow-up is about the recent, let's say, noise about restrictions, further U.S. restrictions, in China. How do you react to that? What is your position? And how could it affect your business? Thank you.
Yeah, I wish we knew. There is indeed a lot, a lot of discussion now on additional controls, so very likely they will come. We don't know precisely when, but somewhere maybe in the coming month or two months or three months, we'll see. Difficult, difficult to speculate. I mean, you've seen in the newspapers that there is discussions between various governments going on, and that all kinds of measures are being mentioned, hopefully not implemented. One is the Foreign Direct Product Rule. That would be the most severe if that were to happen. Another thing that could happen is that more fabs will be classified as advanced fabs. Another thing that could happen is that certain applications might be stopped. I mean, yeah, we've seen these things in the past as well, but yeah, it's really...
We really don't know. If we would know, we'd tell you, but we really don't know. And maybe to give you some comfort, at the end of the day, our growth strategy is not based on China. Of course, China helps, and China has been a good help last year. But our growth strategy is based on leading-edge logic foundry, high-end memory. That's where we see the inflections happening in the coming years. That's where we see growth, and that's what we need to deliver on our guidance. And in China, of course, where we try to focus on in a selective manner, is more and more silicon carbide, power, wafer, analog, which is then more and more, and to a limited extent, where possible, of course, mature logic foundry.
Because we do believe that if new export controls were to kick in, then it is, yeah, likely maybe to hit mature logic foundry. But again, we don't know. It's too early to speculate. We don't think that power, wafer, analog is targeted or silicon carbide, but even on that matter, we just don't know. I mean, so I don't want to speculate, but yeah, we're on top of this, and ultimately, we will be on the receiving end, unfortunately, and we'll see what it means. But again, it's for us not a super major issue, if anything will come, because our growth strategy is focused on the areas I just mentioned.
Thank you, Stéphane.
Okay, thank you.
The next question is from Ruben Devos with Kepler Cheuvreux. Please go ahead.
Yes, good afternoon. Thanks for letting me on. I just have a first question, regarding the TAM, basically for single wafer ALD and silicon epitaxy. I understood that obviously for next year you keep your targets, but for the TAM, how, you know, how should we be looking at that? I mean, you've had more feedback from the market since the Capital Markets Day in September last year, so nearly a year ago. Of course, AI applications, it's been a relatively young and emerging market, but are there any comments you could make on, yeah, on the TAM for single wafer ALD and silicon epitaxy, and how that may look somewhat differently now compared to September last year?
Yes, thanks for the question, Ruben. I mean, it's more or less the same answer that I gave before on the overall guidance we gave for next year. Of course, there is this, there's positive and negatives. There is a lot of moving parts, based on also our Capital Markets Day in 2023, so last year, in which we also have taken into account quite a lot of the impact of AI already. We have basically reconfirmed these TAMs, and also today, we don't see a reason to change that. So I believe by heart, ALD was EUR 3.1 billion-EUR 3.7 billion, and epi in 2025, I think EUR 1.9 billion-EUR 2.2 billion or EUR 2.3 billion.
Given where we were last year, where we reconfirmed that we were actually on track, and that these TAMs seem to be reasonably well projected, we see no reason at this stage to change that.
Okay, fair enough. And then just on the silicon carbide epitaxy, I think the wording is somewhat different from the prior release, but my interpretation is that it reads somewhat more bullish. Could you maybe talk about changes in ordering behavior in the past three months, and how you managed to onboard new customers? And just more generally speaking, there's been, of course, a lot of news around automotive, around the disappointing end demand for EVs, which has even triggered some car makers to prioritize ICE engine car production again. Yeah. What do you make of these developments? And, yeah, how should we think about the end market size? So basically your growth trajectory beyond 2024 for silicon carbide epitaxy.
So, I will answer this question. We remain confident about silicon carbide technology in the future. Yes, we do see a downturn in 2024, because of lower adoption, slowdown in adoption of, EV, especially in, in the States. But, I think medium-term and long-term, EV adoption is, is happening and is gonna continue. So we are very, optimistic, and confident about the growth of silicon carbide. The second thing about silicon carbide, I think what we have done, this year, is actually we have released a new product on 200 mm, which is a cluster single wafer product. We've seen, lots of adoption by our customer, both in China and outside of China.
That's why we have double-digit growth in the silicon carbide market in 2024, even though actually the market is going down. So, we are confident about the market, we are confident about our technology. And, I think in this power electronics market, which ASM has been a player for the past 40 years on silicon power electronics market, we see these ups and downs. It's part of the technology. I mean, and, we have to live with it, but, we need to come stronger, when the upturn comes, and I think, that's what we have done in 200 mm.
Okay, thank you.
Thank you, Ruben.
The next question is from Didier Scemama with Bank of America. Please go ahead.
Thank you so much. Good afternoon, everyone. Couple of questions, and before, very boring questions, for Paul, just wanted to clarify. So you said SG&A this year will be flattish, year-over-year. Is, is that excluding the EUR 8.4 million one-off from Q2? And in R&D, you said it's gonna grow 10%-20%, which is quite wide. So do you expect the, capitalized R&D expenses to pick up dramatically in Q3, Q4? Because you were guiding them to be up in Q2, and they were pretty visible, I think. So is, is that the main culprit? Also, on gross margin, I'm glad to hear that, you know, being above 50% is gonna be tricky without China.
But equally, I mean, do you think like 48% is kind of a floor now for your business, even if China disappears or at least declines materially in the second half? I've got a follow-up for Hichem on the technology roadmap.
Okay, thanks. Thank you, Didier. And may I first add that what we have actually guided from the beginning of the year onwards will be marginal increase up to 5%, and that's still the case. So it's not flattish. I'm not sure where that came from, but must be a misunderstanding. We've continuously said marginal up to 5%, and I think it will be around this 5% for the full year, excluding indeed this one-off tax charge related to accelerated divesting of certain shares. Then R&D, you're right, it's a big range, and you're also right that I already had expected in Q2 a higher increase in amortization. Actually, some of the projects that I expected already to close in May, so I have 1 month of additional amortization in June, did not happen.
They're closed by now, so it will start in July. So in Q3 and Q4, you will see a pretty significant step-up in amortization. And that is one of the reasons, not the only one, but one of the reasons why net R&D will increase between 10%-20%. Maybe next quarter, I'm fine to narrow the range somewhat, but for now, I wanna leave it at that. But there's also, of course, growth in gross R&D, as you've seen, I think this quarter from EUR 104 million to EUR 114 million, given all the opportunities that we see and that we're investing in to support our growth.
But yeah, as I said before, net R&D will increase faster than gross R&D because of this step up in amortization, because some projects have reached HVM, and we're now selling these applications and have to start amortization. Then on the gross margin, as 48% the floor, I wish I could say... Yeah, I mean, I have to go back to 2022, where I think also, not a lot, but some quarters have been below the 48%, so it would be, I think, not right to say that it is a floor. So it could still be below, of course, in specific quarters.
But, yeah, if you mean for a full year, yeah, the safest bet would be around the midpoint, I guess, of the guidance, because I cannot give you any better guidance. If I could, I would. But it's, yeah, it's pretty difficult.
The full year growth margin 48% or you mean like for the second half?
No, no, for just in general. This is not meant for this year. This is just in general.
Ah.
That was your question, I believe, if China normalizes, could it be below 48%? So the question is yes, in certain quarters it could happen because we've seen it in the past as well. And then in general, for a full year guidance, yeah, the best I can tell you is the midpoint of the guidance, but of course it could be high and low. I mean, that's why we have provided the range. And the answer is that I cannot narrow down this range, given the margin profile of our products.
Ah, totally get it. Thank you. One question for Hichem on the technology roadmap. What's your view on 3D DRAM? I mean, your predecessor, and presumably you are quite bullish on 3D DRAM, and we're starting to see now the few R&D papers and, you know, patent applications from many of your DRAM customers around 3D DRAM. So my question is, A, what's the timeframe of adoption? I think initially you were talking about 2027, but it feels like it's a bit further out. And then, B, what's the market for 3D DRAM and that cannibalize HBM? What sort of end markets are we thinking about?
Okay, thanks for your question, Didier. For 3D DRAM, I think if you look at the DRAM technology node right now, as I mentioned earlier, there is a shrinking from the cell structure from 6 F² to 4 F² , and that's gonna happen in the next couple of years. So we see that the 4 F² DRAM node is gonna continue to 2027, 2028 and 2029. And after that, starting in 2030, at the earliest, we see the transition to a 3D DRAM.
You mentioned like 3D DRAM 2027, maybe you heard it, you heard it, but I think for us it was really always that 3D DRAM is 2030, starting in 2030, at the earliest. Customers are actually working on the roadmap right now on 3D DRAM, because it's a newer cell architecture. So we need to start very early to do some R&D, and we engage actually with all players on this, but I think the HVM part will start at the earliest is 2030.
Okay. Thanks very much.
Thank you, Didier. Can we have the next caller, please, operator?
Yeah, the next question is from Timm Schulze-Melander with Redburn Atlantic. Please go ahead.
Great, thanks for taking my questions. Actually, my first one is a follow-on from Didier's question around that roadmap for DRAM, and probably a question for Hichem. Just, if you could, help with some color around the market opportunity, and that kind of served or target available market, how that changes from 6F² to 4F², and then admittedly very early days, but is there any kind of scaling of what that might be, 3D DRAM, and then I have a follow-up?
... I think if you look into when you go from 6 F² to 4 F² , that's mainly shrinking of the cell structure, but also you see some increase in ALD intensity. I think part of it is because the lower capacitance happens and higher leakage, so you need some different ALD layers with low-k value, et cetera, et cetera, and when you go to 3D 3D DRAM, which it's a totally different structure, there are two things happening. When you go to 3D, by definition, you're gonna have a very high aspect ratio gap fill, so that we see lots of ALD layers happening for gap fill. The reason for it is, you know, ALD has very conformal film.
So, that those high aspect ratio feature would tend to favor ALD. Same thing that really what happened in 3D NAND. I mean, you see many ALD gap fill opportunity at 3D NAND. The same thing is gonna happen for 3D DRAM. One big difference in 3D DRAM we see is actually in the epi part of 3D DRAM. I think the structure is actually gonna go to silicon, silicon germanium. So, if you go and you look at the 3D NAND, you know, you have ONO structure, which is oxide nitride, which mainly deposited by PECVD.
In 3D DRAM, you're gonna have lots of channel films, silicon, silicon germanium, so 100-200 superlattice structure, and those will be actually epi, epitaxy. So, the growth in epitaxy in 3D DRAM will be very significant from that point of view.
Great, that's very helpful. Thank you. And maybe just a longer-term question for Paul. When you referenced the profitability of China, one of the things you talked about is how, you know, tools have developed since their initial insertion, and that that allows you a better profitability point. Just as we think about your business, very big picture, China going down, gate-all-around, high-bandwidth memory going up, and that, that mix, you, you're saying with, with China coming down, makes that 50% gross margin level a tough one to breach. If you roll forward sort of five or eight years, does that picture change maybe as the product set matures and pricing improves on, on your gate-all-around and high-bandwidth memory chip set? Thank you.
Yes, it's a good question, and of course, we will try, of course, what we need to do to improve margins, of course, going forward. It's too early to tell where we will be at that moment in time. It will also depend on the mix. So if the products, of course, with a better mix, will further increase relative to today. That for sure will also be a help. So yeah, I'm not saying no, I'm not saying yes.
Mm-hmm.
It's too early to tell. As I said before, I'm looking at a wider range of margins than you guys look at, from higher to lower than the guidance we've given. But we are focused on margin because margin features R&D. So we clearly want to come with differentiated propositions to have good margins into the market, and have the margin that we can continue to differentiate. So that's the philosophy that we try to pursue. But again, too early to tell how it will develop in five, six, seven, eight years from now.
Okay, and maybe just one quick one. Was the book-to-bill above one in Logic Foundry across your business in 2Q?
In total, it was, but I don't know it by heart, to be honest. No, I would really have to look at the numbers. I don't know by heart. I'm sorry, Timm.
All right. Thanks for your help. Thank you.
I'm afraid we are running out of time. We have time for one final question. There are still more callers in the queue, so if you didn't have the chance to ask your question, please contact us after the call. Operator, can we go to the final question? Operator?
Yeah, the final question is from Robert Sanders with Deutsche Bank. Please go ahead.
Yeah, thanks for squeezing me in. I just on, on the HBM orders, are those for delivery in 2025 or 2026? I guess the reason I'm asking is it does seem like most of the greenfield HBM fab won't be ready for equipment until 2026. So I'm just wondering if people are ordering earlier for 2026 greenfield, or are these just conversions that are happening? And related to that point, I'm just sort of interested to understand a bit more about the high-k metal gate transition.
It seems to be quite correlated with DDR5, but I'm just trying to understand if it's DDR5 correlated, you're already kind of 40% of the way through the transition, or are you in fact much less of the way through the transition because maybe high-k metal gate is not used in mobile or PC? Thanks a lot.
Yeah, on the. Let me take, because we did not fully, the line was breaking up a little bit. But I think the first question was on, HBM orders, I think, Rob, for delivery in 2024, 2025 orders today. I think a reasonable chunk, delivery is likely to be in 2024, so next year, 2025, in next year, so 2025. There might be some of that for sure also in this year, but, I think a reasonable chunk, given the lumpiness of what we received in, in Q2, is likely to be for delivery, also, to a reasonable extent in 2025, but also some in 2024, I'm pretty sure. I don't know the delivery dates by heart, of course, but that's the most logical assumption.
Then the last part of your question, we couldn't hear because of, I know there was some disturbance in the line. So can you, can you repeat that?
Yeah, it was just about high-k metal gate penetration. Can you just run me through how far along high-k metal gate is in terms of penetrating the DRAM footprint? Are we in the early innings? Are we halfway through, like DDR5 is? Just checking if you can hear that as well.
So I think I will answer this question. For high-k metal gate for in memory, I think you look into any high-performance memory right now, DDR5 is one of them uses high-k metal gate . So it's not in the first inning, maybe like halfway. I think the adoption is there. It's getting more and more. We see it with every technology node. I think we're gonna see more and more adoption of high-k metal gate ALD. And also HBM memory is actually taking that involved. So any high-performance DRAM being, you know, HBM or otherwise, actually is adopting high-k metal gate .
Got it. Maybe I could just squeeze in one more question, which is just about the Entity List. You know, the U.S., your competitors seem to be alleging that Dutch and Japanese companies are somehow favored in China. Are you conforming to the U.S. Entity List, for example, entities like YMTC? Are you able to ship to them, or do you have your own Entity List that's driven by your own government? I'm just trying to understand if there's any credibility to this idea of, you know, Dutch and Japanese companies having an advantage in China. Thanks.
Yeah. We are also the same Entity List as our, let's say, American peers, so there's no difference. Actually, the Dutch and the Japanese don't have the concept of Entity Lists, but for us, it doesn't really matter. The U.S. regulations has extraterritorial powers, and, yeah, we also cannot, let's say, as you say differently, we have to adhere to the same Entity List as our U.S. peers, so no difference from that point of view.
Thank you, Rob.
Got it. Thanks.
There are no more questions registered at this time. I turn the conference back to the CEO for any closing remarks.
I'm good.
There are no more questions registered at this time. The conference is back to the CEO for any closing remarks. Ladies and gentlemen, thank you for joining. The conference is now over. You may disconnect your telephones.