Good afternoon, this is the Chorus Call conference operator. Welcome, and thank you for joining the ASM International fourth quarter 2023 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, Investor Relations. Please go ahead, sir.
Thank you. Welcome, everyone. I'm joined here today by our CEO, Benjamin Loh, and our CFO, Paul Verhagen. ASM issued its fourth quarter 2023 results yesterday evening at 6:00 P.M. Central European Time. For those of you who have not yet seen the press release, it is available on our website, asm.com, together with our latest investor presentation. As always, we remind you that this conference call may contain information relating to ASM's 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, reports, and financial statements, which are available on our website. Please note that the profitability measures mentioned in this call will be primarily based on normalized non-IFRS figures. For reported IFRS results, please refer to the quarterly results press release.
With that, I'll turn the call over to Benjamin Loh .
Thank you, Victor, and thanks to everyone for attending our fourth quarter 2023 results conference call. I'll start with a few of the highlights. ASM had another successful year. Our sales increased by 13% at constant currencies, despite the softer wafer fab equipment market conditions. This marks our seventh consecutive year of double-digit growth. In the second half, we received the first meaningful orders for Gate-All-Around technology, which will be a significant driver for ASM. We remain on track towards our strategic target and continue to invest in innovation and expansion. I want to thank our ASM people. Their hard work and dedication contributed to our company's robust performance in 2023. The agenda for today's call is as usual. Paul will first review our financial results. I will then continue with a discussion of the market trends and outlook, followed by the Q&A session.
With that, handing over to you, Paul.
Thank you, Benjamin. I will start with, with the fourth quarter results. Our revenue amounted to EUR 633 million in Q4, down 7% year-on-year at constant currencies. Revenue is above the midpoint of our guidance of EUR 600 million to EUR 640 million. By customer segments, equipment sales were led by foundry, followed by power analog wafer, and then memory. Combined logic foundry decreased both year-on-year and compared to the third quarter. Softness in the leading-edge part of this segment was partially offset by continued strength in mature logic foundry, mainly in China. Despite the weakening demand that reflected our Q3 results, sales in the power analog wafer segment increased to a quarterly record high.
As we shift the tools related to the strong orders in this segment received earlier in 2023. Memory sales increased strongly in the fourth quarter compared to the very low level in the third quarter, mostly driven by DRAM. Gross margin of 47.9% in the fourth quarter of 2023 was up by one percentage point compared to the same period last year, but down one percentage point compared to the third quarter of 2023. The mix in the fourth quarter remained favorable, including a continued strong contribution from China sales with above-average profitability. The Q4 margin was also impacted by approximately one percentage point due to a one-off restructuring cost related to the further optimization of our global manufacturing footprint.
Net R&D expenses increased from EUR 70 million in Q3 to EUR 79 million, reflecting a continued increase in R&D projects and also the impact from a EUR 2.5 million euro impairment of capitalized development costs. SG&A increased from EUR 77 million in Q3 to EUR 83 million, including higher variable expenses and specific investments to further strengthen the organization, such as in IT. Our new orders in the fourth quarter amounted to EUR 678 million, up from the EUR 627 million in the third quarter and down 14% compared to the fourth quarter of 2022. Logic foundry accounted for the largest part of our Q4 orders and increased both sequentially and compared to the fourth quarter of 2022.
After the first more meaningful contribution in the third quarter, GAA Gate-All-Around 2-nanometer orders further increased in the fourth quarter, while demand for mature logic foundry in China remained relatively solid. Compared to the relatively strong bookings in Q3, memory dropped in the fourth quarter. Power analog wafer also dropped somewhat compared to the third quarter, and especially compared to the record high level in the fourth quarter of 2022. Now, let's have a look at the full year results. At EUR 2.6 billion, our sales in 2023 increased 13% at constant currencies, outperforming the WFE market, which dropped slightly in 2023.... In terms of customer segments, our total logic foundry sales increased slightly and continued to account for more than half of our equipment sales.
The leading-edge part of the logic foundry sales meaningfully decreased, impacted by end market weakness and the delays in a number of new fabs that we reported in earlier quarters last year. This drop in the leading-edge segment was offset by growth in mature logic foundry, particularly in China. In memory, our sales dropped around 40% against a relatively strong level in 2022. The drop in DRAM sales was relatively moderate, supported by demand for high-K ALD solutions for high-performance DRAM and AI applications. 3D NAND sales dropped sharply, reflecting the market-wide sharp investment cuts in this segment. In total, the memory contribution dropped from 19% of our equipment sales in 2022, to 11% in 2023.
We booked strong growth in the combined power and logic wafer segment in 2023, and compared to a relatively lower level in 2022, sales in this segment almost doubled, including the consolidation of a silicon carbide Epi business, and also in large part fueled by strong growth in China. At constant currencies, our equipment sales increased by 12%. Our ALD sales decreased moderately, impacted by weakness in the leading-edge part of the WFE market, but continues to account for more than half of our equipment sales. Apart from silicon carbide Epi, the vertical furnaces were our strongest growing product line in 2023, in part fueled by solid demand for our new Sonora tool, followed by silicon Epi.
Spares & Services had another strong year, with sales 19% higher at constant currencies and continued momentum, momentum in our outcome-based service offering. Our silicon Epi sales increased strongly and comfortably met the target of more than EUR 130 million in 2023. LPE has been successfully integrated, and we have been stepping up capacity and capabilities in this business to prepare for continued growth in coming years. Gross margin for the year increased from 47.5% to 49.3% in 2023, mainly reflecting mix, including substantially higher sales from China. Despite inflationary prices on our cost of goods sold in 2023, we were able to limit this through commercial price negotiations and increased cost reduction value engineering initiatives.
We remain committed to gross margins of 46%-50% in coming years, which we believe is a healthy and sustainable rate. With a normalization in the contribution from China sales in the full year of 2024, we expect a decrease in gross margin compared to 2023. Net R&D increased by 29%, reflecting a record high number of customer R&D engagements and further expansion of our R&D headcount, up 11% in 2023. For 2024, we project a further increase in R&D in support of the rising number of engagements for new applications, such as for the second-generation Gate-All-Around and new opportunities in memory.
Net R&D expenses will remain at a high single-digit to low double-digit % of sales in the period up until 2027, and in the first part of this period, more towards the higher end of the range. SG&A expenses increased 10% in 2023, which was broadly in line with the indication that we provided at the start of the year. The rate of the increase in SG&A slowed down compared to 2022, as a large part of the investment to strengthen our organization were completed by the end of 2022, and also due to continued cost focus amidst softer market conditions. We expect a moderate increase in SG&A in 2024. Our target for the period up until 2027 is for SG&A to be a high single-digit % of sales.
In 2023, SG&A was still at 11.5% of sales. We expect relative normalized SG&A to decrease once revenue growth accelerates again. The operating margin in 2023 was stable at 26.6%. Results from investments, reflecting a 25% stake in ASMPT, dropped from EUR 78 million to EUR 21 million in 2023, reflecting the downturn in the back-end market. Now turning to the balance sheet. ASM's financial position remained very solid, with cash of EUR 637 million at the end of 2023 and no debt. We generated free cash flow of EUR 447 million in 2023, a record high and up 17% from 2022. Next, to continued solid profitability, free cash flow was driven by improved working capital.
Working capital decreased as we managed good collection from our customers. Inventory decreased slightly in 2023. With the normalization of supply chain conditions, work in progress already decreased in 2023, but we still kept higher buffer inventories to ensure flexibility in our operations. CapEx increased to EUR 154 million in 2023, up from approximately EUR 100 million in 2022. As a reminder, for the period 2024 till 2027, the target for CapEx is EUR 100 million-EUR 180 million annually.
The majority of this CapEx will be geared to the expansion of R&D infrastructure and activities, including the construction of new facilities we announced last year in Korea and Arizona. In terms of shareholder remuneration, we spent EUR 223 million in cash on dividends and share buybacks in 2023. And yesterday evening, we announced a new share buyback program for an amount of EUR 150 million, as well as a proposed dividend of EUR 2.75 per share, up from EUR 2.50 last year. We remain committed to return excess cash to our shareholders. And with that, I have to go back to Benjamin.
Thank you, Paul. Let's now look in more detail at the trends in our markets. We had relatively solid bookings, but overall, market conditions continued to be soft in the fourth quarter. Leading-edge logic foundry spending was still solid at the beginning of 2023, but demand weakened in the course of the year, as we already first reported in April 2023. Demand for advanced node capacity additions continued to be relatively soft in the fourth quarter. Importantly, our customers remain strongly committed to their technology roadmap, in particular, the transition to Gate-All-Around technology, with pilot line activity starting in the first part of this year, 2024. After the first orders in the third quarter, we booked additional and increased GAA or Gate-All-Around pilot line orders in the fourth quarter.
Gate-All-Around is a major inflection for our industry, which will drive further improvement in device performance and power efficiency, which in turn will be key for new end market applications, such as in AI. We expect Gate-All-Around to increase our served available market, or SAM, by $400 million per 100,000 wafer starts compared to the previous technology node, as discussed in our Investor Day. The more complex device architecture of Gate-All-Around will increase ALD requirements and layers. Silicon epitaxy is also an enabling technology for Gate-All-Around to create the multilayer stack of nano sheets.
We believe we successfully defended our leading market share in ALD in the transition to 2 nm, and in addition, based on product tool of record selections with the leading customers, we remain confident that the transition to Gate-All-Around will drive our overall silicon epi market share to the target of more than 30% by 2025. In memory, we had good traction in 2023 with our high-K ALD solutions for high-performance DRAM, which is a key requirement for AI. As Paul mentioned, this could not offset the steep drops in the other parts of the memory market, particularly in 3D NAND. We continue to work with key memory customers on new applications, such as on the expected adoption of epi in DRAM, which, we expect to further increase our share of wallet once the industry migrates to the next memory nodes in the coming years.
The softness in leading-edge logic foundry and in memory market was offset by strong growth in the mature node segments, particularly in China. In power analog wafer, sales were strong throughout the year. As we already discussed with our third quarter results, demand started to slow down towards the end of 2023. Following two years of strong investments in this segment, particularly in the automotive and industrial space, the industry has now started to digest the added capacity. Our revenue was still strong in the fourth quarter, but orders already decreased, and revenue is expected to trend down in the next couple of quarters. Mature node demand in the Chinese market was exceptionally strong in 2023. In power analog wafer, and also in the parts of mature logic and foundry, they were not impacted by export control measures.
We confirm the assessment that we shared in the third quarter earnings call, that the updated export control regulations that were issued by the U.S. government in October 2023, are not expected to have a material additional impact compared to what we previously communicated. While the mature node segments were only a smaller part of our revenue in earlier years, they had a strong contribution due to the substantial size of the investments in this segment in China during 2023. We also benefited from our investment in recent years to expand our presence and customer base in China. Furthermore, our sales in China also increased due to the consolidation of LPE, which has a strong share in this market. In total, we booked strong growth in China in 2023. On the acquisition of LPE, our silicon carbide epitaxy business performed strongly in 2023.
The synergies in leveraging ASM scale and capabilities have already been paying off. We increased development and manufacturing capacity in Italy. We also qualified the latest 200 millimeter silicon carbide epi tools for manufacturing in our Singapore facility, and we substantially stepped up global support. We expanded the silicon carbide epi customer base in 2023 with a leading North American customer and a major European player, as previously reported. Additionally, in the fourth quarter of 2023, two more customers selected our latest 200 millimeter silicon carbide epi tool, with multiple tool orders expected in 2024. We believe our tools offer industry-leading performance and cost of ownership, particularly for 200 millimeter silicon carbide epi processing....
We expect to further expand our position as the silicon carbide industry transitions from 150 millimeters to 200 millimeter wafer size over the next couple of years. Despite the recent deceleration in the automotive market, long-term prospects remain strong for the silicon carbide market, driven by structural increase in electric vehicle penetration and other applications such as renewable energy. Supported by new customer wins, we expect strong revenue growth for this business in 2024. A highlight during the fourth quarter was a landmark ceremony in Scottsdale, Arizona, where we will construct a new state-of-the-art research and development center. It will consolidate our existing operations in the Phoenix area and double available clean room space. Investment in innovation remains a key priority for our company. In terms of manufacturing, we believe we have the capacity in place to deliver on our 2027 revenue targets.
The completion of the second assembly floor of our facility in Singapore and the additional manufacturing in our expanded facility in Korea will in total increase our capacity by 3.7 times by 2025 compared to 2020. Next, I would like to highlight our progress in ESG. A highlight this year was the verification of our net zero target by the Science Based Target Initiative, or SBTi. We target net zero greenhouse gas emission for all scopes by 2035. One of the shorter term goals in our path towards net zero is to achieve 100% renewable electricity by 2024. We are well on track. In 2023, we increased the use of renewable electricity to 88%, up from 73% in 2022. The longer term prospects continue to be strong for ASM.
Third-party research firms forecast the semiconductor market to grow to more than $1 trillion by the end of the decade. Digital transformation will continue to drive semiconductor usage. Artificial intelligence is expected to be one of the fastest growing end market applications. In 2023, the impact on the semiconductor market started to become more significant, triggered by the huge interest in generative AI. Related chip volumes were still limited, but expected to notably increase in coming years. This will lead to more investments in manufacturing capacity for advanced logic devices such as GPUs and MPUs and high-performance DRAM. And that means for ASM, more ALD and Epi steps. Let's now to have a look at the outlook that we issued with our press release last evening. While the economic outlook continues to be uncertain, the broader semiconductor market is expected to recover to double-digit growth in 2024.
However, the softer WFE market conditions we saw during the second half of 2023 are continuing into the first part of 2024. Our guidance, as published yesterday evening, is for the first quarter 2024 revenue to be in the range of EUR 600 million-EUR 640 million, and with a similar level in the second quarter. For the second half of 2024, we expect revenue to be up compared to the level in the first half of the year, but it is too early to provide more specific guidance for the second half. It is also too early to give an outlook for the full year. Looking at the expectations for WFE demand, memory and leading-edge logic foundry demand is expected to gradually recover in the course of 2024, supported by early investments in gate-all-around.
We expect continued gate all around orders throughout 2024. Power analog wafer is expected to decrease somewhat in the first half. Revenue from the Chinese market is expected to be still relatively high in the first part of the year, but likely to normalize in the rest of the year. For our silicon carbide EPI business, we expect a strong performance in 2024. We expect that ASM will benefit from an expected rebound of the WFE market in 2025, and based on this, we remain confident ASM revenue will increase to the forecasted range for 2025, which is EUR 3.0 billion-EUR 3.6 billion. An important driver supporting this will be the move of gate all around 2-nanometer technology into high volume manufacturing in 2025. And with that, we have finished our introduction. Let's now move on to the Q&A.
We'd like to ask you to please limit your questions to not more than two at a time, so that everyone has the opportunity to ask a question. Okay, operator, we are ready for the first question.
Thank you. This is the 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. The first question is from Tammy Qiu with Berenberg. Please go ahead.
Hi, thank you very much for taking my question, and Paul. So I have a question on Gate-All-Around. You mentioning that Gate-All-Around order should be received throughout 2024. I wonder what kind of pattern should we be seeing? And also, is that actually for volume production in 2025, or that's still going to be for pilot production?
Sure. Tammy, thanks for the question. I think what you are seeing is that, of course, customers have placed, you know, orders for pilot line, but the way that they have done this, it varies from customer to customer. Some of them will, you know, just give you a block or bulk order and, you know, for some time, maybe, they will not have any more additions. Some of them, they place it, I would say, regularly, depending on which process or which, you know, process is defined, and which equipment has been decided to be a process of record. So we will see both.
Now, there's also some possibility that you might see that, you know, there's, there's an increase in terms of, number of tools that they might buy, even for a pilot line to increase capacity because of sampling, because of tweaking the yield, et cetera. So we do expect that, you know, we will be seeing gate all around orders throughout the year. The end of the year is where we potentially could see pull-in for high-volume manufacturing. You know, it is, actually, you know, announced and reported by one of the three players that they are trying to pull in 2 nanometer, high-volume manufacturing, you know, into early 2025. And if that happens, they'll probably have to, order and buy the tools already by the, end of the year.
Okay, thank you. That was clear. And also, from your market share perspective on the three players, on the GAA production line, are you basically getting very similar kind of steps within all three chip makers, or it kind of varies between different players?
They, they do vary, because, you, you know, even though the technology is common, it's Gate-All-Around technology, but there are different ways that they, you know, design it. So you do see, differences between, the three of them. So it's, it's difficult to give a generalized, you know, idea, for example, exactly, you know, what is the increase or, or, or how each one compares to the other, but there are, there are differences between the three of them.
Okay.
Thank you, Tammy.
The next question is from Stéphane Houri with Oddo BHF. Please go ahead.
Yes. Hello, good afternoon. Actually, my first question is about the second half of 2024. There are obviously different scenarios. There are different moving parts, like market recovery in logic foundry or analog expected, gate around starting to rise. What about memory? What about China? And on what do you base your assumption that there will be a rebound in the second half? And if you can help us trying to quantify this kind of rebound, that would be helpful. Thank you.
Sure, Stephane, thanks. I think you mentioned all the, you know, factors that I would say contribute to a lack of visibility or maybe uncertainty, you know, the pace of recovery for memory, for example, how much is the normalization in China, mature logic and foundry, you know, any recovery in the power analog, wafer, et cetera. And I think that's the reason why, you know, we have decided not to quantify the second half versus the first half. However, we are confident that the second half we should see that leading edge, logic and foundry, primarily driven by gate all around, you know, will be stronger.
At the same time, we are already seeing signs, you know, maybe still a little bit slow at the moment, but there will be probably more investments coming in, DRAM, especially for high-performance DRAM, because of the drive by AI. And that, that usually is good for us, especially with the high-K ALD tools. So I think with that, we, we kind of are confident that we should see a second half that is higher than the first half, but until we have more clarity, which probably could come in a couple of months, so it might be that we might give you a better visibility or better guidance when we announce our first quarter results. As of today, we will just leave it as second half will be higher than the first half.
Okay. Okay, thank you very much. And my second question would be about silicon carbide, because there has been a lot of, let's say, discussion in the market that with the slowdown in EV, maybe silicon carbide investment will be more moderate. And you didn't give a guidance for your revenues in silicon carbide for 2024. So can you help us understand how it's going to grow this year? Thank you.
Sure. So I, I think the reason why we, we gave a guidance, you know, first, for EUR 100 million, and then we upgraded it to EUR 130 million last year, was because it was the first year, and we just did it, we just acquired them. So I think we wanted to give more color, to everybody, you know, how big this business is potentially going to be. But I think going forward, as we have always, you know, let's say, said, we, we don't give individual, what we call product or, or segment breakdown, because it's, it's very competitive, and whatever we say can be easily picked up by also our competitors.
Nevertheless, we did, as we have in the prepared remarks, we did much more than EUR 130 million for last year. And this year, we do expect to significantly increase that. You are right, Stefan, that EV is slowing down, and there's been some concern about, okay, you know, for silicon carbide devices, is that going to continue growing? I think we do see, you know, a slowdown in the growth. So it's not a slowdown in the market, a slowdown in the growth. So it's, it's growing slowly, slowly, slower than what it used to be, but I think it is still growing. And what is really propelling that, of course, is you see a lot of silicon carbide manufacturers preparing for potentially the ramp that will come when the EV market comes back.
And at the same time, they are also exploring a lot of other, let's say, you know, markets, become much smaller, but they're still meaningful, such as renewable energy. So what we are seeing is really that, despite, you know, the market being, a little bit, going slower, it is still going to result in a significant increase in revenue for us this year. And when it comes back, I think we will be in a very good position.
Okay, thank you very much, Benjamin.
Thank you, Stéphane.
The next question is from Janardan Menon with Jefferies. Please go ahead.
Hi. Hi, good morning. Or sorry, good afternoon. Thanks for taking the question. I just wanted to get a feel for what you see for this year, since you haven't given full guidance, but I'm trying to get some information on what your thoughts are. For instance, SEMI, the industry organization, is forecasting WFE this year to grow at 6.5%. You know, for many years now, ASM has always comfortably outgrown WFE, as you did once again last year as well. I'm just wondering, is there anything specific that you see in 2024 which would cause ASM's revenues to underperform WFE?
Is it that, you know, you have, specific exposure in China or something like that, where, or, or excess in mature nodes outside China, where, you know, could be a slowdown and that could affect you? Or is it that, you know, at this point, you just don't know, but if, if WFE does grow at 5%, you'll probably grow at least at 5%, if not something, higher than that. And that's my first question. I have a short follow-up.
Sure. Janardan, thanks for the question. I think the explanation is actually easier than that. It's two reasons. One is, I think the data research companies and the independent organizations like SEMI, I think the WFE number or growth that they have is still, you know, fluctuating. You know, you still see a pretty wide spread between zero to, you know, some have gone up to at least 8 or 9%. It's kind of quite a wide spread. So I think there's still a lot of uncertainty as to exactly how much it will grow. This is not uncommon.
It's just that we are at the beginning of the year, so, you know, we need to be more careful about that. Second thing, of course, as we have mentioned, we know that the second half is going to be, or our second half is going to be higher than the first half. But we do not want to quantify this yet, because of, you know, some of the reasons that we have mentioned earlier. And, you know, as I said, in a couple of, maybe months, when we have a better view of exactly how the second half is going to shape up, I think, you know, we might be able to come back with a statement on how we will perform relative to WFE this year.
But if you look at the longer term history of, you know, ASM, we have always, you know, kind of outperformed WFE over the last couple of years, and we do intend to outperform WFE again over the next couple of years. In fact, I think we have great opportunities because, whatever is being driven by Gate-All-Around, is actually, you know, helping us, you know, increase, let's say the part where we play in the share of wallet. So we are actually, you know, confident that longer term, over the next couple of years, we will outperform WFE.
Understood. And just on the second half, is the uncertainty on which way it will go coming more from the memory market? And you have some idea on how the 2 nanometer advanced logic side will go, or is it more sort of an uncertainty on how much weakness there will be in China or something like that? I mean, can you just give us some color on where you see more of the uncertainty?
Sure. I think it's a combination of all of them. So let's take it one at a time. You know, leading edge, logic foundry, specifically, 2 nm Gate-All-Around. I think we are seeing already signs. There have been also official reports published that we are seeing acceleration. So there's a lot of competition between 3 customers of ours, and they're all trying to, you know, get to a market as soon as possible, which is good for us as an equipment supplier, because they are trying to accelerate that. But how fast? I think that's still a, you know, a question mark. Memory recovery, you know, is also to some extent uncertain.
I think, as we said, DRAM, because of the AI push, you know, driving a high performance DRAM, that should be relatively, I would say, will gradually grow. But there's no, there's no view now, especially on 3D NAND. In fact, the latest report that came out from one of our customers is that they are still trying to cut production capacity. So that's another one. We have mentioned that China will be, you know, mature logic and foundry will be strong, continue to be strong in the first part of 2024, and we do expect it to normalize in the second half. But how much of the normalization? I think we still need time to, to look at this.
And then, of course, you know, we have mentioned as well that power analog wafer, you know, the order trend has been going down since, probably the end of last year. Is there going to be a recovery maybe at the end of the year? That's another factor that we want to have a little bit more time, so that we have better visibility before we make any, you know, concrete judgments.
Understood.
Thanks for your question, Janardan. Thank you.
The next question is from François-Xavier Bouvignies, UBS. Please go ahead.
Thank you. My first question is on the memory side. So you disclose that now 11% of your revenues and is down 40%, and I was actually surprised by the drop. I mean, it's actually underperforming even the memory market, I mean, at least -40%. That would suggest that you are much more 3D NAND exposed than I thought you would be, and with if you have -40%. And since the beginning of the year, we saw some light on the memory side, you know, if you look at VAT and ASML on the orders. So I was wondering, you talked about high bandwidth memory, but we don't see any, like, much impact, you know, at least in orders.
You didn't particularly mention it, and your performance like last year, which suggests that you have a low exposure toward that. Could you tell us maybe like, give more color about your content opportunity? I mean, not your content, but your incremental market opportunity for HBM, for example, or DDR5 versus traditional one, and how your market share is evolving there? That would be very helpful, because we don't see much, you know, evidence of any traction on your side. So I was wondering.
Sure, Francois. I actually thank you for your observation that we are actually underperforming the entire memory market. You know-
It was not my goal. It's not my goal, too. I was just,
No, no,
Actually. Yeah, yeah.
I know. First of all, I think, you know, memory, you know, especially on the 3D NAND side, has been very, very weak, and the weakness actually continues. As I said earlier, we are even hearing... Actually, there was a report that one of our customers is even going to cut the production even further. What has really, you know, been, in 2022, driving 3D NAND for us, of course, was ALD gap fill. And that was a fairly big win, which was, you know, a very nice chunk of revenue for us. And that contributed, you know, to our significant memory sales increase in 2022. And of course, in 2023, that almost evaporated because there was just no investment in 3D NAND.
Now, of course, if you look at 2023, and actually it follows also from 2022, our big play in DRAM has always been in high performance DRAM, always DDR5. And of course, when you know, generative AI started kicking off and driving a lot of HBM, that helped us a lot. So we already saw you know, additional orders for some tools that are required to make HBM type of memory I think as far back as the third quarter of last year. And we do expect that this traction is gonna continue because the demand for HBM memory devices is just huge. And I think this is, of course, driven by AI.
And, you know, as we go on, we think that we will probably see more orders for this type of tools, because they are absolutely essential for making, you know, HBM. I hope that, that kind of gives you some color between DRAM and 3D NAND.
That's very clear. Thank you. And maybe on your market share there, I mean, you know, I mean, the memory has been more on the batch side for some time, and you saw a couple of opportunities on the single coming through and—but how is it evolving, you know, with HBM? I mean, can you quantify a bit, you know, the number of incremental layers that you can address maybe, or your market share? Anything you could flag there?
Maybe not in terms of layer, but you are correct, Francois. I mean, you know, memory is a very cost competitive, you know, type of commodity. So of course, you know, batch type of tools do have a certain advantage. But as the technology progress and as they go into a very, you know, advanced, you know, devices, let's say, you know, scaling and so on, more and more you'll find that some of these, applications will have to change to a single wafer ALD. And, you know, the, high-K, ALD tools is one example. So a lot of, a lot of, not a lot, I mean, we are seeing increasing demand for high-K ALD tools because these are required to make high performance DRAM, whether it's DDR5 or whether it's HBM.
So that is where it's helping us in this down market, even for DRAM.
Thank you very much.
Thank you.
The next question is from Didier Scemama , Bank of America. Please go ahead.
Yes, good afternoon. Thank you for taking my question. I'm gonna try again on HBM. Could you just give us a sense? It's probably a stupid question, but I just want to understand. So what we learn from ASML order intake in Q4 is that to get the best deals on HBM dies, you need EUV? Is that not mutually exclusive, but does that also mean that high-K metal gate is a must for HBM? And related to that, how are you positioned in Korea? Have you got one customer or two? And I've got a follow-up for Paul on OpEx. Thank you.
Didier, I think you could, you know, try to correlate that because when you look at DRAM, they, especially the more advanced guys, they started adopting EUV, you know, with one layer, and then they went up to four layers. And it's probably more now, I haven't kept track of that. But a lot of this was, you know, used for making advanced DRAM devices, so high performance DRAM. And at the same time, you know, you've got the launch of DDR5 and also HBM3, and they also need, you know, high-K, you know, ALD tools. Let's say tools. So there is some correlation, but I do not think that it is a one-to-one correlation.
It's just that at the most advanced DRAM devices or what you call high-performance DRAM, it seems that both EUV and High-K ALD are used.
Got it. And on Korea, I mean, how-
Oh, yeah.
How... Yeah.
We have both.
And, and-
Both of them are-
To be perfectly honest, I think we are all a bit surprised that you're not seeing the sort of a pop in orders from the Koreans on High-K, or at least not, if not in Q4, in Q1 or Q2, given the lead time differences with ASML. And clearly, the ASML orders, which is spectacular, I mean, I think that's probably an understatement. So I'm just a bit surprised you are not seeing that, or perhaps you are suggesting you're gonna get that in the second half, and I suspect you will. Just can you help us understand, is that a meaningful part of your second half recovery, if not in revenues, at least in bookings?
Okay. I think the easiest way to understand this, Didier, is our peer in lithography, their lead times are probably anything from 15-18 months, especially-
Yeah
... when you talk about EUV. Our lead times are six months, so our customers-
Yeah
... still have a lot of time. You know, potentially, you know, as I said earlier, we do expect to see continued orders, you know, for our High-K ALD tools, you know, until the end of this year. But we are not going to, at this moment, you know, let's just say tell you when it is coming in, but we do expect that the pace of memory recovery, first of all, will increase, and it should lead to increased orders for us.
Okay. Perfect. Thank you. And just for Paul, quick one. I mean, taking into account all the things you said on the mix, should we assume gross margins to trend down in the second half? And I think you've been sort of fairly explicit on OpEx, but, I'm sorry to be stupid, but can you give us maybe a range of OpEx for the full year you're thinking about? Because obviously we don't know if your revenues are gonna be, you know, down 1%, up 5%, up 10%, down 5%, and so it's... The OpEx number is something you've got control above. It's a fixed cost, so that would be helpful to understand. You don't have visibility on the top line.
Yeah. No, thanks for the question. So maybe first on your gross margin in the second half, I think it's fair to assume that we will see some drop compared to 2023. Simply because we believe that China, which is one of the mix factors, will start to normalize somewhere in the later part of this year. The first part, we still expect a strong contribution from China with, as you know, above average profitability. And towards the latter part of the year, we expect that to normalize. So there will be some impact, very likely on the margin, so somewhat lower.
Of course, all the mix elements can have a plus or a minus, but that's always difficult to tell, given the number of tools that we have with different margin potential. On the OpEx, as I mentioned just now, SG&A, expect a marginal increase compared to this year. Not a lot. Of course, there will be some merit. There's maybe some inflation, but don't expect a major step up there. R&D, we will continue to invest, so that will go up.
As we said, also, our, our, let's say, medium-term guidance is high single digits below double digits, whereby, we will be maybe even above, the range slightly, in the beginning period, given that the revenue is not yet growing, to the extent that we foresee in 2025. So we might go a little bit over that range in, in twenty, in 2024.
All right. Brilliant. Thanks very much.
The next question is from Adithya Metuku with HSBC. Please go ahead.
Yeah. Good afternoon, guys. Thank you for taking my questions. I had two. Firstly, just around 2025 demand, around Gate-All-Around, I just wanted to hear your thoughts on what level of capacity you're expecting to be deployed. Is it around 80K wafer starts per month, which is normally what you see at the leading edge in the first year? Will it be higher, lower? If you could give any color around that, that you factored into your 2025 outlook, that would be very helpful. And secondly, just on the High-K metal gate or High-Bandwidth Memory, which you talked about earlier. You talked about High-K as just one of the growth drivers.
I just wondered if you could give some color on what the other growth drivers are for ALD tools, apart from the High-K step in the transistor part of DRAM. Any color there would also be very helpful. Thank you.
Adithya, thanks a lot. I think the capacity is something that, you know, we do, we do have a, you know, let's say, an estimation in mind as to, for example, each one of the three players, how much they would ramp. But I think, you know, at this moment, we, we think that that could change. As I said earlier, some of them are already announcing that, look, they are trying to, you know, accelerate, you know, the move to high-volume manufacturing.
But overall, if we look at just maybe for 2024, we expect that the orders that we are gonna get that we're gonna get from Gate-All-Around for logic and foundry will be larger, will be as, you know, greater than half of all of our logic and foundry orders. So the transition to Gate-All-Around is really happening, and we do expect that, you know, orders will continue throughout the year. And, you know, we still have to, you know, maybe wait a little bit as to how much of the high volume will be done next year. But I think overall, we are positive that there should be a significant amount of investment also in 2025. High HBM and high-K metal gate.
I think, you know, we have tried, or at least my colleague, you know, Hichem, has tried to explain during our investor day that the high-K is only one layer. But, you know, there are other layers inside that are gradually being adopted, that is also helping us. One example, you know, just to give, you know, one of the layers, is perhaps, you know, at the most, advanced DRAM devices, instead of just the hafnium silicate, they are now also adding, for example, lanthanum oxide, and that's also on our tool. So we do have more than one layer there. And, you know, as they get more and more advanced, especially at the latest generation of DRAM devices, it becomes more complicated, and they have to use even more layers.
Fortunately, a lot of these layers are done on our tools.
Understood. Very clear. By the way, is this your last call or you-
Unfortunately, no. I still have to do.
Okay.
I still have to do the first quarter.
Okay, got it. Thank you.
The next question is from Marc Hesselink with ING. Please go ahead.
Yes, thank you. First question is actually a small explanation in the press release. You're saying that your bookings in the fourth quarter are also a bit ahead of your own expectations, a modest for the Gate-All-Around pilot lines. Is there anything that we can read into that, the fact that it's a bit bigger than you expected earlier? Does it mean that also eventually it can become a bit better, a bit bigger and more applications for you, or is that not the way to think acout it?
So Marc, thanks for the question. You know, it just came in, you know, slightly bigger than what we expected. You know, I think customers were ordering, you know, maybe, additional tools by as part of their pilot line. That's why, you know, it was a little bit higher than what we had expected, especially after we had already announced that some of the orders actually dropped into the third quarter. So we were of course very positively pleased with, you know, the additional orders. I don't think it really signifies, for example, does that mean that they're gonna invest more in HBM or anything? It's just that they ordered more tools for their pilot line.
Okay. Okay, clear. And then, the second thing, you obviously added a lot of capacity, and you elaborated on that before as well. But given what we're now seeing with, with AI, and also you stressed it multiple times now during, during the call, is the capacity also enough for you in the... Let's go beyond a little bit, the medium, the medium term. And last time it was maybe sort of easy to add your capacity, given that you already had the floor, the floor ready. Is it going to be that easy for you again to add the capacity, if needed?
Yeah, so our the construction of the the second building in Korea, that will be completed early next year. I think it's in the first quarter or maybe early second quarter. Now, that is still the smaller part of our, let's say, capacity in terms of manufacturing. What we have with that and also with, you know, our facility in Singapore, as we have, you know, shared before, we can easily accommodate all the way up to the high end of our 2027 target, which is EUR 5 billion. So we do not really have an issue with manufacturing capacity. And, you know, we are ready to be able to supply, you know, when the demand comes.
So with the current AI story, there's no scenario that you will get into capacity constraints?
I mean, Marc, it's of course possible if, you know, the demand is so big. But we always have, you know, we're always cautious that, look, we need to track the demand carefully, and should we see signs that it is going to, you know, get to a level where it's going to exceed the capacity that we have, we need to start, you know, let's say planning for that quickly. Because for every new factory that we build, it's easily 24 months. So we are very carefully monitoring that, but at this moment, we have enough capacity all the way until our plans for 2027.
Thank you. Clear. Thank you.
As a reminder, if you wish to register for a question, please press star and one on your telephone. The next question is a follow-up from Tammy Qiu with Berenberg. Please go ahead.
Hi, guys. Sorry. Thank you for squeezing me in. So, I have one on China. So you mentioned that China demand would be leveling off in second half of the year because of the capacity digestion period. I kind of think that a lot of the China investment was actually strategic related, and also, they probably do have that market to absorb incremental facility requirement, given the EV industrial trend. Where's your assumption of China will be leveling off, and do you think there potentially can be some upside from that?
Tammy, part of it is just based on discussions that we are having with customers. And, you know, over the last, I would say maybe 24 to 36 months, there's been a lot of investments into what we call mature technology in China. And we do expect that a lot of these now are going to come online, in the sense that they are finally going to be doing high volume manufacturing and producing wafers, which ends up in devices. I think, you know, your assumption or what you've heard is not incorrect in the sense that within domestic China itself, there is a huge demand for such chips. So it is possible that maybe they still need even more, let's put it this way.
In our view, based on what we are seeing today, you know, based on backlog, based on what our customers are giving us as forecast, is that the beginning of 2024 is still going to be strong. But we have not really great visibility now as far as the second half is concerned. And for that matter, for that reason, that is why we are saying that it might normalize, but how much it will normalize is difficult for us to tell at this stage.
Okay, thank you. And lastly, you mentioned that your, your sort of, you know, process within different GAA players are different. Can you confirm your market share is as strong as within all three players for GAA process or your market share differs?
You know, so when you look at the ALD part, we are very confident that we have at least maintained our market share, you know, from the thin film days. And I would just add that on the silicon epitaxy, we have in fact gained market share.
Okay, thank you.
The next question is from Ruben Devos, with Kepler Cheuvreux. Please go ahead.
Yeah. Yes, good afternoon. Thanks for letting me on. I just had a question around an initiative by a Japanese consortium, which I believe has the objective to transition to mass volume on 2 nm by 2027, via Gate-All-Around. I was wondering whether you've had any engagements with them so far, and I guess there's some activity generally in the Japanese market, so curious to hear what your views are on the developments in this region and to what degree you can see some of the local opportunities. Thank you.
So, of course, we are engaged, but I will not say the name. In fact, I visited them last week. And, you know, it's a very, I would say, highly, you know, focused on project by, you know, the Japanese government to really try to build their own, let's say, leading-edge type of, you know, capabilities. I think they are still very much in the early stages, but, you know, we are, of course, heavily engaged with them and with their partner in the U.S., so that's all going well.
I think in general, when you look at Japan as a whole, the government is focusing a lot now on semiconductors, and they have been, you know, subsidizing or giving incentives to a lot of investments that are happening today in Japan. And I then think this will actually kind of rekindle, you know, Japan as a major semiconductor manufacturing location.
All right. Thank you. That's helpful. Just a final question on the aftermarket would be helpful if you could share a few comments on that. I guess you were trending higher in terms of growth rate for the aftermarket relative to equipment sales. I expect that could be the same for 2024. It's helpful to have your thoughts on this.
Yeah. So I think for 2023, we grew at 19%, which was nice. And I think a part of it is because some of the outcome-based services type of products that we have been pushing, we are gaining traction. I think we will continue to see a very nice growth again in 2024, but we will not give a number at this stage.
All right. Thank you very much.
The next question is from Nigel van Putten with Morgan Stanley. Please go ahead.
Hi, good afternoon. Thanks for letting me on. Just a clarification, Benjamin, on what you said around 2024 Gate-All-Around orders. I've written down, but I'm not entirely sure. That you said, greater than half of our logic foundry orders are for 2 nanometer Gate-All-Around. Is that correct, or did I mishear?
No, that's correct. That's correct.
So that's quite a step up then. I mean, if we model third quarter, fourth quarter, maybe combined triple-digit, but I have the sense that, you know, per quarter, high double-digit, then this should be quite a step up relative to what we've seen in the last two quarters. Is that correct?
Yeah, I think the, you know, you know, the way that you described it is quite a step up. But I think now we are seeing, you know, you know, orders really are coming in. And I, as I mentioned, we will continue to get orders throughout 2024. And you know, again, I want to refer back to what I said earlier, we also see some signs of acceleration. One of them has publicly committed that they even want to pull in. So I think it's moving, it's moving in a very nice way, and, you know, we are actually quite sure that we will see more than half of our logic foundry orders, you know, being 2-nanometer Gate-All-Around.
Good. All right. And then, Paul, follow up on, on sort of OpEx. There's, there's obviously also the capitalization line in cash flow. Is that approaching CapEx this year? So EUR 160-200, or is that maybe a little bit too high?
You can expect this year a, let's say, some acceleration in the amortization, because we have, of course, capitalized quite a lot in the last, whatever number of quarters, even last two years, especially given the, of course, all the upcoming inflections that now start to gradually move into, into pilot and then, of course, into high volume production. So there you would, you can expect an acceleration. At the same time, there's still a lot of engagements going on, new engagements. You can also see it in the backlog of our ALD tools. So also capitalization will trend up further. But if you look at the ratios, if I put it that way, of the last, let's say, two years-
... I think the relative ratio of capitalization relative to growth spend and amortization relative to growth spend will not change a lot. But you will still see a step up in both of them.
So I guess there's a bit of a headwind on the PNL, and the ratio is staying the same. Okay.
There will be some relatively more amortization than this year. That's what I expect. Yes. Yeah.
Understood. Thank you.
Thank you, Nigel.
The next question is from, Adithya Metuku, HSBC. Please go ahead.
Yeah, guys, thank you for squeezing me in. Just two more. Just, I noticed the impairment of some capitalized R&D in the quarter, about EUR 2.5 million. I just wondered if you could shed some light on what happened here. Are these some projects that you expected to materialize that didn't come through? Just, just any color there would be helpful. And then, just as you look through to the second half of this year, one of the potential uncertainties is tied to political change in the U.S., and how that may affect your ability to ship to China. So I just wondered if you could quantify what proportion of your revenues in this year came from China, and how you're expecting that to trend in 2024.
Any high-level color around that would be helpful. And if possible, if you could talk about why your CapEx came in a bit lower than expected this year as well, would be helpful. Thank you.
Hey, Paul, do you want to take the impairment and CapEx, and I do the second half?
Yeah. Let me, yeah, let me take the first one and, and, the second one, the third part of the question. On the impairment, it's very simple. It's indeed, a project that was, was started, where at a certain moment in time, at a certain toll gate, you, you, you recognize that we need to maybe take a different, different turn or take a different approach. That happens sometimes. Luckily, this time we've seen it, on time, so only, between inverted commas, EUR 2.5 million. But yeah, that, that happens sometimes. Not every project that is started is, let's say, successful, from the get-go onwards. So that's the first one. On the, on the CapEx, now we guide indeed for EUR 150 million-EUR 200 million.
Our phasing of CapEx is always difficult to project because you never know precisely progress, invoicing, et cetera. So we are just in the range that we gave. So I would not read too much out of it. But yeah, given the uncertainty, we yeah we forecasted a pretty big range, EUR 150 million-EUR 200 million, and we're just in that range. And at the same token, we've guided for 2027, for EUR 100 million-EUR 180 million. Also, that gives quite some yeah large range again. And that's an indication of yeah different phasing of CapEx in between years. That's very difficult to precisely predict.
Adithya, on your question, on the second half with the political change, you know, China, I wish we have some better ideas of forecasting this, but, you know, it's just not possible. And our view is that, you know, we should just continue, as long as there's demand, the customers want, you know, our products, we will supply them with our products, fully complying to all the regulations of the three countries. But, you know, as to how, you know, this will change because of the potential change in the political situation in the US, it's just almost impossible to predict. And, you know, again, we comply to all the regulations, every single one of them, and we do whatever business that we can do.
Got it. And maybe are you able to quantify what proportion of your revenues came from China in 2023?
Yeah. Well, I can only say that it's a, you know, it's a significant percentage. If you look at 2022, we mentioned that it was about 16% or so. We will not disclose a full number, but, you know, we can only say that in 2023, because of the massive investments also in metrology, foundry, and also power analog, it increased significantly for us.
Understood. Thank you.
The last question is a follow-up from Didier Simama, Bank of America. Please go ahead.
Thank you. Question for Ben, on AI, actually. So AMD have given a AI TAM for, I guess, their accelerator business or accelerators at large, an HBM of about $400 billion in 2027, up from $45 billion last year. Sam Altman is also... You know, there were press reports that were talking about $7 trillion of capital investments to build AI chips, et cetera. So we're gonna put that aside for a minute. But, I don't think many people in the market really believe in that $400 billion TAM in 2027. But if we were to dream for one minute, that would effectively imply a 2027 global semiconductor TAM of almost $1 trillion, which is double from 2023 level. And so the question to you, Ben, is: what's the latest?
If these numbers are even remotely close to reality, what's when does these orders need to come in for semi cap companies? I mean, is 2025 already too late, or can they come a bit later for, you know, those fabs to be built and the revenue to be delivered on that timeframe?
I think the way to look at that, Didier, is, you know, to make, you know, a wafer at advanced nodes, whether it's 3 nanometer, 2 nanometer, the wafer probably has to stay in the fab for six months, you know, six months or sometimes maybe even longer. That is when the wafer is completed, and then it has to be packaged, which is probably going to take another couple of weeks, et cetera. So if you really look at, you know, the demand coming in at the end of, let's just say the demand is required, you know, December 2025, I think they have to already start manufacturing this early 2025. Otherwise, you are not going to get the chips.
No, understood. But I guess the broader question: Is there enough capacity today to satisfy $1 trillion of semiconductor demand in 2025, with basically the majority of the upside coming from AI, so leading edge?
I think that's easier because, you know, the AI chips is already having a constraint in terms of advanced packaging, let's say, capacity. So you look at, for example, the largest foundry making a massive investment in for advanced packaging, and also getting the help of what we call the OSATs to do that. So that is actually the bottleneck. It's the packaging, the advanced packaging that is the bottleneck. Now, if you go to a trillion, you know, I haven't done the math, but my guess is there will not be enough capacity.
Brilliant. Thanks very much.
But I hope we will get there.
Me too.
Thanks, Didier.
Gentlemen, there are no more questions registered at this time. I turn the conference back to our CEO, Benjamin Loh, for the closing remarks.
Thank you, Judy. I would like to thank you all for your attendance today, also on behalf of Paul and Victor. We look forward to seeing many of you in our upcoming investor conferences and roadshows. Thank you again. Stay safe, and goodbye.