Siltronic AG (WAF) Earnings Call Transcript & Summary

Hello everyone, and welcome to the presentation of Siltronic's Q3 2025 Results. Please note that this call is being recorded and streamlined on Siltronic's website. The call will also be available as an on-demand version later today. Your participation in this call implies your consent with this. At this time, I would like to turn the conference over to Stephanie Malgara, Senior Manager, Investor Relations at Siltronic. Please go ahead.

Thank you, Cynthia. Welcome everybody to our Q3 2025 results presentation. This call will also be webcast live on siltronic.com. A replay of the call will be available on our website shortly after the end of the call. Our CEO, Michael Heckmeier and our CFO, Claudia Schmitt, will give you an overview of our financials, the current market developments and our guidance. After the presentation, we will be happy to take your questions. Please note that management comments during this call will include forward-looking statements that involve risks and uncertainties. For a discussion of risk factors, I encourage you to review the safe harbor statement contained in today's press release and presentation. All documents relating to our Q3 2025 reporting are available on our website. I now turn the call over to Michael for his remarks.

Thank you, Stephanie, and a warm welcome also from my side. Let me start with the key messages of today's call. As anticipated, Q3 was the weakest quarter of the year, mainly driven by expected volume shifts into Q4 and FX effects. The market environment remains challenging and wafer demand continues to be soft, in line with the trend we have observed in recent quarters. We managed the situation by taking active measures to control costs and cash. At the same time, we stay close to our customers, ensuring that we align production and deliveries with actual demand and remain a reliable partner. These measures help us to navigate the current environment effectively and deliver on our commitments. Today, we can confirm and specify our guidance for the full year, although we had to further adjust the FX assumption as the weakening U.S. dollar remains a relevant factor for our performance. Finally, as you've probably seen in the media over the past few weeks, there has been some positive news flow from the semiconductor industry. Several companies in our sector have improved their results, clearly benefiting from the ongoing AI momentum, particularly in the memory segment, which has gained further traction in recent months. I would like to emphasize that these effects are not yet visible in the short term for Siltronic as some of our customers are still in the process of normalizing their inventory levels. Nevertheless, we see these announcements as confirmation of the long-term growth trend of the semiconductor and wafer industry, and we are well positioned to benefit once demand is further picking up. Let me give you a broad overview of our development in the third quarter. Claudia will provide more financial insights in just a moment. In line with our expectations, sales and profitability weakened in Q3. Sales came in at EUR 300 million, which has declined compared to Q2. This development mainly reflects the temporary volume shift into the final quarter of the year, as well as the continued FX headwinds. Our EBITDA margin was 21.9% compared to 26.3% in Q2. The decline was primarily driven by lower sales volumes and FX effects. As a result of the reduced EBITDA and because of significantly higher scheduled depreciation related to our new fab in Singapore, EBIT fell to minus EUR 31 million. CapEx totaled EUR 90 million, primarily related to our new fab in Singapore. Consequently, and as anticipated, the net cash flow remained negative at EUR 30 million. Finally, our market share remained stable on a year-to-date basis. This underscores our resilience in a continued challenging market environment. Let's move to the financials. Claudia, please.

Thank you, Michael. A warm welcome from my side as well. I'm pleased to walk you through our performance over the past quarter and highlight key developments. As Michael mentioned, sales in Q3 developed in line with our expectations, totaling EUR 300 million, an 8.7% decline compared to the previous quarter. This decrease was mainly driven by a lower wafer area sold due to a planned shift to Q4 and by FX effects. The euro continued to strengthen against the U.S. dollar reaching an average rate of EUR 1.17 in Q3 compared to EUR 1.13 in the previous quarter, which had a noticeable impact on our reported sales. In addition, we saw slightly negative price effects. Our EBITDA came in at EUR 66 million, down EUR 20 million compared to Q2. This decline resulted from weaker sales and the ramp cost of our new fab in Singapore recorded in the P&L since August. As previously communicated, this will weigh on our profitability for now. However, the impact will ease as volume grows. In the quarter-on-quarter profitability comparison, it is also worth noting that Q2 benefited from a positive non-operating effect related to spare parts valuation. Consequently, the EBITDA margin declined to 21.9%. Along with the successful customer qualifications, depreciation for the new fab also started in August, having a significant effect on our EBIT, which amounted to minus EUR 31 million. Taking all these factors into account, Q3 net income came in at minus EUR 44 million. Let's move on to the balance sheet, which shows a solid and healthy structure. When comparing the 9-month figures, please keep in mind that significant portions of our balance sheet are subject to the translation of Singapore and U.S. dollars into euros and that the euro's appreciation has resulted in FX valuation effects. Following this, total assets stood at EUR 4.9 billion at the end of September compared to EUR 5.1 billion at the end of 2024. The euro strength was most visible in fixed assets given our significant presence and ongoing invest focus in Singapore. Despite CapEx exceeding depreciation, fixed assets declined due to a 6% depreciation of the sing-dollar. Cash inflows from operations combined with the partial drawdown of our syndicated loan in Q2 did not fully cover CapEx payments and prepayment refunds over the 9-month period. Consequently, cash and securities declined to EUR 510 million, yet our liquidity position remains solid. In turn, trade payables, mainly those related to investment activities decreased from EUR 280 million at the end of '24 to EUR 232 million by the end of September and prepayments were reduced to EUR 529 million. Financial liabilities increased as planned by EUR 36 million. Our equity ratio remained stable at a healthy level of 43%. As you can see, our CapEx has been significantly scaled back since the peak in 2023, totaling EUR 308 million year-to-date and demonstrating our continued focus on disciplined capital allocation. However, payments for capital expenditures once again exceeded the invest level, reaching EUR 340 million year-to-date. Such timing differences between CapEx and payments are influenced by factors such as the timing of asset additions during the year, the completion of construction phases or specific payment terms. Looking ahead, we anticipate some rollover effects into 2026, though at a lower investment level. Let's turn to our cash and debt position. Liquidity decreased from EUR 664 million at year-end '24 to EUR 507 million as of September '25, mainly due to substantial CapEx outflows and the repayment of prepayments. Including the undrawn EUR 127 million portion of the syndicated loan, we continue to maintain strong financial flexibility. Financial debt remained stable at around EUR 1.55 billion with initial repayments of around EUR 65 million scheduled to begin in Q4. Compared to our Q2 communication, the overall debt level, drawn facilities and maturity profile are largely unchanged. Net financial debt amounted to EUR 933 million at the end of Q3, marking the peak for 2025. In Q4, we expect an improvement driven by superior business development, positive working capital contributions and an improved cash flow from investing activities. With that, I'll hand it back over to Michael.

Thank you, Claudia. Let's take a look at the key factors that influence our performance. We see a gradual recovery in volume since the sharp decline in 2023. This is mainly driven by the 300 millimeter segment where demand has started to pick up again, an encouraging signal after 2 softer years. We clearly expect this volume upturn to continue beyond 2025. However, 200 millimeter demand remains weak, primarily because the power segment still has high inventories and is suffering from some end market weakness. Regarding prices, we see a continuous price decline outside our LTAs, which is visible in all diameters, but more pronounced in the segments below 300 millimeter. It is likely that we will continue to face this trend beyond 2025. In addition, foreign exchange effects are substantial as the weaker U.S. dollar had a significant impact on reported sales. Compared to the U.S. dollar level at the beginning of the year, the euro has strengthened significantly, which translates into an expected negative top line impact of around EUR 50 million in 2025 versus '24. Over the past 3 years, we have incurred more than EUR 100 million in sales losses due to unfavorable FX exchange rate developments. Let me summarize the main developments we see across end markets. Overall, wafer consumption is expected to increase by up to 8% in 2025, driven by strong AI-related momentum. Recent tariff announcements have created uncertainty in the markets, but the underlying demand trend looks healthy and overall is a significant improvement from the slight decline in 2023 when the downturn began. Server demand remains the key driver, continuing to benefit from the strong momentum in AI, particularly through sustained investments in data center infrastructure. The PC market has shown a substantial step-up largely due to the launch of Microsoft Windows 11, which has significantly increased demand. Overall, the majority of this growth comes from content and only a smaller share coming from unit growth. This underlines that the innovation model for all end markets is intact. Looking at specific inventory levels: memory inventories are starting to come down, which is encouraging. Although, logic inventories are in good shape. However, as previously mentioned, power inventories remain elevated. It is important to note that inventory headwinds persist, although they are gradually decreasing. They continue to dampen the full volume impact on wafer demand. Foreign exchange continues to be one of Siltronic's main external headwinds. Why is this? We have a U.S. dollar exposure of more than 80% of our top line. This makes us sensitive to exchange rate fluctuations. Throughout 2025, this headwind has accelerated driven by the continuous weakening of the U.S. dollar against the euro with a particularly strong impact on the second half of the year. Without this effect, our sales would be roughly at the same level as last year, rather than showing a decline. Let me briefly outline how we manage the current market environment in a proactive and decisive manner. The measures will sound familiar to many of you as we have consistently been emphasizing these priorities over the past quarters. Firstly, we are maintaining strict CapEx discipline. We have deferred selected investments, especially with our new fab in Singapore in synchronization with market demand. Also, we apply restrictive new project approvals to ensure that every euro spend is fully aligned with our strategic priorities. Secondly, we've launched a comprehensive program covering all key cost drivers. Our goal is to drive efficiency and secure lasting savings across our operations without compromising our technological capabilities and customer service. I will provide more details shortly. And thirdly, we continue further cash measures such as closing our small diameter line or working capital management. Together, these actions strengthen our financial resilience and ensure that we remain well positioned even in this challenging market phase. This slide provides a closer look at our capital expenditure. As you can see, 2023 marked the peak of our investment cycle, mainly driven by our new fab in Singapore. Since then, CapEx has come down significantly, and this trend continues in 2025. If you look at the right-hand side of this chart, you can see that CapEx has declined steadily on a half year basis. For the second half of this year, we expect a further reduction, reflecting our strict investment discipline and our ability to adapt to the market environment. Going forward, we will continue to invest selectively and within a disciplined financial framework. But I can assure you, our strategic focus remains unchanged, the ramp of our new fab in Singapore according to the market environment and a necessary steady-state CapEx level to support our global operations. Cost discipline has always been a strong part of Siltronic's DNA, and we continue to build on this strong track record. We have a global cross-functional program in place that targets all major cost categories: labor, supplies, raw material and energy. One example I would like to emphasize, our head count program, which is progressing as planned. By the end of September '25, our total workforce was about 10% below the level of end 2022 despite the ramp of the new fab. Additionally, we are driving energy efficiency measures that have already reduced the electricity consumption by around 5% compared to 2022. All initiatives will run through until the end of 2026 and will contribute increasingly positive to our results over time. At the same time, during the ramp of our new 300 millimeter fab, we are temporarily seeing higher fixed costs that are not yet fully absorbed by production volumes. These costs have been capitalized until mid-2025 and since then are impacting our P&L. This will lead to temporary margin pressure. However, once volumes go up, the burden will significantly reduce. While we remain focused on cost discipline, we also built the foundation for the next growth phase. At the heart of this are our customers, and we are proud to do business with all big names in our industry. Over the past 12 months, we have received multiple supplier awards from some of our key customers. Some of these are displayed in this slide. These recognitions are a strong validation of our technological leadership and reliability as a partner. The strong customer proximity not only reinforces our position as a trusted partner, but also helps us deliver solid results even in challenging market environments. In closing today's presentation, we would like to share our guidance for 2025. We confirm our full year guidance, although we have further refined our FX assumption. Our outlook now reflects a euro U.S. dollar exchange rate of EUR 1.17 for the second half of 2025 compared to EUR 1.15 previously. Sales are still expected to come in mid-single digits below '24 levels. We narrow our EBITDA guidance from 21% to 25% to 22% to 24%. EBIT is expected to show a significant decline, mainly as a result of the start of the depreciation of our new fab in Singapore. Depreciation is now expected to be between EUR 340 million and EUR 360 million compared to our previous guidance of between EUR 340 million and EUR 400 million. CapEx is expected to be between EUR 360 million and EUR 380 million, which was EUR 350 million to EUR 400 million previously. Last but not least, cash flow is projected to be significantly improved compared to last year, although it will remain notably negative due to our continued high investment activity. With this, we conclude our Q3 2025 results presentation, and Claudia and I are happy to take your questions. Thank you very much for your attention. Cynthia, please open the Q&A.

[Operator Instructions] The first question comes from Daniel Schafei with Citigroup.

So, I just wanted to ask you on the competitive dynamics you're currently seeing. So, I'm just wondering, some of your peers are now past the trough and seeing stronger AI wafer demand, whereas you're kind of seeing a slightly lower benefit there, yet you're still pointing to a stable market share. And so I'm wondering what are the drivers which help you maintain your share in this current environment where power is kind of weaker? That would be the first one. And the second question would be on pricing. You mentioned headwinds, especially in lower diameters. You mentioned before that you have a power, edge in 200 millimeter, which should aid you in competing for -- with China in that field. Do you see China being more active there? If not much, then what precisely can they replicate in terms of this power, edge 200 millimeter that you have, and kind of what prevents them of taking more share in these wafers?

Thank you, Daniel, for your questions. The first one, we have always said that we are exposed to all major segments in industry that means memory, logic and power almost with an equal share with a slight overexposure into the memory segment. So what we see in terms of market share is, of course, an average of those segments, an average of multiple customer effects. And here, we are on a stable track. So, it's typically a give and take at certain segments and customers, which holds true for us and our competitors as well. So therefore, there are no major shifts at all from AI demand where we are benefiting and participating as well as some of our peers. With regards to the pricing, yes, it's more pronounced in the lower diameters. Particularly China was the question. I think China is well advanced at lower diameters, that's what we always said. The smaller the diameter, the more is, let's say, the China activity there. The more also the price pressure is pronounced. That was one of the reasons why we exited the small diameter business. So everything below 150 millimeter is history for Siltronic. We concluded that phase out in mid of the year. 200 millimeter, as we said, is under some pressure. Volumes are stable roughly, but not growing. And in 300 millimeter, we see a significantly growing volume trend, which is driven by AI. Here, I would say, the Chinese competition is on a way to localize some of their activities, means Chinese chip manufacturers also buy from Chinese wafer suppliers, certain wafers. And outside China, we see some activities on lower grade test wafers and simple specification. So, there is, let's say, no breakthrough news or any significant change to our statements we made around this situation in the previous calls.

The next question comes from Constantin Hesse with Jefferies.

I've got 3. I just want to start a little bit on momentum and customer conversations into next year a little bit. So I understand that 200 millimeter, I think, that's kind of expected. No major improvement there. Industrial activity is expected to remain relatively sluggish next year. So the focus is obviously on 300. Looking at 300, legacy 300 seems to be flat to improving. And obviously, leading edge 300 millimeter is doing all the work. So, I'm just trying to get a feel for what your customers are seeing in terms of what they potentially expect next year. And within the mix of legacy and leading edge 300 millimeter, could you give us an indication of how much the mix today is -- it doesn't have to be for Siltronic, because I know that you typically don't provide this for Siltronic. But maybe a market indication, how much is leading edge today of 300 millimeter compared to legacy 300 millimeter? Let's start with that.

Constantin, thank you very much for your questions. I mean, as you know, today, we don't guide '26. So my comments around your questions will be a bit more, let's say, high level. I think you are absolutely right. The power segment is indeed the one being still significantly elevated in inventory levels and that will not be depleted very short term. So there will be definitely clear hangover effects into next year, which might, of course, particularly influence the 200 millimeter segment. In 300 millimeter, we see indeed the growth trend driven on leading edge specifications on advanced nodes. It's a difficult task to kind of differentiate really between leading and legacy as there are even multiple definitions in the market. But the volume upswing will be mainly driven by larger means higher specification, including leading edge, which will be, I think, a significant volume growth to be continued. On the other side, when you have in mind our chart with all those arrows, we would also see the major trends there being continuing into -- beyond the year's end. So means the price and mix effects, and particularly the FX effects, will be definitely opposing the described volume trends. So that's maybe as much as I can say today for an early view into the market environment in '26.

Could you maybe, Michael, just a follow-up on that. I mean, let me try it differently. Compared to conversations that you were having with customers at the end of '24, have these -- are conversations today showing a bit more -- are they more optimistic today? Are they perhaps a little bit more positive compared to what they were then? Is there any reason to be a bit more positive about '26 than you would have been at the end of '24?

Yes, I think, by and large, we can say that the memory environment is, let's say, on a more positive tune. In the meantime, some of the customers are back to really healthy inventory levels, not everybody. There are still elevations here and there. So definitely that's, I think, better in terms of tonality and atmosphere. Logic never was that bad anyway. So especially leading edge logic is a driver of this whole development and particularly loaded in the industry. And power, as we -- as I indicated already, is still, let's say, the worst out of those 3 segments, and it's going to hang over beyond the year's end for sure. But for the other 2 segments, we can say tonality and atmosphere did brighten up a little bit.

Okay. good. And maybe, Claudia, going over to you. Clearly, trying to protect cash flow here given the continued sluggish environment. Just to have an idea, can you just remind us again roughly what the repayment cycle is going to be over the next 2, 3 years? And just for 2026 specifically, just sort of, I mean, without guiding, but just to have an idea because, obviously, you will be cutting CapEx further. So your maintenance CapEx runs at about EUR 200 million -- or roughly EUR 200 million as far as I understand. How should we think about the expansion area like? Do you think you can bring free cash flow to at least breakeven next year?

Yes. Let me put it that way. We don't guide for 2026 right now. But of course, I can provide some context around 2026. So right now, we have quite a solid cash position. And taking into account that we also have another -- roughly EUR 130 million undrawn loan, we have quite a good flexibility going into 2026. In Q4, we will start to repay loans with an amount of EUR 65 million roughly, and next year roughly EUR 100 million or a bit more, EUR 105 million or so. But looking into 2027, there's more to come, more to repay. So most likely we will have to make a refinancing next year. Of course, we try to protect our cash, as you said. So we will bring down the CapEx level next year even further. But right now, we do not do any guidance on that. You have to wait until, let's say, February or so. But I can promise that we will bring down our CapEx level next year again. But as I mentioned during my speech, there will be a rollover effect from CapEx payments. So asset additions this year will lead to some CapEx outflows next year. So this is more or less what we can say now about 2026. More details we will provide beginning of next year.

Okay. Maybe just to try and go a little bit further into this view, with regards to the refinancing, are you comfortable on the refinancing? Or do you see any reason for having to go into equity markets here? Are you comfortable that you can refinance it all via debt?

Definitely. So there are several options that you can use to refinance. And there's a high interest by financing partners. They are actively approaching us about refinancing. So we -- there's no doubt that we can refinance what has to be refinanced.

Last one, Michael, over to you. I think this does become quite important now because China is, obviously, becoming a bit more of a risk feels to me compared to the last few years. So I know that you don't break down Mainland China, but I think it's really starting to get important to understand what the risk is for Siltronic if the Chinese start making a dent on volumes overall. So what -- how much of sales could we potentially be talking about being at risk, call it, at 200 millimeter or 300 millimeter legacy? Just to have a rough idea of what the risk could be over the next 3, 4, maybe 5 years if the Chinese really start catching up?

Yes. Thank you, Constantin. Yes, China is, let's say -- so we don't single out details. What we say is Greater China, so that means Mainland China and Taiwan together for us, are 36% of revenue last year. What we also say are both are more than 10%. Both are double digit. More details, we will be reluctant to break down for, let's say, also competitive reasons. What I can say is, we have a more lower exposure to Mainland China than some of our peers. And we see in recent times more a, let's say, sidewards market share development for us. So not a significant pressure or decrease. So it's an attractive market for us. Some of the customers really love us as partners and are kind of building in relation with us. On the other side, of course, competition is building there, its growing there, so we're fully aware of that. But let's say, our exposure is not huge. It's smaller than, let's say, the average of the industry from the Western manufacturing side perspective.

The next question comes from Gustav Froberg with Berenberg.

I have a couple, if I may. First, on sort of the sluggish recovery in wafer shipments. Clearly, things are happening, but slow as has been the case in the last couple of quarters. With a bit of reflection, is there now anything you can point to in terms of maybe structural elements that are causing fab sort of buying to be a bit slower than it has been in the past, maybe elongated fab cycle times or anything of the sort? And is there anything you can point to in terms of a tipping point where you think this might change and properly reaccelerate again, maybe broader adoption of HBM memory or anything of the sort? So that would be my first question. Second is around working capital contributions. You mentioned a positive uptick in working capital cash inflow in Q4. Could you help us understand the magnitude there, just so we can understand a little bit better how to think about the year-end cash balance? And then last one is on loading in Singapore and just how you're progressing with the fab build-out there? Those would be my questions.

Yes. Thank you very much, Gustav. I will take your, let's say, volume effect question and the loading in Singapore and then hand over to Claudia for the working capital related question. Structural changes on volume effects, I mean, volume is driven by 300 millimeter and it's by AI and some, let's say, smaller effects around AI. In terms of high-bandwidth memory, we see that growing significantly. There are those effects you're mentioning, yes, so high-bandwidth memory wafer is maybe processed longer in a chip manufacturing cycle. On the other side, also we see, let's say, yield of those manufacturing is lower than, let's say, standard memory yields. So therefore, there are 2, let's say, opposing effects. Overall, volumes seem to be almost on 2022 levels as some of our peers also reported. So it looks quite a volume growing situation again. Other structural effects, I'm currently not aware of. So when you add more complexity, let's say, the technical requirements on wafers get higher, which allows, let's say, a very close technical collaboration with customers. And which is maybe an additional, let's say, burden for new entrants into this industry. That's maybe another thing where we see advanced leading edge and those kind of colors really have very high-tech requirements where, of course, we are perfectly, perfectly strategically positioned with our new fab in Singapore, having this fully automated and all the state-of-the-art metrology tools in place. I mean, that's now really, I think, a great positioning. That's maybe another effect that there's more high-tech and advanced requirements. And with that, loading in Singapore is progressing as planned. We reduced the initial ramp speeds. But now, of course, loading is gradually climbing up. We have achieved the major milestone mid of the year that major customers are now qualified for advanced and leading specifications there. So therefore, we can more and more use the fab. But as indicated, of course, we're doing, let's say, the further ramp and bringing in new or additional tools very carefully. We have some, as Claudia described some hangover from our orders for the first wave of equipment coming into Singapore, and we are kind of pulling the brake of more as we currently don't need it yet. So we will synch further scale up then really with market demand. And with the qualifications being achieved, we are in a position now with using the capacity, doing some preferred loading there to drive volumes into Singapore, but in line with, let's say, market demand and, of course, a relative restrictive CapEx approach to protect our cash. So for the working capital question, Gustav, I hand over to Claudia.

Thank you. Thank you, Gustav, for your question. As indicated in our guidance, Q4 is expected to be the strongest quarter in 2025. And to ensure timely deliveries in Q4, we built up inventories in Q3. So you see that in our balance sheet and in the cash flow statement. This, obviously, led to a negative cash effect in Q3, which will reverse into a positive effect in Q4 when we reduce the inventory again. So this is more or less the working capital effect that I mentioned. And regarding cash flow in Q4, as mentioned before, we anticipate an improvement in cash flow from investment activities as well. So those are the major factors impacting the cash flow in Q4.

The next question comes from Florian Treisch with Kepler.

My question is a bit more on, let's call it, a very near-term demand. I mean, you reiterated your guidance more or less on implying a clear uptick in demand looking into Q4. I mean, can you maybe add some, let's say, details to it or where your confidence is coming from? Have you already seen some kind of clear acceleration in October? And building on that question, would you be willing to call the bottom in Q3 and expect continued sequential growth in the coming quarters, i.e., would you be willing to say Q1 will likely be up sequentially over Q4 as well, i.e., is really the worst behind us and now you really can all benefit from volume recovery entering 2026?

Yes. Thank you, Florian. Great question. As we are already 1 month into Q4, of course, confidence is growing by the day. As Claudia hinted, some of those, let's say, shifts of volumes from Q3 into Q4 have been already preproduced in Q3 and are now scheduled with very clear, let's say, dates in most of the cases already. I mean the year is only 2 months to go by and large. So confidence is growing there by the day, and that was the reason why we definitely and clearly confirmed our full year guidance there. Is Q3 the bottom of whatever time frame, I don't know, to be honest. But what we clearly say Q3 is the bottom of the year, and that was mainly because of those mentioned individual shifts. As we -- you know, today not to talk in detail about the next year, it would be a bit weird now to make a precise Q1 statement. So I'm really not in a position to do that. But Q3 is the bottom of this year based on very individual customer shifts and decisions by some of our customers rather than anymore.

The next question comes from Martin Jungfleisch with BNP Paribas.

Two questions, please. The first one is on this cost reduction or head count reduction that you mentioned. Could you just provide some color on this one? When this will be implemented and what kind of full cost savings run rate we should expect from that?

Martin, Siltronic has a strong track record of sustainable cost reductions and we implemented that cost program, yes, end of '23, beginning of 2024 in order to deal with the weak market environment. Our cost program, as we mentioned, is addressing all cost categories across all sites and it's definitely strengthening our resilience. You can see that probably already in Q3 with the low sales volume, we still achieved an EBITDA margin of 22%. So I think that speaks for itself. Regarding definite numbers, we are a bit reluctant or we don't want to communicate on how the program is going despite the fact that it's going well, and we do not communicate any targets. But I hope that the numbers, they speak for themselves because you see that we are -- we have reached quite a resilience in our cost position, I would say.

Yes. And maybe to build on this, Martin, the statement that we already came down 10% in head count globally compared to 2022 levels. I mean, it's, I think, a clear message. And of course, there's a counter effect when you ramp a new fab in Singapore, you have to build certain head count. So this is even overcompensated by head count savings in other areas. So therefore, I think, as you know, labor is cost driver #1. It's, I think, a clear proof point that it's going pretty well in that corner.

Got it. So you expect incremental cost savings in Q4 versus Q3 as well, right?

Definitely, it's not continued -- sometimes it's more a step function, but we expect sequential improvements at least until the end of 2026, and then we have to decide how to continue with the cost program. But we do not stop.

Yes, that makes sense. And then secondly, just on pricing. I mean, 300 millimeter, as you ramp the Singapore volumes over the next quarters, I mean, should this have a positive effect on the total 300 millimeter pricing environment given that contract pricing for these FabNext volumes should be significantly ahead of the levels that we are seeing today. So I guess should you see a tailwind on 300 millimeter from these contracts, I think, increasing the mix next year? Or it is not a correct assumption?

So when we -- Martin, when we talk about pricing, it's, of course, a mixture of multiple effects. So roughly speaking, 2/3 of our business is in LTAs and here the prices are as contracted, and there's also no change to that statement. What we see outside LTAs, and that's about 1/3 of the business is indeed some price pressure that is, let's say, increasing. I mean if you have a small effect on a quarterly basis and you accumulate that over a couple of quarters, it's not so small anymore. And when we talk about pricing in overall, it's the average of those 2/3 as contracted and 1/3 that is under some, let's say, price pressure. Going forward, we don't see any, let's say, indication that, that will change because, yes, we are ramping more volumes in the new fab -- but outside the LTAs and outside, let's say, the core of the 300 millimeter business, the price pressure is persisting and is ongoing. So therefore, overall, I don't see, let's say, a very short-term change in this pricing trend.

The next question comes from Jimmy Huang with JPMorgan.

Can you talk about which applications we might be able to raise prices of silicon wafers next year or maybe at least reflecting higher cost to customers? Are only AI-related applications such as HBM and leading edge logic have this kind of chance? And in the meantime, will other mainstream AI market emerge to have spot price erosion pressure next year?

Jimmy, it was very difficult to really get the point of your question. Can I ask you to repeat maybe the essence of the question again? Also the line was a little bit noisy.

I was trying to ask about the silicon wafer price trajectories into next year. Can we talk about the different applications such as AI-related and AI applications? Do we see any different pricing trajectories for silicon wafers into next year?

Okay. I hope I got it. So the question was whether there are different pricing tendencies in different segments, particularly in the AI segment?

Yes.

Okay. Yes. So it's not so much about, let's say, end segments for us. It's more about really wafer specifications. And when we talk about leading edge, what we always said is it's a very clear strategic focus for us. And here, we have indeed higher prices, higher margins and also a slightly above average market share. So that holds without any change. So the pricing is more driven by, let's say, technical detailed specifications in certain projects rather than by end market segment overall. There are certainly also, let's say, more standard AI-related wafers in the power segment, for example. When you do power supply in those data centers also some, let's say, chips are required in the power segment, which they also claim to be AI-related demand and market dynamics, but that would be rather standard wafers being needed for those sort of applications. So I think technical specification, impurity levels, defect levels geometry and physical properties of the wafer and let's say, the increase in these specifications are, let's say, more the driver for the pricing in the end rather than, let's say, an overall segment. So I would not say that every wafer that is related to AI is different in pricing.

I see. Understood. Can I have a follow-up question? Yes. So for 300 millimeter wafers based on your company's own supply-demand projections, could 300 millimeter wafer utilization rate return to 95%, the level that big wafer supplies could have certain bargaining power over its customers? Or just currently, there's no visibility that the 300 millimeter wafer utilization rate could return to such high level based on your customer discussions or based on your industry observations.

So again, Jimmy, so we apologize, but it's really difficult to understand really the question. Was it whether 300 millimeter volume at customers brings us a different power in negotiations?

Yes. Not sure whether it's my device issue, but I was trying to ask about what's your view on the 300 millimeter wafer utilization rate, because we see there's so many industry supply, and we don't think that the 300 millimeter wafer have high utilization rate. But we would like to ask you what's your view on the 300 millimeter wafer utilization rate over the next few quarters?

Okay. Now I think I got it. It was about UT in 300 millimeter in general. So I'm not able to talk about different segments, UT. But what we know is the UT did climb up again. So we have a situation that some of our peers and some of the market analysts reporting already volumes being back on 2022 levels. On the other side, in the meantime, some capacity has been added. A high-level estimate would be that the UT is somewhere around 80% in the overall industry, but that contains, let's say, certain arrow bars. It might be higher in, let's say, in segments where the volume demand is picking up very quickly. It might be lower in segments where the volume demand is rather sluggish over the last couple of quarters. But that's a number I heard in, let's say, in certain communications already. It's not our number, but that's maybe an indication with a significant error bar.

The next question comes from Robert Sanders with Deutsche Bank.

I'd also like to ask about utilization of epi versus polished. I'm just interested to ask around the new facility, FabNext that you're ramping up. Is that going to be primarily ramping up polished wafers? Or will it move quickly to epitaxial wafers? The reason I'm asking is I would assume that the utilization rate of epi is higher than the utilization rate of polish.

Rob, thank you very much for your question. Epi versus polish, I don't think there are substantial differences because both in 300 millimeter are driven by, let's say, leading edge and high-end demand. You're absolutely right in Singapore for the first time in 300 millimeter, we established also the epi technology, which is also ramping at the same time as polished is doing. So we're delivering both sorts of wafers out of our new fab. In the meantime, also when we talk about major customer qualifications, this also comprises both polished and epi wafers.

Got it. And just in terms of China, your Japanese competitors talked about SMIC and Hua Hong not allowing them to compete for business anymore because they are under pressure to buy locally. What have you seen in China so far?

So we also hear and read such news flow. Current geopolitics is not hindering us. So we have customers in China, which we are shipping to unchangedly. We have a team in place that is kind of monitoring new requirements and regulations coming out of Washington or wherever, continuously and make, let's say, assessment, which is not sometimes a simple task. Those guys have to analyze 400 printed pages overnight sometimes. But for the time being, we don't see any major effects on our business development with China from those regulations.

Got it. And just the last question just would be on covenants on the debt that you have. Can you just remind us what those are, whether it's interest coverage, net debt leverage, net debt to EBITDA or anything like that, just so we can understand any risks there may be as you go into losses.

Rob, I take your question regarding financial covenants. Our covenant is net debt to EBITDA. We do not disclose any further details on that, but I think we gave you some hints regarding reaching net leverage, which was a discussion before. But I can still -- we are still confident that we will remain within the financial covenants.

The next question comes from Didier Scemama with Bank of America.

A very quick one, maybe for Michael. I think you mentioned earlier that there were still excess inventories of wafers in the power segment. And in memory, things were normalizing, but there is still some pockets of inventories. So I guess my question is, can you maybe quantify that? What you estimate is the level of excess inventories or level of inventories in general in power and in memory relative to what it should be in a normalized market?

Yes. Thank you, Didier. Let's start with the memory segment first. So they built up inventories, I think, for more than 2 years in the meantime. It's a kind of growing line and substantially above, let's say, healthy levels. In the very last quarter, we saw potentially the peak as those levels did maybe come down a little bit. Now we need to kind of watch out whether this was really the peak or whether it's more sidesways walk on an extremely large and high level. We need to check on further data coming out in the next quarters there. Memory did come down significantly, I would say, on average, almost on healthy levels, which still means some customers are elevated. But they also were significantly elevated in the past, and I think did a good homework both in managing this and also maybe with some advanced demand from the, let's say, particularly advanced specification, high-bandwidth memory side. So therefore, overall, close to normalizing with a few customers still being elevated a little bit.

And in power?

Power, I think, across many players and almost the whole industry significantly elevated and not yet clear indications that they're really coming down, maybe flattening out, maybe saw the strongest growth in inventory, but there's just one data point being below the second quarter. So we need to really see whether those stabilizing and potentially decreasing trends are really manifesting in the quarters to come.

Got it. And I have a quick follow-up just on the memory side as well. I think estimated HBM demand is about -- or current capacity, I should say, is about 400,000 wafer starts per month, the total DRAM capacity of about 1.9 million wafer starts per month. Is that roughly your exposure when it comes to wafers going into DRAM market or a similar exposure to HBM? Or do you think you are slightly underweight to slightly overweight HBM? That would be helpful if you could help us understand that.

I'm not sure where your numbers coming from. But I'm, of course, very reluctant and maybe even must not comment on our exposure to those different segments. Our competitors would love, of course, to know our precise exposure to high-bandwidth memory. I can say very clearly, we are well positioned there. As you know, we are -- have a slight overexposure into the memory segment. But today, I am really not in a situation to give you more details there.

We will take a follow-up question from Constantin Hesse with Jefferies.

Very quick one. Michael, do you think that the volumes that are coming online in Singapore now that have started coming online in the second half and obviously, the base effect next year, could these volumes compensate the loss of the smaller diameter that you closed in '25? That's the first question. Second question, is the price pressure entirely focused in 200 millimeter? Or is there some in 300 legacy? Let's start with these 2.

Yes. Thank you, Constantin. The DST closure is really a small effect. We will see, let's say, slight margin contributions from it as we communicated. And the top line contribution in the full year was mid-single digit. So this year, it's only half a year, so you can pretty much neglect it. And the overall development will be covered by other effects like the volume growth, the price effects and the FX effect. But in that context, the DST closure this year is almost in the noise level top line-wise. In terms of price pressure, it's, I would say, everywhere outside LTAs, yes, the majority of the LTAs are on 300 millimeter, but there is also of course, non-LTA 300 millimeter business. And here also we see some price effect. So it's always when we don't have LTAs, then you are more in price discussions.

Claudia, just a very quick one. Remind me again what the rough reimbursement is of the prepayments. I think you made a comment to that in Q2. I just wanted to get a quick -- just a homework question.

The prepayments, I think it's around EUR 50 million within the next 12 months.

There are no further questions at this time. I will now turn the conference back to Ms. Malgara for any additional or closing remarks.

This concludes our Q&A session. Thank you for joining us today. We'll release our preliminary full year 2025 figures on February 3. Please note that there will be no conference call on this day. The full set of numbers, including our Annual Report will be published on March 12. On this slide, you can also see our next IR events. Thank you, and have a good day.

This concludes today's call. Thank you for your participation. You may now disconnect.