Soitec SA (SOI) Earnings Call Transcript & Summary

Hello and welcome to the Soitec Q3 2025 revenue call. Today's event is being recorded. [Operator Instructions] I will now hand you over to Pierre Barnabe, CEO, to begin today's conference. Please go ahead.

Welcome to Soitec's conference call dedicated to the publication of the third quarter revenue of our fiscal year 2025. This is a quarter covering the period from October to December 2024. I'm Pierre Barnabe, Soitec CEO. Together with me on this call are Lea Alzingre, our CFO; and Steve Babureck, our EVP Strategy. As usual, we will start with a few comments on our figures, and after that, we will open the floor to questions. After the strong rebound we achieved in the Q2 '25, we maintained our Q3 revenue roughly in line on a sequential basis, down 6% year-on-year. Our Mobile Communications division resumed growth on a year-on-year basis, a much better performance than in the previous quarter. This illustrates some seasonal restocking in RF-SOI cross-value chain, and a continued strong traction in POI for smartphone filters. Conversely, our Automotive Industrial divisions continue to be impacted by the weak automotive market. In Edge & Cloud AI, the momentum remains strong in cloud infrastructure as well as in AI-driven IoT applications. That dynamic is punctually offset by a low performance in images, as we are about to phase out our product next year. Our sales have hence been highly concentrated in Q1. Overall, our Q3 '25 revenue amounted to EUR 226 million, a 6% year-on-year decline which breaks down between a 10% decline on an organic basis, a positive currency impact of 5%, and a small negative scope effect of minus 1, which is related to the disinvestment of Dolphin's Design mixed signal activities on early November. Let's now have a look at our Q3 revenue by end market, starting with Mobile Communications. Our division was back to growth in Q3 '25, leveraging the progressive recovery in the smartphone market. Some seasonal RF-SOI restocking, and the continued strength in POI. The EUR 154 million revenue was up 11% year-on-year. RF-SOI inventories across the overall supply chain are trending down overall, but some customers continue to optimize RF-SOI inventory level based on seasonality and market conditions, which will keep driving fluctuations over the next few quarters. We are confident that we will achieve strong growth in RF-SOI in Q4 2025. In the meantime, we continue to leverage our state-of-the-art R&D capabilities as well as our strategic relationships across the value chain to bring leading-edge innovative products to market. In Q1 '25, we announced the extension of our partnerships with UMC to bring to market the industry's first 3D IC solution for RF-SOI technology. A solution that enables the integration of more RF front-end modules into a single device. In December, we announced our commitment to provide GlobalFoundries with our latest generation of RF-SOI 300mm wafers to support the most advanced 9SW platform. As you know, we have a long-standing relationship with GF. GlobalFoundries' 9SW RF-SOI platform offers significant advantages and value for premium smartphones with enhanced RF performance, improved power efficiency and scalability. These features are critical to ensure a superior user experience in high-end devices and evidence the strengthening of our strategic relationships with the key players in our industry. Turning to POI. Sales continue to gain traction, increasing quarter after quarter since the beginning of fiscal year 2024. In Q3, POI wafer sales were significantly higher than in Q2 '25 and Q3 '24, as the adoption of our product continued to accelerate. We now have 10 customers in volume production and more than 10 orders in qualification and we are engaged with all leading U.S. fabless companies. Finally, FD-SOI sales were also higher than in Q2 '25 and Q3 '24. Let's switch now to Automotive Industrial. After a long period of sustained growth, we have been impacted by the weaker automotive market since the beginning of fiscal year '25. Overall, we continue to see increasing adoption of our products and growing content per car, driven by infotainment, autonomous driving, functional safety and electrification. Our division revenue reached EUR 25 million, a 47% decline year-on-year, reflecting different dynamics across our product portfolio. Sales of Power-SOI reached a particularly low level in Q3 2025, as the current weakness of the automotive market is leading to some inventory adjustments at customer level. Going forward, the strong outlook for battery management system support of products roadmaps towards 300mm. Conversely, automotive FD-SOI sales recorded a year-on-year growth as FD-SOI continues to be driven by adoption in microcontrollers, radar, and wireless connectivity. We continue to support our foundries and IDM customer design wins on FD-SOI, such as GlobalFoundries on the 22FDX, or STMicroelectronics recently on 18-nanometer adoptions, with production expected by H2 2025. As for SmartSiC, I'm very pleased to announce a fifth customer in qualification. This is a strong testimony of strong market appetite for very innovative product. We continue to deliver a growing number of samples and prototypes to our 5 customers in qualification and around 35 prospects in evaluation. The current weakness of the automotive market and the longer than initially anticipated customer qualification cycles confirm a delay in the expected wafer production ramp-up by around 2 years, in line with what we already shared in 2024. Beyond the short-term challenges, in the automotive market, we remain highly confident that further digitalization and electrification of the automotive industry will drive a rise in semiconductor content in the new generations of vehicle. Finally, we recorded a mixed performance in Edge & Cloud AI linked to seasonality effects. The momentum for Edge & Cloud AI divisions remains very strong, driven by sustained investment in cloud infrastructures across the industry and growing appetite for low-power computing devices and edge AI applications. The division revenue down 30% year-on-year at EUR 45 million reflect a strong seasonality effect and mixed performance across our product portfolio. Photonics-SOI wafers continue on their high growth trajectory with revenue much stronger than in Q2 '25 and significantly higher than in Q3 '24. We are addressing the industry's requirements for new data center architectures with better performance, lower energy consumption and lower cost. We are accelerating our partnerships with leading-edge fabless to support AI, machine learning development. And we are benefiting from high investment across the AI value chain as key industry players accelerate their co-package optic roadmap. Photonics-SOI enables new data center architectures with better performance, energy consumption and lower cost, a critical challenge for the artificial intelligence value chain. Photonics-SOI is becoming a standard for high-speed and high-bandwidth optical interconnections in data centers. There is no new AI data centers without Soitec technology. Sales of FD-SOI wafers remained as strong as in Q2 '25, but were lower than Q3 '24. FD-SOI technology is a key enabler for AI-driven devices, both in the consumer and the industrial sector with IoT application requiring ultra-low power edge computing. Finally, sales of Imager-SOI wafers remain lumpy from one quarter to the next, dragging the division's performance in Q3 '25. In fiscal year '25, the phasing of sales was very much concentrated in Q1, whereas last year, they were mostly recorded in H2 '24. That summarize our Q3 '25 performance. On a 9-month basis, our revenue came at EUR 564 million, down 12% year-on-year. Turning now to our fiscal year '25 outlook. We continue to expect a strong recovery in Q4 '25, supported by our backlog, a progressive recovery in the mobile market, seasonal restocking in RF-SOI and continued traction in both POI and Photonics-SOI. However, in the context of the ongoing weakness in the automotive and consumer markets, a couple of customers have asked us to put deliveries on hold, which impacted our Q3 and will impact our Q4 '25 as we are rigorous in matching our deliveries with customer requirements. As a result, we are revising our fiscal year '25 revenue outlook to a high single-digit percentage decline compared to flat previously at constant exchange rate and perimeter. As a result, fiscal year '25 EBITDA margin is now expected to be between 32% to 34% of revenue, and we continue to be very disciplined on cost control to preserve our margin and enable our investment in R&D. Capital expenditures will be around EUR 230 million, in line with what we shared in November, reflecting a more moderated recovery in SOI and SmartSiC. Let's say a few words on fiscal year '26. For calendar year 2025, we continue to anticipate different dynamics across our 3 end markets. The mobile communication market is anticipated to see gradual improvement. The weakness in the automotive market is likely to persist through the first half of the year and investment in cloud AI infrastructures are expected to stay at high levels. With the lack of visibility on our end markets for now, it is also too early to provide specific guidance for fiscal year '26. Given current market conditions, we expect fiscal year 2026 growth to be quite limited at this stage. Our fundamentals remain solid, and will allow us to accelerate as end markets recover in the context of this temporary challenging market conditions. We continue to enhance our technology leadership everywhere to strengthen our SOI positioning with both existing and a lot of new customers, to deploy our expansion into compound semiconductors with the acceleration of POI volumes and fixed customers in qualification on SmartSiC, and to carefully manage our profitability and cash generation profile. This ends my opening remarks. Thank you for your attention. We are now ready to take your questions.

[Operator Instructions] First, we have Emmanuel Matot from ODDO.

Emmanuel from ODDO. First, Pierre, you are talking about limited sales growth for fiscal year '26. Could you give us some colors by division? It is disappointing for me knowing that your main division, Mobile Comm, I was expecting it to grow double-digit given that the smartphone market looks healthy, that RF-SOI destocking should be essentially behind us and that customer diversification is a success in POI. So why more growth than just limited what you say for next year? Second, are the difficulties only due to the current uncertain macro environment? Is Soitec experiencing specific problems with its technologies? And as a result of these difficulties, are you planning to reduce your R&D efforts and your CapEx budget next year? And if we can have also a word about your new customer for silicon carbide for the SmartSiC technology. Is it a key player in this segment of business?

Manuel, on your first question regarding fiscal year '26 preview per division, I will recall what I already said on November on the market dynamics and that is reflecting this preview for fiscal year '26. First of all, the mobile market is under recovery, but still low level of growth, first point. Second point, privileging more mid-range, low-range smartphones. And as you know, we are getting a higher footprint in high-end, let's say, smartphones. And this segment is more suffering even in more negative territories as it is today. That's the first point. The second point is that our content growth is more expected in the range of plus 4%, plus 5%, nothing to compare to the plus 15% we are expecting a few years ago because of shrinking models, because of new generation of RF-SOI solution. And third, what we see is the phenomenon we are observing for a few years now, then the customers are stocking or replenishing the inventory in their calendar Q1. They are consuming to be sure that they're going to deliver the smartphone on time and then they replenish in Q1 and so on. And this is the same cycle now for a certain amount of time. And we don't see dramatically increasing the consumption rate. Then that's for RF-SOI. Of course, POI going to grow and going to continue to grow because we have customers, but looking at the observation of RF that is soft and POI is strong, we are cautious on the mobile communication market. And if you listen to peers, I mean, a lot of players in the value chain, we see softness and uncertainties and we are reflecting what we hear and what we observe today on the market. If we look at automotive, automotive, as we said also in November, H1 calendar is going to be weak, and we see clearly uncertainties in this market. We were even believing a possible rebound in H2. For the moment, it's unclear. There are a lot of big lack of visibility in this market. And it is clear that when you listen again to the peers and to all the players in this market, calendar '25 is going to be difficult. And we expect this market to decline and it's going to reflect in our activities, this type of trends. Looking at Edge & Cloud AI, Edge & Cloud AI continued to grow, thanks to Photonics -- Silicon Photonics and FD-SOI to some extent. But we got the confirmation that Imager phaseout is engaged and Imager, we were waiting around 5% of our fiscal year '25 revenue. That means that, of course, we will have to compensate this loss by a big strong in Silicon Photonics and FD. Silicon Photonics strong is confirmed, and we have more and more traction, that's for sure. And this is really by the combination of market trends and the preview on what we see. We are -- and looking at the market uncertainties and lack of visibility that is obvious everywhere. We are looking at this limited growth preview for fiscal year '26. But of course, we'll come back to you more precisely with more flavors and more details on May. On your second -- okay. Lea, you want to comment. Yes. Of course.

Yes. Maybe I can take the one question. Regarding the R&D and the CapEx for next year, so we want to -- we are going to continue to invest in R&D. This is very important for us to prepare for the future, both on our SOI product to prepare the next generation to improve the cost of our product and also to work on new products, new technology for new markets. But of course, we will adapt also at the level of the revenue, we want to protect margin. So we are going to continue to invest, but keeping in mind our profitability and also our cash flow. We will adapt our CapEx plan definitely based on the visibility we have on customer demand. We are going to continue to invest for POI, and we are seeing big traction. And for the other technology, we'll see depending on customer visibility as we have always done.

And we will really drastically adapt CapEx on the business needs. And of course, on R&D, we are keeping a very strong path 10% to 11% of our revenue going to be invested in R&D. In many domains, we'll come back on it, but you know that we have launched some incubators to prepare the ground for next-generation products that are on site. There were also a question in between you asked, and I would like to come back on it is on the -- is there any specific problem in the -- that I want to be super clear on. There is no intrinsic or I would like to say, in-depth problem in this company, okay? Are we losing market share? No. We are not losing market share where we are already, okay? Are we gaining market share? Yes. We are gaining market share in the new products we are launching. We have never been so diversified hopefully. And again, in the $400 million club, we have 4 products today. We expect, as I say, the fifth one as soon as possible, and you know that the candidate is Silicon Photonics to come. Do we have any problems in quality? No. Do we have any problems with customers? No. From some blacklisted customers a few years ago. Today, we have strong relationships, strong contacts with customers at the highest level. And also, we are extremely attached to respect customers' rhythms. I know we had 2 orders put on hold by a couple of customers, okay? Suddenly, they came to us to say, guys, there is a contract, but we cannot run off this contract because we don't need looking at uncertainties and difficulties of the market, the orders because we don't need for the moment this product. They put on hold for further notice. We could have force it. We could have asked, okay, execute the contract now. I believe we need to respect. And also, we are building long-term relationships with our customers. You know the fact that we have announced the 9SW with GlobalFoundries, the 3D packaging on RF with UMC. And you know how these guys are discrete in communicating particularly with suppliers? They have accepted to do it, and you will have other news flows to come because they consider us as a very reliable and strong partners technology-wise, business-wise, competence-wise. Then it's very important to keep in mind that if you look at the fundamentals of this company, they are intact in terms of positions, technological leadership, competence of the people we are reinforcing, quality of our relationships with our customers and also strength of our balance sheet. Then when the markets will get back on track and when we will get win in the sale? I can tell you that we have all the foundations to make strong growth coming through. Then that's very important to compare to it. And there is no -- and of course, you can check. Emmanuel, you can call any customers, any people you want to check. There is no interesting problem of this company. The fundamentals are very strong for the growth story in the coming 5 years to come. It's a very strong conviction I have with all the team. We stabilized the governance totally. We diversified like never. Now we have to navigate in uncertain geopolitics and lack of visibility markets. That's it. And we'll do it by keeping high level of disciplines on cost management and profitability management to be sure we continue to finance our R&D, the needed CapEx, we're going to adapt and to generate cash for the future. That's really extremely important to recall it with strength. On the new SmartSiC customers, this is, of course, very good news. This is a very important customer. This is a key customer. I will stop with the teasing here because I will never jeopardize. As you know, I'm obsessed by that customer relationship by giving more, let's say, indications or information for you to guess, but this is very good news.

But how do you explain that you have those design wins from several customers in SmartSiC at a time? They don't need to order your products because the end market is very challenging. I'm talking about electric cars in Europe and Americas.

Yes. You have 2 elements. The first element is that the qualification phase is long, okay? Then we have -- and we said it a few years ago, we have underestimated the qualification times. Then it's 2 years plus 2 years, 2 years to be qualified by the IDM integrators and so on and 2 years to be qualified within the car. And it's a long process, and we have to accept this long process. That's the first point. Second, you're right to mention that the electric vehicles market is shaken very strongly. And SiC promises in the futures are now back to reality, back to the ground, taking into account that also we have to follow the price war, and we are following the price war by being stringent on the cost structures. Then we made very active work in adapting our cost structures, including, of course, efforts by the MonoSiC and PolySiC suppliers to be in the race -- to be in the race. And the fact we get a fifth qualification -- today, we are in the path, it's not something I promise for the next quarter. But we are today in the path of qualification per quarter over the last year, this year. Then it means that clearly, we have something very innovative breakthrough that is cost effective for the customers in terms of CapEx, OpEx avoidance, features effective for the final customers and competitive from a price point of view. This is a proof point. Taking the profile of these customers, this fifth customer, it's an add-on and a tick in the box in the strength of the SmartSiC line of product.

And we're now moving on to our next question, which is Aleksander Peterc from Bernstein.

I just have 2. The first one is regarding the roll-off of Imager-SOI. So I'd just like to understand if the full EUR 50 million that you referred to is rolling off in fiscal '26, would you already see some effects of this phaseout in the current year? And so if that's the case, what's the proportion of this roll-off this year and how much remains for next year? Then the second question is regarding your comments on content growth. So you say that you expected a few years ago, 10% to 15% and now we're around 5%, 4% to 5% content growth. So is this 10-percentage point drop in content growth a cyclical matter? So it's just the factor of the mix that we have currently, which is negative? Or is there a structural drop in content growth? And if that's the case, why?

Okay. Alex, then on the first question on Imager-SOI, the roll-off going to be total. That means that next year, fiscal year '26, and we got recent confirmation, we are not expecting revenue coming from this Imager-SOI first-generation product. Then let's be clear, in fiscal year '26, we need to compensate around 5% of revenue weight we had, we benefited in fiscal year '25. As I said, we are working on next-generation imaging systems, but they're going to generate first, let's say, revenue in 2 to 3 years from now, then we'll have to compensate. That's the reason why the acceleration of Photonics is helping a lot, particularly on Edge & Cloud AI to offset a part of this big drop and this big phaseout that is quite rapid and brutal. But we need to also face reality, and we need to tell you exactly what is the situation. Regarding the content growth, the 5% content growth is on RF-SOI. If you take, of course, POI and so on, the content of Soitec in a smartphone is between 10% to 15% because you add POI, you add a bit of FD-SOI for some applications like millimeter wave envelope trackers and so on. Then 5%, the limited 5% is on RF-SOI, okay? Why? Because there are some -- and we are working with the customers also on this aspect. There are some adaptations of the architectures trying to shrink a bit the effect. And we have also some features that are taking more time to come. If I take the Wi-Fi 7, the Wi-Fi 7 is taking a longer ramp-up than expected in terms of adoptions by, first of all, the service -- the telecom service providers and the infrastructures. Then this content is also slowed down by the rhythms of features we were expecting quicker. And that's the reason why. But in terms of expansion of Soitec technologies within the smartphone, we're going to continue to expand, thanks to more POI and thanks to more FD-SOI, even if for the moment, millimeter wave, I would like to say, is stuck to around 11%, 12% penetration rate, and we don't see particularly a strong increase. And the last point to add is that if you remember the 5G adoptions compared to years ago, we see a slowdown in the growth just because the infrastructure, the 5G infrastructures are also taking time to be deployed, then it is, let's say, postponing a bit our ambition and our trajectory. But of course, the more 5G, the more Soitec, the more new features, the more Soitec. It is taking a bit more time. That's the reason why we have flattened a bit the content increase on RF-SOI in the different phones. Plus, of course, as you know, the high range versus low range that is also waiting for us and the breakdown is very important to follow.

Can I just have a quick follow-up on Imager-SOI? So it's my understanding that Face ID isn't going anywhere. So did you miss a trick and that's why you're not in the current generation or the forthcoming generation and only in the one after? What exactly happened here? Why do we have this gap?

No. It's -- in years ago, the customers of the customers decided to enter into a more integrated and simpler solution that is not using -- you don't need SOI product to proceed with this solution, then it's really something that going to a bit more simplification and integration on which SOI has no added value. But we need to think on the next generation of imaging systems for the phones. And not only for the phones for IoT and devices, generally speaking. Today, the imaging system is 2D, is a superficial imaging recognition where you take some points on the surface. The next-generation imaging system is going to go 3D, and they're going to use infrared millimeter waves to detect and to bring imaging across in the 3D dimensions and sometimes across the skin. Then this is the next-generation technologies, they're going to be very interested for SOI on another materials application. I don't want to disclose confidential road maps we are working with customers. But you will see that by using different materials, we're going to be in a position to also looking at crystal orientations to use waves to go into a 3D imaging systems and for more than only smartphone face image recognition. And we are working with several customers for several applications today but again, we will come back on it during maybe the presentation on innovation during the CMD.

And up next, we have Sebastien Sztabowicz from Kepler Cheuvreux.

Elaborate a bit on the level of inventory. Today, you mentioned that you are almost back to normal. Can you give a little bit more precise indication where are we standing right now versus the normal level? And where do you see the inventories, I would say, ending maybe 12 months down the road in terms of level? So are we expecting further inventory correction on RF in the next 12 months? And the second question, we have seen these 2 orders cancellation affecting both autos and consumer. How do you see also the inventories in those markets? Is there also a big inventory build ongoing that is likely to affect your demand over the next few quarters?

Sebastien, then on the RF-SOI inventory today, then there is no change compared to what we said. We observed average 12 months end-to-end inventories in the supply chain with, of course, difference and it's a patchy view, of course, regarding the different customers, but this is what we see. And as we said already, compared to the observed inventory before COVID, we are, yes, 2 to 3 months above this level because the level was around 9 months, then we have 3 months excess. And the question is, will the customers go to pre-COVID view or metrics or will they continue to leave with 12, 11 months? This is today unclear, and we prefer to be cautious and to consider that some correction would continue between 12 to 9 months. But this is exactly what we already said on November. No big changes compared to what we said. Just the fact that the orders we expect to get are just replenishing the inventories in Q1 calendar to be consumed in order to provide the smartphone new models, particularly during H2. And then after the consumption, they will restock again in Q1 calendar and so on. And this cycle is now quite stabilized and there is a kind of aplasia in this model. And then as long as we don't see more smartphones, high-end, more content, of course, AI revolution these rhythms would last. But of course, we see a lot more and more innovation and technology to enter into the phones, and we have hopes beyond fiscal year '26 to see back to growth in RF-SOI territories. But for the moment, we need to be cautious looking at the uncertainty and the lack of visibility of our customers and the customer of our customers, when you talk to them, when you look and you read the earnings, let's say, documentation by all these customers, they are also calling for, let's say, cautiousness and the lack of visibility in the end-to-end market. If we look at the inventories for auto, there are some building up of inventories. But as you know, we are working with a few customers for the moment in Power-SOI, then we have a certain visibility with these customers. But these customers clearly has to fight on a daily basis to confirm contract, gain contracts. We don't see the same phenomenon, but there are some inventory building up. That's the reason why also. We are also very cautious on our Automotive and Industrial segments by forecasting for the market, a decrease -- strong decrease in H1, talking about the market and maybe a continued decrease for H2. We'll see. But lack of visibility is today I would like to say, the key words, and we are expecting by -- before summertime, spring times a bit more of visibility within the supply chain.

And for consumer, what is happening there? Is this also an inventory build or issue?

For consumer, we are less accurate directly on consumer market than it -- we are to -- I would like to say, too small to tiny players to give you, let's say, any visibility on the pure consumer electronics markets. The markets where we see and we discuss a lot with supply chain and high-end customers. And of course, customers of the customers is smartphone, automotive and agent for AI activities.

And from UBS, we now have Francois Bouvignies with our next question.

I just wanted to come back on this inventory comments, the 12 months that you described that maybe will come down and you want to be cautious. I mean, last quarter, you said that it would be potentially the new normal, the 12 months saying that you don't expect further down because it's a new normal level. And now you seem to suggest that you -- it could come down. I mean, what is changing since last quarter when you said about the 12 months? It looks like you expect it further to come down. I didn't have this impression last quarter, but maybe I'm wrong. Second is you talked about adaptation of architectures that would reduce the content for RF-SOI. Can you elaborate on that comment? I mean what do you mean by adaptation of architecture? I would love to have some color on that. And then finally, on POI, you had a strong year this year. Can you explain maybe the customers' road map? Do you expect more customers coming through? And should we expect for '26 like a double-digit percentage growth at least? Is the momentum still there?

Then starting with RF-SOI inventories, no changes, except cautiousness within the value chain that is obvious now. And 12 months is a new normal. But the normal we have experienced in the past is more 9 months is exactly what we said. Then if we look at the readings by the different players in the mobile supply chain, we need to look at a bit of cautiousness. Then some customers might be comfortable with 12 months, but others might ask for going down. We have no specific messages. Clearly, there is nothing new in the conversation we had with our customers, except the fact that when you look at fabless conversations and readings coming from the earnings. Yes, we need to look at cautiousness back to the lack of visibility in the market. If we look at the RF-SOI architecture, this is an optimization of the architecture because, of course, you have many, many features to run on RF-SOI, let's say, footprints. And the idea is to manage the different features. It's part of the, let's say, deal and strategic relationships, technological relationships we developed with UMC. Now with this 3D IC packaging for RF-SOI is exactly this type of -- then we are on it. We are in these changes. We need to have more compact phones because the phones are bringing more and more dices, more and more capabilities also to swallow AI revolution. We need less and less energy consumption. This is exactly what we are working on. But content reduction doesn't mean value reduction. And maybe the content is today plus 4% to 5% in RF-SOI as it is today, but we are bringing more and more value because also we are bringing more and more technology. The 9SW product for GlobalFoundries is a revolutionary product for RF, where we are bringing more and more value, okay? Then it means that, of course, we expect to increase step-by-step our footprint and also our value within this. Then it's not -- it's a natural trend. In which we are -- in which we are investing with our customers and in which we're going to bring more and more value because we're going to concentrate more features in a bit less footprint, then of course, it means more technology. And it's going to also give us a better advance compared to any other competition. And that's also a way to lead the pack and to continue to really run -- to be a frontrunner in these technologies. And it's the changes we are anticipating and we are working on very actively and creating value. Regarding POI. Then on POI, as we said, we have today 10 active in production customers, a bit more than 10 now under evaluation customers. The big shift is that these customers now are more in Western world, particularly in the U.S., where we have started in China. You know that China is going to wait again at 20% of our revenue in fiscal year '26, and we continue to wait that level in the years to come. But POI now is moving to Western world, and we have more and more customers under evaluation or qualification in the U.S. to sustain the filtering revolutions in these areas. And that's, of course, a very important trends, POI is expected to continue to grow because we expand our footprint in terms of customers under qualification, industrialization and evaluation.

And up next we're moving to Olivia Honychurch of Jefferies.

A couple from my side. I just wondered if you could talk about the visibility that you have in each of your end markets. It sounds from your comments like mobile is still low. Clearly, auto is still low given the inventory correction the whole market is seeing, not just you. Edge & Cloud AI might be slightly better. But I'm just trying to get a sense of how you expect to gain confidence in building guidance for FY '26 in the context of that limited visibility, which I think you'll be giving us in 3 months' time? And then I have a follow-up.

As we said, visibility is quite low. It's quite low. It's lack of visibility, to be clear, and there is a cautiousness in all the message surrounding us. We have a strong relationship with our customers. We review with them on a regular basis, their view, forecast and so on, even if they are also lacking of visibility. I mean we are lacking of visibility because they are lacking of visibility and their customers are lacking visibility. And it's not something we say out of the blue. It's just you read, let's say, the documentation around us for the last months. Then in this context, we are working step by step on building, of course, a business plan over 5 years as it is our regular process and a budget for fiscal year '26. Then we're going to refine and revisit this exercise, and we're going to come back to you with, let's say, more details on May. And this is everything we can tell you. But I repeat, mobile soft, but POI growing. Automotive in decline H1, I'm talking about market in decline H1 and risk of continued decline on H2 and Edge & Cloud AI growing, thanks to FD-SOI, but mainly Silicon Photonics, offsetting a total phaseout of Imager-SOI business. This is everything we already said, and it's exactly what I said already on November.

Just my follow-up is on Photonics-SOI, which you mentioned there. Previously, you've talked about that growing at around 60%. Can you remind me which year that rate relates to? And maybe talk about how you see that growth progressing over the next few years given that the adoption of Silicon Photonics in AI data centers is clearly very strong, as you said.

Well, Silicon Photonics is today on a track that is, of course, very, very intense because they need to -- we need to deliver to many, many customers for equipping next-generation data centers with transceivers. This is the mainstream today of activities for Silicon Photonics business. Even if we are working on, of course, the next-generation co-packaged optics solution that will come perhaps a bit earlier than we had scheduled with the key players. All in one, we believe that Silicon Photonics will enter into the club of the $100 million revenue BU by fiscal year '27.

And from Deutsche Bank, we now have Robert Sanders with our next question.

I've got 3 questions that I just want to squeeze through. One is, have you let any larger customers allow their LTAs to expire? I mean, have they chosen, I guess, is the right way of saying it, to let those LTAs expire? Obviously, it has kind of accentuated the problem given the minimums embedded in those contracts. So I was just wondering if you decided to maybe change the way you work with those largest customers. The second one is just for Lea. On the H1 versus H2 seasonality, is that still roughly 40%, 60% in terms of the split between first half and second half as it was in fiscal '25? Just wanted to double check that. And then the last one on the China RF business, obviously, quite a lot of new China RF players. Can you just confirm that your share is higher in China versus companies like Qorvo and Skyworks?

Well, I'll let Lea to take your second question. I will cover the question 1 and 3, Robert.

Yes. We see an effect of seasonality for FY '26. It's too early to know the exact trends, we are still working on our FY '26 outlook. But yes, we'll have a seasonality effect at least at the same extent as this year.

Okay. On your first question, Rob, on the LTA. There is no LTA expiration coming or came recently. On the contrary, we signed a few LTAs with customers running over a few years, 4 to 5 years depending. And that is quite new. Of course, it's not generally speaking. But this year, we had LTAs after difficult years before, fiscal year '24 in this domain. Then we start seeing customers coming back with LTAs. It's a beginning. It's soft, but there is something important coming up. Despite the difficulties of the market, despite the uncertainties, it's a very important signal to underline that. But I believe that there are more buying Soitec technologies and reliability, and they are more betting on our ability to really run ahead the market and provide the best-in-class technologies. That's really, for me, a signal of trust and confidence in the long run, again, despite the market uncertainties. Regarding the RF players or RF business in China, we are -- our technologies are today equipping 100% of the smartphone in the world. Then we can suspect that it is the same for, let's say, RF or smartphone made or sold in China. This is everything I can tell you. And the overall SOI segment share for Soitec is 70%, as you know. Then we could imagine that through the way of the supply chain, we are experiencing more or less the same market share for our technologies and for Soitec, but nothing more to tell you at this stage.

We have one last question for today's call, which comes from Daniel Schafei from Citi.

Just to ask on content in a different way maybe. So you mentioned that in the near future that there will be more mid-range and low-range phones versus premium ones, where Qorvo commented something similar as a continuing trend. Do you see the possibility that this will be kind of not only the state in 2026, but also further out?

Difficult to forecast. We need more to look at the dynamics and what are the criteria to bring, let's say, high-end mobile phones in a better dynamics. First of all, the price is a bit dissuasive. We experienced inflation in smartphones and particularly in high-end smartphones. Second, the secondhand market continue to struggle. And people in secondhand market or repair market are buying high-end phones that is a bit cannibalizing the firsthand high-end smartphones. That's very important to understand the trend. And for me, the dynamics that are going to reshuffle the high-end segments of the smartphone is going to be features, applications surely based on AI that are going to make this smartphone unavoidable or, let's say, to get in the pocket. This is for me. The AI revolution entering into the smartphone that's going to make the type of, let's say, devices different from phones and more kind of personal assistant for your daily life. In that case, we would see -- we would enter into a super cycle of -- for the smartphone industry. It should come when you listen to many things, articles and whatever, it should come. We don't know how, when and which magnitude, but that's going to be, of course, a very important topic. And of course, at Soitec, we're going to benefit a lot from this super cycle, but it's not today at all in our model. I believe we covered all the subjects. Okay. Thank you very much. Then this is the conclusion we can go for, and thank you for the Q&A session. Thank you for your interest and for the dynamic of the exchange in depth and your question. And also very important to recall and to repeat strongly that the fundamentals of this group are totally intact. We manage successfully the governance crisis that is behind us. We managed successfully to diversify our activities with many products, many customers, many geographies going on. We are gaining shares, gaining marketplaces, gaining positions. We are developing a very strong technology leadership in all the domains we are investigating and more to come. We didn't talk about -- in details about SmartGaN. We didn't talk about other products we are working on, but we're going to enter more into details. But already where we are, we are leading from a technical point of view, and we have many, many new products in the car box, particularly through the incubators we have created very recently to be ready for launching when the market is going to be ready. And we are working, of course, it's confidential with many customers for that. And talking about customers, we have developed in depth and strong relationship with our customers on the long run. I can tell you that we have today good, if not very good relationship with all our existing and also new customers we didn't know a few years or even months ago. And that's, of course, extremely important and very inspiring for the team, for the engineers, for the technicians, for the operators in this company. They see and they feel that they make part of the history of the semiconductor by bringing to the market next-generation, very innovative engineered substrates. Thank you very much for your attention again. And we'll have the next stack of the agenda for the Q4 '25 revenue and our fiscal year '25 revenue on May 27 after market close. And we'll be hosting our Capital Market Day in Paris the day after on May 28. And I'm looking forward to seeing you, many of you there. This ends our call for today. Thank you very much.

Thank you for joining today's call. Ladies and gentlemen, you may now disconnect.