Welcome everyone, and the theme of today is ready for what's next, and this will become clear throughout the presentations that we'll do today. Today, I've brought with me Jörg Doblaski, who is our CTO, who will talk about powering the future. Luigi Di Capua, who is VP Product Marketing, who will talk about sensing the real world. Volker Herbig, Business Unit Manager, Microsystems, who will talk about all the exciting heterogeneous integrating systems, which is the new trend. This afternoon, we'll have Damien Macq, our COO, who will explain how this all transforms into operations and production. At the closing, Alba Morganti, our CFO, who will explain how this all translates in numbers. The theme of today is X-FAB is ready for what's next.
So I'll walk over, on the one hand, the megatrends that drive our business, and this is with great purpose, and this is really something that doesn't change. It's going for over a long term. We also, in the company, decided to really focus on three technology pillars, where we will be focusing on, and that is the theme of the three presentations this morning. Then I'll also explain that this drives the demand, or that the demand is there, in combination of our technologies and the megatrends. We have the right capacity expansion in place. We have invested a lot, and it's now all coming online. Damien will talk about that. And we have the big picture. There's a lot of things that are changing in the world.
So every day there are new things that we need to take into account, and how X-FAB is prepared for that. So let's first talk about the megatrends. So as X-FAB, we are focused on two megatrends. On the one hand, the decarbonization, the climate change, this drives electrification of everything, and this is an important thing for our society. For the next thirty to forty years, the mission is to come to zero CO2 emissions, and that's still a long way to go, and there will be a lot of electrification, a lot of new things that will have to happen, and we are part of powering the future, and Jörg will talk about that. But also the aging population, this is something which is very important to take into account.
It's—t he population is getting older, needs more medical care. We need to improve the efficiency of medical care, and this all this service of medical care is falling on the shoulders of a ever-decreasing younger population. So there's a lot of challenges in front of that, and we are part of that. There is also a technology trend that we cannot neglect in the last couple of years. It's all about smart everywhere, AI, the age of AI, we're in, and this plays out in two ways for X-FAB. On the one hand, AI creates new opportunities. We're into everybody heard it, AI comes with a lot of power management, consumes a lot of electricity. We're into this power management of data centers, and Jörg will explain about that.
But on the other hand, we are very complex factories, very sophisticated systems that also, with AI, it allows us to further improve productivity, and Damien will talk about that. All right, let's see. We're playing on our strengths, and there are many exciting opportunities where customers come to X-FAB because of the unique offerings that we have, and there are several aspects to it. First of all, we have a very broad combination of capabilities. We have CMOS factories with also unique capabilities in the CMOS factories, but also our Microsystems, and we are combining in the Microsystems and MEMS processing, and we're combining those two technologies into the Microsystems, what Volker will talk about, and this is kind of a unique offering in the semiconductor market.
We are also giving a high level of flexibility. We are adapt to our customers. We co-create with customers. The customers have dreams in their mind, we help realize their dreams with our technologies. And on top of that, we install robust supply chains. Making innovative products or samples, people, startups can also do with institutes like IMEC, Fraunhofer, LETI, and so forth. However, when it comes to industrialization, that's another thing. That's where X-FAB comes in, then with our years of expertise in producing large volume of specialties, we mature those technologies and bring reliable supply chains for that, that allow our customers to produce these products for 10, 15, 20, 25 years. We have products that we are producing already for 25 years.
We have looked at, okay, where our strengths are, and we decided to play on three pillars. The semiconductor industry is very. It is a very large industry, it is very wide, so we can't play everywhere, and we have decided to play on these three pillars. Let me walk you over these three pillars. First is powering the future. As X-FAB, we have a history of making technologies for higher voltages. Started off with automotive products, where batteries of a car are different than the batteries of a mobile phone, and the environment is very harsh. We have a history of making things mature for high voltage processes, and we are building on that strength, and now the whole world is moving into higher voltages, 48 V, 400 V, 800 V.
Eight hundred volts will come in the data centers. So, and this is where we have a particular expertise, and Jörg will talk about that. Then the second pillar is about sensors. Sensors everywhere. We have been working for years on sensors, and sensors is a very big portion of our business, and there is a trend of, yeah, making things smart. So in order to make things smart, you need sensors. Also, the whole AI is based on data. This data comes from sensors, and so the whole all these trends, they will drive further requirements and needs for more sensors in everything around us, for robots, for smart cities, smart homes, and smart everything.
Already today, also, many of our customers, also our largest customer, they are a leader in magnetic sensors, but also beyond that, we have a lot of sensor products already, that we're dealing with. And then third, the, is all about the next level of integration. So what I explained is, so we are, the CMOS, and we have the microsystems. We're combining that to integrate systems on, microsystems on wafer scale. And this is a new trend, and, we see there a lot of demand, things that were not possible years ago. Because of this combination of technologies, new materials, and so forth, become possible and create new demand that comes to X-FAB and, Volker Herbig will talk about, that.
So what we see, there is great demand for what we are offering, and the mega trends, they drive this long-term demand across all the end markets that we serve. As X-FAB, we're also the sole supplier to our customers. So typically, our customers do develop a product with a foundry, do this only once, and so they are all the customers of X-FAB, they source their products that they develop with X-FAB, only from X-FAB, and that means that we have a very sticky relationship with our customers. But that puts also a very high level of responsibility on X-FAB, and we need to make sure that we have the capacities in place to serve our customers. So therefore, also, we decided to invest in capacities going forward.
We have, over the years, accumulated a very large pipeline of different products, and although there might be short-term headwinds because of uncertainties, geopolitics, tariffs, and so forth, the long term doesn't change the purpose and the needs for the products that we produce, and the technologies that we're offering are not changing. As I explained, we need to have the capacities in place, and therefore, we decided three years ago to invest $1 billion expansion throughout our different factories, and we progressed very well. Our Damien will talk about that. Two years ago, our capacities, or one year ago, our capacities were fully utilized. Our customers were in allocation, and this also created a big headwind for X-FAB to develop new business. In the meantime, capacities have come online, and we have caught up with the demand.
We have now sufficient capacity in place to serve the demand that is out there, and that gives new possibilities to develop business. Another important element through, as I explained, through AI, we will also be able to further improve our efficiency in our factories, and that will also have another effect of productivity improvement. The biggest expansion that we did is in our factory in Malaysia, that is now completed. This Friday, there will be the official opening with the Prime Minister of Malaysia, and that's from there on, the factory will have increased capacity and will further grow its output. Another important thing is that semiconductor is a strategic technology, and more and more, all the large industrial states, they have programs to stimulate semiconductor industry, and as a foundry, we're in the midst of this. And although you're—
The market is global, there are trends to try to localize it. However, in my view, it will stay really a global market, and as a European supplier, we are somewhat seen neutral in this geopolitics, and that will help, that supports our business. Also, an important factor, we have manufacturing capacities also in the US. We are one of the very few producing silicon carbide in the US, and that will also helps us with the tariffs. Another important geographical region is China. China has grown a lot, so has driven the EV deployment, is an important region for electric cars, and we, as X-FAB, we are supporting those needs with specific technologies, and Jörg will talk about that.
What we also see is that China is different from the rest of the market. In China, you have, in the rest of the market, you have typically IDMs who serve a lot of the automotive needs, like Infineon, ST, Renesas, Texas Instruments, Analog Devices, and so forth. They're all, serving the China market with their analog and mixed-signal products. There is two trends. Because of the China- for- China policy, we see the two trends. On the one hand, the IDMs are trying to put Chinese supply chains in place. Still, they are Western companies. On the other hand, there is a lot of, fabless companies in China who develop new products for those applications, and they need foundries, and if they need specialty technologies, they come to X-FAB.
Now, there's another trend, or also China is particular, is that in the rest of the world, in the mixed-signal world, there is a lot of IDMs. In China, the market, the business model is foundry and fabless, so there is virtually no IDM in China, and that means there is for X-FAB a very big market. There is also competition, of course, and that's X-FAB, with all our technologies, we're serving these needs. Let me also explain about one particularity because I get very often the questions, "Yeah, what about all the new fabs that are built in China?" Well, there is in fact if you see this on this slide on the different phases of the supply chain or development of a semiconductor product from process to IC sales—
In the cooperation model where X-FAB is running, where we're developing technologies for customers to design products and bring those to the market, that's one business model. There is a lot of the foundries in China, they run a contract manufacturing model. That means they have good factories, they have employees, they can operate the fab, but they don't have a pipeline, they don't have technologies. So they solicit companies around the world and say, "Come to us, we'll produce for you." And that is typically attractive for, like, IDMs who want to set up a supply chain in China. They go to such contract manufacturers and have produced in China. So these are two different models, and the first model, of course, is also much more sticky because it's based on unique technologies of this of the cooperation model, as opposed to this outside sourcing business.
So with this introduction, I would like now to go to the next step and have our three speakers explain about the pillars of strength, and we would like to pass the word now to Jörg, who will start with powering the future.
Yeah, right. Also from my side, welcome to all of you to our Capital Market Day 2025. I'm happy to have all of you here. My name is Jörg Doblaski. I'm the CTO in X-FAB, and as Rudi said, the next twenty-five minutes will be all about power and power efficiency. So how do our technologies help powering the future and help enabling that big mega trend we are all experiencing, which is the electrification of everything? So more and more applications and whole industries are moving away from, let's say, fossil fuels to electrical energy. So there is a huge demand for not only more electric energy, but also more efficiency. This is coming from, let's say, applications like the electrification of transport. Electromobility, for sure, is a very prominent example.
But it's also driven by the need to use more, let's say, renewable energies, wind and solar power, that come online and that need a very efficient energy generation, energy conversion, energy storage, and— Yeah, all of this has to be enabled. And there are new applications like artificial intelligence. You heard from Rudi, the age of AI. This needs data centers, this needs a lot of power and a lot of power efficiency to make that possible. As X-FAB, we have already in the past, I think, made the right choices and invested in the right technologies to serve these markets. And I will explain about two major technologies here, which is our BCD-on-SOI platforms for smart power, I would say, and our gallium nitride and silicon carbide technologies for efficiency and real high power. Let's first move into that first section, smart power BCD-on-SOI.
We are the foundry leader in BCD-on-SOI. So we have believed in that technology already 15 years ago, or even more than that, and believe that this is the right technology for automotive, for industrial, and for medical applications, specifically when you need to handle higher voltages, when you need to handle very robust isolation. Our BCD-on-SOI has built-in features that enable really a much easier design for our customers. So customers can implement the design at the right performance, achieving the right reliability criteria, and they are enabled by the process features to do this in a very area-efficient manner.
So our customers can achieve smaller chips at a better performance compared to bulk CMOS, which is the, let's say, what most other foundries offer as a process architecture. And before I go more into the applications, I think it's worth to explain a bit what is actually the difference between these bulk CMOS and our BCD-on-SOI. So let's look at a bulk CMOS wafer, and as we are microelectronics, we have to maybe cut it and have to zoom in really, and zoom in until we really see the semiconductor devices in the wafer. So this is for sure a schematic, but it just to show the principle. And let's really zoom in until we see an individual device.
So this is a very basic structure, maybe a transistor pair that forms an inverter, maybe some digital analog function, and there will be really millions and millions of these devices on the wafer to build the different functions of the integrated circuit. Now, these different building blocks need to be isolated from each other for their proper function. And in the bulk CMOS, the classical way of doing this is using these deep implanted wells for that. This is, let's say, just a process feature to enable the isolation, and there are multiple disadvantages in this approach. And first of all, there has to be spacing in between those different isolated islands. And depending on the voltage level, the larger the voltage differences, the larger the spacing has to be. So it can really be significant.
Designers can't place any electrical component there. This is just overhead, a wasted space. But this is not the only problem. Those wells are never, let's say, perfectly isolating. There's always a bit of leakage, always a bit of current flowing into the substrate. There's always some coupling through parasitic devices in the substrate. Designers need to find solutions for that, and this is usually expensive in terms of area. And there are issues like ESD, electrostatic discharge, and EMI, electromagnetic interference, where it's pretty hard to design a robust solution in a bulk CMOS process. Also, this costs area, and sometimes the requirements can't even be met. Now, the trick in a BCD-on-SOI, these leaky wells are replaced by what I would call the gold standard for isolation in a semiconductor process, which is oxide.
Oxide-based isolation with a buried oxide towards the wafer and a deep trench in between the different devices. This removes this coupling through parasitics completely. This achieves a much lower leakage, so much easier to design for very low power, and it also enables designers to find much easier and much smaller solutions for proper ESD and EMI performance. But most importantly, you can space all these devices much closer together, which means much smaller chips are possible, much less area is wasted. And this is really what makes it very interesting for our customers.
If we look at the die size shrink potential this application brings, and we just look what our customers achieve, up to 45% smaller chips on the same process node, I have to say, can be achieved when moving from a bulk CMOS to our BCD-on-SOI, which means effectively 1.8 times the number of chips on the wafer. Especially applications where a lot of high voltage is needed or really significant isolation capability between different voltage levels, we can see this really being achieved in that scale. Some examples from automotive here are, for instance, smart motor drivers, smart actuators, or smart LED drivers, where this level of scaling is seen.
Also, battery management systems is a very critical application for achieving the electromobility and also the move to renewable energies. I will speak in detail about that application later on, so this is scaling, but I mentioned there are many other, let's say, goodies that you get when you move to a BCD-on-SOI process, and in many applications, these process features really make the difference, so it's really hard to design those applications in other processes. And to give some examples from medical applications really enabled by BCD-on-SOI, we have a lot of implantables or wearable applications, hearing aids, cochlear implants, pacemakers, for instance, that need really low power as they run on a battery and have to run a very long time, but often they, or usually, they also need some level of high voltage.
This combination, high voltage, low power, is a sweet spot for BCD-on-SOI. There is, for instance, medical ultrasound, where you really have to go very high voltage and still need to sense very low voltage levels. Luigi, in his presentation, will go more in detail about that one. If I look on industrial, there are industrial gate drivers that need a proper ESD protection and the combination with smart logic. There is industrial power conversion, industrial test equipment that benefits from the high voltage and from the isolation, and there's also a nice application. Volker will spend later on a bit time on which is inkjet printers. Really, a tough job to miniaturize that, and a very good application to demonstrate the microsystems approach, all based on BCD-on-SOI.
For now, I would move into battery management systems as really a critical application for EV, but also for, let's say, energy storage. There's a huge growth in the need for battery management system ICs coming from the electrification of transport, electro mobility, and the growth of data, of energy storage, battery energy storage. Already in two years from now, 2027, more than 50% of the cars worldwide will be some level of electric cars, from a mild hybrid to a fully battery electric. They will all need battery management, and in a full electric car, you will see, or you have already, 12-20 of these battery management system ICs integrated, and this is growing while we go forward, so really a large volume. Now, battery energy storage is growing even faster, so there's a huge requirement for more battery storage.
You may all know some area where another storage facility is built. It doesn't even need to be funded anymore, so this really pays off if people want to sell the power to the grid at the time where it makes most sense, and it's really a requirement if we want to make more efficient use of wind and solar. So there's a really significant growth projected over the next years to come. All of this needs battery management. Why does it need that? If you look how a battery pack or a battery that is in the car is assembled, it starts with battery cells, which are not so much different from the cells you may have in your, let's say, notebook, or let's say, such mobile devices.
These cells are, several of those are assembled into modules, modules are assembled into battery packs. Now, if we look at an example, I took the ID.4 here as a, from VW. This car runs on a 350 V battery, with 77 kWh of energy. But still the cells, these are lithium-ion cells, with just 3.65 V at their nominal voltage. Those cells are kind of stacked into modules, 24 cells per module, to reach the right, voltage level for the car, and then those modules are combined in a, in a pack, 12 modules per pack. There will be one battery management system per module, so 12 per car.
For sure, it is of utmost importance that these single cells do not see a higher voltage than this 3.65 V nominal, so they really have to be controlled in the right manner. So let's say the most critical function for the battery management system is protect the battery, the battery cells from being operated outside of their safe operating area. Really critical for safety. No one wants to see an unreliable battery or a battery that is exploding or catching fire or something like that. The other important area, I think as important, is battery optimization. All these many cells in that battery need to be, let's say, operated in a way that their capacity, that their energy that is stored, is used most efficiently.
To make sure also that there is no unnecessary wear out, that the battery life is, well, let's say, as long as it can be, there's no problem with the battery over lifetime. Very important for an EV, that the battery capacity is used efficiently to extend the driving range. Now, while this is important for cars, I would say if you look at such storage facilities, like I took this one as an example from Scotland, this Blackhillock was, for a time, the largest battery storage in Europe, 600 MW. This will be operated 24/7, all year round. At the end, the basic component is the same as in the car battery, so this has to be absolutely reliable all the time. Even more challenging requirements for the battery management system.
As X-FAB, we have been working with leaders in battery management system IC design for more than ten years, and we have been supporting them over their product roadmap for many years. So you see here, let's say, the product roadmap of one of our customers who brought the first product to the market in 2016, with a battery management system managing eight of these single battery cells. And as you see, the trend goes to larger number of cells managed per battery management system IC, which means also higher voltages that need to be managed by that component, and a better isolation capability even. In last year already, we were supplying 10% of the global battery management system, let's say, IC markets share to the industry, serving major car vendors with our technologies.
Now, this is not only automotive. I mentioned storage, so the same customer also supplies to battery storage, not only the big storages I have shown, but also, you know, all these, let's say, storages that are put in the, in as a solar storage in your house, and so on. This all needs battery management systems, and here the trend is very similar, going to larger number of cells, higher voltages. And this is just the start, I would say. There's alternative mobility, which needs battery management. There is the whole new thing, like robotics, all operates on battery, all needs battery management, so really a growth driver. Coming to growth, so how does this relate now to our revenue?
This slide shows our BCD-on-SOI revenue overall, and for sure, BMS is one of the drivers here, but there are many other applications in automotive, in medical, industrial, that show a similar growth path. We had first prototypes in those technologies in 2012, and already last year achieved 14% of our overall revenue in BCD-on-SOI, and this is projected to grow by a factor of 2.5 over the next five years. It's really one of our major volume drivers, BCD-on-SOI for smart power. Now, moving from smart power, BCD-on-SOI, and voltages of maybe 48, 100, 200 V, to really high voltage, several hundred, several thousands, several kV, with silicon carbide and gallium nitride, so efficiency and really high power.
Also here, also in SiC and in GaN, we are a leading foundry, and we are pioneers in both technologies with our silicon carbide technologies in Lubbock and our gallium nitride on silicon in Dresden. And both technologies, SiC and GaN, enable higher power and better efficiency compared to silicon-based solutions, silicon MOSFET, silicon IGBTs, which are still kind of the mainstream technology at these voltage levels. As X-FAB, we have enabled a unique engagement model with fabless companies to open up these technologies for their designs. So this is really making the difference and helping them to compete with IDMs that are very strong in that area. And this helps really our customers to design solutions for automotive, for infrastructure, and other applications like data center.
In the next slides, I will mainly also focus on data center, as this is really a growing application as we see. Data center really has, again, changed the picture a bit in terms of power demand. This is really one of the applications where at least I had not expected that this is growing so fast. So this showing here the data center power demand last year. This reached already the level of France. The whole electrical power in France, the same amount was used for data center worldwide, with AI as a big driver for that. The projection until 2030 is this amount will more than double. This amount of energy just for one application.
And the problem is, I mean, still, with all the advantages that AI brings, this is still a huge carbon footprint. At 2030, we will not yet be fully carbon neutral. So a lot of this is still CO2 emissions, and I think we all have seen announcement, like the one from Microsoft with this, a Three Mile Island old nuclear plant being switched on again for supplying the data centers. I think there is a huge need for not only a fast transition to renewables, but also to really more innovation and power efficiency to make this growth really sustainable. Let's look a bit how to get to sustainable data centers, and we have to keep in mind, data centers nowadays, these need hundreds of MW, up to even gigawatt of power.
This power comes from the grid, from voltage levels, AC one, 10,000 V , 20,000 V, but the AI workloads at the GPU, CPU level, this needs a very stable voltage in the range of one volt. So there's a lot of power conversion in all between, between all these steps, from the very high level down to the one volt, from the grid, or the transformers, the UPSs, the power supply units, and so on and so forth, and while this is still mainly done on silicon with silicon MOSFET, silicon IGBTs, there's a lot of this energy that kind of enters the data center, is lost for conversion, is lost for distribution, is lost in the protection circuits, and these losses are effectively generating heat, so you need more energy, even for cooling of these data centers.
So from the energy that's entering the data center, a large part is going into losses and reducing significantly what is actually reaching the AI workloads at the end. Now, silicon carbide and gallium nitride replace silicon at the different voltage levels where their sweet spot is, these losses can largely been reduced. It's really a significant reduction possible, and the critical parameter why this is possible is on-resistance. For such a power switch, when it's switched on, let's say in a theory, the resistance should be zero, but this is a theoretical switch, which is not physically possible.
Nevertheless, silicon carbide and gallium nitride both bring on resistance significantly down compared to silicon-based solutions, which enable higher power and higher power density, a faster switching, and also smaller form factors compared to silicon, based on the material characteristics of both SiC and GaN. There's really a huge potential by changing the technology here. Now, why are not more, let's say, fabless companies moving into that? Why is it still, let's say, dominated by IDM? The, let's say, one reason behind that we have to understand, to innovate in SiC and GaN, customers need to innovate on the process and on the product.
So in the classical way, what happened, a customer comes with a process, and the foundry in its, let's say, classical way, contract manufacturing way, they would implement the process in the fab for one application, so one process, one product. Next customer comes, next product, next process, and so, and so on and so forth. So this is adding a complexity in the fab, which is very difficult to manage, and it's also slow, I have to say, for each of these products. It's also not really possible to invest in any level of design support to improve, accelerate, and so on, because all different processes and no abstraction layer in between, and it effectively does not really create a scalable approach, so maybe this is one reason why many foundries stay away from that, or not, are not really successful.
Now, what we have done, we have invested in what we call the co-creation model. We take the customer-specific processes and map those to predefined process blocks that we have in X-FAB that still can be parameterized. And this, let's say, process integration kit, that creates that level of abstraction that we then later on need to invest in a bit of design support to make silicon carbide and gallium nitride as easy to design as CMOS. And with that, we can really come to a scaling. We can come to many different applications that can be, let's say, enabled faster with a faster onboarding from bringing the process in, to having the first samples out and having a robust and a stable manufacturing.
And it still keeps the high level of customization also on the process level that is still important to innovate in SiC and GaN. This really makes the difference. And with that, we have seen a very good adoption by our customers. So this chart shows the application areas of SiC and of GaN, compared with the classical silicon-based technologies. And we can see in the different voltage levels for silicon carbide and gallium nitride, all these blue areas or blue dots show applications. We are engaging with customers. We are in the different stages with our customers, from the grid, grid power to wind and solar, charging applications, data center, and the different, let's say, voltage levels, 400 V, 800 V for the EV.
And in a similar manner for gallium nitride, we have applications that benefit from the small, small RDS(on) , low RDS(on) values, and the higher frequency capabilities that are possible for medical sensors, again, data centers, space applications, EV again, and also protection circuits for the grid. So this is really a huge and a fast adoption we are seeing. Now, fast adoption and time to market is one thing, but you also need to achieve a good performance, and this chart is coming back to performance, so leading performance in these SiC and GaN applications. I'm showing you an example from silicon carbide. So this graph shows 1200 V SiC MOSFET. So this is what people would use for an 800 V car, let's say, for the drivetrain.
The y-axis shows this critical parameter on-resistance, so the lower, the better, we have to remember. The blue dots is what our customers can achieve already on our generation two of the SiC platform, and you can see this plays already in the same ballpark as the competitor planar technologies. Now, with generation three, we enabled our customers to bring this on-resistance significantly down, now achieving the smallest form factor for Silicon Carbide, planar MOSFETs, with lowest on-resistance and for best efficiency. It's really, I would say, showing what can be done with that cooperation model together with our customers. Now, I would like to end that part with a customer statement. We are engaging with many different customers in automotive, in infrastructure, and in data center.
And what I can mention is one application out of the data center, of, with one of our clients here. You will all have seen the announcements around the NVIDIA new 800 V architecture. And as X-FAB, we are very proud to support one of the few fabless companies that was nominated for that new architecture, which is Navitas. So, and I'm very happy that they mention us as their supplier for silicon carbide to achieve this. Yeah.
To summarize, powering the future with our technologies, BCD and SOI, gallium nitride, and silicon carbide, we have the right technologies in place in automotive, from the drivetrain to the battery management systems, in infrastructure or industrial applications, from the grid down to the data center, and in many, many other applications, automotive and industrial and medical, and some of those you will later on see and hear about in the following presentations.
So— Thank you very much, Jörg, and good morning to all of you, so I'm very pleased to be here today. My name is Luigi Di Capua. I'm VP, Product Marketing. Today, my presentation will focus on sensing the real world, so a key strategic technology pillar for X-FAB. Sensors and smart sensors are everywhere today, so in automotive to enable safe driving, in medical for healthcare monitoring, in manufacturing for automation, and in consumer and smart devices. Megatrends like, for instance, digital healthcare, electrification of everything, are driving the need for more sensors. Specific sensors require specific technology, and this is why, with time, we have developed a comprehensive technology portfolio for sensing applications. We started with CMOS and MEMS.
We continue to enrich our offering, and we continue to innovate, and today I'm very pleased to introduce our new emerging technology for sensing application. It's called silicon photonics. Sensing the real world is our core business, so we have become a leading supplier for sensing applications by developing or co-developing our technology with some of our lead customers, and this has been highly successful. So as you can see on the chart here, in 2024, our revenue for sensing application accounted for about 51% of X-FAB production revenue, and we expect to further increase this revenue with silicon photonics. Now, I would like to explain you what a sensor is and what it does. So sensors are, in fact, the bridge between the analog world and the digital world.
A sensor is a device that can detect changes in its environment, like speed, acceleration, force, pressure, sound, whatever, and convert them into an electrical signal that can be analyzed and processed. The smart sensors not only detect changes, but also process the data locally before they are transmitted, and smart sensors are really crucial for real-time monitoring. Today, our technology portfolio for sensors can enable a large variety of application, not only in our three core markets, automotive, medical, and industrial, but also across the communication, consumer, and computer sectors. Now, I would like to introduce our technology offering for sensing applications, starting with CMOS. So we have developed four main CMOS technology for sensing applications, so with photodiode, CMOS image sensors, single photon avalanche diodes, SPAD, I will use SPAD in the next slide to make it simple, or magnetic Hall sensors.
Now I will cover all of them briefly in the next slide, and starting with photodiodes. A photodiode is an optical device that convert light energy into an electrical current. A typical application using photodiode is an ambient light sensor. I'm pretty sure that you're all familiar with this application. Ambient light sensors are used in mobile phones to adjust the brightness of the display based on the surrounding conditions. You all know this button. You know when you push it, then when it's in the dark, you know, the display will get darker or brighter if you are under, let's say, sunny conditions. Our CMOS image sensors are widely used in medical imaging, including X-ray systems used in hospital, dental offices, or for computed tomography.
Wafers for X-ray imaging have just a few chips per wafer or even up to only one chip per wafer. So you have a lot of wafers put there on the showroom. You will see that there are many, many chips, you know, on the left there. There are many, many chips on the wafer. For dental, it's a totally different world. For dental X-ray or medical imaging, it's a totally different world. So let me show you an example. So the picture on this slide shows a comparison between two wafers manufactured at X-FAB. So on the right, a wafer for automotive. On the left, a wafer for dental X-ray. On average, a wafer for automotive manufactured at X-FAB is about five thousand chips.
In this case, in this application, this example, in fact, this wafer for dental X-ray has only 24 chips, so really, really huge tiles. Manufacturing large chips is highly complex, and we have developed special capabilities to produce large chips with excellent yield results. Moving on to SPADs. SPADs are highly sensitive optical devices that are capable of detecting and counting single photons. SPADs are used in 3D sensing applications. Today, our SPADs are used in drones, industrial or consumer drones, not military ones, and more precisely, in time-of-flight cameras used in drones. Finally, Hall sensors. Hall sensors enable to send the position and the movement of an object into a system. Our Hall sensors are widely used for automotive applications, and since 2016, we have shipped over 1 million wafers for Hall sensors.
Now, BCD-on-SOI can also enable some sensing application, and I will talk about it soon. Moving on to MEMS technologies. We have been developing MEMS technologies for sensing application for over twenty-five years. MEMS technologies can enable additional sensing that a CMOS technology alone cannot achieve. For instance, sensing motion, acceleration, speed, or pressure. MEMS sensors can be either a discrete solutions, or they can be combined with CMOS chips for smart sensing solutions. Our MEMS technology portfolio address four main types of sensors: so pressure, inertial, gas and flow sensors, and temperature sensors. So I will show some application on the next slide. Our MEMS technologies are widely used in automotive, for instance, in tire pressure monitoring systems. So you will see one product over there on the showroom.
They are also used, for instance, in for airbag control units or to control the air intake into vehicles, and our technologies can also be used in consumer applications, so to stay in line with the trend of wellness, our technology can, for instance, be used in smartwatches to measure body temperature. Now, I would like to introduce Silicon Photonics, so a new emerging technology for sensing applications. So what is Silicon Photonics? So Silicon Photonics is a technology that uses silicon as a platform to engineer and process light, to transport information, or for sensing. At X-FAB, we have been developing Silicon Photonics for more than five years, and we plan to be in mass production in 2028. Our ambition is really to become the European foundry of choice in this field. Currently, we have three technologies under development.
So one technology is mainly intended for biosensing application, a technology transfer from imec, Belgium, and a technology co-development with our partner, LIGENTEC, a Swiss company. The pictures shows on this slide shows an example of structure used to process the light. It's called a waveguide, sorry. And in fact these are four pictures, but in fact so those three ones are just zoomed in from the first one on the top left. Now, I would like to show you that real work is ongoing. It's not just development. So this slide show the main application enabled by silicon photonics. Application highlighted in blue colors are ongoing customer projects.
We expect the main revenue for silicon photonics to come from the two major markets, which are on the left, so datacom, AI and telecom, and quantum. The five other markets address sensing applications. So just to give you some figures. So last year, our prototyping revenue for silicon photonics was about 4% of our total production, total prototyping revenue, and we expect to see this figure to increase to 6% this year. Last year, we shipped 60 prototypes for silicon photonics out of roughly 1,000, let's say, prototypes at the group level, and 50% of this revenue for silicon photonics prototypes were coming from quantum computing applications. Okay, so now I would like to spend a moment to showcase two examples.
The first example, sorry, is a current application, medical ultrasound. I think Jörg just mentioned about it briefly, and now we will deep dive into it. Medical ultrasound is a current application and a business success story for X-FAB. The second example is an application which looks to the future. It's called embodied AI robots, and this is a trend that will create a high opportunity growth for X-FAB. Let's start with medical ultrasound. How do we contribute to medical ultrasound? You just heard from Jörg about our BCD-on-SOI success story in automotive. Our BCD-on-SOI technology is also very well suited for medical ultrasound applications. Why is that?
As you can see from the picture on the left, during a scan, the transmission of the sound waves requires a very high voltage, up to 200 V or even higher. In contrast, the received echo signal is very tiny, just a few mV or even less. So thanks to excellent noise and high voltage isolation, our BCD-on-SOI technology enabled the integration of both the transmitter and the receiver in one chip solution, allowing to reduce the cost and the size of the full system. Medical ultrasound is a business success story for X-FAB. We started this business more than 10 years ago. In 2012, we shipped the first 180-nm BCD-on-SOI prototypes.
Now, we are developing and promoting the next generation, so 110 nm BCD-on-SOI technology, and we expect to ship the first prototype by end of this year. In 2025, the projected revenue for medical ultrasound will account for about 25% of our total medical revenue, and we expect this figure to almost double in the next five years. So why is this business sustainable? So we keep innovating to enable the evolution of the medical ultrasound market, and many of our customers already selected our 110 nm BCD-on-SOI technology for the next generation product. So as you can see from the slide, the evolution of the medical ultrasound market is clearly moving to miniaturization to enable more portability for point-of-care application or home use applications.
Our BCD-on-SOI technology, 110 nm, has already been selected by some customers to be used for new applications, like the ultrasound patches, so the picture on the right, and this new application will drive further revenue growth for X-FAB. Let me switch to the second example, an application of the future, embodied AI robots. The future of robotics will see the rise of humanoids or embodied AI robots that will have a physical presence and directly interact with the real world through sensors and actuators. An actuator is a device that convert energy into motion. You already start to see, in Japan, for instance, of those robots used as companion robots to care for older people, and companies like Tesla or Amazon are really early adopters of humanoid robots. Why are we so excited about AI robot?
The rise of and evolution of AI robots will drive the need for more advanced sensors and actuators, and this will create an opportunity to grow for X-FAB. Morgan Stanley carried out a bill of material analysis on the Tesla Optimus robot, generation two, and quantified the number of sensors and actuators used in this robot. As you can see from the picture, and we just here, let's say picture or redraw just for the upper part, I can tell you, the bottom part is as complex as the top one. You can see from this picture that the Tesla robot is highly sophisticated and is full of sensors and actuators, so about 100 sensors and about 40 actuators. Our growth opportunity will not just be limited to sensors.
I just talked about sensors up to now, but in fact, X-FAB is also a major supplier for actuators, and in 2024, our revenue for actuators accounted for about 14% of our total production revenue. Finally, our credibility, reputation, and expertise position X-FAB as really a trusted supplier for AI robots. Now, I would like to summarize my presentation with three key points. Sensing the real world is our core business, representing about 50% of our total production revenue in 2024, and we expect to see the same figure, and match the same figure in 2025. We continue to drive innovation in sensing technologies, and silicon photonics is now a new emerging technology for sensing applications. We are at the cutting edge of enabling the development of emerging trends, and a trend like AI robots will expand our market reach.
Now, with that, I would like to hand over to Volker, that will talk about our next strategic technology pillar. Thank you very much for your attention.
Thank you, Luigi, for your nice introduction. Before I dive in, in my topic, the next level of integration, I'd like to take you back a brief moment about 10 years ago. Ten years ago, I was running product marketing at X-FAB, and I spotted an opportunity. There was a global sales workshop, and I was organizing some presentations about a topic of adding a few, but high-value MEMS layers on top of an X-FAB CMOS. Back then, 0.6, 0.35, 180 nm. Since then, exactly three things happened. A year later— Oh, by the way, I framed this. I said, "We are putting fairy dust on top of that." Fairy dust. Yeah, I had small kids back then, yeah, and they liked it. Back then, exactly three things happened.
A year later, Rudi asked me and told me, "This is a great idea. Go and get it done. Run the MEMS business. Take the resources." Second, this concept got traction with an X-FAB, or even better, traction with our customers, and that's why I'm standing now in front of you. And this concept, now, the third thing is the concept got traction in the industry. You're all aware of that. I mean, and that's the story what I'm gonna tell you today. That's my story. My name is Volker, Volker Herbig, and I'm running the Microsystems business at X-FAB. Welcome. Advances in the semiconductor industry are happening in two direction. One is called More Moore , and this is about a race, a race to ever smaller geometries.
Right now, the industry is at 1.4 nm, and just a handful of companies is able to follow this race: Intel, Samsung, TSMC, you know them. The other track, it's called More Moore, and this is the integration of features into a seamless process. And features, these are passive devices, these are non-volatile memory devices, high voltage, up to 3,400 volt, lasers, biochemical sensors, photonics devices. This is obviously the track where X-FAB is on. But now, there has been a new playing field emerged between those tracks. This is about integrating best-in-class features, and best-in-class means no compromises at all, means high-performance computing tiles, chiplets. It's called chiplets. You heard about that. It's about integrating high bandwidth memories. These are chiplets, computing chiplets, and the memories, these are the AI accelerators done by NVIDIA, for example.
It's about integration of lasers. It's about the integration of sensors. And the integration is done in a 3D fashion. It's going up vertically, and it's either done on wafers or on substrates, and it's called doing either a system-on-a-chip or a system-in-a-package. And what we see now, that we experience significant market pull from our customers. They're pulling us in this direction. We are engaging with existing and new customers, and also, I mean, this has traction in the industry, and now we have chosen, actively chosen to play in this arena. Okay, let me dive in a little bit deeper what exactly we do. It's all about wafer-level chip-scale microsystem. What does it mean? Imagine that you don't cut a wafer into pieces, and there was 5,000, 7,000 individual pieces, and don't build a system.
That's what the industry does, building up systems, individually with each of those cut ICs and devices. What is happening now, that we are building with 3D integration methods, a system on a, on this very wafer, and that's why it's called, wafer- level. And the size of the system is more or less the size of the chip, and that's why it's called chip scale. And here you have a glimpse what type of features we are integrating. We are integrating, platinum or, or gold layers. We have a microfluidic layer. We integrating lasers, modulators. We are integrating, TSVs, through silicon vias, to have a backside contact. Yeah? This is the things we are integrating, and again, no compromises means also we need to integrate a lot of new materials, which provide a no-compromise approach to certain, performance envelopes.
Why does it matter? We can do much, much, much smaller systems, and I show you, and you can see it by your own eyes, some examples. Systems are also thinner. Those systems have a higher performance. Higher performance in many instances translates into lower power consumption. You have a lower system cost. Also, the wafer with the CMOS, plus the 3D integration, is quite expensive for our customers, and the number of parts in the end to do such a system will go down significantly. That means you achieve a higher reliability. I already talked about this no-compromise approach, integrating best-in-class features. Now, we all talk about silicon. The semiconductor industry is based on silicon. It's a good material, however, for certain applications, silicon is not the best material. You are familiar with, we heard today already, Jörg was talking about silicon carbide, gallium nitride.
Those materials push the envelope for power. But now, what you need to understand, getting those materials into a semiconductor fab comes with a price, and this is really a nightmare for the fab because these new materials are coming on smaller wafer, sometimes larger wafer, thinner, thicker, more brittle. They are transparent. Processing sometimes takes longer, and it's getting worse. Those wafer could contaminate a tool. No, they could contaminate a whole fab. You need a dedicated fab. That means introducing such a new material is a big deal, and you just do it when you get a significant benefit, if you achieve a breakthrough performance increase, and that's what we do in Microsystems as well now.
In my introduction, I talked about that there is market pull, and the best market pull you can think about, that customers come to you and ask you, "Can you do that?" By the way, it's much better than they ask, "How much is it?" If the question is, how much is it, the first question. I mean, they come to us and ask, "Can you do that?" And there's another evidence for a market momentum for Microsystems. You're all familiar with the European Chips Act. 43 billion EUR are set aside for subsidies, and 3.6 of those 43 billion EUR are set aside for Pillar One, which is, in essence, pilot lines for the research and technology organizations across Europe. There are five pilot lines. One is called NanoIC. This is imec, chasing the ever-smaller geometries.
One is for wide bandgap, silicon carbide, gallium nitride. One is for, it's called FAMES. This is about fully depleted SOI. This is the Leti ST cluster in Grenoble. There is a photonics pilot line in Spain, and there is a pilot line in Germany, where I'm really excited about it. Yeah, in Germany, there is a cluster of RTOs, 15 institutes, Fraunhofer, Leibniz, scattered across Germany, working on semiconductor topics, and they are getting funded EUR 730 million to implement technologies for 3D integration, for chiplet architectures, for heterogeneous integration, and system in a package. And I told you that we started to work on those technologies much earlier because development timelines in the semiconductor industry are long. And we have exactly those 3D integration technologies, chiplets architectures, heterogeneous integration, advanced packaging. We have a head start on those.
And compared to the RTO community, where the output are demonstrators, that means they show, in principle, it's possible, and they can do a wafer or two wafers or ten wafers or a couple of chips, and show, in principle, you can implement such a technology. We create, for our customers, high-volume supply chains. And this is really exciting. I mean, this creates a huge business opportunity for X-FAB, and this is exactly the reason why we are teaming up with the Fraunhofer in Germany right now. Talking about supply chains, we heard about the co-creation model, and this is the typical model in the microsystem space. It creates a lot of sense, value for us and for the customer. The customer can innovate in two dimension on the product, our translation is design, but he also can innovate on the process axis, on the process vector. Yeah?
It gives him two directions where he can differentiate in a highly competitive semiconductor market. For us, working with those customers creates a significant amount of commitment, and we can measure the commitment in NRE dollars. Yeah? We can measure the commitment in commitments to help us to finance the new capabilities we need to put in place. But there is also an intangible benefit of working with our customers. Yeah? They provide us deep insights in their application, in their market, and this is helping us in the medium and long run. This is the co-creation model we have heard Rudi and Jörg talking about it. This is where the technology is conceived. But we heard we create supply chain, and what is the meaning of a supply chain? We have this technology, conceived together with the customer.
Then we need to make wafers, not only one, not only 10, we need to make thousands of wafers per month. We need to have a high volume capability. We deliver those wafers on time, according to a schedule we agree with the customer. Every Monday, we deliver the wafers, every Tuesday, whatever we agree with the customer. We deliver according to a quality specification. Yeah? We commit to quality, and it has commercial or financial implications if we don't deliver quality, it hurts us. We agree on a price roadmap, or we agree on price, on a price roadmap in many instances, means we need to have our costs under control. And we agree to do this, in essence, forever. And forever means five years, 10 years, 15 years, 20 years. This is the true meaning of what X-FAB does.
We are a tech company, and we develop supply chains for the technology we either develop on our own or with our customer. This is our true purpose, developing those supply chains. Now, we talked about the why, the what, and the how a little bit. Now, let's dive in and try to understand how the business impact is of what we are doing, why we are so excited. I talked about this at the beginning, that we are applying those few but high-value MEMS layers on top of a CMOS. From a marketing perspective, what it is, it's a reframing. Yeah? We are drawing a new frame around our business. We are using the very same technologies, we're reusing them, which we use for a pressure sensor, for a microphone, for an accelerometer, for a thermopile.
We are reusing those technology and pushing the envelope a bit to do microsystems, to do 3D integration. And as a result, we have, at least by an order of 10, magnitude of order, increase of the total addressable market. We are able to engage more customers and other customers, and we dive in much more applications outside of the MEMS space. And this is the good thing, I mean, by adding those few but high-value layers on top, we are able to increase the ASP significantly. Let's take a look at that in a moment. From a revenue perspective, that's what happens if you do that. We have—
We show, or we have proven that we can increase revenue, or we have increased revenue by about 20% compound annual growth rate over the last years. Looking in our pipeline, seeing our pipeline, we estimate that we can maintain this growth at least for the next five years, at least for the next five years. Therefore, microsystems business will be one of the growth engines for the X-FAB Group. Talking about ASP, this is the concept. This is the concept. We have a CMOS wafer, and adding a MEMS layer, adding a system integration layer, adding a heterogeneous system integration layer on top of that. This will increase the ASP up to four times. Let's have a look at the real-world example, two real-world examples. One is an optical sensor, and we have, we have it over there.
This is a contactless temperature sensor, medical-grade. It's based on the 0.35 CMOS right now, and by adding just a MEMS release layer, we were able to increase the wafer price by 30%. By having those release layer, but by adding a packaging layer, by adding a TSV, by having a dicing process where we open up some pads, we are able to increase the wafer price up to 3.3x. And it's a win-win situation. I invite you to take a look at those dies over there. I mean, if you don't have a good eyesight, you won't see them anymore. Yeah, it's an expensive wafer, but there are so many dies on the wafer, the overall system price is significantly low. Yeah? It's a win-win situation for us and for our customer.
Second example is a medical sensor, where by adding a platinum pad and an interface layer, two main steps, we can in essence double the wafer price for a 0.18 process we do in Kuching. Let me comment. This is obviously a marketing slide, right? And let me do three comments. I mean, it's a marketing slide. For me, it has the added benefit of being true. Everything what is in there, I believe in it. But I just wanna make two comments for you here. We talked about the new applications, and Luigi was talking about silicon-based photonics. Why is this so important? Why, why, why are we so optimistic about silicon-based photonics? Our claim to fame, our.
What, what will set us apart, and we are developing this right now, is what we call a heterogeneous integration layer on top of a silicon photonic circuit coming out of our Corbeil, coming out of Kuching. We heard about the X-FAB offering, but also from sources outside of X-FAB, perhaps from imec, perhaps from Tower, perhaps from Silterra. And why is this so important? In particular, in the photonics, the performance envelope for silicon-based photonics is maxed out. The next step in performance can only be achieved by system integration, by integrating new generation of photodiodes, high-performance photodiodes, by integrating quantum dot light sources, by integrating laser or VCSELs, by integrating modulators based on lithium niobate, integrating modulators based on BTO, barium titanate. Only this heterogeneous integration layer will enable silicon photonics to go to the next level.
And there are just a few companies in the world actually who are able to set up the high-volume supply chains for those technologies. This is a race that's going on right now, and just a few companies in the world are able to set up the high-volume supply chains for those technologies. And X-FAB, my team, is part of that. The second comment I want to make is, I really embrace the approach from Jeff Bezos for Amazon. He told his shareholders, his team, "It is still day one," even if Amazon was really huge. What is behind? This is the enjoyment of innovation, working on cool stuff, working with our customers, absolutely customer-centric, and in an extremely agile environment. That is what is meant. It's still day one. And I keep telling my team, day one needs to, and will remain, for a very long time.
Now, I talked about the exciting stuff we are doing. I have brought four application examples with me I wanna show. Two of those I showed and mentioned two years ago, if you had a chance to see my presentation there, but the difference between now and two years ago is that two years ago, those applications were in an NRE stage, in the development stage. Now they are in high-volume production, and make a difference for my top line and for my bottom line, and two applications are in development. Let's start with this automotive headlight application. Coming, by the way, out of the Fraunhofer environment, got the German Future Award, which is quite prestigious, by the way, in Germany, and this is about an automotive headlight. I'm not sure who has a newer car.
In those newer cars, you have typically an LED matrix headlight with 16, 32, 64 LEDs. Here, you integrate gallium nitride LED array on top of an X-FAB CMOS wafers. And this gallium nitride LED array has 26,000 LEDs. T his is not a headlight anymore. This is a projector, and you have two headlights. You have at least two projectors then in your car, and you can project guidance for your car at night, you can project traffic lights. If it's not, would be forbidden, you can project a movie, a black and white movie. I mean, it's highly regulated, this market. It's not allowed, but you can do a lot of fancy things. Yeah? This is high volume business, it's one of my largest customer now. Super exciting.
The other one is, again, a wafer-level chip-scale microsystem for infrared sensor. This is the contactless body temperature measurement. And here, and I invite you to check this out over there, what we have done, by creating a microsystem, we were able to bring the volume of the sensor down by almost two orders of magnitudes. You see the first generation and this, and the second generation here. The third generation is over there, and you can barely see it anymore. Yeah? And we increased the performance, and what it does is now enabling in your personal health devices, whether it's a wristband, or it's a watch, or it's integrated in a phone, to enable a body core temperature. Body core temperature is a super important vital sign.
If you're in a hospital or if you go with your kids to a doctor, the first thing they do, they measure the body core temperature, the body temperature. Yeah. Having this at your wrist, having this on your phone, it's not available yet, but it's super important. Yeah? What does it do? It enables for us a huge untapped market application space for us. Yeah? Yeah. Again, all the features, the benefits are on the slide. I mean, even it's a super expensive wafer, we lowered the system cost for our customer. Next application is a wafer level chip scale microsystem for inkjet. Why is this so interesting? Many things I have talked about come together in this application. It needs to start with a BCD-on-SOI wafer. Jörg has described this technology, and we use a voltage above 200 V.
Then we put a microfluidic layer on top of it. We integrate an actuator based on a new material, aluminum nitride. And we have some MEMS processes to integrate that. Why does it matter? For our customer, it creates a high resolution printhead, much higher resolution. The next one, it sounds not impressive for somebody who is not an inkjet guy. I mean, this inkjet will be capable printing everything on anything. It's not a bubble jet, where you heat up the ink and it's exploding, then you're pushing out a droplet. You have some limitations there, just water-based inks. Here, you can print almost independent from the physical properties of the ink, anything on any substrate. The inkjet guys are super excited about that. Yeah? It reduces system cost massively, and for us, again, it's an expansion in an untapped market, long lifetime market.
It's also some sort of consumables. Inkjet printheads need to be replaced. It's super exciting for us from a technology perspective, from a business perspective. Last application I wanna show, it's about wafer-level chip-scale microsystem for photon-counting computerized tomography. It's a CT, right? It's a CT. And here we are, it's also got the German Future Award, Deutscher Zukunftspreis. Our customer is Siemens Healthineers. I'm allowed to mention the name. And we are working with Siemens Healthineers on the next generation of this, of CT. It's about photon-counting computerized tomography, and what it is, it has two main features. It's increasing the resolution by a magnitude of order. It means you can go from a more or less black and or we can go from a black and white picture to a 4K color TV experience.
It makes a huge difference for the medical community. Everybody wants to have this. Everybody wants to have this, and the whole CT world will migrate to this technology in the next ten years. This is gonna happen, and this migration will accelerate. And on top of that, there is a market expansion happening for our customer, because this technology can now be used for screening. So far, if you had an indication you needed to have a CT, there needed to be an indication, because you got an X-ray exposure, and this is something you don't wanna have too much. This technology brings down the exposure dose to 30%, a 70% decrease of exposure dose, and this enables screening. If you have a problem with your lung, do a CT screening. It's impossible right now. This is the technology. If you have a problem with your heart
Do a CT with this technology. Yeah, it's super exciting, and again, long lifetime business, medical business, we are super excited about that. Now, let's close out. I mean, I mentioned that there are significant public investment in microsystems in Europe, across Europe, which creates a significant, a huge business opportunity for us. By reframing our MEMS business, using the technologies of our MEMS business, and expanding a bit on that, we could expand our total addressable market by an order of magnitude. Come on. I foresee that based on our current development pipeline and production pipeline, and I gave you a glimpse, we will maintain our growth rate. We will maintain a 20% growth rate year-over-year, over the next five years. We started early, quite early, and we believe that this gives us a leg up in terms of competition.
We also have the infrastructure in place for this already, those dedicated clean rooms, and tools, and so on. We have already achieved a robust profitability, and by scaling this up, this will improve significantly. NVIDIA, TSMC, is my guidance there, by the way. I love that very much, and we have a lot of good applications we are solving. This really makes us awake at night. I mean, solving the application, generating purpose. It's still day one. Thank you very much for your attention.
I hope this video gave you a taste of what is X-FAB about, what we do, why we do it. My name is Damien Macq. I'm the COO of X-FAB, and I want to give you an update on our capacity expansion that we announced here, like, a bit more than two years ago. So we came here to talk about a $1 billion investment plan in the different factories of X-FAB, and my ambition today is to explain you how we deploy this plan, and how successful we are with offering the capacity that our customer want to see. So I start with a slide presented by Volker earlier today. So Volker explained that X-FAB is capable to produce, and to build, and develop supply chains. If you look at this slide, you can connect most of the dots to commitment from the operation team.
It's commitment about managing the diversity of process that we have there. It's commitment about innovation. It's commitment about long lifetime support and quality, and commitment about price and delivery on time. The investment plan that we spoke about two years ago, and that we have implemented since then, talks about the delivering of enough capacity at the right level of flexibility. So moving forward, what I can say, as I said in my introduction, the $1 billion investment is done. Basically, we have installed in our factories all the equipment that were supposed to be installed and hooked up. We have built from ground up, new factory in Kuching. We will talk about it in a few minutes, and we have built also additional clean room in existing factory, like in Lubbock and in Dresden. Now, this is done.
What is still to be done is the qualification of the tools. So, the tools are qualified one after the other. I will also explain the dynamics there, and once the tools are qualified, the capacity in terms of output wafers per month becomes available. So by Q3 2026, the full capacity of the $1 billion expansion plan will be available, but as of today, the whole investment that we wanted to have on our factory and our tools is done. Now, we talk about flexibility, and there was an interesting question this morning as well. Flexibility is about flexibility on the site. Flexibility on the site means once we want to enable a new technology, we have a park of tools running, for example, a mature CMOS process.
By adding a set of new tools, just a few new tools, we can utilize this fully depreciated set of equipment and offer new technologies capability. So our investment is not about replacing old tools by new tools. Our investment, in this case, is about adding capability by adding specific tools that are missing here and there. So that creates flexibility, because that allows us also to manage the transition between some technologies that are somehow ending their life, and some technology that we see are more attractive for our customer, and so attractive for us. That's the flexibility that we have on the site, and I will come back on this. Besides this, there is also flexibility between sites, and there we talk about second sources. Second source was a feature that was a kind of luxury in the past.
When you have high-runner products, people wanted to have the possibility to, you know, manage their supply chain by sourcing this high runner from different factories. What we see now more and more is that customer want second source for any product. Why is that the situation? Why, you can read the press, the geopolitical tensions are exacerbating the instability in the supply chain, and we want to make sure that we can offer a solution for this instability. So sometimes you will see that we have dual source in Asia and in Europe, and that is one level of confidence. And this needs to be complemented by the statement made by Rudi earlier, that for many of our product, we are single source.
So if we don't deliver, the machine cannot be built, the car cannot be built, and the medical equipment will not be available for customers. So flexibility within the site, flexibility between the sites, quite important. Now, we will continue to improve our operational performance. This doesn't require a lot of CapEx, but requires some CapEx. And here we talk about automation. So automation, I will present one slide on where we stand on our automation journey. This has been a motto for the past few years, making sure that we can automate our process to get the maximum utilization of our assets. And last but not least, we manage actively our legacy processes. So every process has a lifetime to some extent, and this lifetime needs to be properly managed.
We can observe that for some processes that are really there since 20 years, some suppliers stop to supply essential features or essential chemicals, and then we have to manage the end of life of this process. We announced a year ago, more than a year ago, we announced already the end of life of our 0.6-micron process in Kuching, in Lubbock and in Erfurt.
Erfurt. Thank you. And this management is combined with a push also for our customer to help them to prepare their next product, preferably in its fab technology, but also to prepare enough material. So we see a kind of a big increase of the volume in this technology to stock, to store the necessary equipment, and then the transition can take place. So this active management help us also in terms of cost. It helps us also in terms of preserving customers, because in general, we want to make sure that our customers have solutions for this end of life. Let's now zoom into the process that we have used to share the investment across site.
I think you have seen this morning a lot of technical and marketing presentations about the different area where we believe we have growth opportunity. So you can see on this table, on the vertical side, the different technologies, the 180 nm, 110 nm. This is the fundamental technology that is used also by microsystem and photonics. So any investment in 180 nm or 110 nm helps also microsystem and helps photonics. And we discussed also this morning about GaN and SiC. Each of these technology is connected to one or two sites, in general, two sites, but for the emerging stuff, some like a power device, it could be one site. And you've seen also this morning, how the technology pillars in powering, in sensing and miniaturization is served by our technology.
So we try to combine all these things together. This is a bit of a symbolic view on where the investment is going, but you can see that the share of the $1 billion, the lion's share of the $1 billion investment goes to our 180-nm and 110-nm. And just to repeat myself, the investment that we do there is helping also microsystem, is helping photonics, but it's also helping the customers that today utilize the technology as is. So in numbers, you saw the map in the movie, and you have the same map of all our different sites. You can see where the technology is produced, depending on the technology we spoke about. So you see microsystem in Itzehoe and Erfurt.
You see the silicon carbide in Texas, and the two big fabs doing CMOS and BCD, also on SOI, in Kuching, Malaysia, Corbeil-Essonnes. Now, the amount we represented there, this amount of dollar allow us to create capacity, and it's really capacity corridors in terms of wafer out per month, so WOPM on this chart means wafer out per month, and you see at the end of the investment program, how many extra wafers we can produce per month on every site. To give you an order of magnitude, we will produce by the end of 2026, 50% additional wafer versus the start of the program, and we will add 50% more value to these wafers, because these wafers become more complex, the process are more complex, so this two times 50% is, for us, an opportunity to create additional value.
Now, where are we on this journey? This is shown on this slide. So you see that the execution is done, is over on certain site, like in Dresden and Itzehoe. On other site, like in Kuching, for example, we barely started to create additional capacity. And the reason for that is that on certain site, we were just installing an additional equipment, and we get the capacity once this equipment is qualified. While in Kuching, you have seen in the movie, we started from ground up. We had to build first, a new clean room, and then we started to install equipment on the clean room, and now we are qualifying equipment. So the capacity in, in, Kuching will become gradually available once the whole equipments are fully qualified. Let's stay in Kuching for a second.
So, if you look at the Kuching site, we installed six thousand square meter of additional clean room. I'm really proud to say that the team in Kuching did a tremendous job. I mean, if you look at what it takes in general to build a new fab from ground up and to produce material in this factory, the two years that is there is really a top leading edge. I would say they did really a fantastic job. Already mentioned, later this week, in fact, for us, later today, we are flying to Kuching to congratulate our teams for the performance that they achieved there. So if you go there, you saw the new clean room in the movie.
It's really fantastic, and in fact, they are running production material in this clean room already since March this year. So gradually, this will become available. It's also important to mention that so there, all the tools are connected and hooked up, so basically we have paid the bill, or we are paying the last bill for Alba. Maybe, we are getting close to that. The clean room is larger. We installed enough clean room for fifty thousand wafer out per month on this site. But in terms of equipment, we limit ourselves to forty thousand wafer out per month. And there, the flexibility comes into the picture. You see that we offer flexibility for the 350 nm, and we offer flexibility for the 180 nm between Kuching and the fab in Dresden, and Kuching and the fab in Corbeil-Essonnes.
So this is really a differentiating element for many customers for the reason I explained earlier. I wanted to give a bit more details about the tools installation. Why does it take so long to have really the capacity? Why does it take years to get what we want? I explained in Kuching, we had to build a new clean room from ground up in two years. And you see that from Q3 2024, we started to move equipment. It was not really in the new building, it was in an existing building, where we also get some additional clean room. But you see that the bulk of tool installation and in production is really starting to pick up from the beginning of this year. You can see also there are two curves there.
One curve is the plan, the other curve is the execution. In fact, you don't see the difference between the two curve. The team has done really a fantastic job in executing the deployment of this tool on the factory. And if you look at our capacity, you can take the same curve or similar curve and shift it by the few months that it takes for the material to move to the new tools, because we count capacity in wafer out per month. Let's now go into France. France was a bit of a challenging situation over the past few years. So basically in France, we achieved, as of, you know, maybe one year ago, the full transformation on the site from a consumer site to a site delivering core technologies for its fab.
More than 95% of the technology delivered from France are now made out of its fab technologies. The transformation on this site is a bit more challenging than what I presented for Kuching, because there, every new piece of equipment has to be installed in a factory that is fully loaded. So basically, you have equipment in the factory, and you have to insert a new piece of equipment in between, where you see space, where you have opportunity. So it's a pretty challenging environment, and this was aggravated by the crisis that we had in the semiconductor for automotive. So between 2021, 2023, we were really at the peak of the crisis, and so there were constantly challenges between the priority on qualification of new tool or the priority on production.
I wanted to show a similar curve as the one that was presented earlier. You see here two zones. There is one zone where you see a difference between the plan and the execution, and it's in red, and it's the place where we had some challenges, because we had to make choice. It's not that the people did a lousy job, but we made a choice, the conscious choice, to focus on production. That means the qualification is delayed, and you can see on this curve, there was up to a quarter delay from time to time, and that, during an allocation situation, can be very problematic. Since then, we completely recovered. You see the top of the chart in blue, and you see there that we are now ahead of the plan.
Basically, the whole plan was executed on time, but clearly, we have seen two phases: a challenging phase till mid-2024 somehow, and another phase where now the site is delivering according to the plan and has the capacity that will allow us to move forward. Now, on the site, we did other things. Basically, we installed on the site our 110-nm BCD-on-SOI. That's the technology that was already announced. This was developed globally, but mostly only installed on this site in France, and this is now in full production. We have presented this morning also the photonics. Photonics is essentially today a Corbeil-Essonnes story.
So, we moved to install there a technology from imec, and, on the site, we have in fact LIGENTEC, which is our partner in photonics. They are nearby and are working in close partnership with our production team and our product development team to develop the next photonics technologies. The fab automation is a focus in France. I think it's a focus for all the fabs, so I will repeat that again and again. But, you know, this is something that is essential for us to get really the most capacity out of our existing assets. And the journey in France started a bit late, so we have really to accelerate the automation in France. In terms of flexibility, we have also flexibility. So the 180-nm from France is available in Kuching and vice versa.
So that gives us really the opportunity for critical products to offer dual source for our customers. The overall capacity, if I combine the 110, 180 nms from Kuching and from France, you see that from 2022 to 2026, we multiply this capacity by nearly three, so it's a substantial increase of capacity. And I repeat myself regarding the utilization of this capacity. So it's not only about providing a light sensor or light drivers or things like that, it's also some wafers that will go, that will receive some photonic additional step, or that will also be utilized for some of the process that Volker described this morning. So basically, the growth that is there will be eaten by some of the sensor product, will be eaten by some of the power product.
We talked about battery management this morning, and will be eaten by microsystems. In terms of revenue contribution, if you look at this process overall, they utilize in the range of 42% of our revenue. So basically, it's really the main piece of our technologies moving forward. And within that piece, the BCD over SOI, which is one element, presented this morning by Jörg, is the one that is growing the most. Let's go now to Erfurt and Itzehoe. So these are the microsystem sites. Of course, they utilize wafers, CMOS coming from other sites, but they built on top of it, features and capabilities for our customers. So on this site, we invested EUR 55 million CapEx.
On this site, with this CapEx, we managed the transition from 150-mm wafers to 200-mm wafers. Volker spoke about it this morning. The added value that this can generate in terms of value per wafer, but also in terms of value for our customers, because to some extent, you know, there is additional material on every wafer that we sell. In Erfurt, I mentioned we are engaged in the end of life of our 0.6-micron technology. So this is a bit of a challenge because right now we are really running in full steam to allow our customers to prepare the stock that they need to have when the technology will be over.
We are helping also the same customer to migrate to our more advanced process, like 0.35 or 0.18. As presented by Volker this morning, the capabilities that we have on the site will continue to be enhanced. It's in cooperation with RTOs to prepare proof of concept. Once we have something that works, we buy the equipment, we put it in Erfurt. All these are analyzed case by case, looking at ROI, looking at what makes sense, and proceeding like that. Also in terms of investment for this equipment, in nearly all the cases, the customer is co-investing in the machine.
So basically, the customer, as Volker explained, comes, "Can you do that?" And then we say, "Yes," and then, you know, the customer is ready to pay for us to do this. So the co-investment goes together with co-operation or co-design with customers. So in terms of capacity, you see that the growth is relatively modest, but you have to realize that behind the wafer that we produced in 2022 and the wafers that we produce in 2026, there is a world of differences. The value of every wafer that we produce, Volker explained this morning, is multiplied by the fact that we can add additional layers on top of each wafer. So the slide tells you where we are with this investment. We will have 60% additional wafers.
In terms of revenue contribution, we are in the range of 10%-12% today on microsystems, but Volker explained we have a double-digit, more than 20% growth in this area. So a lot of ambitions, a lot of trust also in the fact that we can execute this growth with the product that were presented this morning by Volker. Let's now go to Texas and later to Dresden. So these are the two sites where we develop power devices. Texas started quite earlier, so by adding a few machines to an existing technology park, we can develop and offer silicon carbide technology, which means compared to people who want to invest from scratch in silicon carbide, the majority of our tool is depreciated, and we add just a set of equipment to create this silicon carbide capacity.
So it was a great deal for us in terms of ROI, because we could position ourselves as a leading factory in silicon carbide. As a matter of fact, we are the only factory in the U.S. that is delivering silicon carbide, and this attract these days a lot of focus and attention. Cannot disclose the details on this, but we see that the location in the U.S. is definitely an asset that will help us for the future. On this site, also presented this morning by Jörg, we have developed also next generation of our products. So Jörg presented the added value of the third generation of silicon carbide. So it's the technology that has today the best-in-class RDS(on).
So this is a technology that this is a level that we can reach because we work hand-in-hand with multiple customers, and we take the best out of it to offer a generic flavor of this process to our customers. New customers can go much faster in the integration and utilization of this performance. In terms of capacity, these are the numbers. You see also some comments about the fact that the capacity today is not fully utilized. In fact, compared to what was announced in June 2023, we have delayed some of the investment that were linked to silicon carbide, so we were more ambitious in the past on silicon carbide.
I think everybody knows that SiC has seen a quite challenging past year, and right now, at this moment, we saw really from our side a bottom situation earlier this year, and since then, we see gradual recovery of our SiC business. So for us, we adjust our investment to the business, but we have, at this moment, enough capacity. We can produce 10,000 wafers per month, and we utilize a bit more than 3,000 wafers per month. So we believe that with what we have in hand in terms of CapEx, we are safe for a few years to come.
Just to mention, there is an opportunity there to utilize the money from the CHIPS Act, so we went quite far in the approval of a CHIPS Act file to the U.S. government. But right now, we are kind of pausing these whole efforts. We want to see how the SiC market will continue to evolve. We see good indications. We don't see yet the hockey stick that was mentioned earlier. It's not yet NVIDIA from our perspective, but the press release that were announced this morning give us also a lot of hope that we will see some significant growth to come on the silicon carbide. So we are ready for what's next. Yeah?
Dresden, last site, where similarly to Lubbock, we created also an opportunity to have flexibility between a CMOS process that we produce there, the 350-nm, and the GaN process. The GaN process is a bit less mature than the SiC process today in X-FAB. We have some customers who are working with us since already many years to develop and come to the right level of performance on our GaN technology. And you see that with the flexibility that we have there, we can activate corridors of GaN by exchanging the capacity to between GaN and 350-nm. So this flexibility is absolutely an asset for us.
Same story, we add a set of tools that allow us to prepare the future of GaN technology on the site, and we utilize the asset that were there present for the 350-nm. Capacity in Dresden, so we have there a factory that will be able to produce a 200-mm GaN, while TSMC was still in a 150-mm. And in terms of capacity, if you look at it in a full 350-nm node, we can go from 10 to 11. But as I said, we can utilize the flexibility we have between 350-nm and GaN to produce the capacity that to offer the capacity that the market wants to see there.
So if you combine now, what we do in Dresden and what we have in Lubbock, I think Jörg presented the value added of GaN and SiC. They are complementary technology, and they profile really X-FAB as a supplier, a foundry supplier for this type of technology. So this is positioning X-FAB in a very strong, very strongly on the market. Important to mention, I mentioned automotive, sorry, automation. Automation is a critical item, for us, so for our factory. I presented already this slide earlier, but for us, it's a journey. So basically, you see on this chart the different levels that a company could reach in terms of automation. You have seen in the movie some robots moving in a factory. This is not yet in X-FAB, these robots that you saw moving stuff.
It's not with us. But we are getting there. So basically, we see so much progress, so fast innovation in robotics, that we believe we have to keep monitoring what's happening there, because this will come faster than we think. But let's start with level 1 . Level 1 is about making sure we automate the next move of our material in the factory. Level 2 is about making sure all the equipments are automatically set up for the material that will come in front of them. So these two levels are really the key focus for all our factories. We want to make sure that in all the factories, the systems decide on what should go next to which tool, and making sure that each tool is automatically set up. The technologies exist. The deployment of this technology—
Can be here and there challenging because we are deploying this on legacy factories, and so what we want to make sure is that from every site, there is enough synergies in terms of baseline set of tools that are utilized from one site to the next. We don't want every site to come with their own solution. So, that is a process that is ongoing. The right choice on MES, the right choice on scheduler, and these tools allow us to deploy the foundation of automations, which are level 1 and level 2. Right now, if you look at Kuching, for example, they are already at level , and level 3 is about making sure that you have automated transport equipment that will move your material from the place where the tool, the step is finished to the next step.
You have storage places, and you have automated transport between the storage places in the factory. And then the last step, you have something, can be a robot, today it's a human, but can be a robot, can be a machines. If you look at the 300-mm factory, with all the tools nicely aligned, you can utilize very nice machines. In a factory that is fab, you need a bit more versatile robots, but the versatility of robots becomes super cheap now, if you look at the evolution of robotics. But from our perspective right now, level 1 , level 2 is the priority everywhere. In conclusion, I wanted to demonstrate that we are ready for the next step. We are ready for what's next.
So I'm really glad to say that this $1 billion investment plan that we announced here looked a bit ambitious compared to the size of the company. Would say a company that is not yet doing $1 billion revenue, why the hell do they invest $1 billion in CapEx? The reality is that we have done it. We have done it because we believe the markets are there, the demand is there for the technologies that we need, and by mixing some of the technologies, by utilizing also the flexibility that we have within a factory, we can add value for every wafer that we deliver.
So based on that, we believe we are now ready with our technology, especially, if you look at BCD, BCD-on-SOI is a very demanded and promising technology, but also in microsystems and wide bandgap, silicon carbide or GaN, we have the capacity we need. Mission accomplished there. No need to further invest. We are ready for a few years to come. Moving forward, in microsystem, we'll continue to explore all the opportunity that will come in our way. So it's quite important to mention and to remind that this exploration comes with customers who are ready to invest in capabilities in our facilities. And they come often with a proof of concept that they have tested in RTOs, et cetera.
And so for us, this gradual increase of capability is making us also stronger and stronger as a specialized foundry. So this is something that will continue, of course, at a level that is affordable. And moving forward, we'll keep focusing on our operational efficiency, so automation, the foundation of automation, level one, level two for sure, level three when it makes sense, when the ROI is there, and, you know, let's build further foundations. I want to thank you for your attention. I want to thank you for coming also from far away in this meeting, and with this, I conclude my presentation. Thank you.
Thank you, Damien. I hope you all enjoyed all the presentations that were done so far. Last but not least, yeah, I'm Alba Morganti, CFO of the group of X-FAB. I will now, for the next 20 minutes or X amount of slides, talk about how all you have seen so far translates into financials. Since we met at our last capital market update in December last year, we have continued to work hard, and we increased our capacities and capabilities, as you have seen in Damien's presentation. We have come a long way. As you can see in this graph, our revenue consistently grew through the various cycles with a compounded annual growth rate of 10.4% until last year.
Alongside this top-line growth, together with our strong operational leverage, we were able to also steadily improve our profitability. This is particularly visible in the down cycle, in where we were at 6% in 2019, and where we are today at 23% of EBITDA margin, and if we take the midpoint of the annual guidance, we should at least reach 25%. Now, what's next? We are all set for top-line growth. You have seen in different presentation how this will translates into future growth. In-house capacities and capabilities are there to grow our business to an annual revenue of $1.2 billion. Our core business reached a stable share of more than 90% of total revenue, and going forward, we will further increase the share of even more complex, higher value-add business. We are all set for long-term growth.
However, the short term is more uncertain. Geopolitics, tariffs, a change in our customers' ordering behavior, and ongoing de-stocking is reducing our visibility. Inventory corrections are taking longer than expected as a result of stockpiling that was partially related to the long-term agreements volumes customers had to commit to. But these long-term agreements will expire by end of this year, so we are now returning to pre-LTA ordering behavior that is more closely related to short-term needs. All this considered, how does the $1.2 billion revenue that we can achieve with our current capacity compare with our long-term outlook from December last year, when we were projecting reaching $1.5 billion? On the one hand, the $1.5 projection included a $100 million of outsourced business, but this is not required anymore at this stage, as we have enough capacity in-house.
Anyhow, we keep the door open for this flexibility in case of need. On the other hand, there is another $200 million deviation related to the SiC business due to two factors. Firstly, the recovery of the SiC business is taking more time, and it's progressing much more slowly than anticipated. And secondly, the proportion of the consigned SiC wafers has increased to more than 90%, compared to the 50% that we budgeted. Of course, this reduces the top-line growth. Now, what are the growth drivers going forward? Let's have a look now at our key technologies we invested in, how they play in our strengths, and build our strategic pillars to deliver the future growth of the group. On this slide, you can see how our technologies will contribute to in terms of revenues. Let's start with CMOS. We expected our—
We expect our 180 and 110 nm CMOS technology to grow on an average in a range of 7%-9% until the end of this decade. As you have seen from also Volker's presentation, our microsystem, and then our SiC revenue presented also today, both are projected to grow on an average of approximately 20% in the period from 2025 to 2030. Overall, our larger node CMOS business is expected to remain flat, because on the one hand, we expect a CAGR of 3% for the 350 nm CMOS business, whereas the CMOS business with geometries higher than 350 nm will decrease. This is mainly due to the termination of some 0.6 micron CMOS technologies in 2027.
The end-of-life scenario for these technologies has been boosting demand, as explained by Damien, as customers were filling inventories to secure supply. In 2025, this portion of our business will come in at roughly $90 million, and then it will phase out in 2027. The affected sites, being Erfurt in Germany and Lubbock in Texas, are both in a transition phase. Erfurt will focus on X-FAB Microsystems business, mainly explained by Volker, and Lubbock will focus more and more in the SiC business going forward. Today, Dresden has two businesses: the 350 nm CMOS business, which will grow more slowly, and the GaN business, which will further grow, even if it's still on an early stage, and will become on stream. We only recently announced. Sorry, I'm fighting with my mouse. We will.
We only recently announced the introduction of GaN on-silicon foundry services at our platform in Dresden. However, after looking at the growth drivers and how the different portions of our business are expected to evolve, let's have a look at our profitability. At X-FAB, we have four key drivers. First of all, the top-line growth and the increase of utilization. Our business is characterized by a strong operational leverage, as you can see on the next slide. And now let's visualize where our strong operational leverages come from. As you can see, and already explained, our fixed costs count for two-thirds of our total costs, which gives us a strong operational leverage, especially with the important additional capacity we set up in the last years.
In other words, by increasing capacity in the still high demanded products, we planted the seeds not only for future growth, but also for future improvement of our profitability. For every additional dollar sales, it will translates into $0.60 going directly to the bottom line. Secondly, our profitability will be driven by economies of scale from the increased capacity at our existing sites. These effects will become visible over time as more capacity will be online. We also aim to improve our efficiency through our automation initiatives, largely explained also by Damien. While it's not yet possible to quantify these savings, as we are still in the process of implementing them, in the long term, they will help to increase productivity and output without additional investment or headcounts.
At this stage, it's all about cost control and reduced spendings, and we are now at the point where the major CapEx program is finally behind us. As you can see in this graph, the major CapEx program has been completed, and going forward, we will return to a CapEx ratio of 10%-15% of annual sales, so I'm partially already responding to a previous question. Whereas CapEx in the next few years, starting in 2026, will be at a low of 10% before it will return to a normalized level of 15% of our revenue. CapEx is equally split into maintenance CapEx and capacity CapEx, and in addition, there will be leverage with CHIPS Act fundings, as explained already, other fundings, but also some customer prepayments for specific projects.
Such a major CapEx expansion was driving significant financial needs, which we have been able to limit to about $375 million of new credit lines. Also, thanks to the about $300 million prepayments we received from customers through the LTAs, and also thanks to the $425 million of operational cash flow. In other words, despite the high level of CapEx, we kept a good balance between in our financial structure, using from financial leverages but not abusing from them. Let's now have a look at our financial liabilities. On this slide, you can see the evolution of the net debt in dark blue, and the financial liabilities from the prepayments we have received from our customers under LTAs in light blue. The net—
The net debt includes our two revolving credit facilities, each of an amount of 200 million EUR, which were required to support our CapEx expansion, and which are now not still fully utilized, and some also lease and sale and leaseback we also put in place recently. The first credit facility will expire in 2026, and the second in 2029. We don't expect to repay these credit facilities before maturity date. Actually, both have the option of a one-year extension, which we might use, and which provides us even with more flexibility. Of course, should we decide to repay them earlier, of course we will do. We still have the option.
In terms of the customer prepayments under long-term agreements, we have already started to repay a small part of them, and, at the end of the second quarter, liabilities related to the customer prepayment were amounting to $276 million, which means that the portion of $16 million has been already repaid. The vast majority will be repaid from now on until 2027. It's a steady program. While we have a large amount of debt outstanding, we are in a good position to manage the repayment program of both the credit lines and the customer prepayments. Our business is naturally hedged. Revenues and cost match in terms of currency splits, as shown in this slide, with the revenues and cost by currency for the year 2024.
Therefore, exchange rate fluctuation do not impact our EBITDA, which is extremely important to protect our margins, in particular, in light of the weakness of the fluctuation of the euro/dollar as you have seen already, we significantly increased our bank loans, both in US dollar and in euro, to support all the CapEx you have seen. As our balance sheet is stated in US dollar, we need to reevaluate our euro-denominated debt, which has an impact on our financial result. Due to this reevaluation of euro, the euro-denominated debt, we reported an unrealized foreign exchange loss of EUR 17 million in the second quarter this year. However, it's important to note that these are unrealized non-cash items, and we partially also hedged this risk via US dollar/euro hedging contracts.
Any risk on the short term has then been hedged, while we are still having some flexibility for the long term. Let me conclude this section by sharing our financial targets. Where are we today? 2025 is performing above expectation, which is good news. Our guidance for 2025 remains unchanged, with revenue in the range of $840 million-$870 million, and an EBITDA margin between 24% and 27%. However, given the, what is going on around us, with global turmoil, economic crisis, the near term is looking more uncertain than we could have anticipated. Even if the long term is looking good, we have decided that we will not guide towards specific numbers, as we prefer to remain cautious. Our long-term ambition is to achieve an average revenue growth across cycles in the high single-digit range. We also want to grow our profitability.
Our EBITDA margin is expected to gradually increase to more than 30% over the long term, driven by our strong operational leverage, increasing utilization, and economies of scale. We invested a significant amount of money. Not doing so would have been a mistake, as already explained by my colleagues, because we would have lost our customers, and we would have also lost market shares. So we are very well pleased with how we have set ourselves up for future growth. Clearly, we are at a challenging point in this cycle, but we have made a strategic decision to focus on our key technology pillars, all of which have a high value, and we believe we will deliver the growth for our group.
In my eight years journey with X-FAB, I have seen a lot of market cycles already and faced quite a nice number of challenges. But I have the feeling that we did the right thing so far, and we have prepared ourselves for future growth, allowing our customers to grow as well, and we remain a strong, stable, and committed group. My role as CFO is to make sure that we remain financially, stable and strong also during down cycles, and, also that we remain an employer of choice for our teams, and to enable our group improving quality and security of life by producing terrific products you have seen in many other presentations today, life-saving devices, such as chips, enabling, for example, liquid biopsies.
It's teamwork to get there, so I take this opportunity to thank our teams, but also our banks, our investors, shareholders, customers, suppliers, for contributing to our successful journey, and of course, to all of you, being here, for your interest and for your presence, here today. And by this, I pass the word back to Rudi.
Thank you, Alba. So what you heard today is that, as X-FAB, we're mainly focusing on two mega trends: the global warming and electrification of everything, the aging population, and improving medical care, but also service and so forth for the elderly people. And this falls very nicely with, on the one hand, the markets we know very well: automotive, industrial, and medical. These are markets that we are familiar with; we know how to play with. We have a reputation. Our customers know us. They recognize us for what we're doing in these markets. And then we have the three pillars that were presented today. These are our strengths.
We want to play further on these and invest where we are at our best, and that these three things, they fall fantastically together for the mega trends, the end markets, and our strengths. I feel very strongly about the future, and what I see is that also our customers, they recognize that. They come to X-FAB for these specialty technologies, the service, and the customization and the co-creation that we do. So what we see finally, we have good demand and interest for what we are doing, and that gives a lot of confidence in the future. Damien explained and we are very glad we took the decision to invest in capacity because we were in allocation. It was really hindering our business development.
Now we are ready for growth, and when there is a downturn, we're actually in a downturn, we are ready for growth. So when it turns around, we're ready to respond immediately and grow with our customers, but also widely with all the new applications that we're working on. We also talked about the geopolitics as the number one European foundry. We are well-placed in the whole geopolitics. We're somewhat seen as a neutral to supply all geographies in the world. We're progressing well in automotive applications in China, where the electric vehicles growth is the strongest in the world, and so we have the right technologies to serve that. And the ecosystem there is in need of specialty foundry technologies.
All these elements fit very well together, and Alba explained when the growth is coming. Our CapEx program has come to an end. Our spending will drop significantly. The leverage, so for every dollar of extra revenue that we generate, $0.60 falls through to the bottom line. We're very well-placed for the future. And with this, I would like to thank everyone.