ASM International NV (ASM) Earnings Call Transcript & Summary

Good afternoon, ladies and gentlemen. I'm Victor Bareño, Head of Investor Relations at ASMI. It's my pleasure to welcome you to our Investor Day. Thank you for taking the time to be here with us today, both to our guests here in Amsterdam and to our investors and other stakeholders on the webcast. We are very pleased that we -- that it was possible for us to organize this event in a hybrid format. Our in-person meeting here in Amsterdam is in accordance with the COVID measures of the Dutch authorities, and we provided for additional safety measures such as a meeting room set up with a 1.5-meter distancing. Let me now walk you through the agenda and introduce you to the senior management team members who will be presenting today. We have split the program in 2 sessions. The first session will be opened by Benjamin Loh, our CEO, who will talk about our opportunities, the unique positioning of ASM and our growth strategy. Next, our technology -- Chief Technology Officer, Ivo Raaijmakers, who will present our R&D strength and the exciting new technologies on the future road map. Ivo will be followed by Han Westendorp, our Corporate Vice President of Corporate Marketing, who will talk about the outlook for our key markets. After these presentations, we will have a 25-minute break. The second part will then be kicked off by Hichem M'Saad, Executive Vice President of Global Products, who will provide an update on innovation in ASM products. Next, our CFO, Paul Verhagen, will give a financial update and talk about our targets. After we wrap up by our CEO, we will have a Q&A session of 45 minutes in which our management will be happy to take questions from both the audience here in the room and from the conference call. We will have one combined Q&A with the whole panel. So if you have questions during the presentations, please make a note and save them for the Q&A session at the end. Before handing over to Benjamin, let me remind you of the forward-looking statements. All of the presentations and the Q&A today may contain information relating to ASM's future business and results in addition to historical information. And please also note today's presentations are webcasted. A recording will be available after the event. And with that, I'll leave the floor to Benjamin Loh, CEO of ASMI.

Thank you, Victor. I hope the mic and everything is working well. I want to start off by saying, repeating what Victor has just said, a very warm welcome to our Investor Day. I understand there's quite some anticipation about what's going to be shown. And we hope that with the information that we have put together, we will be able to share with you more about the company. We have a lot of information actually that we're going to present to you over the next 2, 2.5 hours. But I'd like to, first of all, start off with a quick video. This is a new corporate video that we have put together. And this is -- we are showing this in public for the first time. [Presentation]

I hope that was good for everybody. And I'm going to start off with the first presentation, growth through innovation. And in this presentation, I'm going to talk about a couple of things. One is what do we see as the market and the opportunities are in front of us, where are the strengths that ASM has. And finally, how do we convert the opportunities that we have in front of us into growth for the company. And before jumping into it, there's a couple of key messages that we would like to convey to everybody. And we hope that this will be the key takeaways that you will bring with you at the end of the session. One is, of course, the strong digitalization growth which is based on the secular trends that we are seeing that is driving rapidly increasing usage of semiconductors and is going to lead to growth in the WFE market. Second point is, we are the ALD leader, and we are fast expanding in epitaxy. Both of these technologies that are very critical, they will outgrow the WFE market, but at the same time, they are enabling technologies for further advances in semiconductors. Our strengths are in our innovation, in our early customer engagements and our ability to differentiate our products, leading to the most cost-effective solutions for our customers. And in terms of strategic objectives, we want to maintain our leading position in ALD, in logic and foundry. We want to expand in memory. In terms of epi, we want to increase our market share actually fairly significantly as we will show you later on. We want to grow our spares and business -- service business. And at the same time, we want to increase and accelerate our focus on sustainability as we have done with the statement that we have issued this morning in our press release. All of this is going to lead us to a target growth where we are expecting to reach in terms of revenue, between EUR 2.8 billion to EUR 3.4 billion by 2025, which represents a CAGR of 16% to 21%, while achieving an operating margin of 26% to 31%. Before jumping into the first topic of market opportunities, I just want to maybe do a quick recap of the key highlights over the last couple of years where are we coming from? So in terms of innovation, we continue to do that. And we have always been at the forefront of advancing a Moore's law. We are the leader for ALD. In fact, we have achieved a share of about 55% according to our own estimates. We increased our share in epi from low single digits to about 15% at the end of 2020. Expanded manufacturing capacity, first of all, 2018 in Korea and at the end of last year in Singapore. We have grown our revenue over the last 5 years at a 22% CAGR, resulting in revenue per headcount increase of 44%, improved operating margins by about 11 percentage points. We continue to focus on our R&D and the buildup of IP. Our IP portfolio has increased by about 40-plus percent. And finally, we did our part in terms of sustainability -- small part. We reduced greenhouse gas emissions by about 15%, and we're going to do more as we have announced this morning. So market and opportunities, what do we see in terms of the market over the next couple of years? And how does this translate into, for example, addressable markets for ASMI? First of all, I think we all know, digital transformation is going to drive significant increase in the usage of semiconductors. And with that, there will be quite some additional capacity that needs to be added. The megatrends like 5G, electric vehicles, autonomous vehicles, AI and so on, I think are all well known to us. In terms of WFE spending, this is going to translate into additional capacity that needs to be added to make up for the increase in usage that is required. Secondly -- and I will show this a little bit later on, there will also be technology inflections that are going to drive additional capacity. And last but not least, there could be sovereign government efforts that might also add further investments. But that's something that is still in the works, nothing has been fully confirmed yet. Another way of looking at how the increase in WFE is going to look like is to look at the difference between the advanced nodes and the less advanced nodes. And we see that a lot of the investments will be in the advanced nodes. In fact, that is structurally the fastest-growing part of the market, which is the 7-nanometer and below. And that plays to our strengths because ASMI as a company plays mostly in the leading edge, especially with our leading-edge ALD and epi technologies. In terms of the trends, we look at major inflections coming, whether it's FinFET to gate-all-around, whether it's further 3D stacking, whether it's new materials and one which has already actually taken place and is happening, high-k metal gate in the DRAM periphery. We have always shared with you that all the 3 major DRAM manufacturers are already using our ALD solutions for this particular application in high-volume manufacturing. And of course, even though it's a little bit further out, planar transition from -- for DRAM into 3D DRAM. New materials, et cetera, this again will be covered in more details later on. So we are very well positioned to address these trends. Which leads us to where or how the addressable markets are going to evolve. And this is where we have done our projections. So for the single-wafer ALD market, we expect that it is going to grow from around EUR 1.5 billion at the end of 2020 to about $3.1 billion to $3.7 billion at the end of 2025. Now in terms of market share, we -- based on our own analysis, we have a 55% market share at the end of 2020. And we aim or target to grow our share so that we will be greater than 55% by the end of 2025. For epi, we think it's going to grow from $0.8 billion to somewhere between $1.5 billion to $1.8 billion. And we are going to increase our market share from about 15% today, in fact, to double it to greater than 30%. So that's where we see the addressable market size changing for us and how we want to address and how we target ourselves in terms of market share in these key markets for us. Next, we move on to strength and innovation. Innovation has always been at the heart of ASMI. And I will share with you a couple of our key strengths as we go to the next couple of slides. Strengths. So we are very focused. We only focus on deposition, primarily, especially in ALD and epi. So we do not have many different, let's say, distracting kind of business. Our focus really is on deposition technology, especially in ALD and in epi. We have a strong track record, a history of innovation, and my colleagues will later on show you what we have done over the past. And this is going to continue to help us be the innovative leader in the areas where the technologies are most needed, which is ALD and epi. We will continue to lead with innovative products, innovative solutions and applications. And that's something that we have continued to do over the last 5, 6 years or maybe even longer. And early customer engagements, which are really, really important for us to start working at a very early stage with our customers to make sure that we have the right products in place when they go into high-volume manufacturing. Now what has that really given us over the 2016 to 2020 time frame? We have maintained our position as the market leader for single-wafer ALD. In fact, we grew our market share during the period by a couple of percentage points. We expanded our market position in epi. And compared to the overall WFE market, we have grown 1.5x faster than the market itself. History of innovation. We have a highly globalized network R&D model. We work with a lot of partners, and the partners include customers. As I said earlier, some of the development work that we do is actually done with the customers. And in terms of the technical capabilities that we have, especially in materials, in chemistries, we have more than 2 decades of developing that, and we have built up and accumulated a lot of experience. At the same time, we have accumulated a lot of experience in ALD, epi, reactors and processes, and we will continue to focus on innovation going forward. Leading products and applications is a little bit of -- a lot of tax here. But in terms of ALD, the key thing that stands out is really we have the broadest portfolio of products and applications. And my colleague will show you later on what this means. And all this is leading us to be able to offer our customers the best performing films that comes from our ALD with the best cost of ownership. For epi, we have been very innovative in terms of the processes, the hardware. And one of the things that stands out for us is when you look at the newer, let's say, epi applications that require very stringent control of monolayers, we are able to do that probably better than anyone else. And in fact, that is really critical when you go into the channel structure for the new gate-all-around transistors. And we shared a couple of months ago that we have already been selected by one of the leading logic/foundry players as the production tool of record. So when they go into high-volume manufacturing, they will be using our solution. And of course, last but not least, the surface clean capabilities that are becoming more and more important for both ALD and epi. And again, this will be further explained to you at a -- later on. Early customer engagement, I would say that we are probably working with all the major logic/foundry and memory customers. And some of these engagements are for, let's say, processes or solutions that will be only used 3 to 4 years down the road, but we are very early there to build and to develop solutions for them. To develop the critical answers that they need for the forthcoming inflections and their manufacturing requirements. So with all the opportunities that we have, with the strength that we have, how are we going to translate that into growth for our company? And that comes our strategy, which is the last part of my presentation. So first of all, let's look at what are our strategic objectives. We want to maintain our leading ALD position in logic/foundry, and we want to expand in memory. We want to increase our share in epi and actually, we have a fairly aggressive target to even double that. We will selectively grow our vertical furnace and PECVD product lines in areas where we have differentiation. So we play in the niches. We are going to grow our spares and service business because it offers us another avenue of significant growth going forward. Now with all the business activities going on, we also want to accelerate our progress in sustainability, and the first step that we have taken today was in the press release this morning where we are committing to net zero by 2035. Last but not least, we will continue to drive strong financial performance going forward. So these are the strategic objectives, what about the how? We're going to invest in R&D, and I will show you later on what is going on, especially in ALD and in epi, the 2 fastest-growing segments of the market. We want to make sure that we continue to engage with our customers early and align to their road maps so that the solutions that we are going to create for them will become high-volume manufacturing adoptions later on. We want to strengthen our culture and develop our people. People are our biggest assets, and we have a firm belief that, that is something that we need to do and focus on. We also want to accelerate our commitment to sustainability, as I've described earlier. We have a growing installed base that we want to maximize the potential. And at the same time, we think that we can grow our spares and service business also by introducing more outcome-based services, which we have just started last year. Last but not least, the expansion and optimization of our manufacturing capacity and supply chain is, of course, important. And we have done part of that last year, and we will be doing more of that as we go forward. Expansion in R&D capacity. We have already been expanding in Japan, that's in progress. And we will continue to expand in Helsinki in U.S.A. and in South Korea, not just in terms of headcount, but also in terms of infrastructure, such as metrology tools, lab development tools. Leading with innovative products and applications. On the left, you see our ALD product that was launched in 2018 that has now become the industry workhorse. And on the right, my colleague will share a little bit more innovations in hardware in process that enables even more stringent epi applications. Strengthening culture and developing our people. So we want to strengthen our culture based on our core values of, we grow (sic) [ we care ], we innovate, we deliver, at the same time, develop our people to take on bigger responsibilities, increase our engagement and also increase inclusion and diversity. And in terms of sustainability, we are going to accelerate our progress. And as you have seen in the press release this morning, we are committing to a very stringent or, let's say, aggressive challenging targets, and we will continue to do this driving forward. Growing our spares and service, I already explained, we are going to optimize and maximize from our installed base, growing installed base and at the same time, transform some of our offerings from a transactional to an outcome-based offering. And in terms of expanding our manufacturing capacity, we moved into the new facility in Singapore end of last year, and it's been a real good timing for us. At the same time, we have also started work to start the design on the spare floor that we have with an intention that by 2023, early 2023, that flow will also be ready for manufacturing. In terms of our supply chain, optimizing our supply base and also increasing our multiple sourcing. So to end my presentation, I want to just maybe go through again the 5 key messages that we would really like to convey to you. One, the strong digitalization growth, driving a stronger semiconductor growth and increases in WFE. ASMI being the market leader for ALD and fast expanding in epi to enabling technologies for future technologies. Our strengths in innovation, in early customer engagements and product differentiation, allowing us to bring to market, to our customers most cost-effective solutions. Our strategic objectives of trying to maintain our leading market share in ALD and in epi -- sorry, in ALD and expanding in memory and wanting to expand our market share further in epi, growing our spares and service business and increasing our focus on sustainability. And last but not least, financially to reach a revenue target of between EUR 2.8 billion to EUR 3.4 billion by 2025 with an operating margin of 26% to 31%. So that's the first part or the end of my presentation, and I'm going to hand you over now to Ivo Raaijmakers, our Chief Technology Officer.

Thank you, Benjamin. Good afternoon, ladies and gentlemen. Good afternoon, good evening, good morning also to the audience online, whenever time zone they are. Driving innovation. That's a title of this presentation, and that's also what ASM is about. So I, as the CTO of ASM, will tell you how and what about is driving innovation. Following are the key takeaways from this presentation. Global R&D network, already mentioned by Benjamin. An R&D organization, driving innovation, more than 2 decades of ALD experience and more than 3 decades at epi, I should add. Our current R&D portfolio is very well aligned with what our customers need and what our customers need in the future to extend Moore's law. And then finally, I will address some very important inflections in the industry, which will lead to ALD and epi opportunities. The first point, I will show you again a small video about our R&D organization explaining the how. [Presentation]

Well, that was a nice little video how we do our R&D. You see here kind of a summary. The big balls you see there are our R&D sites where in the U.S.A., in the Netherlands, in Japan and Korea, we have our product development centers. And then in Finland and Belgium, that's most important for our R&D and materials innovation. That's where we have a big effort on really leading the industry. And those sites are co-located with important R&D partners, University of Helsinki in Finland and IMEC in Belgium. Just a few numbers on our R&D effort. More than EUR 170 million spending in 2020, which is about 13% R&D intensity, more than 600 permanent R&D employees of 26 different nationalities, and of which 44% have an advanced degree. More than 2,000 patents in force already mentioned by Benjamin. And this is just a number, 2,000. But if you would look to the strength of these patents, they are very, very strong. Independent analysis by innography and other engines show that we have one of the strongest patent portfolios in the area of ALD. So let me tell you a little bit about driving innovation and aligning our R&D effort with the customer road maps. Little bit looking backwards in the past, a few of the things where we have lead the industry in adopting change. 5 things here summarized. Low-k dielectrics. You may have the best low-k material, already 8 generation, successive generations in mass production in the industry. Selective epi, [ source train ] this is one of the first selective technologies in the industry going into production. ASM did it with a customer. And this was for strain, which was the bigger of them. High-k metal gate already mentioned in the video. What's more important to realize is that high-k metal gate, if that would not have been done with ALD, there would be no FinFET. This is essential for FinFET. Further ALD use in patterning. At that time, EUV was not ready. We were ready to step in with ALD and do double patterning spaces, blocking layers and so on and so on. So we extended immersion until EUV was mature enough. And then last, but not least, ALD for gap fill and low temperature silicon nitride, a little bit more recent. A lot of new technologies developed with our PEALD process. So we will continue to do this, driving new materials, driving new processes. So how do we align with our customers? This shows a few high-level transitions aligned with our customers. The top line is N, N+1, N+2, N+3, N+4, those are technology nodes, roughly every 2 years, some customers move a little bit faster, some are -- move a little bit slower. This is about the pace in the industry, continuing till 2030. Three important transitions, planar DRAM to 3D DRAM, I will explain you more in a few subsequent slides. 3D NAND is going from single to multi-tier 3D NAND. So already the 3D transition has happened there, and now we're stacking it together. FinFET scaling from 5 to 3 nanometer. And then transitioning to gate-all-around for about 2-nanometer technology node and later to a so-called fourth sheet gate-all-around, which is a small modification for the FinFET. So our current R&D. Well, our R&D really spans this whole spectrum, from supporting HVM to the external R&D and scouting. The center of gravity of our spending is in the N+1, N+2 regions where mostly Hichem will tell you a lot more about in the pilot production, development, PTOR and DTOR selection. But at all times, we make sure that we have enough money allocated towards the future node. That's what drives our innovation effort. So putting this graphically. Benjamin already showed these 4 pictures, FinFET to gate-all-around. Now we have to put a gate-all-around the silicon. This is very complex. It's much more complex than doing it in a structure like this. DRAM, high-k metal gate. DRAM periphery will move to high-K and later maybe to FinFET. Planar to 3D NAND already happened and now just stacking and upcoming planar to 3D DRAM. So if I summarize this, almost everything is going to 3D to enable scaling. It's no longer shrinking this way. It's going up and going 3D. And that requires conformer thickness, composition, electrical properties, et cetera, and that requires more ALD. We see a lot more thin films, much thinner, complex stacks, more complex materials. So we need atomically engineered surface clean and preparation technologies. Many new materials and bottom-up selective deposition, I will tell you more about that later, to supplement a more top-down deposition and edge. So these 3D transitions will really build on our strength and drive a lot of new opportunities for ASMI in especially the ALD and epi markets. That's one thing I want you to take away is that the 2 technologies which are most important for the future of semiconductors are ALD and epi. What is ALD? Let me show you that in this very quick animation. You see a surface covered with atoms with bones sticking up. There's a new precursor coming in, attaching to the surface, then a next precursor comes in, a reactant we call it, taking away the stuff we don't want, and we have one layer of atoms formed on the surface. And we do this a number of ways and then we build upper layer. So it's very digital, very controlled and very easy to make these balls all in different colors and different materials. So the big advantage of this technology is shown in this little animation. You see the blue layer grow in each of these 3 animations. For CVD, sometimes the step coverage is not okay. So the layer is thicker on the top than it is in the bottom of the structure. The bottom is almost no blue, top is all blue. Well, you can fix the step coverage, some CVD processes have a very good step coverage, but then it's very difficult if you make a complex material to make the composition or the properties uniform in this structure. So even if the thickness is uniform, the composition is not uniform. So only ALD can do this and can make this structure, which is like a 3D NAND structure to make it uniform step coverage, composition and properties all the same in a very deep, very complex structure. So we can do that with a lot of different materials. This shows the periodic system and about -- sorry, about 50% of the elements will be used in the industry in the coming 10 years. and we need so-called precursors, the chemical development. We need that to make these new materials with ALD with the right qualities. This is where the core strength of our lab in Finland is. We can make these precursors, we can use this for ALD and invent the right process. So let me go to the next section of the presentation, which inflections do we seek and what opportunities do they create? For example, and this is not exhaustive. Four examples I will present: gate-all-around, DRAM, V-NAND scaling -- 3D DRAM and selective ALD. So let me first go to gate-all-around. I already mentioned that the -- we pioneered the transition from 65-nanometer, which was planar to high-K metal gate. That's the second picture here. And then there are just a few ALD layers, maybe 1 or 2. There was some epi also there. The epi for the [ source train ]. Then we went to FinFET. Today, there's already 6 generations of FinFET we support with ALD. And I mentioned earlier, ALD really enabled the FinFET. So there's a lot more ALD and a lot more epi in FinFET. And then if we go to gate-all-around, basically, we need even more ALD and even more epi to make these complex structures. And I'm not going to go in detail here. This shows a process flow. It shows a picture on the left of a so-called fork-sheet. I mentioned gate-all-around is first, and then there is fork-sheet gate-all-around. The only difference basically is the dielectric wall in the middle. So that's another 2 extra ALD steps. But look to the process flow in the middle of this chart, the small bubbles are the ones where we already have ALD and epi implemented, the big bubbles are where we will see a strong expansion of ALD and epi use to make these type of structures. If you put that graphically, planar just one or a few layers, epi and ALD, high-K metal gate calling substantially, FinFET, another growth step and gate-all-around, you see a lot of layers will be to be made with ALD and epi. It's not only that, it's not that we need the same materials. We also need new materials to do that. If you go to gate-all-around using the same materials is not going to suffice. You need to complement this with more materials to form the work functions, et cetera, et cetera. Hichem will tell you a little bit more about that. So this will drive a significant SAM expansion in our ALD and epi markets. Memory, 1 slide about memory here and then going to 3D DRAM. DRAM, basically, the capacitor doesn't scale anymore. That's a problem. V-NAND, also there. We have already gone 3D, but it's very difficult to etch this deep hole. So what our customers do, they just stack 2 of these stacks together, and then we have to fill them. So even though these holes cannot be etched anymore, we can still fill them with ALD, and we can still form the new materials in these much deeper structures. So common between DRAM and 3D NAND is higher aspect ratios. The other thing is if you have these higher aspect ratios and more cells per unit area, you see increased peracetic. Peracetic is a difficult word for higher -- too high resistance and too high capacitance at the places where you don't want it. So these things need to be tackled. Now if you look to DRAM, first to the left side, you see that a large area of the DRAM is consumed by the periphery, the red area in this green, red diagram. So if we tackle that, we can continue to scale the DRAM in surface area. And that's exactly where high-k metal gate comes in, an epi channel in DRAM to further scale this periphery down. New ALD materials are needed to improve the electrical performance for that, better conductors for the word and the bit line, better insulators to create air gaps at low-k insulators to get rid of the capacitance. On the right-hand side, you see multi-tiers. So Tier 1, Tier 2 stacking together and our periphery is being put under the cell to gain surface area. Also here, structures are more complex. We need a better vertical channel material, because the channel is getting longer. It needs to have better conductance, otherwise, you can't reach the bottom of this channel. And it has to have better conductors for the word line and better gap fills I already explained because the gaps are a lot deeper and a lot smaller. 3D DRAM. This is something which is imminent to happen, and I'll tell you why. 3D DRAM is going to follow the same trend as V-NAND. V-NAND went from planar, we turned it upside down. And we went from bottom to top. 3D DRAM, we can do the same, use similar technologies and go from planar to vertical. That allows us further scaling. Why is this going to happen? Two main reasons. One, if you look to a DIMM package, it has a certain size. And you cannot cram more than 64 gigabit of memory in this size. So if you want to go beyond that, then you need to do something different. Second reason is that it's eliminating some expensive EUV steps. If you look to that graph on the right-hand side of this picture, scaling with 2D DRAM and continued use of EUV will indeed drop the cost per bit, which is a very important parameter in the memory market. It will drop the cost a bit, but not as fast. So if you introduce a 3D DRAM, and this picture looks the same if you do for V-NAND, a few years ago, you see this trend going down in cost per bit much more rapidly, starting with 64 stacks, then 96 stacks, then 128 stacks. We expect the transition to be in between 64, 96 stacks. around 25, 26. And what's important is that the channel is going to go here to epi. We will need 64 layers of epi here. And those will not be 64x the number of systems, and we will do that a little bit smarter. But just to say epi will be a very important technology here. And also, since we are going vertical right now, we have more space for the capacitor. And we can start changing the dielectric of the DRAM to form a higher-k material. So that will continue to drive the DRAM road map. Selective ALD, two slides on that. Top down, litho-etch, that's what we call down, you pattern something, you etch out and you throw away what you don't need. That's very inefficient. It's much more efficient to grow things bottom up and grow things on the places where you want them. So that's exactly what selective ALD is about. What this does, it delivers an expanded toolbox for very smart process integration strategies. We can develop that together with our customers. Early engagement is very important here. The expected customer benefits are reduced cost because it potentially eliminates some of the litho-etch steps, and an improved die performance and yield because you can eliminate some of the edge place materials. We can do these 2 different ways, topological selectivity, which means that we, in this case only deposited on the flat surfaces and not on the side walls, only on the flat surfaces, so the red layer in the picture. And surface selectivity is where you deposit it on a certain material and not on another material. So in this case, the red on the orange and not on the gray. Where are we here? We're leading in the adoption of selective ALD. Topo selective silicon nitride is already in production in 3D NAND. And we have a JDP with an important customer for 2-nanometer foundry logic for what we call dielectric on dielectrics. So it's a certain dielectric layer grown on dielectric and not on metal. And that eliminates a lot of edge placement errors. And interesting to say is the technology here, the dielectric on dielectric is developed also with some atomic layer etch steps, which play a critical role in maintaining the selectivity. So to talk with Richard Feynman, there's a lot of room, plenty of room at the bottom. So selective ALD, I think, really will present a very large SAM expansion for us going forward or could present a very large SAM expansion. So summarizing our road map. The white is what you have already seen. The other bars show you where we start R&D and where we expect approximate manufacturing adoption. So epi? Second -- sorry, second bullet here, enabling gate-all-around channel -- instead of 1 epi layer, we need now 3 epi layers, 4 epi layers. So that's going to be very important to form this gate-all-around. Gate-all-around also need multi VTs. So many different metals and many different dielectrics dipole layers, as we call them, to tune to VT. So here, you see that our ALD technology is going from high-k metal gate, enabling FinFET and now enabling gate-all-around with a lot of different materials. We have enabled -- or we have extended immersion until EUV was ready. Now we are capturing layers where we need to support the EUV introduction, because the resistor is going to get a lot thinner and that means we have to drive for new materials also in the patterning steps. I didn't show you this for the sake of time, but I just want to mention that. Metals for memory, new metals, better conductance, surface selective ALD, I already mentioned, topo selective being in production and surface selective coming into production for 2-nanometer 3D DRAM, I explained. And 2 things here, I didn't explain yet. They're a little bit further out, but I told you we're allocating R&D also to the far out nodes. 2D channel gate-all-around or 2D materials. That's a bit confusing because I've talked about 3D all the time. But these are 2D, very flat, very thin materials, which prefer to grow in a 2D way, and we apply them in a 3D structure. So this transition is of similar magnitude as the high-K metal gate transition. We championed that in the industry, and we intend also to play a big role in this next big inflection in the transistor. The bottom, you see new memory materials. And I refer here to different types of memory, which are used for artificial intelligence. We need a different memory action, a different way to remember the bits in a device. And we are working here on several new materials which can enable that. So all in all, if I look to our R&D portfolio, we are targeting a multibillion euro ALD and epi SAM expansion. So again, the key points, I showed you our global R&D network, early R&D collaboration with partners and with customers globally. I showed you our R&D organization, how it drives innovation, how it starts with road mapping, how we accumulated over 2 decades of ALD experience and more than 3 decades of epi, how we align the R&D portfolio with our customers. And I showed you a few examples of inflections where we think we can play a big role in capturing those and helping the industry moving forward with Moore's law. So that brings me to the end of my presentation. And then I would like to introduce Han Westendorp. Corporate VP of Global Marketing, and he will present us the market outlook and opportunities.

Thank you, Ivo. I also want to welcome you very much this afternoon, this morning, in your evening wherever you are in the world to our Investor Day. And like Ivo mentioned, I will address the markets that we're active in. Second topic is I will address our performance in recent years and give you some more background of what drove the outperformance in those years. And then third, I will talk about our growth opportunities based on what Ivo already mentioned and what Hichem will present on products later on. So also in my presentation, some key takeaways I'd like to share in the beginning and recollect at the end for my presentation. So first of all, like Benjamin already mentioned, strong digitalization trends are ongoing, are driving semiconductor sales to reach $1 trillion level by 2030 and also driving a strong WFE market. Second, we grew faster than the market in the last 4 years since 2016, and I'll explain you what drove that. Then third, we released this morning our view in numbers, at least on the single-wafer ALD market size. By 2025, we are coming to an assessment of $3.1 billion to $3.7 billion as the market size that we think that market will reach by 2025. We've done a similar analysis for the epi market, which will, in our view, grow from EUR 0.8 billion in 2020 to somewhere between $1.5 billion to $1.8 billion in 2025. And then finally, we have done an extensive analysis of what gate-all-around that Ivo talked about will mean for the change from advanced logic, the size of the advanced logic market from FinFET -- mostly FinFET-based last year to what will be likely mostly gate-all-around based in 2025. For ALD and epi combined, that's going to mean a market increase of $1.2 billion, and I'll give you some more background on that. So first on the markets. The markets as you know, artificial intelligence, security, many elements, autonomous vehicles are driving a stronger use of semiconductors in general and are expected by 2030 to reach $1 trillion level. So roughly doubling from where it is today. That will drive a strong growth in our view of wafer fab equipment. We're sharing here a VLSI research forecast through 2025, split between logic/foundry and memory. So basically, we -- there is a common consensus that the drive in wafer fab equipment will remain robust because of the strong drive for semiconductors in everything we do in the coming years. Specifically for wafer fab equipment. If you look at the technology nodes, like we have shown also in previous occasions, structurally, we believe that the largest percentage of that growth comes from the more advanced technology nodes. All of the inflections that we have seen in the last 10, 5 years going from planar to FinFET now to gate-all-around is going to drive basically the advancement of semiconductor performance. And also the equipment needed for that will represent structurally the larger portion of growth in wafer fab equipment. Let's now look at our performance in the last 4 years. So I'm showing here on this slide the -- our equipment revenue in 2016 compared to 2020, where a CAGR of 24%, 2016 to 2020 was reached, where at the same time, over that same period, wafer fab equipment CAGR was -- sorry, the deposition in wafer fab equipment markets grew at a CAGR of 15%. So our equipment in the period grew by roughly 1.5x faster than those 2 markets. So the question is what drove that? What drove that stronger growth? In this slide, you see on the left side, the ALD single-wafer market development. So the single-wafer ALD market, we forecasted that in 2016, at the time, it was $0.7 billion, $0.8 billion. We made a forecast at that time, 5 years ago, in the General Shareholder Meeting, that we expected that market to go to $1.5 billion. And sure enough, last year, it reached that level. So compared to the growth of the deposition and WFE market in the same period, that is a faster growth, not only were we participating then and leading in a market that grew faster than the deposition market, but we also gained share in that market because we continue to focus on our strongholds in logic/foundry and expand it in memory. So we basically -- those 2 elements drove the -- were an important factor in outperforming in that period. On the right-hand side, we look at the epitaxy market. The epitaxy market in 2016 was $0.68 billion. It reached a level of -- was $0.6 billion. It reached a level of $0.8 billion in 2020. And that's a relatively low CAGR, and that's only -- it's a little bit lower level because in 2019, actually that market was stronger. We saw a downward period in the 2020 epi market because of a slowdown in analog and power. But in our case, we used those years to actually gain share, to triple our share roughly from 5% share in 2016 to close to 15% in 2020. We introduced, and Hichem will talk more about that, the Intrepid ES product in 2017, and we gained a lot of traction with that product in the advanced CMOS logic epi market, and that was a strong contributor to our tripling our share in that period. So both on the ALD and epi side, we performed well, and that drove in aggregate, our outperformance in those years leading up to 2020. If you now look at our growth opportunities in the coming years. So let me show here first, basically some example applications that are expected to drive our growth. And I've listed here basically a selection of applications where we feel rather confident that based on our current position and whether it is already in high-volume manufacturing or in early development and early engagement with customers that they will contribute these applications quite strongly to our growth in the coming years. So in the top left, you see ALD high-k gate moving from FinFET to gate-all-around. So we have a well-entrenched position there with all of the leading logic makers, and we are well positioned there to also benefit from the more intense use and the further propagation of ALD high-k gate technology in general when the market further develops. We have also developed a strong position in ALD dielectric gap-fill. We have very little externally talked about it, but ALD electric gap-fill and deep trenches require specific technologies for ALD, whether it's thermal or plasma-enhanced ALD. And there, we have developed technologies to, in a seam-free way, fill certain gaps in -- to fill gaps in deep trenches, whether it's in memory or in logic devices. ALD metal is another category where we have been participating in high-k metal gate for many years, and there is a demand for new metals in general, in memory and in logic, and we are also participating there in early engagements with a broad range of customers. On the -- in the patterning area, the lower left corner, some of you that have followed us for many years know that we participated and led actually in the early years of PLD spacer patterning. We have basically kept and further developed that position. So also with the advent of EUV, which requires different types of patterning layers, we are strongly engaged with the leading customers to develop spacers, specifically for patterning using EUV. CMOS epi is an area where Hichem will talk more about where we're also well entrenched and have developed a much stronger position in recent years. And also in the upcoming changes in an advanced CMOS well entrenched in the developments for the future nodes. And then Ivo mentioned selective ALD in the bottom right, also a technology where we have experience with topo selective silicon nitride already for years, and are further participating in when that technology advances in further generations of 3D NAND, DoD, the JDP that Ivo mentioned is also making use of our, let's say, strong experience in ALD to further help our growth in the ALD space. So these are 6 example areas where we feel we can strongly -- let's say, further develop our position in ALD and at least maintain share or further grow our share in that growing market. And that is what is shown here on this slide. So on the left in black, you see what I showed earlier, the 45% to 55% share increase in the recent years. And with our, let's say, our experience and insight in how devices will further develop and the ALD technologies that are needed to basically enable the move from FinFET to gate-all-around, we have done, let's say, a thorough assessment of where that market will further -- how it will further develop. And that has led us to a growth of $1.2 billion participation of logic/foundry in the growth that we are announcing today for the single-wafer ALD market from $1.5 billion to somewhere between $3.1 billion and $3.7 billion, so that logic/foundry segment is expected to contribute $1.2 billion in that growth. So the key elements are high-k gate, metals patterning, high aspect ratio gap fill just to name a few, and there's more specific applications that we are working on. In memory, that in absolute sense, the growth participation will be somewhat smaller. But also there, we are expecting, based on our insights and participation in that market today, that, that segment of the market will develop for further growth by EUR 0.7 billion by 2025, totaling a growth of EUR 1.9 billion from EUR 1.5 billion to a midpoint of USD 3.4 billion by 2025. I know that a lot of you that have followed us for many years, we're eagerly waiting for an update from us on the size expectation of that market, and we're happy to share it with you today. Then on the epitaxy market outlook, I showed to you already 0.6, 0.8, the number of the sizes of that market in 2016 and 2018. Our analysis of that market has led us to believe and to report to you today that we expect that market to grow to somewhere between $1.5 billion and $1.8 billion, driven by various elements, certainly, CMOS epi, so advanced CMOS whether it's the channel application or [ source train ] contact application. We are deeply embedded with leading customers. So we have developed a fairly good insight in where that market might develop. That's an important element of that growth that we are forecasting. Memory will also, as Ivo mentioned, see more use of epi. So it's also an important element of that. And then there is a power analog and wafer segment of the market that is also making increasing use of epitaxy and also the overall general drive of more semiconductors fundamentally will require more -- let's say, more capacity and further advancement of capability for epi specifically also being a part of that growth of the epitaxy market to the $1.5 billion to the $1.8 billion by 2025. So we're happy to share with you today both updates of the ALD and epi market, where we have strong positions, leading in single-wafer ALD and growing and epi with the ambition in epi based on our current view that we are happy to share with you a target today of a market share larger than 30% by 2025 in that growing market. So if we share them as the -- let's say, last assessment I wanted to share with you in this presentation is specifically the ALD and epi market growth in advanced logic. So we explained to you already that in the development of advanced logic, more ALD is going to be used, more epi is going to be used. So if you look then in 2020 at the invested capacity, that the capacity basically being purchased for advanced logic, which is mostly FinFET driven last year. And based on, let's say, our insight in how the technology will develop, how the various customers will ramp and the timing at which they will ramp gate-all-around or advanced logic in general in 2025, we have made an assessment that, that market will grow by $1.2 billion by 2025. So roughly, you can see it on this chart, doubling the segment of that combined ALD and epi market in the coming years. So in conclusion, I'd like to share with you that we -- that a strong digitalization, just recapping the strong digitalization is expected to drive more semiconductor and more wafer fab equipment in the coming years. So I hope I have explained to you what drove our outperformance in the last 4 years, give you a little bit more color so that you can better understand what drove it. Third, the size of the single wafer ALD market, our analysis has led to an assessment of a growth to $3.1 billion to $3.7 billion by 2025. A similar assessment we have done for the epi market, growth of $0.8 billion -- from $0.8 billion to $1.5 billion to $1.8 billion by 2025. And then finally, the assessment of the specific advanced logic segment of the market from mostly FinFET-based last year to what we expect to be mostly gate-all-around driven in 2025, ALD and epi combined a $1.2 billion market growth. So that concludes what I wanted to share with you today. Hichem will in his next presentation, give you all the insights about, let's say, competitiveness of our products. But before I give the floor to Hichem, I give the floor back to Victor because he will likely announce a break.

Yes. Thank you, Han. And also Ivo and Benjamin for the interesting presentations. So this concludes the first part of the program. We will now have a break of 25 minutes. So to the participants on the webcast, please be sure to be back in 25 minutes. And to our guests here in the room, you can go to the room next to the welcome reception for coffee. [Break]

Okay. Welcome back to the second part of the program. We will first start with Hichem M'Saad, our Executive Vice President of Global Products. He will walk you through our products and innovations in products. This presentation will be followed by Paul Verhagen, our CFO, who will give you a financial update and talk about the financial targets. But first, floor to Hichem.

Thank you, Victor. Good afternoon, everyone. It's my pleasure. So I'm going to wait. So good afternoon, everyone, and good morning for those of us who are joining us from the United States. It's my pleasure to talk to you about the products and application that are really leading our growth. First, I'm going to start with ALD. And our leadership in ALD is a reflection of the products and the application that we have developed in the industry. So I'm going to show you some example of those products. Second, we have grown our Epi market share in the past few years. And I'm going to show you that the growth in our Epi market share is due to some innovation we have brought into the industry, especially in the reactor design. And with the Epi innovation that we have brought in, we are very well positioned in the gate-all-around transition. Third, I'm going to talk about that for both our vertical furnace and PECVD product lines, we really are selectively moving into this area. We are selectively growing our offering. Fourth, because of our drive to sustainability, we have actually embedded sustainability in our product life cycle. And with it, we can actually reduce energy usage and also we reduce precursor usage. Last but not least, in our service business, we have moved from transitional type of service business model to an outcome-based service model to provide better value to our customer. So let's first start with ALD. I'm showing you with this graph that shows the ALD evolution from 2003 to 2021. As you can see in 2003, there are about 3 to 4 layers of ALD. Nowadays, we have over 70-plus ALD layers being used in the industry. The other message from this presentation, you can see that the past few years, the ALD intensity has increased. The growth in ALD has increased a lot. This really reminds me of my days in the mid 1990s as a process engineer in CVD. During those days, CVD was growing in the same way. We have -- at that time, we have PECVD and we have LPCVD technology. Then all of a sudden, many CVD technology have come to pace. For example, HDPCVD, high-density plasma chemical vapor deposition; SCVD, subatmospheric CVD. We had also ECRCVD or electronic cyclotron resonance. And with this new deposition -- CVD deposition technique, we have many new layers. For example, antireflective coatings, advanced patterning films and also many located materials, starting with FSG or fluorinated silicon glass at a k-value of 3.5 to all the carbon-doped oxide film that ASM has been part of with our Aurora technology. So the late 1990s is what we call the golden age of CVD. Fast forward 20 years, and we are in the midst right now of the ALD Golden Age. That's why the title of this slide is ALD is a new CVD. And not only ALD is -- has the same growth that CVD in 20 years ago, but actually, this growth is even stronger. Why? Because of 3 different things that sets ALD apart. First, device shrinkage, when you shrink you want to have thinner films. ALD is the best technology to deposit these very thin films, and we see right now in the industry that many of the CVD layer are actually being transitioned to ALD. And to a lower extent, some PVD layer are actually moving into ALD. The second inflection point for ALD is the device structure. As Ivo have shown you earlier, the device structure is moving from 2D to 3D. For example, DNAND, NAND has moved from 2D NAND to 3D NAND. DRAM has -- is actually right now is a 2D DRAM, but by 2025, we're going to have 3D DRAM. In the larger part of the industry, we have the FinFET and the FinFET is moving to gate-all-around. In all what's common between all this technology is that we're going vertically. And when you go vertically, then that really favors ALD because ALD is the best technology to -- for deposit very high aspect ratio gap fill. It's good for gap fill and also provide very good step coverage. The third thing that makes sets ALD apart is that right now, Moore's Law is being driven by new material and ALD is the best technology to deposit new materials. So let's show you these -- the products that we have that are driving this ALD Golden Age. First, I'm going to talk about the Pulsar. This is -- the Pulsar is the industry workhorse for depositing hafnium oxide high-k gate. Every customer in the industry uses Pulsar to deposit hafnium oxide. Next to Pulsar, we have the Emerald and the Emerald deposit carbides and nitrides, these are the metal gate, the thick aluminum carbide. Then we have -- in 2018, we have developed a new architecture. We have developed a new architecture that we call Synergis. And the reason we have developed Synergis is to provide the best reactor technology to the industry with the best productivity. So we took the best reactor design from the Pulsar, Emerald, we put them into Synergis. And we have with that, we also improved our productivity. With that capability, we can't deposit any kind of material we can think of. So for Synergis, we have different materials. We call them the metal oxide. These are the patterning films like aluminum oxide, hafnium aluminum silicate and so on and so forth. And then we have the metal nitride like dinitride and tungsten nitride. Also, we have Synergis for metal. We can now deposit metal in our Synergis reactor. But when you see for ALD, we see there is also a trend in ALD to deposit thinner and thinner films. Right now, we even deposited film as thin as 5 Angstrom. We're talking about 2 monolayers. So when you go very, very thin, then the surface becomes very important. And the surface really controls your film -- the ALD film property. So in order for us to have a very good film, we need to make sure that the surface is clean. So that's why we have introduced an ALD based surface PEALD technology, which is actually integrated with our Synergis ALD processes. Next, I'll talk about this new architecture, which we call the DCM and QCM architecture. This architecture is actually geared more for our dielectric film. So DCM stands for dual chamber module. QCM stands for quad chamber module. So this architecture is very key in depositing the nitrides, the oxides for memory, for patterning amplification, a sacrificial layer. Will also deposit the nitride and oxide as spacers in the -- as spacers and liners. So as you can see, we have a large portfolio of ALD products. But what's common between all this ALD technology that we have is our unique reactor technology. It's common between all the products that we have. And what's unique about our ALD reactor technology. We need to make sure that the reactor technology provides the best ALD film with the best step coverage. So things about our ALD technology that sets us apart. One of them is that the chamber design is made to provide very good uniformity. So view graph on the right side here shows you that we can achieve a uniformity of 0.3 angstrom on a very thin 48 angstrom film. 0.3 angstrom, that's extremely uniform. The second thing that we can do with this reactor design is that the reactor is made to be flexible to deposit 4, 5, even 6 elements. And why that's important? It's important because Moore's Law is being driven by new material. So we can deposit many materials that are unknown to humanity right now with this ALD technology. It took about 2 years to develop this kind of flexibility in our reactor. And here, you can see ternary phase diagram showing that you can -- any point you want to think about, any materials you can think about, you can deposit it with our ALD technology. Third thing for the ALD technology is that we can actually -- our reactor is very small volume. So we can have very fast cycling times. And also, we can purge the precursor very, very fast. What's that important? What's -- why is it important? It's important because for gap fill application. So as you can see we can achieve very good gap fill -- void-free gap fill with ALD. At the same time, what Ivo has talked about earlier that we can also deposit liners in very high aspect ratio memories with a very good control very good uniformity. And this really shows that this liner here in 100:1 aspect ratio. This film is very uniform. So when you see here that the same what it's ratio between top and bottom, it means it's the same film, same quality film between the top and bottom. So that's really some of the improvement for our reactor technology. Next, I'm going to go move into Epi and showing you why we are growing in Epi deposition. First, I'll start with 2015. In 2015, ASM was only addressing the power electronics part of the market with our E2000, the Epsilon 2000 and the E3200 offering. In 2017, during SEMICON West, we have released the first reduced pressure Intrepid ES cluster tool platform. And then earlier this year, in July of this year, we have released the first cluster tool atmospheric pressure Epi, which we call the Intrepid ESA. With these 2 offerings, now we can address the totality of the Epi market, including the advanced CMOS and also wafer makers. So when we moved into this new technology in the advanced CMOS we were faced with the dilemma. Number 1 thing is that when you go to CMOS, advanced CMOS, Epi has to be deposited at much lower temperature than power electronics, with deposit film at 700 degrees Celsius. For power electronics, you have to deposit film over 1,000 degrees Celsius. So as this low temperature, the surface on which the Epi sits is not very clean. You can have oxide from etch, you can have residual, carbon residue from etch, you can have defectivity. So in order for us to deposit pristine film for the advanced CMOS, we have to remove this impurity. And that's why we have developed an ALD-based surface clean technology, which we call Previum. So this is the Previum, and we have about 3 different reactors for Previum with V3, next and VP. So this reactor would monolayer by monolayer remove any defectivity before we deposit Epi. Then we need to integrate this preclean with the Epi deposition. We cannot break vacuum because if you break vacuum, then you're going to have some oxide on your wafer again. So we have to come up with a platform, which we call the -- our platform, the XP4 and XPE, which allows us to integrate both the preclean and also the Epi deposition on the same platform without breaking vacuum. So we have flexibility for us to deposit to have preclean and Epi deposition on the same reactor. And these kind of reactors, some of them are actually in high-volume production. The third thing about Epi which was a dilemma for us and challenge is really how -- why we're going to get into this very advanced Epi process where the competition is much entrenched and where the barrier to entry is very high. We made a deposit -- we made a decision early on that the only way for us to penetrate into this market is to provide technical value to our customer. We need to provide them with technical differentiation. So for that reason, we have invested heavily in Epi. We have innovated in Epi. We have come up with many innovative feature, which I'm going to talk about some of them right now. The first feature is what we call isothermal reactor. What does this mean? Epi is a thermal process. So having exact temperature is very important. If the temperature changes, your process is not uniform. So we have made an isothermal reactor to make sure that every wafer at any time, in any reactor, any 1 of the world sees the same environment. With it, we can achieve very good repeatability for high-volume manufacturing. The second innovation that we first introduced in the industry is called AEGIS, which stands for advanced gas injection system. This is the first MFC-based delivery in Epi. With it, we can dope Epi very easily and very uniformly. So for SIP application, you can introduce phosphorus with extremely good uniformity. And also for compound Epi like cynical germanium, you can achieve very good germanium uniformity. The third innovation that we have is we call VERACE-CL. VERACE in Italian stands for accurate, CL is closed loop. So we have an accurate closed-loop control of the temperature. So not only we can measure the wafer temperature directly, but we also can control the wafer temperature. Why this is important? This is really important for the advanced Epi application in the gate-all-around. So let's look into these advanced Epi application. So if you look into the gate-all-around picture over here, you see many silicon-silicon germanium nano layers, is deposit silicon and silicon germanium on top of each other. So in black, that's the silicon in white here, that's silicon germanium. So we have 4, 8 or 8 of this layer 1 on top of each other. That's what defines the channel. Next to it, you have the source drain -- and the source train here like mass source drain, for example, what it means, it means you have to deposit Epi in a gap-fill, not only deposit Epi, but you have to get gap-fill Epi. You have to deposit Epi from the bottom, from the top and so that you have no seam formation. And that's what we call 3D Epi, deposit Epi X direction, Y direction, Z vertical direction. And this picture shows that we can actually deposit Epi between all these nano sheet structure. So why gate-all-around is so important in the device. The reason is really in the next picture. As you can see in the FinFET, you have -- the FinFET channel thickness is defined by 2 things: litho and etch. So you deposit your silicon, you litho it and then you etch it, and you form your channel thickness. If you flip that around by 90 degrees, you get the gate-all-around. So what you can see here, this is the channel film, the silicon channel film that I showed you earlier in this picture here. So right now, the Epi is actually what defines the structure. That means that Epi thickness is very important. That means the channel thickness is right now defined by Epi. So you need to make sure that the Epi thickness within the whole wafer and layer to layer has to be extremely uniform because the channel thickness pretty much defines your transistor performance. So going from FinFET to gate-all-around, Epi -- we go from litho etch definition of the gate to Epi definition of the gate. So how do we -- so in order for us to be able to get very good uniformity, that's really where VERACE-CL comes around. So if you take this layer to layer thickness and you try to measure it with the spec of plus or minus 1 angstrom with the common technology right now what you see here is that at the edge of the wafer and a bit at the center of the wafer, you lose the spec. You cannot meet your specification. And then when you go to VERACE-CL, whereby you have a pile on top measuring the wafer in the middle and in the center. And not only that, those piles are really control the wafer temperature to about -- to delta of 0.1 degree Celsius. All the conventional technology before control the temperature to plus or minus 1 degree. We have improved our temperature measurement accuracy by 1 order of magnitude. And with it, you can achieve very good uniformity of this channel thickness in the middle, center, middle and edge. Why this is important, not only we can achieve the spec of the customer, but we can exceed the specification by 50%. So what does it mean? It means we can control the channel thickness monolayer at a time. This is really enabling technology for Epi going forward. The second thing that for Epi, I wanted to give you an example about our reactor technology. This is for the Intrepid ESA. This is the atmospheric Epi. And here, this is what one of the users that's happening in power devices. We all know right now in electric vehicle, you want to do fast charging. So you want to need Epi needs to be thicker, very thick Epi for fast charging. Also in solar energy, we have inverters, and the solar energy is going from 1,000 volt to 1,500 volts. So with doing to higher voltage you need thicker Epi. So right now, you need Epi as thick as 100 micron. But it's very difficult to deposit Epi, 100-micron Epi. Why? Because as you get thicker and thicker you -- on the wafer, you also depose on your chamber wall, then you can have defect and residue come on the wafer. So you have to stop your deposition. So the industry right now, this is what happened. In order for you to deposit 100 micron, you'll go, for example deposit 20 micron, you take the wafer out, you need to clean the chamber, dry clean it, give the wafer back in, deposit to 20-micron and so on and so forth, about 5x. What does that mean? It means it takes a long time. It's very costly to do that. But most importantly, once you divide your deposition, you have what we call interfaces and interface by definition is a crystal defect and Epi is a crystalline film, monocrystalline film. So once you have a defect, then your device performance degrades. So it's very important that we can deposit 100 micron in 1 step. And this is the capability of our reactor. We can deposit 100 micron in 1 step, not only that is that once you open your reactor, after this 100 micron, you can see the courts of the reactor it's very clean. And this is really an achievement in ATM Epi. Third part of my presentation is going through our other product lines, which is our vertical furnace and our PECVD. And if these 2 product lines, we have decided that we're going to selectively grow in this area. For example, for furnace, what we have done. The last tool that we have or the last product that we had was about 20 years old. So we have made a decision to update our product line in furnace. And to update our product lines, we need to make -- we offer value to our customer. Value number one, the tool needs to be more reliable. Value number two, the tool has to be lower cost. Value number three, the tool has to have lower throughput, higher productivity. That's why in November of 2019, we have really -- which during SEMICON Europe, we have released our A400 dual. This is a 200-millimeter new platform. And with this release of this platform, it really allowed us to expand our market share in China a lot. We have win many markets in China with this tool. And we expect the same thing as we develop our 300-millimeter platform. For PECVD, what we looked into PECVD is really going into a niche application. So what I'm showing you here is in -- so we looked into the low-k and the carbon-based films. For the low-k, what we see, we see as the device shrink in logic, the number of layers of low-k, the 3.0, which is our Aurora film, which is an adversary in production for many years, actually, those layers are increasing. So as the device shrinks, our low-k revenue is going to increase. What you see on the right-hand side here is our foray into carbon-based films, PECVD carbon-based films. And as you can see, these are PECVD films, carbon-based from PECVD that have very good gap fill. If you ask me about this, when I was in PECVD about 15 years ago, I tell you that's impossible. But right now, it's possible. So why this is good. This is good because this PECVD film would replace this spin on carbon. And where the offer, the value here is that the PECVD films have better planarity than the SOC or the spin on carbon. The second thing is that once you deposit the film with a plasma those films have higher density and better selectivity, and that's what you need for patterning. All our products that I've shown you today have something in common. It's our drive to sustainability. So we have a very aggressive sustainability target. And I have -- we have used sustainability. We have embedded this sustainability of our company into our product -- product life cycle in the area of innovation, in the area of design, in the area of system operation and this area of equipment refurbishment. So in innovation, we have innovated in 2 areas and engineering innovation and process innovation. In the engineering innovation, we have -- for example, we have used the green chiller. We use chiller for some application. And with it, we actually can reduce energy consumption by over 4%. In process innovation, we have focused on cleaning chamber clean. We have been able to reduce the precursor usage, the clear precursor usage by over 16%. In design, we have looked into both our product lines. We have 2 product lines. You can define them in 2. We have thermal processes products and plasma products. For our thermal products, what's important is temperature. So we actually have come up with new materials, new insulating techniques and new materials that can reduce the heat losses. And for plasma products, which use RF circuitry, we actually worked with our supplier to reduce the electrical usage of their power supplies, but not only that we also have reduced the RF leakage in our systems. In the system operation, we have partnered with a [indiscernible] company to reduce the affluent usage. We also have developed smart idling techniques to reduce energy by over 15%. In the refurbishment of the tools, we actually have an active program in our company to recycle all the equipment and get it back into production. The same thing, we actually are extending the lifetime of the component through development of new cleans. And we actually have an engineering team that tried to develop extended lifetime of products of components to reduce the cost and also to reduce the energy usage. The next slide here shows you that how we have in the design part, we have been able to reduce our energy usage. I talked earlier about our DCM and QCM in ALD. So when you go from DCM, QCM, you actually have more reactor per tool. Of course, your throughput increases. You have higher productivity when you move to the QCM, but -- and you lower your cost of ownership but most importantly, it's actually you reduce your energy usage. So for a liner process, you actually reduce your energy usage by 16% and for a gap fill process, you reduce your energy usage per wafer by 19%. This usage is per wafer and the reason we can achieve that because we have more component -- common component in the QCM. Last, let's talk about our service business. What we have made a decision to move the business model of our service from a transactional business model to an outcome-based service model through -- and the way we can achieve it is actually through some engineered solution. For example, we can go to a customer and we can commit a certain COC or cost of consumables. And we drive ourselves to reduce the cost of these consumable through developing new cleans, having new coatings and new refurbishment techniques. For example, also we can go to our customer and we can sign up for an uptime of our tool. We can sign up for a certain uptime. Customers are very happy with it because they don't have to worry about it. But for us, we can drive ourselves to come up with a new CIP or continuous improvement programs to achieve this. In all cases, ASM actually manage both -- we manage the inventory and the spares ourselves for our customer. So to summarize my presentation, I hope I have shown you today that we have a very broad product portfolio for ALD, which drives us to new heights. I hope that I showed you that for Epi, we have a new technology, a new innovation with Epi, which has positioned us very well for the next gate-all-around transition and also for many use of Epi in memory. I hope that I have shown you today that for our product offering, both vertical furnaces and PECVD, we are selectively growing there very smartly and very precisely. Fourth, we have embedded innovation and sustainability in all our product lines to make sure that we can reduce energy usage and also reduce precursor usage. Last but not least, we have transitioned our service business from transactional business model to -- from a transactional business model to an outcome-based business model to provide better value for our customers. Thank you very much for your time. Right now, I'm going to give the floor to our CFO, Paul Verhagen to talk to you about the financials.

Yes. Thank you, Hichem. Welcome. Good morning, good afternoon, good evening wherever you are, people here in the room. Please stay with me for another 15 to 20 minutes. You got a lot to digest this afternoon. A lot of interesting updates, I believe, the growth to innovation strategy, the market outlook and opportunities, the technology road map, the product road map. And what I will try is to share with you what that means in financial terms. I have 4 key takeaways for you. The first one, as you've seen, I think, in multiple presentations this afternoon, ASM has grown to be a leader in ALD, creating significant value for all stakeholders. ASM has made it possible for our customers to move to more advanced nodes. You've seen some of the key inflections that the company has managed. We have been growing tremendously, offering more opportunities to our employees, also focusing more on a more inclusive and diverse culture in the last years. We have created a lot of value for suppliers who have grown with us. And last but not least, and I will come back to that, of course, I believe the company also created a lot of value for the shareholders. We have a revenue target. You've seen that of EUR 2.8 billion to EUR 3.4 billion by '25, a CAGR of 16% to 21%. And with that, outgrowing the WFE market. An operating margin of 26% to 31%, which, of course, should put us in a position to deliver and generate strong free cash flow. And last but not least, capital allocation policy unchanged. Investment in growth because there's a lot of opportunity in growth remains the key priority and excess cash one way or the other, as ASM has also done in the past, will be returned to shareholders, and we'll come back to that later in the presentation. First, a few highlights, and then I will take you a little bit deeper into a few financials of the last 5 years and thereafter I will share the projections for the coming 5 years. So first, from '16 to 2020, an annualized total shareholder return of 39%, which compares to our weighted average cost of capital of around 9%. So this is a pretty decent return. Total cash returned to shareholders over that period, EUR 1.4 billion and free cash flow EUR 411 million. Return on invested capital improved 11 percentage points over the last 5 years, a CAGR of 22%, gross margin up 3%, operating margin up 11% and -- and Hichem just shared the focus on sustainability. We've also been able to reduce greenhouse gas emissions by 15% per euro R&D investment we've done over that period. So overall, I believe ASM has delivered a very strong performance in the last 5 years. Further, as said a little bit more detail on the recent 5 years and then again, I go to the projection. The 39% that I just shared with you was annualized -- annual return. This is the accumulated return but not until the end of 2020, but until the 22nd of September 2021, more than 1,200% now EUR 123,000. So a pretty good return. Outperforming the IX, which is the main index in the Netherlands, outperforming our peers and also outperforming the SOX index. Also dividends, we've seen a strong growth from EUR 0.40 in 2010 to EUR 2 in 2020 and even a special dividend in 2019, again, which is fully in line with ASM policy to return excess cash to shareholders. Revenue over the period, total revenue, Han showed earlier on the equipment revenue. There was a CAGR of 24%. Total revenue, a CAGR of 22%. Basically, the company more than doubled from EUR 600 million in '16 to EUR 1.3 billion in 2020. And you've seen all the reasons, there was growth in end markets, but equally important, ASM also managed to increase share, in particular, in ALD and in Epi and ASM crude spares and service businesses. Net earnings on the right-hand side does not show a consistent picture with the revenue increase because you would expect, of course, a decent fall-through of the increased revenue to the bottom line. You're not seeing that and that has everything to do with all the lines below, let's say, EBIT mainly. That is the income from ASMPT. There's material fluctuations from year to year, which make this at least at face value look a little bit strange. There is currency translation differences that caused this trend that you see, not fully consistent with revenue and actually also in terms of tax accrual in particular in 2019 and '20, and I'll come back to that. There was -- were actually the first years where there were significant accruals accounted for in the net earnings. Gross margin, as you know, ASM guided low to mid-40s. Now over the last 5 years, 4 of the 5 ASM has been within that guidance and it was not a linear line up. You see that. It started at 44%, came down a little bit and actually the last year, moved outside of this range to 47%. And this is very important to understand because there's a few structural drivers that improve margin like operating leverage and productivity improvements. I'll come back to that a little bit more later on the presentation and also supply chain improvements, but there's also application mix. And the mix can change from quarter-to-quarter. You've seen that in Q1 and Q2 of this year, but the mix can also change from year-to-year. So there will be differences in margin depending on the mix development from year-to-year. And last but not least, especially in 2017, but even more so in 2018, new products were introduced with an initial lower margin than the running products, which then in subsequent years through engineering changes, through commercial negotiations with suppliers and through other actions, ASM managed to improve these margins again. So sometimes new products come in at very high margin, but sometimes a somewhat lower margin. But then, of course, there is opportunity to improve that going forward through the actions that I just talked about. On the right-hand side, you see the EBIT, The operating margin, gradually improving 14%, 15%, 15%, 20% and 25%. You see that it did not fully follow the gross margin. That's mainly because of the growth of revenue, which resulted in a positive operating leverage, in particular, in SG&A and R&D. Free cash flow. Yes, this is a little bit of a volatile picture, and it has everything to do with changes in working capital. As you know, this is a measurement per year-end. If I would make this picture for January 7, it looks very different again. So that's one element that makes the swings from year-to-year. The other element is CapEx, of course, which brings these changes from year-to-year. From '16, '17 to '18, actually in all the years, the operating cash flow, so before changes in working capital, before CapEx was significantly higher than 20% except for 2018. The decline that you see in particular, of course, in '16, '17, '18 is significant increases in working capital, at least relative to the revenue, which was relatively small at that time and an increase in CapEx, which I will show in subsequent slides. In '19, almost everything, all the pieces of the puzzle fell in the right direction. Profitability improved with higher revenue, even working capital came down. And also in that year, CapEx was lower than 2018. So then you see 18% popping up. And in '20, the opposite happens. We saw almost EUR 100 million increase in working capital for only a few weeks because there were some customer payments that just came in after year-end. If they would have come in a few weeks earlier, also '20 would have looked much better than what you see here. On the right-hand side, you see the cumulative cash. This is over the longer period, 11 years, EUR 2 billion has been returned to shareholders through share buyback, through return of capital and dividends. R&D. This is important. This is the lifeline of the company. This drives innovation. You've seen the presentations. It increased from EUR 100 million in '16 to EUR 170 million. And going forward, I'll show you in the later slides we plan to grow this even much more. It drives innovation, and we grow through innovation, we grow through enabling our customers to move to the next nodes. You see the growth in the net, the most relevant number is the growth investment. The net number, of course, is including a capitalization of costs. And the difference between these two is somewhere between 2% to 3%, typically almost every year, as you can see here. CapEx, again, moving quite a lot from year-to-year, but what is immediately visible is that in the years where we add capacity, manufacturing capacity, that's when you have the peak in CapEx. And here, it's 2018, when we completed our facility in Korea and 2020 when the facility in Singapore was completed. And here you see the swing in working capital. As I said already, the first 3 years, it went up, down in '19, despite a revenue increase and then significantly up in 2020. You see the change in working capital days. 82 was the highest, 42 was the lowest. In 2020, 1 day is around EUR 4 million. So 40 days difference over the period is like EUR 160 million. And 1 day of working capital measured in euros, of course, with the company growing will only become larger and larger. But this is mainly driven, as you see the swing by change in the DSO, which moved from 70 plus days to 52 at the lowest back up to 72, inventory gradually improved and payables remained more or less stable. Okay. That was the past. Let's now go to maybe the more interesting part, which is the coming 5 years. Quite a lot of these targets you've seen, the revenue target, the CAGR, gross margin. We have lifted to 46% to 50% from low to mid 40s. SG&A, we plan to not grow that as fast as revenue and think that should gradually move towards the high single digits. R&D will be high single digit to low teens and an operating margin of 26% to 31%. Furthermore, we don't only have financial targets. We also have nonfinancial targets, which I believe are in today's world almost equally important. It's, I think one of the top issues that the world is facing today. It's a top issue that our customers are facing today and as a result, is also a top issue, of course, that we are facing. And I firmly believe that investments need to become more sustainable. Companies who would not recognize that put the medium- to longer-term prospects of their company at risk, I believe, it is vital that again, investments become more sustainable. And we have decided and announced today that we will step up and accelerate investments to become more sustainable. Net year by 2035, including Scope 1, 2 and 3, which is upstream and downstream. So this is an ambitious target. We have a few interim targets, so 100% renewable electricity by '24 and already 100% renewable electricity in '21 for our key sites. And furthermore, we also plan to have by the end of next year, the assessment ready of, in particular, Scope 3 and targets, what do we need to agree with our partners to enable us to move to net zero by 2035. The other nonfinancial targets that I want to highlight is safety. That's maybe the most important because we want our people to come home safe in the evening. We work with a lot of chemistry, a lot of materials, things can happen in the lab. So we will do whatever reason possible, of course, to protect our people and to make sure that they can go home safe every night. Revenue. So as said, we grow to EUR 2.8 billion to EUR 3.4 billion, a CAGR of 16% to 21%, which means 1.5x the wafer equipment market at the low end and double actually the growth of the wafer equipment market at the higher end. The assumptions that we've used to model this are listed below, and you've heard them a few times today already, which is maintaining market share in ALD, logic foundry, growing in memory, growing in Epi, selective growth in vertical furnaces and PECVD, and, of course, growing our spares and service business. That all should happen to make this happen. Gross margin. We were at 44% in -- on average over the recent 5 years. We believe that we can manage that to 46% to 50%. Actually, in 2020, we were already there as you see. And as I mentioned before, there's 2 structural drivers which is operating leverage and supply chain improvements and one that is structural, but also can fluctuate from year-to-year, quarter-to-quarter, which is the mix. Operating leverage in itself will not do a lot from year-to-year because our manufacturing capacity is to the very large extent variable. We are an assembly capacity. We don't really manufacture, we assemble. But still, over a 5-year period, it starts to add up. So the growth that we project and the revenue that you just saw will ultimately contribute to over this 5-year period to improve the margin. SG&A, as I mentioned already, there, we need to make sure that we don't grow that in line with the revenue. We should see clear productivity improvements. We will set clear productivity targets. It will move gradually to a high single digit. It might not be a linear line precisely because certain years, we might advance certain things a little bit more than another year. So it will not be one straight line. But over this period, it will come down to around the level I show on this slide. R&D. As I said, we are now at around 10% net, 13% growth, and that's more or less where we want to keep it. So with the growth in revenue, applying these numbers, that means a significant increase in R&D, which we believe is very important, again, is growth through innovation. You've seen the examples and you've seen the presentation of my colleagues. This is a lifeline. This provides auction to the company, and we plan to invest going forward. The net will be, as I said, high single-digit low teens. That's very hard to say what it precisely will be. But I think the key message is that we will invest in R&D. And the vast majority of these investments will go to ALD and Epi. Of course, also selectively, we will invest in vertical furnaces and PECVD, but the bulk will go into ALD and Epi. And as Ivo also said, a certain amount will, of course, be allocated to the rise and beyond the midterm targets that we talk about. By the way, for capitalization, I know it's a question. We have assumed a similar ratio as what you've seen in the last few years. So no change there as far as we can judge today. Now that results in this margin improvement, 26% to 31%. Of course, the improved gross margin contributes and in particular, the improvement in productivity and benefiting from operating leverage in SG&A. Then another important element is tax. You see '16, '17, '18 even very low tax rate, and that has everything to do with the fact that ASM used to have quite some net operating losses. They have been fully consumed by now. If there's one disadvantage make a lot of money, it might be this one. They have been fully consumed. So going forward, that means that the tax rate will increase. Actually, already in this year, as you can see on the slide, we expect the tax rate to go to around high teens and will most likely move to low 20s going forward. Obviously, we will see if we can further optimize this. But of course, within the spirit and the letter of the law and also recognizing, of course, that on the one hand, tax, as you have heard many times, is the cost to the company. But at the same time, also a contribution to the countries in which we operate. So we have to, of course, find the right balance there. We, of course, will also see if we can qualify for new tax incentive programs. There's a lot of let's say, news about potential government support, et cetera, et cetera. That's all still very unclear what is going to happen. That would only be speculation now to say something about that. But guaranteed that we will look into all possibilities to see whether or not we can qualify for new tax incentives or other forms of support. But based on what we know today, this is the most likely trend going forward. Working capital. I talked about it, you saw the swing, 82 to 42 days. Our target will be to keep that in a more narrow range, 55 to 75 days. With the current revenue projection in 2025, 1 day, working capital is around EUR 9 million. So you can imagine 10 days is EUR 90 million. So from a cash flow point of view, depending on working capital performance, that will swing a little bit from year-to-year. So it's better to maybe to look over a 2-year average or 3-year average, but with the profitability that we talked about, yes, we should be able to deliver a pretty healthy cash flow and you can do the math yourself because all the ingredients that you need have been presented here in this slide. CapEx, EUR 60 million to EUR 100 million, again, in the years where we have the capacity, it will be a manufacturing capacity will be more than in other years. But as Benjamin already showed in his presentation, we have to expand some labs. There will be new equipment needed in the labs. We are growing. So the level of investment will be higher as we can see it now than what we've seen in the recent past, which was EUR 60 million per year on average. And last but not least, the capital allocation strategy, something else shows up on my screen, sorry for that. The capital allocation strategy, so priority one invest in growth. There's a huge amount of opportunity ahead of us. We want to capture that. We invest in R&D. We support it with CapEx. And of course, we scanned the market for M&A opportunities, and we'll act if they are there. And if they're not there, of course, not, maintain a strong balance sheet. We target to keep EUR 600 million gradually up from EUR 300 million, which was, let's say, the company communicated cash balance before, but that one was, let's say, set at a time when ASM was much smaller. Sustainable dividend payments is, of course, important to us. We don't want to switch on and off every year, sustainable payments. And certainly, last but not least, excess cash will be returned to shareholders one way or the other, as you've seen, and I've shown examples of that. These are my key takeaways. I'm not going to repeat them. You've seen it many times. Thank you very much for being here again. And I would like now to hand back the floor to Benjamin who will do a very brief wrap-up, and then we'll move into Q&A.

Thank you, Paul. Thank you. So we are now to the end of all the 5 presentations that we have prepared. I hope you will agree me that this is a lot of information. And we just want to maybe have the opportunity to share with you as much as we can. Maybe it's a little bit too much. And we do expect that in the next Q&A session, we are going to get bombarded. We have a lot of questions. But I think together with my colleagues here, we are prepared to do that. I just want to do a very short ramp up which is basically the same 5 messages that I have conveyed at the beginning. One, of course, is that digitalization trends is going to drive semiconductor. It's going to drive WFE. It's going to be good for ASMI as a company because that is our business. Second, ALD, we are the market leaders, and we are fast expanding in epi. Two, enabling technologies that are absolutely required for future advances in semiconductors, and they will grow faster than the overall WFE market. Our strengths are in our technology innovation. My colleagues have explained to you what we do in terms of technology and innovation. In terms of customer engagements, we believe in the early customer engagements. And last, but not least, I think my colleague Hichem has shown to you how we differentiate in terms of products, especially the explanation on the way that we got into the epi market was by technology and product differentiation. In terms of strategic objectives, we want to maintain our leading share in ALD, in logic and foundry and expanding memory. We want to expand our share in epi. We want to grow our spares and service business. And at the same time, we want to accelerate our progress and focus in sustainability. And last, but not least, just to follow from where Paul has ended. We target to grow to about EUR 2.8 billion to EUR 3.4 billion by 2025, representing 16% to 21% CAGR, and we want to achieve a 26% to 31% operating margin. Those are the 5 key messages. We hope you take them away with you. That is all that we have for today. And I'm going to hand you over to Victor, who's going to prep us for the Q&A. Thank you very much.

Thank you, Benjamin, and thank you, Paul and Hichem for the great presentations. So we will start the Q&A session in a moment with all of the presenters. We will take questions from the audience here in Amsterdam and also from the conference call. So for the persons on the webcast who have registered for the conference call and have not yet dialed in. Now is the moment to do so, please? We will prepare the stage for the Q&A. So please bear with us for a moment. Okay. We are now ready for the Q&A session. [Operator Instructions] Operator, can you please provide the Q&A instructions for the conference call.

[Operator Instructions]

Okay. Thank you, operator. We will now take the first question from the room here in Amsterdam.

Maybe we move from the left to the right. I think Nigel has a question.

My name is Nigel van Putten with Kempen & Co. I have one maybe topic that wasn't covered in depth today. Maybe the war for you guys said I forget the exact numbers, but a lot of advanced degrees in your R&D department. How do you compete with some of your more formidable competitors for maybe that same talent.

So I think the question was how do we compete with some of our more formidable, maybe larger peers. And I think the way that we have been competing is, of course, we do have competitors, competition. We do respect our competitors. And what we have always been trying to do is to make sure that we are able to provide something that is better for our customers. And a lot of that is in the innovation engine of the company. We pride ourselves with a lot of innovation, and my colleagues have kind of give you a flavor of the innovation that we do. And at the same time, trying to really help our customers resolve what we call high-value problems that they have, especially problems that are related to manufacturing for advanced nodes that goes out a couple of years. So this is where early customer engagement comes in and is very important. I think Ivo has mentioned that -- It's not just the partners that we have that are the institutes and the research or universities, we look at actually our customers as a very key partner for us because a lot of the solutions that we have, we have to develop with them. And at the end of the day, we have to provide our customers with something that is a better performing solution and a cost of ownership that is better for them. That's how we have been able to compete over the last couple of years.

Okay. We will take the next question from the conference call. Operator, can we have the next question, please?

Yes. Robert Sanders at Deutsche Bank.

Just a question on gate all around. I was just wondering if you derisked your estimates for potential delay in the adoption of gate-all-around from 2024 to 2025.

I believe the question that came from Rob Sanders, and I think the question was, do we see a delay in the adoption of gate-all-around from '24 -- 2024 to 2025. Is that correct, Rob. I believe there was a question. Yes, I believe that was a question. My -- our take is that I think that the transition from FinFET to gate-all-around, there's a lot of work that is being done. You have the 3 major players at different stages, but they all are aggressively trying to develop the process or the technology that is required for gate-all-around. In terms of when they actually will go into high-volume manufacturing, I think that's still left to be seen. Our point of view is that we are heavily engaged with all 3 of them. Some of the stuff that they are doing, we have already seen the initial successes. For example, we have a logic foundry customer that we have already been qualified as PTOR for the gate-all-around channel structure channel application. Some of them are still in the process of finalizing their development. So I think it's a little bit of too much of a speculation for us if we were to say whether it's delayed or not. But we do think that in terms of timing, it should happen during the course of the planning period or the '21 to '25 period that we have just laid out in front of you.

Okay. Let's take the next 2 questions from the room here.

Yes. I am Michael from [ Petcom ]. I have a question for Mr. [ Verhagen ]. There will be extra investments in the next 2 years to expand the Singapore factory. So what it comes down to is the second floor will be made ready for production. If you reach your 2025 sales target. Is that entire extra capacity absorbed? Or can you even grow beyond 2025 with that investment?

Thanks for the question. It's a good question. What we believe is that with the current, let's say, expansion, so the utilization of the second floor, which was still not utilized in the new building in Singapore. We should be able to deliver on this plan as we have presented it. At the same time, for growth -- assumed growth beyond 2025, most likely, as we see now, additional capacity will be needed. And that, of course, might trigger some investments already earlier on because there was a throughput time, of course, between starting and ultimately having a facility ready for production. But for this plan, on Korea, now the second floor, utilization in Singapore, which will be ready early '23, we believe we have enough capacity in place to at least deliver on this plan.

Yes. So this is Stephane Houri from ODDO BHF and ABN AMRO. I have a question on the services business because you made a statement that you were transitioning from a transaction to outcome-based model. I'm not sure. I understand the way it changes, the way you're going to get paid, if you think you can get more and what kind of profitability you can extract from this services business new model?

I will maybe answer and then Hichem can elaborate on that. So basically, what we are trying to do, and we actually only started this from last year is that instead of just doing a break fix and when the customers need the parts that they call us or when the customer need service, they call us. I think what we are trying to do is to provide a much more higher value-added service to the customers. For example, using the same example that Hichem has mentioned, we go in and put ourselves into, let's say, longer-term arrangement or contract with the customer at a certain, let's say, commitment in terms of reducing the cost or the prices of the cost of consumables. And one of the ways that we do that is one of the -- I mean the customer could be instead of trying to clean certain parts that gets quoted or gets dirty, they have been doing what we call, they buy new parts. And through the technology that we can introduce, especially with cleaning technology ways that we can prolong the life of this part, we actually help them to reduce the cost. At the end of the day, it's about how much value we can bring to the customers either helping them to have higher productivity or reducing the cost of, let's say, operations for them. So that is one way. And usually, this would require that we enter into a longer-term, I would say, service contract or contractual agreement with them. Maybe Hichem can add on to that.

Yes. On the same wave length, I mean, if you look into the spares, I mean, what we're trying to do is provide value to the customer. So there are many ways to provide value. First thing, like we sign up for specs, which are actually more aggressive then the customer can achieve himself. For example, we have to keep the uptime of the tool, to keep the tool up. I mean we have our engineers who are very well trained that understand the hardware very, very, very well. They can easily much faster with -- to really fix the tool. So that we can provide lower cost to the customer. Example for spares, before a customer can just have his own inventory for spares. Right now, he doesn't have to worry about that. We take care of the whole spares inventory. We can buy -- because we cover all our customers, we can have a much higher volume to buy the spares, so we have more leverage. We can actually get them cheaper than the customer can get themselves so that we can really -- that kind of also value that we can provide from that point of view. Third, but also on the uptime. We can keep the tool up better than anybody else because we know the tool very, very well. I mean the other thing is really in this -- what we have seen in ALD, for example, ALD is all about new material. We have so many new material. And the material not only get deposit on the Wafer, but they can deposit everywhere on the process kit. So we know the material more than anybody else. So we know how to remove those materials. And we know how to take that part and clean it better than anybody else. Some customers, after you run like 10,000 Wafer, they take the part and throw it out because they don't know how to recycle it. We know the chemistry, we know what's there. We have very good chemistry folks in our team, so we can really recycle that part and put it back to the customer. That's really lower cost for them. So that's -- at the end of the day, it's value. It's lower cost to the customer. It's better uptime for the tool.

And let's take the next question from the conference call. Operator?

Sandeep Deshpande, JPMorgan.

My question is on the epi market. I mean you have the new tool shipping now and you announced a win in this tool as well. How do you see -- I mean, do you see yourself being able to create with this in epi and gate-all-around the kind of market share positioned at least in that segment, say, with gate-all-around or any other layer that you were able to create with ALD.

Thank you, Sandeep. And I'm going to try because it was a little bit hard to listen. But I think your question is directed at how do we see ourselves being able to, for example, win or create more market share for ourselves in epi in the gate-all-around inflection.

That's correct.

Yes. I think I will take a step at it, and perhaps Hichem again can give you more insight. As we have explained, we have a lot of engagements with all the logic foundry customers in terms of developments for gate-all-around, and that includes not just ALD, but also in epi. And the thing that is, I think, that has been shown about epi for gate-all-around is that you need to use more epi as far as gate-all-around is concerned. And I think Hichem had this great example where you turn in 90 degrees. And the channel structure is actually going to be epi-defined, whereas in the past, it was [indiscernible] litho defined. So that's going to drive that. But I believe that there are also other layers, in fact, more parcels of what we call that will require epi, and perhaps Hichem can share with you a little bit more.

Yes. To add on epi for gate-all-around application. I think what I've shown in my presentation that we really -- we have some differentiation in epi technology for this application. And we started into this business in the advanced CMOS business in 2017. Since 2017, we learned a lot. Our first engagement was on some with one customer on some application. And then right now, we have learned a lot. We have provided many new technology, many new innovation. And this data around transition is an area for us, which is transition. It's a disruptive transition. It's an area for us where we should position ourselves from the start. So for this time, for the gate-all-around, we actually position ourselves in the gate-all-around with all 3 leading customers from the start from the get-go. So we are involved very much with them on this technology. We -- I cannot mention market share from that point of view, but what I can tell you, and you have seen in the presentation today that we have a differentiation in our tool, in our technology, in our productivity. And I think that will speak for itself.

Okay. For the next 2 questions, we move back to the room here.

Marc Hesselink, ING. My question is actually about how did you get from the assumptions on the market growth or the EUR 100 billion WFE market, to your addressable markets on ALD and epi. Do you know how many layers they will for any of those products? Is that the discussion that you have, is how good is that visibility little bit a few looking on [ what ] you get from the top to that addressable market per product?

Marc, thanks a lot for this question. It's very complicated. So I'm good to let Michael Han try to explain to you how we do the modeling, Han?

Yes. Yes, Marc. Yes, so we have indeed a detailed model using the, let's say, the knowledge we have about power chip is being built, the different layers. We make certain assumptions about Wafer starts per customer per technology node by time. So that by itself will also lead to a certain assessment of the Wafer fab equipment market. So part of the whole model is what will the Wafer fab equipment market bottom up eventually be. So the numbers that we showed today are based on a range of WFE in 2025 between $90 million and $110 billion. So the midpoint is $100 million. But of course, our interest is an assessment of the ALD and epi markets, which we shared. So it's using our knowledge that we have gathered over years of working in the chip industry, specifically in ALD and epi. And so we're basically also scaling the assessment on a certain number of WFE by 2025, right? So if investments by that time would be higher, then the ALD and epi market would be higher. That's also why we give a range for ALD and epi by 2025.

Yes, we get back to you, Marc, for a follow-up. But first, let's move to the next question.

Maybe I can add a little bit. Maybe I can add a little bit to what Han just said. Within the office of the CTO, we have a number of technologies who really try to understand what is the process flow of these customers. I showed you one example in gate-all-around with a small green bullets and the larger green bullets. So we really try to understand what the specifications are way out. And that's our departments collaborate to give their input to, yes, as best as possible forecast the sum. Yes.

It's Peter Olofsen of Kepler Cheuvreux. I have a question on the competitive dynamics. If you look at the bigger players in the deposition market, they have a broader offering than what you have. At least one of those bigger players have been talking about integrated solutions where they bundle different technologies and can then process the Wafer all in infactuum. To what extent do you see that as really something differentiating and that's something that could be a disadvantage for you? Or do you think that with subsequent notes, your clients will continue to go for best-of-breed solutions.

Sure. I will maybe try to answer that question and have my colleagues add on to that. I think when we talk about integrated processes, for us, our focus, of course, is on deposition. And that doesn't mean that we only do deposition blindly. I think my colleagues have shown that. For example, the more advanced ALD and epi deposition process, surface clean is very important. So we have developed that technology on our own. And it's really being clustered together on our ALD and epi tools and being delivered to our customers. Now one thing that perhaps is also a characteristic of the industry is that you find that the different process steps to be integrated. That is really the core know-how of the semiconductor manufacturers, the chip manufacturers. And will they really want to buy something that is bundle integrated. Actually, the preference for them actually is to perhaps make sure that they buy the best of each type of equipment, and they do the process integration themselves. I think maybe Hichem can also give you some idea of what he sees in his discussion with our customers.

Yes. I think I showed today in my presentation that some examples of integration. I think what's important is that we are a deposition company and we play in the deposition market. And everything that can help our deposition process, we use it. For example, when I mentioned ALD becomes very, very [indiscernible]. You have to have the surface clean. So we have developed a reactor, which is ALD-based item-level etching based reactor to clean that surface. And that technology, we have both technology. We have radical based technology, which is a soft way to clean the surface. We also even have a plasma-based technology. We have the know-how. Since we are plasma deposition company, we have that plasma knowledge to apply it to remove material. I mean we -- okay, we are plasma. We also have ALD knowledge. ALD is Atomic Level Deposition. And what's most important ALD is chemistry, and we are leaders in the chemistry. So instead of removing -- instead of putting something, you want to move something, it's very simple for us. So that's how we can use ALD. But we use that ALD for deposition purposes only. So that's really the things that we have done. And right now, even for epi, like the premium reactors, we have hundreds of those in production right now in the industries. So we have the knowledge for that, not only technology but also HVM manufacturing. Same thing -- other thing that we -- you cannot see here, but many of these processes that you see this huge gap fill [ 100:1 ] even [ 200:1 ] gasfield capability. We actually -- you put your Wafer in and get it out. But during that ALD process, there are many things, like there is like what we call like institute integration. So we integrate other things is actually in our reactor. I cannot say much about it in this audience. But those are the kind of things that we do. So integration, yes, we can do In-Situ integration. We can do Ex-Situ integration. It's very normal for us. Anything we need to do to allow us to be leaders in the position we do after and do it.

Okay. Operator, can we have the next 2 questions from our conference call participants.

Krish Sankar from Cowen and Compnay.

I had a question on the clarification. The question is on 3D DRAM. You spoke about it being up introduced in [ 2026 ]. So do you think for DRAM is a temporary solution? And when you move to 3D DRAM, what are the steps that will be really critical in 3D DRAM. And then that classification, Hichem, you spoke about ALD. Do you have an ALD integrated tool already today with customers? Or are you developing it?

I think the question came from Sankar from Cowen. And I got the first part of the question, which is our view that 3D DRAM or we mentioned that 3D DRAM is probably a 2025, 2026 thing. Is that going to be replacing? Or is that going to be a bad for EUV. And what other steps could ALD play in 3D DRAM. I think I got that part of the question, but not the second part. But let's try to answer the first part first. And maybe this is something that Ivo you would like to take.

Yes, that's good. So we think this transition, as I explained earlier, turning things on one side. It's getting more defined by [indiscernible] than defined by lithography. It's very similar to what happened in 3D NAND. However, we went from planar to 3D NAND. Here we go from planar to 3D DRAM. So I think, yes, whether that will affect the EUV steps, you should ask that tomorrow at the ASML conference, I think. But certainly, there will be more dimensions in the device defined by the layer thickness and how we do the deposition and directly by lithography.

And just a clarification, do you have an integrated ALD and ALE product already today?

Well, I explained in the selective ALD effort we have, there are some steps that you might call ALE. And indeed, they have integrated, yes. So it's a little bit what Hichem already said. Whenever we need a technology, whether it's surface clean, whether it's promo layer etching to remove a certain component of the surface. We develop this as long as it supports our deposition process. That's our strategy.

Thank you, Krish. Operator, can we have the next question?

Dominik Olszewski at Morgan Stanley.

I appreciate we are still some time away from gate-all-around introduction. But do you have a sense of what portion of the $1.2 billion of advanced market growth that you flagged around the transition that ASMI has already won or is in pole position and strong position to win.

Dominik, really sorry, I didn't catch the question. Anybody caught the question?

I think the question was on the $1.2 billion increase in 2025 for ALD and epi for advanced logic how much have we won already of that? Or are we in a good position. Was that the question?

Exactly.

I think you should take that one.

Yes. Yes. So of course, as both Ivo and Hichem and myself explained we are deeply involved with all key players in that space, working on that transition from FinFET to gate-all-around. We have won certain applications effect we have shared a few months ago for one player. But it is, of course, too early to say and be more specific on specific wins in that space. But it's -- we are well positioned to develop with the customers that technology. So it is definitely an area where we expect to basically play a key role and have a key share by 2025.

So we can say we are in good shape.

We're in good shape.

Okay. Thank you Dom. For the next 2 questions, we will move back to the room here. Valentina on the second row -- Sorry to make you move Valentina.

It's Basin Rockman from [indiscernible]. Maybe on segmentation of your revenue target in 2025, and the growth. If I understand correctly from the presentations, it seems to be more logic driven as your percentage in revenue is already more logic. And then after -- or in 2025, there's a big inflection with the 3D DRAM. So maybe in the future, there are more opportunities for memory. Do you see like a difference in the coming years for your segmentation between the two?

Sure. Our view is that you will see both growth in logic foundry, and also memory. But logic foundry being a larger part of our business today, we still do believe that logic/foundry was to be the bigger part of our business in 2025. Now we hope that as we go on over the next couple of years that we will increase our share in memory and maybe that difference between logic and foundry and memory get smaller. But we do believe that by 2025, we still would have a larger portion of our revenue in logic and foundry. Now having said that, I think your question was, is 3D DRAM going to impact. 2025, we don't think so because probably at best is just maybe the beginning of 3D DRAM. As Ivo has mentioned, it's probably a '25, '26 event. So it will take some time. But having said that, we also need to look at how much investments are going into 3D DRAM, and how much or how fast they're going to accelerate the transition. Because a lot of investments right now is still being put into a plan DRAM. So we have to see how that plays out.

[indiscernible] you want to increase your cash to -- from EUR 300 million to EUR 600 million. Should we assume a gradual step-up of some EUR [ 60 ] million per annum. Just also to calculate the excess cash. And actually, why would you? Because you still have over EUR 1 billion in ASMPT? And linked to that, what's your current thought about having your interest in ASMPT? And linked to that, what kind of shareholder engagement attitude do you have towards ASMPT since if I look at it's performance, it's clearly lagging it's peers.

This maybe, Paul, do you want to take this?

Yes. No, that's fine. Maybe first on the cash. Indeed, as I mentioned, we want to bring it up from EUR 300 million gradually to EUR 600 million. The key reason is financial flexibility, like it was in the past as well. We are a knowledge intensive company. There will be good years and there will be less good years. And of course, the moment and we want to pay a sustainable dividend as well. So at the moment when that happens, we don't want to immediately trigger a major restructuring effort in our R&D organization because these are key people that we need to retain to, of course, continue the innovation that we talked about today. On the other hand, if opportunities come by, I also talked about M&A, we don't know, it will help to support and to act quickly on opportunities as well. So the key word is financial flexibility. And actually, I can say the same about ASMPT. The important part here is there is a historic element. And we didn't buy that stake just to have that stake. It's something from the past. ASM has reduced the stake now to 25%. We see this as a financial investment. It will also provide financial flexibility. You're right. And at the right moment in time, of course, we might act on that. We don't know. We have no plan for that. We are pleased with the current holding, it's a financial investment for us from the figures I've shown. I think it did well in the past. You saw the cash return to shareholders. A big chunk of that came through dividends from ASMPT as well, of course, reduction in our stake in ASMPT. But again, we'll review it regularly. And for now, we're pleased with the stake. And as I said, don't forget there is a historic element to it. Then in terms of performance, yes, we have limited influence. We have 25% stake. We have, of course, board seats. So we, of course, participate in that actively. We see what's going on. We supervise what's going on. We help support with the strategy. We clearly recognize that the share price performance is lagging and the company is taking actions to improve that. And I think the recent performance actually was significantly better than what you have seen in maybe in the years before that. But unfortunately, the share price didn't move, but that's not in our hands. We can only try to manage the performance of the company.

All right. Operator, can we have the next 2 questions from the conference call.

Tammy Qiu from Berenberg.

So I have a question regarding the TAM. Do you have any idea about pace of the TAM expansion, i.e., it's foundry logic on expansion will come ahead of the memory TAM expansion. And have you stress tested any WFE scenario versus your term expansion, i.e., if the cycle ever become lower than we are today. Will your time expansion story be delayed as well?

Tammy, thank you very much. I just want to make sure I got your question. I think your question was on the -- do we see a difference in terms of TAM expansion between logic, foundry and memory. And if there is going to be, let's say, adverse movements in either one of them, is it going to impact, for example, our market share significantly, should that happen?

Yes, exactly.

Okay. Thank you very much. I think Han, this is another one for you.

Yes. Yes. So the TAM we shared that the growth in the ALD market is for the majority in logic/foundry, but also in memory. So the pace in both is quite gradual. We are in both areas deeply involved with customers. And so also in memory, there is some solid development for us. If there is some, say, delay in one, so in 1 of the 2, we have a stronger share, a stronger position in logic/foundry. So it could theoretically be in one instance if memory, all of a sudden, theoretically, would move a lot faster, that we could see some, let's say, some setback. But again, our recent share in development in -- or potential development in memory is also quite healthy. So the pace of both is quite the same. There is -- if we put it out in time, we see both markets developing quite well.

Han, maybe just to add on that. I think you had one chart that shows that the increase in the market size for ALD about $1.2 billion of that was going to be attributed to Logic/Foundry at about 0.7% for memory. So based on that, the logic foundry piece is still a larger piece for us.

Yes. Yes. yes.

Tammy, I hope that answers your question.

Okay. Thank you, Tammy. Can we have the next question, please. Operator?

Adam Angelo at Bank of America.

It's actually Didier Scemama from Bank of America. Can you hear me gentlemen?

Yes. Yes.

Great. I've got 2 quick questions. My first question is with regards to the design win visibility you have. You mentioned, obviously, your strong design win momentum in [indiscernible]. Can you just give us a sense of your visibility, EG, how long do you think you will be designed in for sure in the recipe of your customers? Is it maybe 3 years, maybe 5 years, maybe a little bit longer? And then secondly, very briefly, the question for Hichem. I love the analogy of ALD is the new CVD. But that means that you will be able to address a substantially greater portion of the deposition market by, let's say, 2030. Because right now, you addressed 50% of the position market. So maybe just give us a sense of where you could be there.

I believe -- I think the second question was how much do we believe in the, let's say, ALD is the new CVD growth or potential and will ALD maybe perhaps even overtake CVD by 2030, I think that's the second part of the question. And the first part of the question was how much visibility do we have in terms of designing, I think that was the question. So maybe I will try to take a stab at the first question and then Hichem can talk about the CVD, ALD, let's say, question. It really depends. There's a wide range of, let's say, visibility that we have, because it depends not only on customer, it also depends on the kind of work that we are doing for them. It could be that even for some of the, let's say, more nearer-term work that sometimes it's just maybe 9 to 12 months, and they want to pull in something that we have been doing for them or doing with them into high-volume manufacturing because they have an issue that they need to fix. It could also be that it's an application that perhaps they know that they will need, but it's 3 to 4 years out. So it's kind of difficult for us to quantify that. But I think what we can say is that we are working with the customers, both in logic and foundry and in memory on a wide spectrum of, let's say, such applications. And if you look at every quarter when we announce our results, there is one item on the balance sheet that tells you about e-mail tools or evaluation tools. And that's something that has been growing for some time. It's basically because we are doing a lot of evaluations and development with the customers, but that's a good sign because that once they are accepted, is actually an indicator of future revenue. So maybe Hichem can explain or talk to you more about the ALD, CVD.

Okay. So when we talk that ALD is a new CVD, what we're trying to say here is that we see that the ALD has the same growth path as the CVD 20 years ago. At that time, in CVD, we have many deposition techniques, many new films. Right now in ALD, we saw the same thing. You can deposit ALD in many ways, and we have many new materials coming with ALD. But I think the key here is that the technology is moving from 2D to 3D. So if you were in the position company, you're in good shape. So CVD is going to grow a lot because going to 3D. ALD is going to grow because we're going to go to 3D. So both of them are going to grow. So both CVD and ALD is going to grow. But what we are trying to say here is ALD is going to grow much higher pace than CVD.

Okay. Thank you. Let's move back to the room for the next 2 questions.

[indiscernible] from KBC Securities. I had a bit more of a higher level question. Touching back up on what you said very early on in the presentation talking about government initiatives driving the Wafer fab equipment market growth. A few weeks ago, we heard Ms. [ Van de lan ] talk about the European Ship Act being one of the objectives of supporting development of European mega fabs. So I was curious whether you could provide a bit of your thoughts on yes, what that would mean for ASMI and then also in terms of the supply chain, bringing those perhaps close to home would be the additional opportunity for you.

Sure. I think this is something that is still being played out. There's a lot of discussions by various governments. And as I said earlier in my slide, that could actually be additional investments because governments want to secure, for example, the semiconductor supply chain. So they might be building fabs. But I think we are hearing, of course of discussions or initiatives in the U.S., in Europe, Japan, Korea, even India, but I -- I don't think that there is anything that is fully decided or concrete yet. Of course, we need to wait until what the U.S., for example, the chips act and so on, what's that going to do? How is that going to help? -- let's say, companies in terms of investments in semiconductor manufacturing. And similarly, the EU, I think they are having a lot of discussions at the EU level. But have we seen any kind of firm statement yet? I think you read a lot of articles, but there's nothing that we can confirm yet. So we, as a company, of course, whenever fabs are built, especially fabs at the leading edge or advanced node is going to be a significant opportunity for us. We try to track and follow as closely and as much as possible. But at this moment, we are like you. We only know what we read from the papers. Yes.

Yes. Hello, this is Stephane Houri again from ODDO BHF and ABN AMRO. I know it's a CMD about 2025, but you have also given an update on your order guidance for the quarter. So could you maybe tell us what it changes for your outlook for Q4? Because you just said that Q4 would be at least at the level of Q3, should we expect that Q4 will be significantly above Q3 for Q4? And what can you say about 2022?

Sure. So we -- our guidance for Q3 has been, I think, between EUR 400 million to EUR 430 million for revenue between EUR 510 million to EUR 530 million for bookings. And we have already seen NRC in continuing. I mean, the quarter is not over yet, but we have really seen very strong order intake. And that's why we kind of revised our guidance for bookings for Q3 that this would be above EUR 600 million. And for the revenue part of it, we stick to our guidance, which is between EUR 400 million to EUR 430 million, we do not see any change there. As far as Q4 is concerned, of course, it still is heavily dependent on I would say, some of the supply chain lingering supply chain issues that we have. And as we go into October, I think we will have a better view. But we do believe, especially if you look at the kind of backlog that we have, that Q4, if we do not have, I would say, crazy supply chain issues, our Q4 revenues should be higher than Q3. So there's no change in that. Now as far as '22 is concerned, of course, with the backlog that is building, you see a kind of a lag between bookings and revenue. So of course, some of the bookings are already going to be delivered in '22. It's still a little bit out. So visibility for the full year is too difficult. But I think the beginning part of at least '22 because of the backlog, I think we are in good position. And until we are much more clear about whether there's going to be any more supply chain issues, we would refrain from giving any guidance about that.

Okay. Let's now go to Marc Hesselink for his promised follow-up.

I think it was also discussed a little bit in the discussion afterward. But to really get the idea on what the -- how do you know how many layers you will have in this 2025? Is that a discussion you have with clients? Is that your own interpretation -- How good is your visibility on the number of ALD layers, number of epi layers in 2025?

I can take that question. I think we have -- if I can answer this question, we have good visibility with our customer. So I wish we can have like a magic formula to give you, okay, from one generation to the other, the number of player increases. It's really different from a customer to the other. The reason it's really the implementation of technology from one customer to the other is very different, especially in ALD. ALD, if you have different integration scheme, then you can reduce the number of ALD layers. So really ALD layer depends on patterning scheme that you use. Depends on whether some customers are more aggressive in using new material in ALD, others are not. So we have extremely good visibility with our customer, customer by customer of their road map for on the ALD side, on the epi side. And based on that, that's how the number that was presented by our CFO and our revenue is really based on customer by customer, bottom down understanding of their technology road map and the integration.

Operator, we take one more question from the conference call.

Aditya [indiscernible] from Credit Suisse.

Yes. So 2 questions, please. Firstly, just looking at the selective ALD that Ivo talked about in the session earlier, you said that it wasn't included in the forecast. So I just wondered if you could give us your thoughts around how we should think about this opportunity from selective ALD, just any building blocks around how we could think about this opportunity would be very helpful. Secondly, my question is for Benjamin. So I just wondered if you could clarify how big your revenue from domestic Chinese customers is currently? And how do you see the risk of ALD shipments to China being blocked as gate-all-around increasingly uses ALD in the channel is increasingly determined by ALD steps and EUV litho steps. And then I have a follow-up if I'm allowed to ask it.

Sure. Again, I think the first question was on the selective ALD and the opportunity. So I'll perhaps let Ivo answer that first, and then I will do the part on the domestic Chinese.

Okay. Yes. As I told you in selective ALD, you have different ways of implementing that. One of the waste to a selective silicon nitride specifically is something which is today already in production. We see potentially several new or additional applications of selective, toposelective ALD. But many of these still need to be qualified and developed so that revenue will not materialize before '24, '25. If I look to surface selective ALD, that's a little bit more complex. We need a little bit more technologies coming together. That's mainly logic/foundry driven. What we see there is 2-nanometer adoption. So it depends how fast our customers will introduce to 2-nanometer technology node, how fast we will see that developing. But the major part of the revenue will be beyond 2025.

And maybe on the situation in China, especially with the domestic Chinese customers and maybe a lack of access to advanced ALD or app tools, I think the situation is still very much a one that is in flux in the sense that the restrictions have not gotten worse, but at the same time, they have not gotten better. So there was a time when the restrictions were hitting us like or new restrictions or new regulations from the U.S. government was coming once every 2 months or so. The last, I would say, half year or so has been quite normal. We don't see really new stuff coming up. But at the same time, it's not gotten any better. Our view from ASM is that we follow all the regulations of the U.S. government. So if our products are impacted to the U.S. legislation or export control, we will follow that 100%, maybe even 101%. But the fortunate thing for us is, the majority of our products actually today, they are not required to go to the export license requirement. We have a -- we do have a small minority of products that need an export license. And for those, we are following the same procedure. In terms of China, I think you look at China. And last year was a great year for us, first time that revenue was more than 10% of our total equipment revenue. And this year, we are seeing a continued growth it is actually a great market for us. The big thing or the good thing about us China as we are less dependent [Audio Gap].

It's Nigel van Putten with Kempen. Maybe just tacking on to some of the questions answered earlier. How do you think about cyclicality in the industry now? I mean, clearly, there is a strong model underlying your assumptions. But to what extent is that a straight line? And also maybe to complement that question rather than asking another one. To what extent are there significant milestones in the next couple of years where you see maybe your revenue increasing disproportionately?

Paul, maybe I do part of it, you do part of it.

Yes.

I think cyclicality, as we know it, for some of us who have been in the industry a long time. These days, our cyclicality is so much less compared to 15, 20 years ago when you had the boom bust cycles. Does that mean that you're not going to see quarterly or even yearly variations you will. But I think by and large, we look at where we are today in 2021, and we look at the growth in the semiconductors, and we look at our customers, having made multiyear investment announcement, it seems like the demand is very strong. And that's why they have the confidence to do the kind of investments. So I think over the next couple of years, it's going to continue to grow in terms of, let's say, fab or WFE whether you have one year that is slightly lower than the previous year or so, I think that's possible. But overall, the trend is it is going to grow. And in terms of milestones for our plan, and maybe Paul can share that with you.

Yes. I think the major -- we've talked a lot about it. The major milestone is the gate-all-around, to high-volume manufacturing start. I mean that's a major inflection for us. You've heard the benefits for ALD for epi depending on the timing when that happens, if it all comes in one go, customer by customer, it will be a major boom, maybe it goes in steps, but that inflection is an important milestone for us.

We are running out of time. So over to you, Benjamin, for any final remarks.

No, I really want to express our appreciation for everybody here in the room and also who have taken the time to listen to our webcast, bringing the first time for us. We hope we have done a reasonable job in trying to give everybody more information. And so that you have a better understanding of ASMI, our, let's say, opportunities, our strength and our strategy. And we certainly look forward to continuing to update everybody as we go along investor calls, some fireside chats that we get involved with. We hope to continue doing that. Once again, thank you very much today for taking the time. I know it's been a long session, but we hope it has been a satisfactory one for you. Thank you very much.

Thank you. Operator, can you conclude the call?

Are you going to say that? Or do I have to say that? We have prepared something which is on the second floor. Please join us. Please join us, yes. We are also -- the 5 of us are quite happy that this is done. Trust me.