Good morning, ladies and gentlemen. Welcome to BE Semiconductor Industries' Analyst Presentation 2018. We've prepared today some hot topics about, of course, the market, about technology, and especially focusing a bit more this time on where the business will come from in the next foreseeable future. Some more about strategic initiatives, also about our cost reduction programs, capital allocation, and a few words about our acquisition strategy. Let me start. First of all, a simple summary. All of you who have followed us in the past years maybe recognize a little bit some of the statements here. A disciplined focus on executing our strategy in the high end of advanced packaging. Clearly focus on those areas where we make better products than our competition, and by doing that with ever lower cost and closer to the end markets, i.e., Asia, and more and more in China.
This has led to a company which, first of all, grows faster than its competitors and peers. Moreover, financially has demonstrated performance which is above any measure in this industry so far. Let's go into some details. If you look at this market, and talking to some of you before the meeting, everyone believes that this industry is no more cyclical. If you look back in the past few years, we recognize certainly some cyclicality. At this moment, we are still in a wonderful period of strong growth. If you look at 2017, VLSI in the past few days has just further increased its estimated market behind us, and that represented in 23.9%, 24% growth over 2016. That leads to, if you take the same number for 2018, our last statement was 12% growth. If you increase the comparison, it leads simply to the 7.7% growth.
Anyway, these numbers go up and down every quarter. More importantly, how does Besi perform in this whole landscape? That you see here. Last year, significant growth of close to 58%. If you look at the first quarter comparisons, it is again 40% more. Anyway, VLSI is the only one who details out the back end segments. If we believe it or not, these are the numbers out of VLSI Research this world. Those comments then, if we look at the latest updated VLSI 2017 market estimate, 2018 growth will be single digits if they don't adjust their 2018 forecast. That would then mean to 8% growth. 2019, all industry experts are in agreement that that should be a cyclical downturn year with always an upturn predicted for the year thereafter.
That's still far out in a world where feasibility is mostly one quarter and maybe maximum two. If you look at the growth so far in the industry, there's always a low and a mid and a high end. There's still certainly growth, especially in the low and the mid end LEDs, but also low-cost semiconductors, driven also by strong investment programs in China. We've seen so far a slowdown in the high-end smartphone capacity investments. More details about that later. In other words, customers are digesting a significant capacity increase for all kinds of new features into the high-end phones last year. So far, the 3D imaging has not been a game changer yet. Sometimes that takes a bit longer for this technology to roll out to other phones as well.
Basically, you can say that the mobile internet slowdown has affected Besi's revenue development, so less growth than what we had last year, but still growth. That is basically the explanation of what happened in the market. Where do we go from here? Of course, long term, there are always significant growth drivers. All the big data. Later on, Ruurd will give some more details about that. Also very important for the development of packaging requirements. Ever smaller designs require ever more complex interconnect solutions, and that's where Besi is a significant leader. Even more so, assembly has become an ever more critical step in end applications. Simply think about smartphones, but also photonics in that matter.
Ruurd will spend some more in detail, technical, to explain to you why those drivers are of significant importance to Besi's future growth. Apart from that, already earlier mentioned, China's big government-supported investment plans, where we are benefiting already since several years significantly from, but also with our ever-increasing manufacturing capabilities in China, we will definitely benefit more in the years ahead. More and more local for local, so China for China, and that is driving our revenue growth for the longer term. Also, the cloud to memory logic world, everything around cloud servers, higher memory demand, more complex memories stacked. Also, technically, Ruurd will tell you some more about those latest developments. However, this industry is like other industries, but we've witnessed this many times very directly. GDP is an enormous, big influence, both in the up and in the down, to what's happening in our industry.
So far, GDP is excellent. U.S. for many, many years. Growth, that has supported also a strong automotive business development, where we are also benefiting significantly. The three drivers again for Besi: computer world, communication, automotive. All of them today are wonderfully intact. Then the ever-expanding spares and service, especially simply imagine last year we installed over 1,600 machines in the world, and they all need spare parts, they all need service. That business is also supporting our growth in the future. Strategic initiatives to drive down cost, I will spend some more about where we are and what is to come to be expected in the near and longer-term future. Very important, not only leaders in technology, but also leaders in our operational model, and that is cost and time to market, so manufacturing throughput time.
We have demonstrated so far that we can faster accelerate than anyone else in upturns, but also in downturns, we can control our cost. Now nine years in a row profitable, 32 quarters, that should further continue, but also can improve. I will spend some time explaining that part. This is the beginning, the introduction. Ruurd, stage is yours.
Yeah.
Please tell us more about these beautiful pictures.
Yeah. Richard, thank you. It's always a pleasure to be here and explain a few more, a little bit technical background. I start a little bit with the digital society. What is our business? First of all, of course, is this mobile world. Secondly, the whole interconnect over the whole world. Big data centers. The automotive world is really changing at this moment. Also, this is another field. The volume is not that big in there, but it is DNA sequencing, for example. The whole healthcare issue is going up. There is a lot of computing power needed for that also. There's even, I was two weeks ago at a conference from imec in Belgium, there was a nice presentation of replacing semiconductor memory by DNA memory, because DNA can keep much more information than a normal semiconductor memory. Now, that's 30 years down the road.
We don't have to put attention to that now, but that's some of the thinking. Then in the environmental area also, there's still enormous drive to reduce energy to make devices more efficient. Now, people say, "Oh, there's no downturns anymore. Things are going up and up and up." I started 1984 in this business, I've heard this before, so I'm very skeptical about that type of remarks. If you look at the usage of data in the internet area, PC, in the beginning, the last till 2010, '11, we were in the, I would say, still in the gigabyte area, then we get in the exabytes, and at this moment we start to go to zettabytes. If you drive from Amsterdam to the north of the Netherlands, you pass close to a little bit past Purmerend, you see enormous big green building.
At first I thought that is a facility to dry grass, a farmer's thing. It's the biggest data center for Microsoft. Enormous data center. If you look at all this data is driven by 5G is coming up, artificial intelligence, how to handle big data. Robotics takes a lot of data in transportation, environment, healthcare. Now we talk about zettabytes, but these are areas where you start to-- What is a zettabyte, huh? First of all, a zettabyte is a billion terabytes. You can ask what's a terabyte? One zettabyte is about 1,000 data centers. One zettabyte is also the energy of 180 million homes. This gives an indication that in technology we have to move forward because you can never handle all this type of energy. You have to continue.
People sometimes say, "Oh, Moore's law is dead, it's not happening anymore." I don't believe that. I believe the world will continue. Yes, it will become more difficult, but it will continue. If you see in the technology areas, there's a little bit differentiation here. You see for the IoT world, for the connected devices, you need devices that have very, say, an average performance, but very low power. This is the power axis. This is performance axis. There you need certain devices. In the middle area, you have devices which have reasonable performance and reasonable power usage, because one of the biggest problem in your mobile phone is still the battery. At the high end, power is a little bit less an issue and you really go for performance.
To do this and to keep on in the technology, there are three key areas. One is dimensional. This is where ASML is trying to get from 10 nanometers to 5 nanometers to 3 nanometers. That's dimensional. Secondly, you start to come in new devices. FinFET was the last word. You start to go to quantum, to nanowires, 3D stacked devices. You start looking at can I put cells more easily together? The next step is putting it all together in like a big Lego box. The last thing where people are puzzling with at this moment is, the computer today is what is called a von Neumann architecture, you have a computing part and a memory part, people are thinking about new setups for the whole way we compute.
You start to hear things like, "Oh, quantum computing is going on." These things will take time. There's a lot of development going on. I firmly believe that this world moves forward and will find solution to these things. If you look these changes in, say, technology, where you go from 10 nanometer to 5 nanometer all the time, you have to have new processors, new equipment. If you look at the front end, we got the FinFET, people are looking at what is called nanosheet or complementary. A big change in the last year was what is called the stack mount, where you have at the front end a whole layer structure builder. This is what drove [the lines, the applications]. That was really where the big business for them was in these memory devices here.
Today, with the limits of materials, you start to see new materials coming up. Applied Materials just recently introduced a cobalt wiring that is for the first time a new wiring contact technology to speed up the devices again. All of this in the front end and driven partly by companies like ASML, drives us to new equipment also, because otherwise you cannot use this. The second element that is very good for us is that this becomes indeed more difficult. The time to next generations, it takes more time, more effort. There are fewer companies doing this because the investments become very high. That puts more attention to the back end, trying to solve some of the problems that are difficult here with new structures where you stack things together, where you couple things directly together.
Also photonics starts to play a role because all this data transfer, at a certain moment, you cannot do this with a copper wire anymore. You have to do this with light, basically. That means for our world that we go to higher and higher accuracies. We are now talking equipment that has 200 nanometer accuracy, so that you start to come in nanometers. It has to be super clean. We have start handling photonic elements. We do more in photonics than most people think because we do already a lot in photonics. Of course, you got TCB, I think I spoke last year about it. Direct bonding is coming. People try to get rid of substrates. Fan-out is coming and in some cases, you want to go to very large panels with a high accuracy.
This means for us, all kinds of new possibilities in getting equipment in the market. ASML has always been at the higher end, so this is an estimate. Where is our business? You see, here you see the nodes. 28 nanometer node, 10 nanometer node. Here you see the required accuracy for that technology. In the 10 nanometer node, you need about 3 micrometer, 2 micrometer accuracy. Most of our business is in this area, in the higher end. You see 55% is in the 7 micrometer accuracy and nanometer, 70 nanometer sub-nodes. We want to continue that. We want to stay at the forefront of the technology. Here you see in the different markets, almost everywhere we are in with our technology.
If you look at where we get our revenue, because at the end of the day, we need to sell machines and basically earn money with it. The biggest is mobile internet devices. Second biggest is everything with computers. Automotive is a very important one. Industrial, more general, about 10%. We do LEDs, but only high-end, high-power LEDs, very specific LEDs. These are our three largest and most rapidly growth markets, this area here. Our internet share, our mobile share went up to 35%. It was typically at 30%, it went up to about 35% this year. Also in this area, in the computer, we have a stronghold here. Services and spares is about 11%, and that's a good business because it always comes back, people need this. That's good.
Richard indicated already last year we did about 1,600 machines, they add basically to the base where we can get services spares revenue from. Now in the mobile market, Richard indicated already it's slowing down a little bit, and yes, that is true, that it is a little bit slower than we all had expected. The paper about the 3D. This is the 3D device to do your face recognition. You have to put a beam out, you have to measure it. It's a fairly complex structure, this was an important one for us. You start to see second-tier telephone companies also starting to introduce this, but that typically takes one or two or three generations that follows. You see more cameras in it coming.
Augmented reality, you see a lot of. Are there any gamers in this room that play games with these telephones? I'm also old for that, but if you see how much gaming is used and how much augmented reality is in that in the meantime, that you are in this room, and you suddenly see a world coming out of a window here and strange things. That's a part, of course. I spoke last year already about 5G. 5G is a whole next level of networking, and people just think, "Oh, it's just going to get faster." There is something new coming in 5G that is millimeter waves, that is very close by waves so that you have very quick local connections, and that requires a lot of new technology in the phone itself. I think last year I indicated that a little bit.
We see that more and more coming. Of course, also the normal things are going better, processors, more sensors and so on. Here you see our growth in this mobile area. This was a real strong thing. We still think this will grow. It's not that heavy anymore, but for us, the absolute number of phones is nice. What is really important, are we getting into a next technology jump in these phones? We still get really new things in these phones. That will continue.
If you go into why our phones are important for us, if you simply go to TechInsights and you ask for a teardown of a phone, take a Samsung or an Apple or whoever, you get all these kinds of nice pictures where you see this is just one of the part of the phone, and you see how much semiconductor content is in such a phone. It's not only the processor where companies are very proud of, but you have memory in it, you have power management in it, you have sensors in it, you have Wi-Fi in it, you have receivers, transmitters in it, antenna switches. There's an enormous amount of semiconductors in these phones, and that only gets more because in 5G, you have to add another part into it. Here we indicated in 2012, we were in this part.
In 2017, we were able to get more parts in. You see here through that camera part, for example, you see here who is manufacturing this or designing this and who is then manufacturing. You see companies like Qualcomm. They are very much into the receiver part, for example, but also in a part of the processing. Qualcomm does not make anything. This is done by companies like Amkor, ASE, or maybe a TSMC. For us, it's very important that we track that very carefully, that we say, "Hey, this device, where is it made? Who is making that?" So on. Also what we do, and I think Richard indicated this. Every week we have a call on the market with the top management. What's happening? Where are we? We do that on Monday morning, 10:00 A.M.
We are not allowed to take holiday at that moment. Everybody has to call in. Friday we do it more on inventory and other things. You see sometimes that there's a PO hanging around. We get a request from, say, from this company. Amkor maybe from this company, from that company. You think, "Oh, lot of market," but often it is only one project, but they may put it there, or there. Tracking these things is extremely important, and yeah, we have a very good handle on that. Mobile world stays extremely important. There will be more content in here. People will try to put more together in one block. In our molding business, we see this, for example, where people say, "Hey, can we mold this whole thing at once?" If we save there.
For die test, we still have to put the dies in place. Second part, the computing world. Of course, the high-performance computing, the deep learning, artificial intelligence. If you follow nowadays in the car trying to track the traffic, you need an enormous power to do these calculations. It's not only, say, data centers, but it's also that's why I put automotive also in it. We sometimes have a little bit difficult to say, "Oh, should we put this in computer or should we put it in automotive?" The video processor might go in a computer thing, it might go in a game, it might also go in the car. Sometimes it's difficult to separate it. You see also here we do also very well. This is continuing growing very strongly with the server market going fast forward, storage market.
Tablet PCs is of course going down, but this will compensate nicely for that. This is also the area where it's most important that we keep advancing with more Not more, but with the technology going to the smaller nanometers, because you need that for the computing power. This is a typical example. This is an NVIDIA processor where you see the processing unit, you see the memory closely connected with an interposer here, with a substrate. These dimensions, this is typically a TCB application with maybe 40 micrometer delta between here and here. That means you have an accuracy of about, simple rule is if the pitch is 40 micrometer, the accuracy needed is 4 micrometer. It's about 9 to 10 factor. Here you see an example from an imec device where they copper layer more things. Now all these things have to be mounted.
This is all our business, mounting these devices very precisely. This is a graph showing the number of storage quantities needed. That is still strong growth. You see the different areas also. For us, this is also a growing market with a lot of high level names in it. Volume wise, it is in machines always a little bit less, but they're very expensive machines. In number of machines, a little bit less, but expensive and high-end machines. Automotive is a different world. Automotive, first of all, we see of course, this whole hybridization, electrification coming up. That means much more power devices in the car. You see more safety features, more sensors.
E-call, the new cars, they have to have a device in it that if you have an accident, that the thing will call itself and say, "Hey, I'm here and I have an accident." This is by law. I think next year this will come. Connectivity, energy efficiency, entertainment. Another element that is really new is car lighting. There's more and more trend to replace the headlights by LEDs, by steerable LEDs. You'll get a high-power LED assembly, and then by switching on one or two LEDs, you can actually steer the beam. Now you have often that you fall like this, and then you can have by counter traffic, you can not blind the other side. Most accidents actually happen overnight with bad sight. That's an area we are in.
Another element that is always very important in the car industry is the zero defect thing. Companies like Infineon and ST are forced now to show that they have simply zero defects. Not 10-6 defects or one out of a billion defect. No, simply zero defect. That requires for us, for our equipment, we have to prove that everything that came out of our machine was okay. You see here also the growth in this area. We grew very nicely, 23%. Automotive plus 11% is the forecast. We had a call a few days ago where the companies like ST, they have increased their automotive forecast. They see much more growth even. That was positive for us. The other thing in automotive is your phone. You switch maybe now with the iPhone X, maybe a little bit less because it's so expensive.
Normally your phone, you switch every one to two years, you buy a new phone. Now cars typically four or five years, and the car has to run 15 years without really problems. That means that also once you are qualified in the automotive, you stay in a long time. We are qualified with many of the big companies. If you take the Bosch, or maybe today Bosch is not a good example. The Continental, the Infineon, once you are in, these companies also prescribe to their suppliers, if you do this business, you have to do it with a Besi machine because that's qualified. Once you're in, that's very stable, good business. See here, this is from an Infineon presentation. The content of electronics in the cars strongly growing in the coming years.
This is basically fivefold increase in the coming 10 years. The other element that is strong growth, this is the part of expected hybrid cars. You see we are now at a few percent, this is also growing strongly. These are both two long-term drivers for this part of our, I would say, customer base that is very favorable. You see all kinds of devices here, whether they're power devices, silicon carbide, for example. These are combination devices, but also sensors. Also in the car, you get more and more general processors with a high level of security in it. These are the reports you got in the past, "Oh, I can switch on your car." You don't want these things, that level is also increasing in the cars.
Also for the automotive driving, you see very powerful processors getting in there. Very favorable terms for us. We're almost with all our equipment, we are in this. These are some of the customer base, but all the top guys we are supplying. Last part I want to spend a few seconds on is the industrial area. Everybody knows I'm a boater, so I hate these windmills in the North Sea now because they're planting everything full of windmills. It's good for the environment and robotics of course. Here you have a lot of power conversion, so you have to go from AC, you have to go to high voltages and so on. Here you have a lot of power and controlling, and smart homes are coming.
Appliances, everything. If you have-- you're all too young, but when I had the first electric drill, now I'm going really back, it was just an electric motor with a switch, correct? If you now buy an electric drill, everything has electronics in it. It can go slow. It can go fast. You turn it around and so on. Everywhere you see these type of applications coming in. Also this is also a real good growth area. Often in the power devices, you have an area where you have to connect power to something with a control. You have a lot of handling in here because you have to combine high power elements with a control. You have to stack these together. You have to isolate this. You see here some examples where you. Oh, let me go one back.
Where you see a whole stack of devices stacked together, combined together. That is our world. This is a nice one from Infineon where you have a digital power controller. It's digital, but it also controls power. That combination has a lot of handling for us in it. You always see in a device like this, you see all our activities. You see the molding in here. You see that we have to bend these leads. That's trim and form. Of course, we have to put a device in it. Here you see a little bit the last customers we have and also the equipment we are putting in here. This whole technology, and this is really our world, started with a simple package where you have a device, you put a wire to it, you have a molding around it.
We start to flip chip this, and we started with these ones. We start with simple wafer level integration. In the mid end, you go from more to a substrate where you have much more connections here on the bottom. Going to a flip chip substrate, you go to a package on package structures here. You go to an embedded die or an embedded die closure structure. This is all handling. This is what we do. We place these devices, and we package them. All of this is growing. The most advanced ones, you see stacks like this coming up now. You also start to see where you have two dies. You start to see two dies with an RDL layer. In these substrates, you start to see where we try to make connections here.
In the sensors, in the past, you had one sensor for one function, like pressure sensor or maybe a temperature sensor. Now you start what is called sensor fusion, where you get all kinds of sensors in one package. Now you have a one by one millimeter sensor where you have an accelerometer in it, you have a pressure sensor in it, a temperature sensor, and that means also more integration, more handling for us. One term you hear often now is fan-out. What is happening in the world is it started, what I said, is you put a device here. With epoxy, you connect this. Then you wire bond. That's not an area where we are in. You mold it. That is where we are in. The next step was flip chip, but still on a substrate.
People still say, "If I want to make this thinner, I have to take the substrate away." That is what's happening now. You start to make the whole, I would say, connection layer here. You start to make directly on the device here, and then you shave again 0.2, 0.3 millimeter of thickness. That's important, for example, in the phone market. How do you do that? You start with a wafer-like carrier. On the carrier, you make this whole structure here, all these electronic parts, that's a lithographic process. Then you put the die on it. That's what we do very precisely. Then you mold this whole wafer, and then later on you cut it in pieces. That's in a short mold fan-out. You're not handling individual.
You make a whole either wafer or even a panel full of dies, then later on cut them out. The big advantage is you can completely forget the substrate and do it directly. In practice, this looks like this. You try to replace interposers with this. Here you have an interposer. Here you have the level directly on the chip. You see that here. Here is this whole structure, and then you connect like this. That area is an area where we are, I think, one of the masters or the master. You start to see this everywhere. Here is the substrate here. I have drawn this thick to show, okay, these are the electric lines, but these are very thin layers.
You start to get combinations where you have a package on a fan-out package, where you may still have this interposing, but there's a strong drive to bring this fan-out to the world. Let's talk a little bit about machines, because at the end of the day, it's all very nice, this technology, but we build machines. First of all, I try to explain this, we focus strongly on this advanced area. In the low-end area, at a certain moment, you cannot make much money anymore, then you leave it. We very strongly have to work with the top players because the developments are more and more. You see companies like the Samsung, the TSMC, the Intel, they're all starting to refocus on back-end. Companies like TSMC, they have a 1,000-man team now for back-end.
In the past, it was just front-end, but they came to the conclusion our front-end is growing slower and more difficult. We have to do also back-end. You see the same development of companies like Samsung, Intel, and that's very good for us. That means also we start talking to a different class of customers. These guys are different. They have a different way of thinking than the typical OSAT companies. We try to stay leading edge. We refresh our products every one or two years. We have to do that for two reasons. One is if we sell a machine, H-Epsilon, and we sell it next year with better performance and we still call it H-Epsilon, we have to give it also another flavor. It's going to high speed or going to call advanced.
Step by step, get a better price for these machines. Yeah. EWLB, fan-out, TCB, direct bonding, ultra-thin die bonding, large area, wafer level molding strongly, and also plating is coming back. On many areas, we now start to see a renewed interest in plating, especially in the automotive area because of the demands for reliability. We have not only plating, we have pre-plating processes where you get better adhesion of material later on, so that the reliability increases. Our product portfolio, die attach, die bonding, multi-module attached here. Multi-module means you can do multiple types of dies in a device. Flip chip, including TCB and fan-out. Die sorting, lid attach. I think I mentioned it last year, but all the processors from a specific company in the world are mounted on these machines.
This is very precise placement of this little metal deca, how do you call it? The lid that you put in there. We can do this so precise that you can increase the power, the calculation power of the controllers by 5%-8%, simply because the cooling is better with this very precise placement. In the packaging arena, of course, the molding machines, our trim and form machines, singulation machines, and in the plating area. Also there we're quite active. These are the, I would say, the high runners of our machines. This is the EVO. Was developed originally in Austria. Production fully in Malaysia and China in the meantime. In China, we expanded our Leshan factory substantially for the Chinese market. This machine is, first of all, it's high accuracy and speed.
It can handle multiple types of dies and wafers, and it's very fast from R&D through production. What happens a lot, people develop a new feature on the phone, they develop it on this machine, and immediately then they need 10 or 20 or 30 to put it in production. Yes, it's been a very successful machine. We are continuing developing this machine. This year we introduce 3 micrometers accuracy. Those who were last year here, I said 4 micrometers. We made again a step. We introduced a 6-axis bond head for active alignment of camera systems. Also an area we're working on is 3D topology measurement. Our customers ask us more and more, "If I put a device here and I see the blue line around it, is it really there where it should be?
Can you show that not only by camera but by a 3D picture?" We are the first ones to start introducing this. Target. You see the marks where it is in mobile. This is the first time we show these graphs, by the way. This is usually the mobile part, computer part, automotive, also LED. We do high power and high special LEDs with it. In our case, photonics, we put in the computer part. We don't separate that which is in the computer part, but it's becoming a very nice part. High-end LEDs, I mentioned already. LiDAR and radars for the cars, we are heavily in. Of course, cameras, not only for the phones but also for automotive. A decent car has now-- Actually, I got a BMW yesterday evening, and it did not have a rear camera. I was so disappointed.
This is a good car, though, no way. The number of cameras is really increasing also there. Second machine that's a real high runner for us is the epoxy machines and also high-speed flip chip machines. These were developed in Switzerland, in Steinhausen, and this machine is also completely manufactured in Malaysia and also in China in large quantities. A little bit more spread out in the markets. This machine is the highest speed die attach machine for epoxy. This is with all our systems, but I highlight it here again. Why do people buy with us and sometimes not with a competitor? Because if we install one machine and they develop a process and they buy number 2, 3, 4, they just can put in the process and the machine runs exactly the same. This is a statement of quality in the machines.
Many other machines, they have to tweak the machines all the time, and if you have 50 or 100 machines in a row, it's unpleasant to do that. People can buy a system with us, and the second one will run exactly the same. Furthermore, yes, we lose also sometimes orders, but customers typically come back after one or two years. Initially, I was always like, "Oh, we lose this," but now I'm a little bit more, "Okay, they'll come back." Because our machines, after two years or three years, our machines still run at exactly the same specifications. It's like buying a car and, oh, now it can run 180 kilometers per hour, and after three years, it only does 150. Ours will still do the 180. That's a very strong selling point.
Sometimes it's a little bit difficult to bring this because people, "Oh, but I have to learn this." Existing customers know this. New customers, they have to learn it, but it's a very strong point of all our machines. New developments in this machine. A new dispenser system, epoxy volume correction. This has to do with that the world gets more and more precise, so if you have a volume, a little epoxy cylinder that you empty out in the beginning, it may empty out different than at the end. You have to compensate these things. Lead frames become more and more difficult, so we have to put a lot of vision system in for what is called low contrast handling. This is something that has to do with the plating.
We roughen up the material, due to that, we get less contrast, we have to do it here better. Bond level correction. Nowadays, in the past, if you put a die there and it would be plus or minus a few microns, it would be okay. Now we have to say, "No, it's exactly 3 microns." A good example is, nowadays you get watches you have on your hand and it can measure the blood pressure, for example. It can measure your heartbeat and so on. These distances where these sensors are working is extremely sensitive that in the package, this sensor is exactly at, say, 3 micrometers height, and we can correct that. Another element we see more and more, we call it Industry 4.0, is where these machines become fully integrated in the automation systems, but also become less operator-dependent.
We have to put much more self-control features in the machine. Like, oh, I change something, it automatically corrects everything again. Yeah, that's another area. Automotive MEMS sensors, also in the NAND area, in the computing area. Normal computer, automotive high-end ICs, also a number of power packages with flip chip on lead frame, where you actually flip chip on a lead frame. This is the most advanced machine from the accuracy point of view, this is what we call the 8800 series, the Quantum and the CHAMEO. We have this in flip chip, in fan-out in TCB. Since 2008, in fan-out, the world standard. TCB, I must say TCB is still good, it hasn't grown as fast as we thought. It's more complicated than what is different at this moment. You see everybody develops its own story.
Yes, we are still happy with it, I had expected more from it. What is going very well is the fan-out part in here. TCB, I mentioned it here. We are working still on this, still going on. What is also coming is new interposer-less technology and also MicroLEDs. Because this is also a term you hear, new LED technology, we're also in that. Here I put a few new things in here. One thing is, this is a real highlight. In this world, we see that the gantries are coming to a certain. We have very good gantries, at a certain point, you simply can't make them fast. What we now start to do is we start to transfer, instead of one die, we start to transfer four dies.
You cannot do this simply by picking up four dies and put it somewhere. That's still inaccurate. We developed a system where you pick up four dies, then by individually bringing this head down and adjusting it, you can still adjust every single die to the correct position. This will speed up the machines very substantially. We will bring this to the market this year. It will always take some time, this is a real nice development. We have a number of patents on it also. What we had to do is make a multi-die ejector. You have to have multiple dies coming from the wafer. You have to pick them up. You have to inspect them much more precisely, we had to work on this. The bond head had to be changed to this four-head bond head.
I keep putting a lot of R&D in them, you have to talk with all the top players to see, are we developing the right thing? The next thing where we're working is, this is also extremely interesting, is in the high end. What you start to see, those who were before in these meetings, this is a typical TCB contact. You have about 40 micrometers in here. You have a solder layer in here. You cannot make this much smaller. If you go to 10 micrometer, 5 micrometer, you have to go to a next level of technology, where you do what is called direct or hybrid bonding, where you bond these things directly without actually a solder in here. You take 2 devices. It's a cold process. These devices are atomically flat, so they are very, very flat.
You put them together, and it's a Dutch guy who organizes that. It's Van der Waals. That was a Dutch physicist who found this. Van der Waals force, it's called, and they stick together like this. That sounds very nice, but in practice, it's much more complicated because if I take this and put it down like this middle part will not go anymore. You have to make a certain movement in it to get everything precise. This is the next generation of tools where we need to go to 200 nanometer accuracy. This is for devices with a pitch of about 1 micrometer, and this will be in the next generation of high-end processors. We have to ship the first, I call it, pre-production system this summer. The production tools, they have to go out basically early next year.
Yeah, we are very proud of this. We have a few nice patents on it also, and we can track the device till the last moment till closure. That's the way how we get to this 200 nanometer. Competitive systems, they have the device here, and then they lose sight, and they go somewhere. They don't know where they're going. We have invented something that we can go here and steer it till the last moment. We have some piezoelectric actuators in there that work in the nanometer range. Until the last moment, we can steer the device to the exact position. On top of that, here we have to work also with very, very clean concepts because if you have a particle, say, of a micrometer and particle of a micrometer, this is 40 micrometer. It will be here. You don't see this.
If this is 1 micrometer, the particle will be as big as this way. It has to be also very clean. We build these machines already in a clean room. Otherwise, you cannot get to the clean room level. There are some companies that help those other companies, that also help companies like ASML in this field. That's one also a very interesting new development. The third one I want to mention this time is, we get a lot of requests nowadays for the same technology, but then on a panel level size. Instead of a wafer, we even have to go much bigger. We developed a tool for that, where we can do 720 by 650 millimeter. A normal SMD machine, companies like the Siemens group or an Assembleon, these sizes are well-known, but not at this accuracy level.
We can do this at an accuracy level of a few micrometers, 2 micrometer global, as we call it. We had to develop a new gantry system for that. This gantry system here has on both sides an additional separate measurement unit in here. That gives us if a current typical gantry is, say, 3-5 micrometer accuracy, you see things like this. With this new gantry, we almost come to the nanometer level of accuracy. I showed this on a conference last year, October, I think, in Shanghai, and there were two or three top companies, real big companies there. I had only one line in it. I said, "Oh, we're developing a large tool with high accuracy." They all jumped at me. "Oh, wait a minute.
You have to tell them." We have to tell that we're working now with these companies in perfection. For you, it may look like, oh, it's a machine, but it's extremely difficult on this large area to get that level of accuracy. There's a few very interesting areas that is high-end interposer-less substrates, and if you dive a little bit into it, you can see which customers that will be. High-end MicroLEDs. MicroLEDs is the next generation of display technology, but you have to put very small LEDs, 10-20 micrometer LEDs. You have to put very carefully side by side. We're working very hard on that, and one of the customers that was interested, he immediately took one of our machines with him home, literally.
He said, "I need that machine now in the U.S. immediately," basically, we packed it and we shipped it. Okay. This is about what I want to tell about new development. Very interesting area. General fan-out, of course, also in here. Soft soldering is continuing. We're working on higher bend forces for diffusion bonding and also this so-called product traceability to show that every device that came out of our machine was good. This is a machine that's really in the automotive and the mobile, especially automotive area, power devices. Molding, we're continuing working on new molding machines. One of the things we're working on is a 120-ton molding press now that we can handle larger areas. You simply need more force for that. Double-sided molding, where we do molding from the top and the bottom side at once.
You start to get devices that are molded from both sides at once, not two times, but at once. Mainly mobile, advanced chip applications, automotive. Second area is, of course, our molding area for wafers. We are really good in including exposed dies. Here, I see. In the trim and form. Handling complex frames, low conversion time, product traceability. Again, this machine has been very well the last couple of years. Here, you have to bend everything, also, we do laser printing on it now or laser reading also that you can put a little mark on it and say, "Hey, I can trace it completely back." We put a new efficient system in here so that you can check every little mistake on the device. You can sort it out immediately, say, it doesn't go into the production line anymore. Mainly automotive and industrial. Our plating machines.
People don't realize it, we have about 800 of these platings machines in the world, these are machines that are between 10 and 15 meters long. The longest one is 140 meters that we ever built that actually had to be zigzag because it didn't fit in the building. Almost all the lead frames in the world are plated on these machines. We now introduce a new line called a deflash line, continuously deflash. Wettable flank plating, that's a little bit of technology term, if I go back, these devices, what you see here is if you coat this with plating, the top levels go very well. Sometimes the sides go difficult because they are more difficult to wet, we found a solution for that also. Again, driven by requirements for the automotive industry.
We also started an R&D program on panel plating on this whole fan-out area, people have a high interest in that. The normal fan-out machines, they come from the front end, are extremely expensive. The back end, so a normal PCB plating is too simple, that middle field, we are working on it this moment, we get very high interest for that. The last machine I show is solar and also battery plating. In solar, the classical solar cells, we replace the silver lines by copper has two advantages. One is simply cheaper than silver, that it has a little bit of better conductivity than screen printed silver, then you get a little bit more efficiency out of the cell, it is a big cost advantage.
Step by step, we're getting more of these machines. A new trend is double-sided cells, where you have a cell that's active from the front, it will also take light from the backside, it increases. The last thing that is coming up is what is so-called HIT or multi-junction solar cells, where you have a classical solar cell with a thin film solar cell on top of it. You get another 5%-8% efficiency increase, there we have supplied already the first machines for that technology also. Battery, we are in batteries, it's slow to get into that. It's a lot of work, basically, that I think that will get to volume later. That's it. Basically, I hope I convince you that we are still in a phase where more and more backend equipment gets into that.
For us, that's more is going slower. It's only positive. It will require more backend solutions. Of course, the front-end part are the most advanced part we are always in. The smaller form factors, that is all driving us. If we go to 5-nanometer devices, we see we go to the 200 nanometer accuracies now. We are working on that. What we also do, I think Richard will talk a little bit more about that also, is we have a very strong program where we really pick the winners, be with the strong guys, we more and more focus on those we see this concentration coming up. These are also very big boys. That's where I think I leave it, then I think we go back to Richard.
Any immediate questions for Ruurd? Shall we have the question and answer at the end? Maybe you're all a bit dazzled by all the technical stuff. Okay, let's continue. Strategic revenue initiatives. If you look back, it's a bit amazing, but if you realize that this market in the last four years grew on average by 11%, and Besi by even more than double that. How did we do that? The key is always pick the winners. That's not enough. Pick the winners is a nice slogan, what's very important is, of course, to demonstrate at the right time that it works, that it works better than our competitors, but even more so that you can deliver that on time and in the quantities the industry needs. It's a complete business model, which we have expanded further into China, as I explained earlier.
With that world being the strongest growth world going forward, that should help us to maintain an above average growth in our very exciting world. If you look at that, our addressable market, what we just mentioned, also simply look at the cyclicality. It's a world with ups and downs, highly focused on end market applications. Where is it coming from? Mobile internet devices, computer worlds, automotive, and a bit of others. That simply will continue. Here are some details. Some of you have asked those details for quite a while. Viavi released those numbers in the last couple of days, where you can see that our market share in the addressable market has grown from the high 20s to the mid-30s and a bit more. Die attach, amazing. Also packaging, strong growth. Plating for years were somewhere around the 80%.
These numbers are hard to verify. We also have indicated that many times. Viavi has their own models. They also change assumptions in models. Anyway, the big picture is market share growth. Although, and I keep saying that for us, market share is not a goal. The goal is to be successful at the winners, which offer above 50% gross margin returns, because it's all in the end about shareholder returns. If you look at growth opportunities, some have been mentioned, our customers already for years. Some of the names you see here, we can expand our positions very strongly where we have smaller positions today, those focuses on the winners in the future. Also there, you have to be very careful. This world is not a roadmap fixed in time and in technology. It all depends on processes before us.
Simply think about EUV development roadmap, that has a major impact on the requirements for interconnect, at the same time, what's happening in the world. Basically, our focus is in the same way as we have done so far that we carefully expand in those areas where there is growth expected in the years ahead, with margins comparable to what we have had so far. China market, to keep emphasizing that, we have positioned the company in an excellent way to benefit from that. Step by step, we are able to build all those machines, which are high volume used in the Chinese market, locally in China, in our own facility, closest to the market, easiest to maintain, that should put us in an ever better position. The objective, what you see here, is 50%. Last year it was about 21% of revenue.
Q1, it was 14%. It varies, of course, with customer orders. If you look at this number, our peers, ASM, KNS, competitors, are close to that percentage, so we still have ways to go. Again, market share is not the first objective. The first objective is profitability. This shows you a bit headcount increase, the growth of our service and support staff. You see in the different colors where they're at and the strongest growth, of course, is in China, doubling that. Supported heavily out of Singapore. Our technology center in Singapore has increased significantly, and those are major parts of the strategy supporting these complicated applications ever closer to the customer, supporting his production output. The strategic cost initiatives, to mention that again, it's nice to have technology leadership, it's useless if you cannot produce that better than your competitors.
That's something we are learning better over time, and still, we can say that we are far off from a perfect organization. First to start, top line. We have expanded our capabilities in Asia, theoretically, to be able to do EUR 800 million revenue with a similar product mix as we have currently. We are further developing the supply chain step-by-step to Asian suppliers, but we're also very careful with that, simply to control not only our cost, but also our technology. There are certain conditions for Asian suppliers. We've come a long way with European suppliers who have moved to Asia in a similar way. Where we have at least control over what's happening to our modules.
As 50% plus 53%, 57% through the cycle gross margin, it only makes sense if you invest all this beautiful capital in R&D if the returns are better than that of your competitors. Very margin-driven business models. What can we further improve? In the end, it's all cost. EUR 15 million per year cost saving to be further implemented within the next 3 to 5 years. Where is that coming from? Supply chain, very critical. If you simply understand EUR 600 million revenue, we purchase half of that, slightly less, but EUR 300 million per year from all kinds of suppliers. We've moved a lot to Asia in the past years, qualified many suppliers, but you can easily imagine that it is far from perfect and completed. There's still a lot of cost which we can take out of a further improved supply chain.
Also, the quality of the suppliers. With growing revenue, we have access to larger suppliers, which are also able to maintain certain inventories over a longer period of time, and all that is critical with a cyclical world in capital goods and in our industry. Headcount savings, where do savings come from? West to East. I'll touch a bit more on that later. As many of you know, we've moved production to Asia. More and more, we have set up support in Asia, support center in Singapore. Some of the admin has moved, but certainly not all. Also in the supply chain, strategic supply chain, spares and tooling. Many of the support functions, also the back office, is moving gradually step by step to Asia, that has a major cost-saving aspect.
Last but not least is the ever greater challenge to develop systems using common modules and common parts. Electronic cabinets, easy to imagine, bolts and nuts. All these things are still in our world in an infancy phase. We build beautiful machines. Sometimes you can say pieces of art. They are far from developed in a common philosophy, driving the cost down to a minimum. That's where continued cost savings will come from. We see that here a bit more in detail. Where does the EUR 15 million come from? Supply chain, increased outsourcing, but also move to Asia, increased Asian sourcing, but then with very carefully selected suppliers and for certain products. West to east transfer, as mentioned, and product design, very critical.
Materials cost, supply chain actions, just simply to spell them out for you again, is a major part of cost savings going forward. We've come a long way, west to east. Some of you know these numbers. We have presented that year-by-year, and we see a gradual, of course, decline in Europe because those functions are moving to Asia, increased in Asia. Also in Asia, we have different developments, production in China, the light blue. Of course we have Malaysia as our center, but then strong growth in Singapore. The whole support, software development, software support specifically, but process development for the ever-growing Asian installed base. Over 200 people now. You see in Europe overall, since 2013, a gradual decline from over 600 to 500 now.
Production, clearly today, over 80% is produced in Malaysia, but very rapidly growing from 2014, zero, only tools and parts, to the first machines. We see the target is very simple, is already prepared for that 50% of our revenue we can produce locally in China. Redesign common modules, common parts. Of course, lower unit cost, but also lower inventories, easier requirements for service support, but also spare parts. Imagine there's a major cost saving to be achieved in further rolling out common modules, common parts. Baseline OpEx, also an important measure if all things go up. Of course, our cost also goes up, but far less. We have an enormous operating leverage, and the strategic initiatives will help us to maintain our OpEx baseline somewhere close to levels which we've had in previous years. Capital allocation. We always mention partly our return to shareholders.
Far in the last years, since 2011, we've returned EUR 457 million back to shareholders in dividends, around EUR 10 per share before the split, which is amazing, also an ongoing share buyback program, which we will simply continue. The baseline for this is very easy. We maintain 20% of our revenue in net cash. With a revenue of close to EUR 600 million, it's about EUR 100 million. Everything in net cash we have above that EUR 100 million we distribute to shareholders, partly in dividends. Our dividend policy is to pay a maximum 100% of our net profit in dividend, and the remaining part is used to buy back shares in the market, and we do that on a daily basis. If you look at the dividend return, our yield versus the peers, it is only a factor of four higher than ASM Pacific Technology.
Anyway, it is what it is. If we then look at our liquidity, end of March, it was gross cash EUR 571 million. Of that, net cash EUR 290 million. In May, we paid out a dividend to shareholders of [ EUR 170.4 million], and basically, we generated EUR 50 million net cash in Q1. That's our total picture. EUR 300 million of the EUR 570 million is the convertible bonds, one in 2016 and one in 2017. That's our cash backbone. A few words about what's happening in this market. We all know that this world is consolidating like many other worlds. What happened in our little world in the past two years, the top four suppliers, while I would have chosen this sequence a bit different, anyway. Apologies for that. ASML, ASM PT, Besi, DISCO, Kulicke & Soffa, alphabetically is nicer. 55% now over 61%.
It means that the others clearly have decreased in size. Survival of the fittest, however you want to call it. What is far more key, and Ruurd went in a lot of detail, maybe for some of you too much detail. The technological requirements to survive in this industry is growing exponentially. For that, you need to simply imagine who will be able to maintain that momentum and also different technologies. There is different technologies. Simply if we go to wafer level, if we go to fan-out, different materials. Those are all technologies which are sometimes not within Besi's world today, and those are key aspects which we will need in the future. What drives Besi's business today? First of all, GDP trends. Although we are all convinced that the world's growth is forever, that means that the semiconductor industry will grow forever.
That is simply very important to understand what do you do, how fast can you ramp, how fast can you ramp down, is critical to maintain. Those business models, we explained a little bit in this presentation, that is in our view, the key to benefit going forward any direction in this industry. Customer roadmaps, as we said, pick the winners. The key is to understand every day in the earliest stages what is happening, not only where it's growing, but also where it's not growing. Where do we have to slow down or even stop? IDMs. Some analysts here have always criticized Besi. "Your market share at the subcontractors is lousy." We've always understood that IDMs are driving the car and not the subcontractors.
In any case, as Ruurd explained, I'm repeating that again, the key is to not only develop market positions but sustain market positions with those IDMs who are driving the bus. Not only mindshare. Mindshare is a nice word, who is developing what? What do they need? Can we offer those solutions in time? You will also imagine that we are a little company, EUR 600 million revenue, we have companies like Samsung, Intel, TSMC. How can they trust that a little company like us can step up to the plate for the most significant success for their future products? You simply have to realize that you are always in that position. Of course deliver the specs, reliable performance, better performance, competitive cycle times, production scalability. 60% up quarter per quarter, then down and maintain your gross margins.
Those are the key elements of what drives our business going forward. Back to the short term, what's happening? We provided this guidance end of April. We'll see what happens. We still have two weeks to go. Always critical at the end of the quarter. Basically, margins in similar level. Growth, operating expenses down compared to the first quarter simply because we had a one-time cost in certain remuneration packages. If you look at half-year comparison, it's still a very beautiful growth world. That brings me to the end. I hope we provided a bit more background about our positions at this time, but moreover, where are stronger positions expected to come from in the next foreseeable future? Where is our focus technology-wise? Where is our focus cost-wise? Where is our focus for investors?
Those are the key elements of the messages we wanted to share with you today. Are there any questions? Peter, subcontractors. Oh, no.
Peter of Jefferies. Maybe starting with the market share development in 2017. A pretty strong performance there, with especially a very strong performance in die bonding. Maybe first to clarify, does that include the die-attach products?
Of course.
Trying to understand what's driving the share gains in that particular part of your business. Well, you've touched on a number of elements within your strategy, the technical capabilities of your products, the focus on the winners, the IDMs. Is there one particular element that you can single out that has driven this very strong performance in die bonding? Is it just a combination of factors?
No. I will explain that. Why are we good in die-attach? Why? The answer is very simple. The only thing that matters in placing a die is the accuracy over time. Anyone can claim, "I can place a die within a certain accuracy," but the key question is, every customer builds millions of parts or 100,000, whatever you want to name it. What is the consistency in that process? Whether you glue a die or you solder a die or you flip-chip a die or you fan-out a die, it's all about repeatability. The next step is speed. How fast can you do that? Anything which we do with a die, a chip, falls into the category of die-attach, because we attach the die to something.
That can be to a leadframe, can be to a substrate, can be on top of each other, stacking dies. All that fits into the category of die-attach.
Okay. That's clear.
To add to that, over time, this has only become more critical because the dies have become thinner, the dies have become bigger in some instances, dies have become smaller. A moving target. That is why die attach, but this goes for the whole assembly, has moved back into the IDMs. Because the risk for an IDM that this does not work, cannot be relied upon outside of that customer. That is behind ever increased importance of IDMs as opposed to subcontractors. Still, subcontractors maintain a volume position in this world. Once processes are qualified, are in a state that they can be expanded in volume, then subcontractors take care of lower CapEx investments for IDMs for that product. Same why we use subcontractors.
Okay. I have a few questions for Ruurd, just to get a better understanding.
I'll stand behind.
You talked about a number of types of advanced packaging, like a fan-out and et cetera. You also referred to solutions without interposer, and you mentioned InFO.
Yeah.
Is it just your big U.S. client which is working on interposer-less solutions, or do you see
No,
broader adoption?
Generally, first of all, we always have to understand that the high end of the market is technologically extremely interesting. Volume-wise, machines, it's always a little bit lower because it simply is a smaller part of the market, but the price per machine is very high. Volume-wise for us, it doesn't automatically translate to many machines. It translates to high financial volume. What we see very strongly is that this interposer is a very expensive piece in the device. These interposers are very big. They're expensive to make. Everybody tries to find a solution how to get rid of the interposer.
That is either done by companies like Intel. They introduce something that is called EMIB, where they put a small connector, a very precise connector in the substrate as a solution. Others are trying this fan-out solution, where you completely eliminate an interposer or a substrate and put a RDL layer directly on top of the chip. Actually, you first make the RDL layer, and then you put the chip directly on it, and then you take it from the carrier. It's a general trend we see that people try to get rid of that is wherever possible. One time is cost for the interposer. Secondly, it is also against space because it's another level in there. We see it in different areas, and EMIB is one of them.
It is heat.
Yeah. The heat dissipation is also an issue because with these 3D devices, one of the bigger issues is you still generate a lot of heat. In a device that is horizontally spread out like an EMIB, you get a little bit better heat dissipation. That's the reason why people look in that direction, but also to eliminate another layer in there to get the heat spread out better to the devices. Yeah.
Okay. You also mentioned a number of end markets that you've not discussed that much before.
Yeah.
Like photonics and MicroLED.
Yeah.
Just trying to understand which products you have for these particular.
Okay
end markets. Is it for the photonics is mainly the very highly accurate die bonding, or is it other products?
Yeah, in the photonics area, especially the EVO machine, one of the first machines I showed, is quite extensively used there. What you have to do is you have to connect a fiber system to a chip because basically, you have to go from electric signals, you go with a conversion to an optical signal, then you have to take the optical signal from the chip, and you have to mount optical, say, fiber devices, very exactly on a chip. Now, that is what we do with that machine, so that main area is there in the EVOs. Those are the that have all kinds of connector of interfaces. We do that with all the large guys.
If you go over the photonics, you have a few big players and a lot of small players that have technically very beautiful things, but actually, the volume is with a few big players, and there we are in. We don't advertise it that much, but it's good to know because everybody's a little bit, "Oh, what are we doing?" We are really in there, yeah.
The MicroLEDs?
The MicroLEDs are something that's upcoming, the [inaudible] for companies like-- Now that Samsung has this, the OLED telephone screens. People are looking at better screens, the next generation is a MicroLED, where you have a very small LED making the actual pixel. That technology is just, I would say, starting. We had already some touch with it initially. It's not easy because now you have millions of pixels you have to place. You hardly can do this pixel by pixel. You have to develop new processes for that because otherwise, it's too slow. I think in the coming two to three years, we'll see that, yeah.
You would address that particular market with the next generation fan-out product?
Yeah, it's not fan-out in that sense. It's high precision placement on a large area. This big machine, for example, can do that. The demands for that are accuracies below two micrometer on very large areas. At this moment, there are no machines that can do that. We are working hard on that, and we got very high interest in that.
Okay. A final question relates to spare parts and services. What kind of growth rates should we anticipate for that part of the business? How does it compare to the rest of the business in terms of profitability?
Well, first of all, any spare part business in the world for any product is a higher margin business than the original product. For a reason that you need a certain inventory, you need a certain handling cost. You need additional cost, but basically, the margins are always higher. With machines becoming more complicated, the reliance on the supplier of this service is growing and has grown over time. 10 years ago, it was below 5%. Today, it is over 10% of our revenue, and you can expect that to grow towards 15%, maybe even 20% of the business. You can compare that to front-end companies who typically have spare part and support of between 20% and 25%. It has a link with the complexity of our machines. It is simply a market position which requires that your customers run 24/7, days a week with your equipment.
Bottom line, a margin-increasing business. Marc. Oh, Hilco. Hilco, yes. Yeah.
Yeah, Hilco. I have two questions, one about China. You expect the objective of 50% of your total revenue. When do you expect and which time frame? What can you tell us about the revenue split of the local Chinese customers and the foreign, your big customers already. The second question is about the acquisition strategy. You have now more than EUR 500 million at the bank. You said we are looking for different technologies. Can you tell us about which one? Are you looking for one or more acquisitions? You always told us we only buy in the downturn. What can you say about the multiples you want to buy?
Excellent. Well, perfect questions. First one, China. China, again, always bear in mind, it's not a market share game, it is a margin challenge. Analyze the business opportunities based on margins. Because in the whole development programs in a company, we've learned over all these years, focus is key. You can't focus on everything. You have to choose. That choice also then determines which applications are available for us for a sound business opportunity. Why 50%? If you look at the world, one expects that for the next five years, China will invest in this industry at least double that compared to the rest of the world. Assembly backend has always been a focus of China. There are a few giants which are longstanding customers. More and more, they are successful in higher-end applications.
With that success, our position could further grow because cost, we're not in a disadvantageous position. More and more, our cost is comparable and even in some cases, advantageous compared to others. Technology-wise, as long as the leaders in the world are outside of China certainly wants to develop along those leading applications in a way that the world accepts that. That is still, in clear terms, a world developing with, how do I say that? A lot of unknowns. The world is becoming more and more skeptical and careful on what to allow and not to allow in China. From our position so far, we definitely can benefit much more than what we have done in the past. Is the 50 a goal, hardcore? No, it's not. The only hardcore is to have business with margins well above 50%. That's that answer.
On the acquisition front, we've always said several things. If you look back over time, we have been able to acquire different processes in the assembly world, which always have improved our margin potential. We've always moved upstream with potentially higher margins. Number two, we have established stronger mainstream market positions. Those acquisitions, six over time, have simply moved us higher in the complexity and margin opportunity, plus more into the mainstream of the world. Because size does matter in the end, because you have to generate your profits in order to invest in the ever-increasing R&D requirements. Those simple business fundamentals is driving our business. Everything's for sale today in this world. It's very interesting. That always happens in upturns. Everything's for sale. The key is then to really understand what brings Besi shareholders, in the medium and certainly longer term, a higher return.
Capital cost is of course, one part of the equation. We have always prepared our capital position and the cost related to that at times that that is advantageous to shareholders going forward. The key is then to understand, are there opportunities which fit into the requirements as just explained. We will simply continue to focus on that as we have done always. If opportunities come, we will let you know. That's as simple as I can explain it. We're not going sideways, we're not going backwards in a sense. What I mean to say is we're not buying market share because it won't help us. Every next generation, and that's what Ruurd tried to explain, you need different machines. To buy today machines, unless it's something which can improve our margin dramatically.
Those are all the elements you have to take into consideration.
Can we expect one or more acquisitions or deal?
No, let's first expect one.
Just one.
Yeah.
We get a lot of because we are in a good situation where we get a lot of people knocking on our door, "Oh, you should buy this, you should buy that," and so on. We always try to look, does it really make sense, as Richard said, long term? Does it fit the technology? Is it something we're already doing, and it would only make life more complicated for us? It's not easy to find something where we say, "This really makes sense." Almost every second, third month, we have somebody knocking on the door.
Every second month.
Every second month, yeah.
You can expect in a few weeks, SEMICON West in the U.S. It's not just a nice statement, today everything's for sale in this industry. It's very dangerous because you can use multiples. Look at the beautiful multiples here.
Yeah.
In a year's time, in a major downturn, it can look completely different. Anyway, with care. The first element is always, does it improve our gross margin potential? That's number 1. Be patient. Thanks. Marc, you had a question.
Thanks. Marc Langeveld, Econopolis. First question, some industry sources talk about order push-outs for memory, affecting the likes of Applied and Lam Research in the front end. Given what you've showed us in terms of momentum in the various end markets, I can hardly imagine it is a demand issue. Do you agree that then it's ramping or phasing of the ramp-up of capacity in that market? Do you see that differently? The second question I have is, you mentioned a slowdown in mobile, which is one of your main markets, obviously. At the same time, obviously, you also showed us that there's still a rising IC content per handset. How do these two balance each other in terms of your outlook, or how do you see that? Thirdly, last question is on Applied Materials' announcement on cobalt.
Is that a threat or a big opportunity in terms of new systems that you also have to develop because is cobalt so much different than copper? Those are the three questions.
Excellent. Well, the first, memory, since ever, has been the most volatile part of this little world. Purest technology-driven, timing is critical, and also the CapEx expansion and the fab cycle times, to say it in those words. Still, the canary in the coal mine is memory. The big question is today, is it a warning or not? Although everyone says end markets are intact, and look at Ruurd's second slide, growth to eternity. That's the big question. We have always been very careful with memory. Memory is maybe 10% of our revenue. We've always more focused on the more complicated logic, power, analog, those parts of the world. Yeah, that's memory. On your second question, Ruurd, maybe you can share some light.
Yes, we see in general the introduction of the iPhone X, for example. Everybody now says, "Oh, it's going slower" and so on. Technology-wise, there's a lot of new things in that phone. These companies, they typically tend to invest a lot and then absorb everything and see how it goes. Then you are in the next cycle, maybe one or two years ago, you get another bump of the next level. Now this technology is percolating slowly, but certainly in the other phones that bring something. 5G, we are still puzzling with what impact will it have for us. We're still working on that because you get new class of devices in there. Sorry. What we see is that for us, the numbers of the phones is often not so important. It's much more do we get into new technology cycle.
For example, one thing that's happening at this moment is people are putting more cameras in the phone. Today, you have companies like, I think, advertising the phone actually not anymore as a phone but as a camera. Actually, if I see how much I use my phone, it's also more a camera nowadays than the phone element. Those elements are modern. We are very close in touch with, I would say, the top players in this industry where we focus also on the ones who really make money, because there's a lot of people making phones but not earning anything. That part, I'm still very optimistic in the coming period that new things will get in there. It's difficult to estimate because they decide first on the technology, then we have to put it in place and so on. I see that positively.
Your third question was about the cobalt. My remark was there, in the front-end business, to bring the devices smaller and smaller, for the first time in basically about 20, 25 years, we see a replacement of copper by cobalt now. That's a major step. Will everybody do that today? No. It will get step by step into it's a driving element or it's an enabler to get smaller dimensions, and that for us positive. It means that our machines have to get more accurate also again. It's actually not a threat to us, it's only positive for us. That's why I put it there, because it means this drive. Everybody says, "Oh, Moore is dead." It's not happening anymore. We believe, yes, it will go slower, it will continue. The world will not start turning backwards.
It will keep moving on. For us, it's only a positive development that that's another thing enabling more accurate technology, and that will give questions for our machines. Yeah.
Thank you.
Any other questions?
[Jonathan Tame], ABN.
Yes.
Okay. Jonathan Tame, ABN. I have a question regarding the EUR 15 million cost savings program.
Yeah.
Could you elaborate a little bit more on, I would say, timing? Of course, it's geared to three drivers. I would argue that product design is very important. Standardization, you mentioned. Does it imply extra R&D cost? Is it more or less included in all your guidances?
Well, the best way to answer that is, maintaining gross margins above 50%, mid 50s, the only way you achieve that. Well, I can say it even more basically. Every year, headcount cost, if you have the same number of people, increase because there are salary increases. In the supply chain, materials cost increase. Customers, of course, there is a price pressure in every world, the competitive landscape basically results in the price of your product. The only way what you can do to maintain margins at those levels are continuously developing those machines, better output, higher output for the same cost or slightly higher if we can accomplish that, but that's a daily battle. Number 2 is take out cost of our entire organization. In order to stay at those margin levels, you have to take that cost out.
Timing is, of course, dependent upon where you are in a cycle. Last year, with a ramp of 60%, you can imagine there was very little possibility to structurally change the organization. Also, we hired 400 people last year. Simply look at the headcount development. That target, EUR 15 million, we have detailed to individual persons, all these programs have owners, and simply there's progress every year and depending upon what is feasible. Same with these suppliers. We are constantly evaluating new suppliers in Asia, different suppliers, simply to make small steps and in the end adding it up to those cost savings. The 15% comes from a simple analysis we did, and we always do that in detail every year. Where can we take out those costs?
Those targets are, there are owners to execute those programs to come closer step by step to those savings.
Maybe I can add a little bit flavor to this. ASML has grown out of a number of acquisitions. All these people have developed beautiful machines. They all work with, say, a camera system. They work with a computer. They work with this, they work with that. Originally, they were all developed separately. Now you see, here we're using a vision system. There we use a vision system. There we use a vision. What you try to do then is, can we commonize that? Can we say we go to one vision system? We go to one there. Sometimes that's very well possible. Sometimes technically it's not possible because you don't want to degrade the performance of a machine because you get a component in there that doesn't [meet the spec like the other guy]. That's what we're continuously doing.
Step by step we implement this, because customers are also not easy in that respect. Some of our customers, they say, "You may actually not change anything in this machine because it's qualified for production, and now you have to make every machine exactly the same so that you have a good idea," and you cannot always put it in. Step by step, we do that, and we have, as Richard said, for a lot of these areas identified, we have a program step by step introducing this in the machines. It's very practical work. It's just looking at, "We use this, we use this. Let's put it together. Can we make a common element out of it, yes or no?" Could you start with a blank sheet? You can do it again differently.
We have to step by step introduce it, but there's still a lot of potential in there. It's not only.
After this program as well.
Sorry?
After this program as well.
We are already working on this quite some time. We are not done with it at all. We still have many parts in the machines where we say we can go to one. We also recently did something on PCB board. Their supplier there, it can be combined as in one package. You get another couple of % out of the cost. This cost element is something that's very much in the blood of everybody saying, "Hey, why spend the money here?" Always in the combination with the performance and the quality of the machines has to stay high level.
Any further questions? Yes, please.
Michael Barghoorn. Question about your capital allocation. I don't have so much a problem with companies that buy back their shares, you are now actually buying back shares at a rate which is higher than your conversion price for your first convertible bond issue. Could you once again explain to us why you have the merits of issuing convertible bonds, once you have a very strong cash flow, et cetera?
It's very easy. We started to buy back shares, 2001, 2002. That is, as I explained earlier, return of capital to shareholders. You buy back shares. You pay out a dividend. In relation to the convertibles, the beautiful thing is, the shares we've bought back, simply support, number 1, if we have to issue shares. In other words, if the convertible is converted, we have the shares on the shelf. At the same time, that is anti-dilution. At the same time, convertible has the beautiful aspect that there are no strings attached. Any loan will have certain strings attached, whether it's dividend or whether it is whatever ratios, a convertible does not have that. Number 3, if you look at the coupon, the last convertible, 50 basis points and a conversion just below EUR 100.
Well, somebody still has to explain to me why that is not a good way to finance future growth. You may disagree. You may say, "Well, why not pay out all your cash and put some leverage on the company?" We have demonstrated we have a huge cash flow generation. It's a decision. It's a concept we have chosen.
Do you still envisage to make or to issue additional convertibles?
Depends on acquisitions. We've always said these convertibles total EUR 295 million. We will use for strategic steps, acquisitions going forward in the next 6.5 years. 7 years at the beginning, 2.5 years ago.
There's a question. Last year, you mentioned that you were targeting a gross margin of 60% in current, in the same setting over here. This year, you're mentioning between 53%-57%. What has changed?
Company has gone downhill, as you can see, from 60% to 53%-57%. Of course, our target in gross margin is above what we accomplish at any moment in time, because that's human nature. There are things which have changed. If you look in the past year, the dollar rate, and this is a dollar business for your reference, the dollar was 1.05, 1.06, and now it is 1, what is it? 1.20. If you simply see that 80% of our business is dollar based, the fact that we still maintain gross margins over that same period of time above the mid-50s is one of the reasons why maybe this level is already unique. Then you can say, "Guys, why don't you develop your model that you would have expected 65% gross margin, maybe.
No, 60% gross margin is very impressive.
It is impressive.
Last year you mentioned a target of 60, and now you mention 53-57, I was just wondering.
It is very disappointing, right?
No.
Anyway, the dollar is the major culprit. To hide behind the dollar is not what we do. We simply look at today's structure in the business. The dollar is the backbone of the semiconductor industry. That's the main reason. Still, where we are today with 15% top-line pressure on those margins is still something to think about. It means a lot of the cost saving has also come through in the last 12 months, and improved product market positions. You see that in the market share gains. Okay, any further questions? Willem.
Willem Burgers at Add Value Fund. One specific question on slide number 65, which tells us that the assembly equipment market is consolidating. Do you expect this trend to continue into, say, the next three to five years? If so, that would mean that Besi could also increase its market share. You also said we do not accept every order. As long as the gross margin is not sufficient, we will not accept that. Can you give us a somewhat more detailed view on how these percentages will develop? Will there be an accelerating downtrend in the other assembly suppliers? That looks to me as obvious because of the reasons you mentioned below.
Would that mean that the position Besi will take into the consolidation process may change, because it may be difficult to hold on to the higher margin levels if other competitors start to compete on a harsher tone. What are your views on that?
Number one, we tried to explain today, Ruurd in particular, that you can expect in the next three to five years, a significant technology challenge in moving the interconnect requirements down to three micron, two micron, and even below that. That focus, one of the examples mentioned was the MicroLED, but simply for all the other applications, if we do that right, that will increase our margin potential. We simply choose to put our eggs in that basket as opposed to trying to get more sideways or even downstream. That's a deliberate choice. Will this trend continue? Certainly, because the R&D requirements for those next steps will further increase the distance between the smaller and the larger companies. Also, one of the reasons why we issued convertibles is to be prepared for those technology investments.
Whether you develop that yourselves or you have to acquire data, that is number one to maintain. Along with the simple, I said that at some point, the market above us is consolidating. Take NXP, Qualcomm. Anyway, this whole world is consolidating. You have less end customers. For us to maintain that position with less customers is always the first challenge to accomplish, not to spend your attention in a broader. You can say it, in other words, solidify our technology positions in the next three years is our number one goal. Along that, you see that in the market share development, if you do that, your market share gain will come as a result of that. It's not the other way around. I hope that answers your question. Next question. Peter.
A follow-up on Hilco's question on China. If you look at China, on the one hand, we have the Chinese operations of your international clients, and on the other hand, there is the domestic companies. Are you particularly exposed to one of these groups, or are you servicing both? Where do you see most upside for you?
Well, we are at both. Let's say it this way, years ago, the big ones took us by the hand into China, with stepping stone, also Motorola, helped us in 2002 to start from scratch where we are. They have also offered to the local Chinese community all kinds of opportunities, that is how we have made our way into those local Chinese companies. It is very simple. More and more, the volume production of many devices, but also modules, have been established with Chinese partners, you could say, subcontractors of the major IDMs. That is still 80% of our revenue in China. In a simplistic way, those major IDMs tell their subcontractors in China, "You have to buy a Besi machine." Even to the extent that all conditions related to that are specified.
The other part is the smaller part, China for China, that is growing beautifully, that may become 50/50 at some point. We have a huge opportunity to grow in China. Yeah.
A good example is, for example, China is extremely strong in electric cars. They are really pushing for that. It means also that the power device and everything that goes in there has to be basically high quality. Then initially they start at local suppliers for machines. Then they say, "Yeah, we cannot do it with these machines." Then they come to us. Our goal and our objective is to stay ahead all the time because at a certain moment, the local guys will catch up, then they make maybe a machine of a lower spec, that will do. We have to keep moving on. A lot of those companies, they start looking, "Oh, we actually need this type of equipment." It is positive. I personally visit a lot of them, whether it is local Chinese ones or IDM-coupled ones.
You see the trend that they say, "We need to get to this world-class level. We also need that type of equipment.
Yeah.
They appreciate also that we start building this in China because it makes life easier for them. Yeah.
Next question. Thank you very much. If you have any further questions, don't hesitate to contact us.
Thank you.
Thank you.