Yeah, good morning, everybody. Welcome to the analyst visit. We're still setting up. We're waiting for a few more people that are just coming up from the reception. Feel free to settle in, get your coffee when you're online. If you're here, you get a coffee from us.
And strawberries.
And strawberries.
Yeah.
No champagne, though.
All right.
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language]
[Foreign language] . Who still wants a business card? You probably have one from Barcelona, right?
Yeah. Anybody? They might get rare, you know. I got a lot. You guys all have them, right? If you want... All right. Oh, I'm okay, Olivier. Everybody has a coffee and croissant or something to drink? Very good. Welcome, everybody, to INFICON's analyst visit in Q2. We do this two times a year just because there has been so much interest recently, we thought we'd do an in-between event. Obviously, in Q1, there is already a number of conferences and also earnings release and AGM, we thought this is something to do. We have an amazing amount of nearly 20 people joining, this shows, again, big interest. That is fantastic to see. Thank you so much for the interest. I want to quickly check online. Everybody can hear us okay, right, while I speak? Yeah?
Yes, all good. Good morning, everyone.
Yeah, so.
All good. Good morning.
Good morning, Reto. We do this as usual. This is for you. I will have a little bit of an introduction, but then most of you will be interested in Q&A, I assume. Otherwise, please tell me how much of what you would like. I'm very happy to go and follow a little bit what your needs are. This is not a capital markets day or a tech day with a fixed agenda. It is very much a visit, so you can use it to your best to fulfill your needs best. All right? Also online, just speak up if you would like to ask a question or give feedback. Maybe Dimitri, you can monitor a little bit if somebody raises their hand so I do not miss it. This is work for everybody. If I give a quick introduction, then we jump relatively quickly in Q&A.
Good also for the folks online. I know some of you guys online know us a little bit less, but I hope this is still okay. Otherwise, please let us know your opinion. Right?
I think this sounds like more or less like thumbs up.
All right. Very good. Just to make this formally correct, a quick safe harbor from my side. This is not a formal analyst meeting nor a capital market day. INFICON will not disclose any new information in this gathering. All information shared today is already in the public domain, as it was discussed either at INFICON's full year 2024 results conference in March or at the first quarter webcast in April. As we have received numerous meeting requests, INFICON has decided to pool these meetings for efficiency reasons as best as possible and also to give a bit more insight for everybody. We invite interested parties several times a year to meet at INFICON with the management in person or online. You can join.
For those that are here, they not only get the coffee and the beautiful view, but they also will get a tour afterwards to see manufacturing here in Balzers. The oral statements made by INFICON during this session may contain forward-looking statements that do not relate solely to historical or current facts. These forward-looking statements are based on the current plans and expectations of our management and are subject to a number of uncertainties and risks that could significantly affect our current plans and expectations, as well as future results of operations and financial conditions. We undertake no obligation to publicly update or revise any forward-looking statements, whether as a result of new information, future events, or otherwise. All right. So much the boring part. Any questions at this point about the visit? I have to also quickly apologize. Not everybody was aware, right?
You, for instance, told me it's good always to get feedback that you weren't aware of these visits. We started them last year, but this is a new idea just to cater a bit better to the information. We will do better in communicating them online and whenever we speak also at the earnings release, so you always see the two dates. The current idea is that we have one in Q2 and one in Q4 to fill a little bit the information gaps. We are very open to feedback. All right. Thank you very much for coming. Now my quick overview. INFICON is a global leader and innovator, $650 million plus revenue, 1,600 plus employees, largely self-financed. In most of our markets, we lead the market with technology and market share. If not, then we have an aspiration to become number one or two.
We had a compound growth of 11% the last five years. We are in a diversified market, but with a strong focus on semiconductors. We choose our markets, though, very deliberately. They need to have a synergy with our technology, and they need to have a growth profile that is 5% +. In the average, most of them are cyclical, so that is always going to be a CAGR of 5%. The profitability is around 5% +. Currently, we do not have a focus of inorganic growth, while we constantly look at inorganic growth and have about 150-200 targets in the pipeline. This is not so much here. This is mostly about technology bolt-on. We'll get back to that later. What we see is significant organic growth opportunities, and that's how we built our long-term strategy of 5 to 10 years to double the business.
Of course, if you have so many cyclical markets, it's a bit hard to say what exactly the timeline of it is, but we know it from designing wins when we win market share and how this scales up when it goes into HBM with the customer, be it chip maker or also tool main. At the bottom here, you see all our different products. And for those of you that are in the room, they're on the table as well, so feel free to touch them. It's not a problem. They give me the broken ones anyway. You don't want to go to our customer, so you can drop them. No, don't do that, but you can totally touch them and look at them. We have some gauges. This is a mass spectrometer. That's two flagship products. Over there, there's a leak detector in microscopic form.
Exactly, that is the spray gun, an innovation, a recent innovation, which produces just a couple of bubbles of helium when you want to go test a vacuum tool. This is the other end of it that pumps out the vacuum tool and then feels as soon as one of these helium molecules goes into the tool when you go along the tool and find a crack somewhere or a leak, right? There is the handheld, what you just touched there. That is the handheld leak detector for refrigeration and air conditioning. There is the nose there that you can bend. There is also for combustion. The yellow thing there that has an apex rating, that is why it needs to be yellow. This one and the stick that you see in the corner, you can sniff gas leaks of gas pipes from 2 meters of earth.
You walk around and check gas leaks because of regulated quality assurance. Yeah, there is all kinds of different stuff here on the table. For those of you who are online, you see the pictures below. Sometimes I get the question, "Wow, you've got so many markets and so many products. How do you do that with this size of a company manage this complexity?" That is a good question, but it is well managed, actually. The idea is that we have about 15-20 core technologies, such as pressure measurement in vacuum or gas analysis, where you identify single gases or single molecules in the vacuum chamber. This technology translates into many applications. The R&D is synergetic. The manufacturing is synergetic. A lot of the customers are synergetic. Definitely, the approach, the go-to-market approach is always a key account approach for us.
Let's say the sales methodology is synergetic. What is really the difference is, and I give you the example of the mass spec here, is how you package it for the customer group. This mass spectrometer, for maybe those online a little bit more difficult, but for those here, simple principle. You see here a tiny filament. Maybe later you can look at it. This ionizes the gas, makes it magnetic. It goes through this pipe. It has four rods where, with magnetism, it's filtered. Only one mass goes through. It hits here a Faraday cup where it produces a tiny signal. Then you know how many are coming off a certain type of gas weight. With the weight, you can then define what kind of gas it is. This is the electronics to evaluate this.
There's already quite some horsepower here to understand that. None of our sensors are trivial. All of them have quite some computational power. Data analytics has always been part of it. What I want to say, this is a tool to analyze gas. When you look at the different pictures online, this is definitely what goes on the tool for TSMC or Intel or Samsung to understand what is happening inside the vacuum chamber when they make a chip. It also goes to the tool makers for integration in the tools. It is part of leak detection. There are different tools that we use to check leaks in car batteries, where we are also number one. There we have sometimes an inline system, so you can do parallel checks in a system that produces 100-200 battery cells a minute. A good format is the 4680.
If you think about the batteries you have at home, a little bit bigger than that, the AAA or AA, or there's also a newer invention that I just mentioned was Tesla. There is the one from BMW, 4620 or 90. When you check this, it's inline, but it's still the same technology here with mass spec. The Hapsite, which is our online lab, I'm showing it with the cursor online here as well. I don't have a model here. With this, you can go identify chemical warfare, pollution, narcotics, explosives. It's the number one tool in the market. Also uses a mass spectrometer to analyze. It has a few more extras on top to quantify. You have them also in stuff like the new market, such as space. One of those is actually the one that we sent to the moon about six months back.
It is now stuck on the moon. Sniffing is for prospecting on the moon. You see this is quite versatile, such a technology to use in many different applications. The application is really where the translation happens and where you could say we manage the complexity. That is a quick explanation. Just quickly about our history. We have been founded in 1969. We celebrated 50 years in 2019. Of course, different pieces of the company came together. Most notably in the year 2000, three parts came together, a Balzers part in Liechtenstein, a Cologne part in Germany, and a Syracuse part in New York, US, that formed the current structure of INFICON. We did our IPO in that year. We are listed for 25 years now. All right. A quick overview. You see here pictures of the different buildings.
Mostly, you will meet us in Balzers. However, we were thinking maybe also in New York to do such events potentially in the future. In the end, it is these three big locations, right? Syracuse with the complicated intelligent sensor solutions, Balzers focused on measuring pressure in any shape or form, and then Cologne measuring leak detection in all different applications. When you look at the global footprint, you see this actually. These are the three main centers that I just mentioned. There are eight other specialized competence centers. There is Finland, Sweden, Massachusetts, Texas, Colorado, and Kansas in the US. Then we have Shanghai and also Malaysia. The last two are mainly manufacturing sites. The others are more focused on R&D as well, next to manufacturing or very specialized competencies.
We have more than 20 offices across the globe, the northern hemisphere between Japan and California to serve our customers. Many of them are pretty sophisticated application centers. Again, this is this translation for the customer. Some of them are also doing some part of innovation because an application can go into product variants and building systems out of sensors for customers. There are different flavors of those. Maybe to our sales team, I should say that's over 400 people, but the largest part is actually application engineers. And an application engineer, you do matching that. The most sophisticated ones, the ones that live on Taiwan, in Tainan, next to the most advanced chip fabs, they go in there every day. They have a batch. They innovate together with the customer in R&D or in the manufacturing plant. They have often a PhD.
They have 10 years experience in semiconductor or more. This is quite advanced patient development. All right. One other aspect is important. We have a lot of smart sensor. That is the bulk of our business on the left side. We have developed for about 25 years also in AI, as we would call it today, data analytics. We have tools that the semiconductor fabs use from us to analyze their processes inside of tools, wafer health. Of course, on the right side, aggregating everything together, you can make a digital twin not only of a tool, but of an entire fab. We currently run over 60 semiconductor fabs with our scheduling and management processes. This is focused always on sensing and then analyzing it and putting together a model to predict and support the productivity, optimizing the yield.
Now quickly to the markets. Again, today, unless you would like that, I will not talk too much about quarters, but absolutely we can answer any questions. I want to a little bit more talk about the four end markets that we are reporting. You see on the left, the semiconductor market. This is about half of our revenue. You see there was constant growth. In the meantime, the last two years or so, we had a bit of a down cycle. This is only possible when you are winning market share or you have a diversified semiconductor portfolio with different key accounts, different types of key accounts that you are serving. We were able to grow 15+% the last five years in the semiconductor. When you look at the next markets, there is the green one and the orange one.
They're about the same size, about 130-150 million annually. The first one, the green one, the automotive refrigeration air condition market, this is a lot around leak detection. We're also here the number one in leak detection, for instance, refrigeration refrigerant detection. The most recent development, as you know, was that the auto market itself was not very dynamic. There were fewer units sold. The EV transition also was a bit slower. There was a big jump in 2023. There's a bit of an effect there of the big backlog that we had. China opening up a lot of these businesses in China. That's why 2023 probably has some revenue that would not belong to 2023 and got a bit delayed in 2022.
Also there you see a rough growth of 9+% in CAGR in spite of these difficult years that the market had. Also there, I would think we were able to win market share. I think most notably is that we adopted our China strategy. We've been in China for over 30 years. During COVID, when there was a full lockdown, it was very difficult to go to customers. We were struggling against local competition that also pushed, especially in these two years there. In the meantime, we have really doubled down on innovation. You've seen from our communication that we opened up an innovation center in Guangzhou that has made real good progress the last two years. We're expanding it.
I was just there two months ago, the executive team to further develop our China strategy and talk to partners and review the innovation center. We further pushed that. I believe that is one of the drivers that we have been able to perform well in China. The largest part of our competition is now Chinese, not Western anymore in China. For us, we are 100% committed to the market. Again, we've been there for a long time. We have strong partners, good partnerships. We see some difficulty, but we have full commitment to it. This is a little bit about this automotive refrigeration air conditioning mark. Maybe one more note is there. Air conditioning is tightly connected with also because a lot of the air conditioning work is also for automotive.
Heat pumps at one point of time was a big push also in 2023, but slowed down as well, actually. What steadily is growing is the refrigerant, the new refrigerant detection. This is new regulations that drives it. Also, for instance, this handhelds that you have over there, the new generation of that has some tailwind because of that. All right. We move over to the orange market, the general vacuum. This is about 20-plus smaller markets. This has been more up and down. It has the same effect in there because China is a big market with this 2023 backlog deduction after the opening. The general industrial development was not as dynamic as maybe we would have expected or as other markets, as you know, the last two years. You see that here.
We jumped up to this 196 in 2023 and then went back down on a normalized level of 160 something, similar like 2022. The comparison of the most recent quarter is still on a quarter that is a higher level, which is the first of 2024. I believe now we are kind of steadily building back up after having been going through rock bottom. One chunk is going to miss also for this year. It is solar, which was a big driver in 2023. This is still in consolidation and struggling. I think in China, where most of this market is, or most of our customers are, there is still overcapacity. There is less dynamism on the customer side currently. I think in some sustainability initiatives, the focus has a bit shifted these days. I think it will come back, same as with EV transition.
In there, there is also a lot of industrial vacuum applications. There is life science in there. There is also big science in there, such as ITER and CERN. There is space in there. This is a pretty broad bag of markets. Hence, this is a bit more dependent on the GDP. It can grow GDP plus. I think that is also what you see when you look at the CAGR. Most recently, all regions were a little bit slowed down. All right. The last market, the smallest market, this is security and energy. This is largely driven through government programs. There you have seen, especially the new Hapsite rollout, with a couple of first phases from the DOD. There is more coming. You see this nice growth. It goes in cycles. We already said this year will not be as high as last year.
While the pipeline is good, security budgets are up, especially in Europe, tremendously up. There is optimism for that. It is also very long sales cycles. I think we started selling for the most recent rollout there in 2023, 2024. We started selling five to seven years before that. This is also constantly ongoing, but it has a time lag. There is this rollout of new energy as well, biomethane, hydrogen. It is a bit similar, but it is also a bit slowed down, sustainability initiatives, especially in the U.S. Long term, we think this is a good market, but it has a bit of a different dynamic. This year is not going to be as big as last year. All right. This is the big out here. The stars you see where we had a record year last year.
Anyway, so then two or three more pictures, and then we should probably jump into Q&A to do everybody justice, the ones that are a little bit newer and the ones that know as well. Here's the global distribution. You see that we have roughly a quarter with some swings between North America, Europe, Asia, and China. Asia is about half, half. You see also how China, and not China, but Asia, I want to say, was growing over the last five years. You only have three years here, but in the end, at one point of time, we had a third, a third, a third. The footprint in Asia is extremely important for us. The markets are over time growing much more. This is a similar picture that no doubt you've seen other places too.
We are very committed to Asia, also to China, but also the rest of Asia, specifically Korea, Japan, Taiwan, also important markets. Obviously, we have a little bit expanded in Southeast Asia as well with this new location that is also an application sales location now in Kuala Lumpur. You see that Asia was growing also in Q1, while Europe and the Americas is a bit slowing down. We see a bit of a slowdown of investment. We are CapEx driven to some larger degree. The US has been a bit moving out projects. I think it's a general trend. We hope that with normalization of this liberation day tariffs, this will get a little bit more dynamism again into the market. Here you see, just for illustration, the many markets we are in.
On the left, the semiconductor industry, we diversify there, as you can see, iCaps, memory tools, display logic, and even down to sustainability topics in the subfab. We're also in air conditioning on top. On the right side, a number of other industries. I touched on most of them already, so not much details there. Yeah. Maybe I make a stop here at this point. I can talk a little bit more, but I also want to make sure that I serve your needs most. Are there questions at this point?
Yeah. I mean, I just want to ask about last bit of the small manufacturing and sustainable subfab. I mean, you said more recently, so things about sustainable subfabs, I guess. What is that? And then other questions. All right. First of all, the definition, what we think what subfab is. Yeah. Sustainable subfab was the driver.
Yeah. Absolutely. Yeah. In the end, you have a ballroom floor or a main floor of a fab where you have all the tools, where you in the end have the processes running in the semiconductor tools. To cater to them, you need to bring gas to there, and you need to bring vacuum to there, and you need to take away the exhaust or the waste. That is happening in a floor below. It sometimes is next to it, frankly, right? Everybody has their own designs. In Korea, they build big towers, even several fabs on top of each other. You see them from Samsung. In the US, where you have a bit more space, you can also spread out, and sometimes the subfab is next to it. Technically, it is the basement, if you want, where you do all that.
There are pumps down there where you create vacuum, and you prepare gases that go up there, and then you abate gases, meaning clean it up so that you can go out. There is also sometimes the infrastructure there for air conditioning, water preparation, and these kind of things in there. Obviously, when you are smart with how you use resources down there, you have the biggest lever of optimizing a fab sustainably, meaning the simple version of our software is, in the end, just switching on and off the right tools in the right problem in the right moment, depending on the process that runs on top.
Obviously, when you have a process-aware sensor on top, to put it simply, that then feeds into a software, and it understands when does it need what, the tool, when does it need vacuum, when does it need this gas or the other gas, or when does it need to clean up a clean process, then you can switch on and off these components in the subfab. Through that, save quite a bit. We published a white paper that we could save two-thirds of CO2, for instance, also other gases. In the end, the same for energy. In the end, this came out of a collaboration as well. It turned out to be mainly a software solution to become sustainable. The sensors sometimes are needed, but it's more about bringing the data together and being smart about it.
I think we have a few more future stages there of how we can push it further. We do basically rather simple optimization at this point still, because you could go, when you include scheduler, you can even plan ahead a day or two, and then you can shut down things that you will not need for half a day and take half a day to switch on. Now it's more about a couple of minutes switching on and off, ramp up and down. This has more potential. Yeah. I hope this answered. Yeah. Yeah. And then just a minute on software.
Yeah. I was kind of sorry. I was reading a little bit about kind of the small manufacturing tools. And it hasn't been used to sort of describe.
Is it sort of fair to say that your small manufacturing tools are more so sold to the kind of 200-mm wafer fabs?
Yeah. I mean, there was definitely a push in 2022. If we go mentally back to that point, the automotive industry slowed down real fast, and then was speeding up again, and the capacity was not there. Specifically, the bottleneck was the chips manufacturer. This is companies that are focused on automotive, NXP, STM, TI, also power chips. It is all a bit related there. They needed to have additional capacity. That gave a real boost to this smart manufacturing in the second tier. Our aspiration is also first tier. Sustainability, we do a lot, actually, with first tier. It is a mix.
We also need to say in first tier, to dislodge a homegrown solution they have for 20 years sometimes, maybe invented by a young smart engineer in the 1990s. Now, this maybe young smart engineer is slowly going towards retirement. That is a little bit when then they are looking for off-the-shelf solutions. That is often when we can flip it a bit. It is a slow selling process, I have to say. In the tier one, it is a bit less agile in a sense on that IT landscape. We are making headway.
In terms of, you would say, within that sort of small manufacturing, the share of your addressable market that you think you can get? I mean, presumably, TSMC has their own thing that they run. That is probably not going to change, I would say.
Yeah. Yes. No, I guess. I mean, we have everywhere software.
The question is how many of the modules they buy. I mean, there are no comparisons to SAP. Maybe just as an illustration, there is so much in these solutions that SAP offers. Once I was in a company, we launched SAP, and in the end, it was just launched for HR management. Do we have SAP or not, right? You can say yes or no. Sometimes we have, or let's say always we have the software that goes with the sophisticated sensors for the data analytics. Looking at the tool, looking at the process, right? Being process-aware. How far it goes beyond that is often the discussion. We are in there. We have the relationship, and we work with them, but it still needs probably some time to break into further places.
It is also a market that is still defining itself a bit. The autonomous fab is a long aspiration, I guess. This working group in the SEMI org is around for time, but only now, the really last couple of years, they are really starting to push forward. It is kind of a long. I think so. On software basis. That makes up 10%-20% of the SEMI. The software only is 5-10% of the overall. Yeah. We do sell a lot of software modules with this or that. That is not what you are asking, right? You are asking about the software only.
Yeah.
Yeah. Exactly. That is that part. Otherwise, software is much more entrenched. I mean, most of this have software included. You have software upgrades to it or additional modules you can buy, boosters, more analytical packages, more integration.
Even these sensors here have different interfaces and things like that. So it's kind of always been part of our DNA. What you're asking, right, is software only, which really is a bit more on the level of where MAS operates, while we are always on data analytics and that, not competing with MAS, but it's a bit of a different level in terms of. MAS is just a manufacturing execution system, meaning the system that steers all the switches and valves and so on. It's not what we're trying to do.
Then that complex, there is no exclusivity now in the new bipolar world order. Delivered to Chinese, delivered to Americans, maybe produced then in Asia as well. Who does the data belong to? Do you have to make sure that it's kind of ring-fenced for the Americans or the Chinese or?
It's not easy, it's never been really easy, but it's gotten a bit more complicated. However, what we need to do is always focus on one customer, and the data belongs to that customer. We also run it in, yeah, we run it in their cloud or on their premises. Yes, we have access to it. For that, we have a security protocol, obviously, for support, but also for upgrading and expansion. In the end, the data belongs to them. We need to ask for data, and then we get sanitized data. We can learn from the data anyway, right? We should be able to beat homegrown solutions always because if you are in almost every fab, then you can go and learn from all of them. We need to be careful of how much we take out.
Yes, this trade war makes it harder a bit. Yeah, it's possible. I think it's just more complicated. We had to ramp up our clients' processes, our regulations. Sometimes product delivery, now I'm speaking more general, is a bit more complicated how it needs to flow to the customer. It's not helping anybody, right? This is all eating in everybody's margin, trade wars, but we can manage it.
Does the tendency, customers become more reliant on you than they were like 10, 20 years ago? Yeah, for sure. Sold like hardware solutions. Now it's the bundle, and now there is a certain dependency increase. It's stickier. I would agree. Would that continue for the next 10 years?
It should. Yeah.
I mean, also that we have been able to raise our profile to be more of a solution provider where we really can build senior strategic partnerships where in the past, really in the past, 15 years ago or so, which I know from my predecessor, we went more through the backdoor to the procurement category manager and then sold a certain piece. Now we have much more these conversations where we talk about, yeah, what is your problem in the next generation or the one after? Where can we help you? What's the problems you want us to solve for you? How far could I go? Could we go as far as that you would profit and monetize a year in movement for this? We sometimes do that. Yeah. It's not so easy because there's so many factors, right?
When you go in there, what we often do is also change the processes because their processes is maybe if you know software implementation from anywhere, literally, maybe you could go back to SAP. Two problems you always have. The data is not up to standard, and the processes are not harmonized or standardized. We have the same problem. We go in there and we work with them, and that's why this selling period takes so long because you make a proof of concept and pilot and then convinced, and then you take the next one on, next one on. Yes, we have models where we show how much you can improve, but it's a lot of factors, obviously. They often turn the whole company around. That's also why we need to talk to the higher level of the management.
Otherwise, it would not work. Yeah. Good questions around software. I want to stress, as excited I am as a computer scientist originally, about software, we have different product lines that are as exciting as this, also new sensors or things like that. I just want to make sure there's not one trick pony for us. We also do not think that hardware is, in general, a disadvantage versus software. Most recently, that's why the ratios stayed the same. The sensors have been growing nicely too, actually. That kind of growth that you've seen in the vacuum is 18% year- on- year. That is commercially sensors rather than like an implementation of software that's driven it more. Yeah. I think software was a bit slower in, if I recall this correctly. It has its own dynamic. That is what I want to also show with this slide here.
I think in 2022, if I recall it correctly, during this automotive supply chain crisis, it really got a boost and it jumped up. Then specifically, this gauges group grew nicely. That is also, if you remember, our investment program, you for sure do. We talked about this big investment program we did at the time. That was a lot with this. You will see today in the tour also where we invested in automation. It flipped to this. Most recently, last year, sophisticated sensors took a jump, specifically this one, the Masspec. I think this year, this one has, again, a good speed. While software went a bit down and is now gradually increasing.
Just to give you a little bit of flavor of how we look at it, we have about 20 such product lines, and they're really behaving quite differently, which is, in the same market, they're quite different. You have diversification of big customers. We can talk about the big customers. Almost each one has a bit of a different story. Just think about TSMC, Samsung, Intel, how different they are. You have diversification over time because you sell an OEM tool, which pressure sensors typically are, at a different point of time than you at the end want to have a fab built. You sell the sensors. These sensors you sell either early on as part of the design, R&D for a process, or when they have problems in manufacturing, quality problems that they want to work with, so they add it on later.
Then you have maintenance, I want to say, like the leak detection. This is somewhere at the back end. You have a few at the beginning when you start ramping up, and then really when you up into HBM, that's when you stack up more leak detection. If you think it's even over the timeline, diversify. This is what helps us, I guess. How much do each of these tools cost, I guess? They have a very wide range of costs. Pretty much, yeah. They also have different sales channels. That's why sometimes we talk about gross margin and tell you it's a bit hard because of these fluctuations that they have different gross margins. This would be $1,000, $500-$1,000. This very simple sensor for pressure would be much cheaper than here's the sophisticated one. This is a heated one.
This is the unheated one, the direct pressure gauge, the CDG. So they're more like $1,000. This one can go from $7,000-$8,000 to $60,000, or we have the big brother of this is this big, it's like half of a phone booth. When you want to do analytics or you build a real big machine for lithography, then you buy these kind of things. Over there, leak detector, $20,000, something like that, $30,000. It depends on the extras that you want. You have another sensor here, for instance. It's also from Syracuse, from the US. This works with plasma. This is an illustration. This is a plasma cell where you create the plasma like Northern Lights or what you had in the lights earlier. Depending on what gas goes in here from the chamber, the color changes.
Through this window here, the optical sensors look at the color mix. You can imagine this Northern Lights effect. For that, you need some sophisticated AI to understand then what makes this that. This one would be about $20,000. This is pretty popular in Oregon and in Taiwan, for instance, chip makers.
How durable are tubes like zero iodine here by harsh gases or things like that? How long do they last?
This is always a fight. What you want at least is from one PM to the next, meaning from one maintenance cycle to the next is six months, typically. You need to make it at least six months. That is not going to impress anybody yet. I think two, three of these PMs you should survive. Four. It depends, right?
Some of them run forever if you have a simple chemistry, nothing aggressive in there, no crazy pressure changes. It is also about how you run it and how well you do the maintenance. Some most extreme ones, for instance, when I was the US president five years ago, we had a massive problem with this one in these very new processes. We developed, sorry for the jumping, we developed this thing here, which was the next generation. This is actually this, but with a whole system. It has a pump, it has a reference, it can plug into different sides of one tool with different pressures and switch around. It is a system. This one is able to survive aggressive gases of a certain chemistry. Almost for all the different types of chemistry, you need a little bit of a tweak.
That's why I say this is one of the fights and that's where we are leading. That's why we have 85% of the market is because we need to constantly adapt. This chemistry is changing. On the right side, you see a little bit of a chart. On the Y axis, different gas amounts, and on the X axis, you see the timeline of about 20 minutes, just as an illustration. Then you have these different variants that you see on the left side, and they constantly evolve. Yeah. I still have here the prior version. This is the MPH, and this is the APEX or APX, so the different generation. I mean, the system is in its core very similar. Yeah. Rapid prototyping is very important for us with the customer. That's also why we need this collaboration.
We want to go in there. We want to stick a first variant in there, I believe, can survive. Then we try it out, then we adapt, and we adapt, and that's how you can do it best. We do have our own investment also to simplify this. Sorry for the jumping. Maybe it's not here. Here. We have, for instance, a mini fab in the US where we have all the different tools. You see there are ALD, PECVD. There's also an edge tool in there and so on. That was still during COVID. That's why they had bearing masks. Where we got the tool, and normally the team calls votes on the name. So that one is called Darth Vapor. PECVD too. They're all a bit nerdy. I think it's a good thing that shows they're creative, right?
Anyway, in there, we can test also sensors with new chemistries, right? We try to limit the amount of test tool time that we use in a factory with testing as much as we can in-house while we do go through these fast iterations. Nonetheless, in the end, the proof is only proving the pudding is when it does run on the tool in that specific process, how they do it. It goes through these rapid cycles. On the left side, you see our innovation process there that goes in through these fast spinning cycles. This is the prototype cycle here, the blue one. This is the making a product cycle where you have different variants and versions then after. There are a lot of labs there that you see where we tinker around trials.
These circles are the same size, but I can imagine that there is a shortening in life cycles. Yeah, it's fair, actually. You could say that the green cycle is probably bigger because that's where you make the full product that has all these quality standards and obviously full manufacturability and serviceability requirements and all of that. Whereas the blue is more really go quick, what I described earlier, between our internal mini fab and the customer try out a new sensor. And sometimes you see aluminum foil and duct tape and a little bit crazy like this. The picture in the middle there. Here, this one. This is a picture of this prototype that we made for ASML, the next generation, this EXE 5000. That's about five years ago now. The early days when we got their requirements and then tried out the first version of it.
It looks pretty terrible, obviously, right? There's all kind of ad hoc piping. But that's how you quickly learn and quickly eliminate issues and then move.
Is it fair to assume that the cycles over time have become shorter as well thanks to the deployment of AI and other things?
Yeah.
Management as such.
Yeah. I mean, I'm really an innovation nerd. I come from a background from software engineering from an agile approach. I did that already before at Mettler Toledo two times to take an organization that also does hardware into an agile mode. That's always not so easy. Software is quite easy. Just if you imagine you want to have a continuous integration, meaning whenever you change something, you want to immediately compile a full product that is tested through and that could be sold and used.
It's a bit harder with hardware because you cannot just make hardware every time. It takes some sophistication to find out what exactly an organization needs to do to stay agile and have these right iterations even with hardware. Yeah, this pushed a lot the last five to seven years. I would say definitely we have increased the innovation cycles. We have to. If you want to go into this fast prototyping with customers, you need to be quick. We talked weeks, right? You launch a new variant in months. That idea, not five years development, and then you come with the new thing and you unveil it. Nobody is interested in that. We hardly do it. I mean, it's something that I sometimes regret because then for you guys, it's a bit hard to see what is going on. It's this continuous rush.
This is really what we work. You see all our labs are full of this stuff where they just try, try, try. And then when it's a new variant, it's launched right away and it's sold. It is much faster, six months, 12 months. When we needed to speed up the Chinese story that I told earlier in China to compete in leak detection with automotive, and we installed this innovation lab there, we had four launches that year of that product. Every time a step further and a variant and so on. Just because you need to speed up, you can. Chinese customers specifically cannot say, "I have something in three years." It hangs up, phone is gone. It's not interesting. Go with something that is 60%. It's really 60, not 80. We had to try to also learn this.
I mean, you see how much more important the Asian market is for us. We adapted this. It's a best of two worlds, right? The European first time right, meaning take a long time, all eventuality, make a fantastic product, small, it's slick, but takes a long time. It might be off the market. If you combine this with super fast hacking it together, which sometimes is really terrible in product quality and reliability, you combine it the right way, then you win. That's what we tried also. Now, that's a couple of factors that I mentioned why we were speeding up. AI, you asked as well. We do for 25 years AI. I like to call it data analytics, honestly. Which always had machine learning or statistic methods and so on in it. We apply them also to our own product development.
I think in a tech day and also last year when there was this Kepler Chevreux AI summit in Zurich, I talked about how we use it for our products, but also how we use it for our internal development. Yes, it does speed up. I think it's been kind of building up gradually. I would not say it's like a big bang. We have really built up our AI team now, meaning people that really have this extreme background and have this focus, have a PhD, they go after the most difficult problems that we see. This we really scaled up. It's a fun amount of stuff that comes out of it. We'll talk more about it. It's happening. You could just see that they are basically over weekend, they did a couple of these LLM bots.
Ask Chen, ask Innie, there's another third one. They just do that on the side. That's not even hard for them, this bot stuff. The hard stuff for them is, for instance, this scheduling optimization, which still is an NPR problem, meaning a problem that cannot actually be solved without assumptions and heuristics, or this data time series where you have a trace of a measurement from a sensor and you try to understand, is this now good or bad? Is the wafer healthy or not? It's not a one-to-one relationship. It's more a wafer dies by a thousand cuts, meaning a thousand tiny things go wrong and then you have the yield go down, right? You need to kind of think over several tools and over several processes. You see a picture. Sorry for the flipping around, guys. It's back here again.
On the left side, you see that's something we launched about four years ago, this Smart FTC. It shows in this blue the envelope of where we think the wafer is healthy. The green one is the trace from the sensor in a very simplified version. Obviously, there are many different traces, many different measurements, and many different envelopes. What we developed here is to show when it's in the healthy range and when it isn't. Over time through a scoring system, you can say this wafer is going to die soon or this tool needs maintenance and things like that. This we can also apply to our own sensor development, obviously. Things like that, how you optimize the parameter for measurements and so on. That does help. We're really boosting this.
We should ask you whether you have stability in those teams, these probably international teams. I mean, that's crucial to keep them, to lose them too. I mean, it's not so easy because a lot of them were originally sitting in Austin. And Austin, as you might know, is the Silicon Hills, meaning they're not in the Silicon Valleys anymore. People also leave. They go to Texas. There was a lot of competition with Facebook, Google. They were all there, Amazon. That's already now three years ago. Things changed, right? They staffed extremely during COVID times. Then they overstaffed, and then they really cut jobs quite a bit. That was two, three years ago. We have no problems to find people.
It's also sometimes, I know this sounds a little bit elitist, but I mean, do you want to go in semiconductor and solve truly difficult physics problems and chemistry problems, or do you want to optimize how many clicks you have on an ad? You also need to ask yourself, what is making you happy? You always get a bigger salary in the end when you go optimize ads, Facebook and such. That's not maybe for everybody. We find people that are really focused on this. There are people that have, typically we find them through papers and conferences. This is this operational excellence, industrial engineering paired with data analytics people. There's a little bit of a breed now developing. There are some institutes and some universities that work on that with a focus. That's where you coach them. We also don't need 500, right?
We have maybe 100, 200 software engineers, and maybe 10% of them are really hardcore, that kind of, yeah, I don't know, PhD, data analytics, and then operational research type people. Some of them post on LinkedIn from our guys. For instance, Igor is one of them. Not everybody does. Some people are a little bit more. They have a lot of introverted people, right? They would rather like to. They very much enjoyed COVID because they could stay at home in the basement. They are still the same people, actually. Go to market, as you described, happens like almost week by week. Immediately. There is no fair cycle or conference cycle. We do not do big laser shows, smoke, and unveiling of a big machine. We do not even have big machines.
Software, you follow us on LinkedIn, on the website, and you'll see when we launched Ask Chen. The story maybe for those who don't know, we acquired FabTime last year. One of the main experts in this group was Jennifer. She's 20 years in the market helping fabs, optimizing the cycle time, meaning how long does a process run through one tool and then several tools. She wrote many newsletters. It's a very famous newsletter that people read. She wrote white papers. These guys, nobody went there and said top down, "Hey, you must do this." The idea is always you create an environment and people just go, just try and break things and make mistakes. They just over the weekend, I know, I talked to these two guys six months ago, eight months now.
When I met them, yeah, we wanted to be the first one with some LLM type thing. We just did that in a weekend, two guys, Igor and Hollander, the two guys. They just launched Ask Chen, which is basically just a smart language model, meaning limited, only on her white papers and other materials she created. Now you can add the tool, you can buy this additional module, and then you can ask, "Is what do you think with this data that we see in our tool? What should I do? What levers do I optimize?" There are different phenomena that they describe. They can suggest, "Look, this is probably this type of pattern you have, and hence that solves it." They just did that. They launched it. I'm not going to go and sign off on stuff like that.
They need to go. Then where you see this is basically for everybody else outside of the direct customer relationship, you see it basically just on the website quickly. They would go, our interaction with the customer is marketing is not so important, the market communication in that sense. What is important is that we go to the conference, talk about it. We have our own conference as well around smart manufacturing software where a lot of the customers join. Almost 80% of the customers in the market, they're regional, join, and they have these discussions, and then they specifically talk about this stuff. It is very plugged in. Yeah, it is small. Then you go there. Yeah, when I go there myself, I mean, my old computer scientist's heart is beating. I do not understand so much, but it is just they are the same nerds.
Interesting enough, they can talk across companies about these topics. They would also talk about roadmaps, what should we develop next, what is important, what does not work. This is relatively quick. That stuff there that developed in two months, it is a small thing, right, to be honest. We do more complicated things like that. That is an example of where maybe a simple AI tool was made. All right. Beginning, you showed a chart with the sales. Is this the picture where you show where you sent the tools or where you really invoice? Yeah. It is a complicated picture. I mean, yeah. We try to be as accurate as we can, but we really not always can. Especially GenVac, half of the GenVac business, so meaning $80 million or so, is basically going through distribution partners.
I mentioned to you in the beginning, we are key account management focused. That is either a large customer or it's a large channel. Obviously, we have these products, and you'll see it when you walk through our manufacturing. We have them in different colors, do private label, and so on. Then it's hard to know where it goes. With the key accounts, we know it a bit better. With the end user, it's very clear because we ship it to that factory. With the tool maker, again, it's not entirely clear. If you also look at the North American OEMs, they may be 10% or 20% of their products are actually going to the US. We manufacture maybe in Europe or in Asia, and they ship to Asia. This product never goes to the US.
Maybe it is listed as a US product in some cases, but it is not actually ending up in the US. It is a bit more blurry than that. These numbers, I would not 100% guarantee that they are accurate. They are good enough. I think maybe that is a question for our CFO, or maybe Dimitri can also help, of how accurate they can be. This is based on headquarters and other shipments? No, it is both. It is more intelligent than that. I can tell you, this is not so transparent, honestly. When we also were to talk about trade tensions, it is very relevant, that number, though, that I mentioned, that only 10%-20% of the products that go to US OEMs actually end up in the US. Very little is actually affected from tariffs, potentially, meaning from Europe.
Right now, we are in this state where it's only 10%. That's not impacting trade, at least in our space, much. If it were going higher again, then that would be a relevant number. We have been talking about different scenarios and so on in the earnings release of year one. Nothing more to add. No, nothing more to add. All right. I got the blessing here from the finance folks.
All right. In terms of pricing as such, I mean, Geberit shows this one chart where you see indexed 2021, 100, that they're still like they went up at 235 and then down to 22, and that stayed there. There have been price pushes due to salary increases, inflation as such, on raw materials, and so on. I mean, how much of your growth that looks nice, prima visa, was pricing related?
Yeah, there was some of that. Yeah, when was the big exercise? The exercise was in 2023, I think. It was, yeah, mostly, and then in 2024. That was mainly to digest this wave of inflation in the different regions. Mostly a one-off. Yeah, it was mostly a one-off. It was a long, silly negotiation with each customer and each supplier where everybody roughly ended up at the same margin with a lot of wasted time. That is why I call it silly. We might end up with the same thing again next year, right? Probably not, but it might be because inflation might come back for a moment. In the end, this is not how we can expand margin with long-term partnerships. It is also relatively transparent. We were certainly not losing, though. We made sure of that.
I was really pushing also to our professionals our pricing capabilities, and we'll also be better in the future. But when we talk about pricing power, maybe that's a question. We typically gain this through innovations that we can price well. So innovate something, set it at a good price, and have a better margin. And then some of the back-end productivity gains and things like that. So there's some of this in there, but it was a one-off correction, right? In the U.S., it was don't take it for granted that I know that remembers the numbers correctly. Maybe it was 4% or 5% there. In Europe, it was something a little less, but similar.
Overall, it's between 2-3% overall.
Yeah, exactly. It was obviously Switzerland was different, and Asia had its own dynamics.
Yeah, there was some of it in there. If you look at the CAGR, it's a chunk, but it's not material. It shouldn't negate the whole story.
Sort of maybe small margin in the long term, just from the kind of mix of end markets like divisions over time growing at different rates.
That they grow different? Yeah.
Some of them might have a higher or lower margin, and maybe the higher margin one grows faster. That was kind of natural.
Yeah. I mean, it's been pretty balanced out in the sense that over time, what I before explained is waves. I couldn't say that there's a real big trend. One thing that we talked about quite a bit is when gauges grew quite a bit. That was this leap. I think this was 2023.
That is what I mentioned earlier with this CapEx program. This has a lower gross margin. It has a similar EBIT like the others, but it has a lower gross margin. It is 40-50% versus we go up to 80 plus. These sensors are more like 50-60%. They need more application engineering. That is just what makes the bottom line similar, but the gross margin different. There you could see that our gross margin went down. That is also where I think we had most questions about what is the development on your gross margin. This old question, do you go to 50% or not? This is a bit hard for us to say because it comes with this mix. Make sure that folks online, is there any raised hands? Guys, feel free to raise hands online.
I know it's not as easy as for the folks here to ask a question, but feel free to. You can also write it into the chat, right, Dimitri, or raise a hand.
Is that changing for the software in terms of subscription? Is that most put down and?
It's quite similar. Yeah. Subscription is certainly established now. So it's a good mix. One third, two thirds, or 50/50, somewhere in between there. It's a little fluctuating. I pushed the subscription more. Yeah, it's an advantage. Financially, but we are pretty balanced already. I think we always would go with what a customer preference is. It's also part of a larger deal, right? What do you pack in there? If you pack all in there, would you like rather this as a one-time license and this is a subscription? You can play with it.
We do not have fixed guidelines, let's say. When you said the max gross margin of 80% is for the software, right? Yeah. Yeah. Why is it not 100%? I mean, there are some services included that we do installation and so on. There is also hardware included at times. We supply actually the hardware for, let's say, a server that goes into the shop floor. We would supply that because it has its own specific requirements. Typically, when it goes into a data center, they will give us a virtual space.
In terms of stuff I was reading, it sounds as though it is able to capitalize on problems within some of your competitors' supply chains in selling a vacuum coating in 2021, 2022. I was just wondering, what is it?
The difference in the way you can organize your supply chain, that may have been the case.
Yeah. All right. Good question. Maybe then I'll quickly show you the world map. Sorry for the flipping around, guys. Don't look too much so you don't get dizzy. Here. This is the chart that I showed at the earnings release 2021, 2025. It's the normal chart of our locations, plus the red circles are the ones where we currently have reconfiguration ongoing. Maybe an explanation about this. What we tried to build over time, and no doubt this was also historically how we came about, we were a distributed company always, right? If you imagine we were these three companies coming together or three divisions of other companies that came together, actually, it's more accurately maybe. We always had the three centers of gravity, Balzers, Cologne, and Syracuse.
We did acquisitions, and we did some certain expansions, for instance, the Shanghai factory we did over 20 years ago. This state-distributed organization, the only thing that they always had as a principle in the past is make one thing only in one place. Have a competent center in one location. Because also how we started out was actually we had the products in different locations at the same time because different companies came together. Actually, at one point of time, this one here, the MassBIC, was made in Cologne, in Syracuse, and in Balzers. Three different variants because it was three different companies. Obviously, we standardized that a long time ago. That was a little bit the former history. Now, about five to seven years ago, when globalization kind of slowed down or even went backwards, right?
That's basically the trend that we have, that we have more regionalization. We pivoted with our strategy. We also strengthened this team to think more strategic. Now it's not the simple rule, one thing is in one place, but now you need to make it where it makes most sense. That entailed a few things. You needed to make products that you can shift around. Much less tribal knowledge, much less there's a bunch of experts that can make that. No, you need to standardize this, that you can move this from Cologne to Shanghai, or from Shanghai to Kuala Lumpur, or you can send it somewhere else. That's what we worked on the last five to seven years.
The COVID time actually further emphasized this because when we all looked at our supply chains more in the aftermath of it, of the supply chain crisis, we were thinking through this dual sourcing, and there was a lot of nearshoring that people wanted in the US and in China. It was a very strong push for that. All of this together got us to the point where we now have the ability in these different competence centers to move product lines from one place to the other. Not everything needs to be able to go everywhere, clearly, right? The smaller locations are not made for receiving more, typically. What it is, is the three big ones plus the two factories are where we can go manufacture. The three big ones, again, that would be Germany, Liechtenstein, and US. Then we have China and Malaysia.
You can actually maneuver all the four regions really well. I'm talking four regions because it's a little bit how we look at the world. There's America or North America, there's Europe, there's China, and there's the rest of Asia. You need to be able to kind of have a footprint in each one of them to play best. Also, things like China Plus strategies, you can then implement. Based on this strategy, I know it's a very long answer, I guess. Maybe it helps answering a little bit how our DNA works or how our operating model works. The most recent announcement came at Liberation Day. This was not news for us necessarily because we had these projects, the red ones there, already going because we could already see things. That's not because we're particularly brilliant.
We're just listening and adapt, much like with innovation, relatively quick. We knew that US and China is decoupling. Everybody talked about it for four years. Since the last Trump administration, this was a push. We just started to rearrange it all over time. All we needed to now do is speed it up because 20% tariff is different than 140%, obviously. That's how we navigate it also in the future. Without.
That's more brokered costs, which was a big issue. Yeah. Just before. Will they go up again with the trade tariffs? Or do you have a solution there?
I think we did, but never say never in this world, I guess. Yeah, I almost forgot about that pain. Thanks for reminding me. That was at one point of time, if you remember, we communicated, I think, around $15 million impact in broker cost.
That was maybe 2022 still. I'm not sure. Please, maybe you guys can confirm me, but it will be in the public domain when we talked about this transparently. Just to name the issue, the problem was there that we couldn't get our hands on our chips. This one uses, for instance, an automotive-type chips because it needs industrial, it needs long life, it needs to be robust. Automotive, that was all the rush. We are not the first ones because we buy smaller pieces, even though all of those guys are our customers. With some, we tried to go and phone up the same people we make projects with, "Can you give us a batch?" Hey, I don't know what exactly happens because we saw serial numbers that look weird.
It felt almost at some point of time they were purposely selling it through brokers. There was an immense fee on top, and then only we received it. I do not know what happened. We never really found out the full truth. In any case, we paid a lot extra, sometimes 3x, 4x of the chip in the end of the PCB. He went away as he came. Nowhere. There is just a bunch of people, there are in-between middlemen, folks that probably really made a lot of money that time. Now we have to try to go and get closer to our customers, better planning, more second source, all that stuff. Some obsolescence as well because maybe a product is not obsoleted, but at the back end of its life cycle, so it is a little bit scarcer.
You want to be rather in the bulk of where it's high in the life and running. We did all kinds of things like that. We also diversified our sourcing for PCBs to have US and Asia more. Yeah, we're in a much better place, but I don't know what happens next year. I can assure you that through this network and this thoughtfulness of how we set ourselves up, we are way more agile than others. We don't have to phone somewhere to look for a plot of land, build a factory, make local connections, and hire HR and finance management and all that stuff. We have locations where we can easily scale up and down. That we do have. The same with supply chain.
There's obviously each one of these five big locations, a good supply chain team that works with the same methodology as everywhere. There's also something that changed. In the past, it was a lot of local sourcing and a lot of local logic and methodology. Now we standardize it so they speak the same language. They can also throw each other balls and things like that. Yeah, did a lot of work on it. Hope it worked. At this point, we don't see this as a problem yet because it came a bit with this automotive crunch. Yes, there was another problem, though, the supply chain crisis. We all remember still this picture of the 500 chips outside of Shanghai. I mean, we have a little bit something like that, though. That worked itself through the system in 12 months.
That's only when we saw it. I will say never say never, but it should be different this time. It has a bit of a different nature. As it presents itself to us, at least. Is it Andreas? Oh, yeah. Reto? Hello. Yeah. Maybe with the distributed expertise all over the globe now, how do you make sure the expertise doesn't leak, that it stays within the group? Do you have some particular policies? Yeah. No, that is a part of this strategy. There are some things like I'm jokingly called, as an example, the watchmakers of our sensors. These are the guys for those who are physically here. You will see them later. Those that make the innermost of the sensor head, tiny, fragile little wirings there. This is, for instance, something where we believe we have a little bit of a moat.
Only a few people can do it, and you need a very long time of training, one to two years for basic, and then five, six years more. Such capabilities, as an example, we try to be protective and place them only in places where they make sense. You can do that because this is not much value, right? Whatever tariffs and so on, it should work. Others or software code that we obviously try to lock down. At this point, we do not have software openly in China for most of our products. There is an exception. Some of the products, we actually will start increasingly to innovate in China because, I tell you, it is there, the center of the world of innovation. It is not here.
Our strategy is now not anymore protect stuff in China, but go run as fast as you can with the local competition and then export into the world. I think you need to think dual these times. Where maybe 10 years ago, everybody was going there and said, "Okay, IP, we're going to be careful." We have this decision by product line. When we went there this year for the strategy workshop, we specifically looked at these places of where we want to innovate in Asia or in China and where we do not. It is not easier. It is more complicated. It is a good question, Reto, because it is moving quickly, and these processes, they move quickly. How can you go and make sure that you stay safe? You cannot always. I think we always a little bit lean towards speed versus protection.
I think in our industry, in our world, that is more important. It does not mean neither. It does not mean not the other thing, but it means a little bit more rather speed if you had to choose. Goes almost for everything. Okay. Thank you. All right. Sure. Good question. Thanks. Is there any other questions from online? Just making sure that.
Any questions for you, Oliver? I have been CEO for some time now. And I remember the first time we met, you things that you were potentially going to change, the things that you have seen in the organization that might need adjustment after so many years of the same management, which is not a bad thing. Of course, you are a new person. You have been in the company for quite some years, so probably you have had some ideas already before.
Maybe now looking back since when you started, can you elaborate a little bit the things that you have changed or adjusted and the things you still see probably that need adjustment? Also, the R&D has increased since you have arrived. I'm not sure if that's connected to you or if that is just the time that we have right now. Maybe I can elaborate a little bit on since you started to today, what have you actually changed and what you think still needs to be addressed?
Cool question. That was asked a lot when I started two and a half years around there. Yeah, it's good to check in, I guess. We do our obviously annual cycle of checking in, which is always Q2 and Q3, and see a little bit how we're progressing.
We produce normally a three-year plan where we detail out what has worked and what has not worked, where we need to double down and where we may stop. Also important. Yeah, maybe to outline quickly what my strategy at the time was. When I came here in 2022, I was together with Lukas for some time, and then we had the split. He was still running stuff. I was still running a little bit my prior location, the New York location. At the same time, I was also allowed to go and formulate a little bit of the strategy, the future strategy with the management team. We had a couple of workshops around that. What came out of this was basically a two-pillar strategy for the next 5-10 years.
This is also the strategy where we said we can double the business organically because we saw the building blocks specifically of how we can grow each piece. It has aged pretty well, even though, oh my God, what happened the last three years really was not what we have foreseen. It is okay. I mean, if you are agile and you like a challenge, it is not bad either. The two pillars are one is around a couple of market initiatives. Obviously, there is semi in there. There is new energy in there. I will talk about more what that is. There is the number of growth markets that we see and maybe the markets that we do not want to do anymore. That was one half. The other half was about 10 initiatives to professionalize the company and make it scalable to support that.
One example in there is this global operational strategy, the operational excellence strategy, which I think you heard a bit more how it materializes. Maybe quickly back to this three-market strategy. One is the semiconductor, which the idea there was to make sure that with all of these top accounts, we end up with a strategic partnership. We go and innovate on their next generation or the one after and build that up further. Yeah, we made good progress there. Not done yet. This is long-term relationships, but I think we did a lot of great big and small moves that move us in a very good direction. There was also about finding new markets within semi and strengthen that base to further diversify, but always with being selective about which markets we go to.
It needs to have the right growth and profitability profile, the right synergies. There we also did some progress. I mean, subfar, we talked a little bit about today. There is more that in the works, and I think we did not progress as much backend, but we did do some stuff. There is prototyping of new products backend of the line of semi manufacturing. There, semi has developed quite nice specifically look at these top accounts in our partnerships. I was also personally very engaged there with our executive team. For me, this is a senior executive job to go and make the handshake on the top. Also, specifically when shit hits the fan, you want to be there and you want to get a phone call. If they are unhappy, I want them to yell at me. That is a good thing.
Because if they don't, they go somewhere else and they're unhappy. There was a good amount of this. We formed this new section, new energy. Maybe at one point of time, we might change our markets. I don't want to confuse everybody, honestly. New energy is everything that we've seen that drives—maybe I can show a picture. Sorry for the flipping around again. Everywhere where we can move sustainability forward. As it happens, INFICON in its DNA is actually very strong in helping other customers do sustainability. The core of our internal sustainability strategy is actually so-called scope forward, as we call it, not official, is how can our product move the customer's sustainability. One example, upper left corner, is this subfab optimization. I a little bit went into detail earlier.
Obviously, battery is one, solar is one, biomethane is one, hydrogen is one. There is also other stuff. I even published something when we worked with carbon capture companies, for instance, Climeworks. There are a few more ideas in this space which are further down the line. Maybe might never come or might be big opportunities. This has been part of our new energy initiatives, big markets also for us and growing markets. This was about building out our foothold in those, but also exploring new ones. It is ongoing. I mean, battery, I believe we have really a strong place. Solar also, I think, and PeresGuide then brings the solar business back. We are ready for that. We have the products. We have the collaborations. Biomethane and hydrogen, similar, that we can go into this detail.
It's a bit more explanation maybe what the dynamics is. I'm optimistic there. We have the growth markets, which is especially what you have in general vacuum. It's a long list of different markets. There we really try to understand each market of them, 20-25 markets, and categorize them where we want to push, where we want to retreat, where we want to be optimistic. I think we have a pretty good feeling now what we want to do there and have adjusted our strategy. That's the whole side on the market, one pillar. The other side on productivity, I think a couple of big topics that were in there was building the team for the future and further build our culture. We can talk about culture one point of time. It's pretty ingrained, has always been for us.
Also, that's maybe our quirkiness that you sometimes notice. We try to really focus on what is key and not do what is not key and no formality. Anyway, so there's culture stuff that we do. We build the team for the future. Digitalization is a big one. So there's all kinds of optimization around that that gives productivity gain. The operating footprint and excellence I talked about. Branding was in there, and I think we're almost done. We feel that we are now. After we had to catch up for a couple of years, we're in a good place. Of course, sustainability is also part of that, these enabler initiatives. That's the quick rundown maybe on my scorecard here. We're continuing. Pretty happy how it's going, actually, even though there were stumbles and troubles and difficulties and still needs a lot of attention.
What are you doing on the businesses to discontinue?
Yeah. I mean, we have some products that are obsolete for 10 years, and we really tried to go convince the customer, "Don't buy it." Still buying it. Some of the sensors. Yeah, we're a bit more proactive in how we do that. It needs smart solutions. Sometimes we sell them a license to the design, and then they make it themselves. We help them make it themselves because there's, I don't know, maybe a company that needs 100 of those, and it's just financially not viable for us to do that. Some we have deinvested, smaller stuff. Some we have just terminated. I just don't want too much distraction as we maybe had in the past.
Because what I described early on with this complexity of our products and the many different markets, yes, it has a risk to dilute our efforts. We need to constantly, I don't know, it's like brushing your teeth or weeding the garden. You need to always look, "Okay, does this still make sense?" Saying no to a thing is harder than saying yes, especially in our company where everybody always wants to try something new. There are very quickly a lot of stuff. We need this process of cleaning out the old product lines and old markets. We try to find good solutions there. It's not just sending a letter and saying no. It's typically not. It's a small world. People need actual solutions. One question online from Michael Feld. Oh, yeah. Michael. Yeah. Hi, guys.
Given that you talk about markets and you also showed the Contoura product earlier on, we have not talked about food packaging for a long time. If you could just give a short update on where you stand there, whether the opportunity is still attractive for you, whether it is still a growth market, maybe some perspectives on that. Thank you.
Sure. Yeah, Michael. I mean, that is maybe one of these examples where people went in there with a lot of enthusiasm, built a fantastic product, and then pivoted five times. Some of it we boost, and some of it we classified as more opportunistic with small efforts. What is that? I do not have a picture right now of Contoura, I believe. I have to scroll up and down. I do not want to do that to you.
What it is, for those who do not know, this is a relatively big product like this. It has a big opening chamber, and it has different flavors to it. You can put some packaging in there and find out, does it have a leak? It has this foil that goes around and then is able to go and find out, is there a leak? In combination with a mass spec, it can also qualify what comes out of there. This product was originally then thought for food packaging because food packaging is still this water bubble test. For those who do not know, I was in food once with metal collider. You take a bag of something here, and then you put this underwater and you look if there are bubbles. That is leak detection in food.
Okay, it's about shelf life mostly, not maybe unhealthy, but sometimes it is. We can, of course, do that better because we can go and actually count single molecules that come out of that leak. We felt, "Hey, guys, why are you not switching?" Most of them, they just don't see even this investment of this $10,000 product, not even on the quality side, right? Inline, not and also not outline. They still do this stupid water bath. That's okay. I mean, yeah, that's okay. It's a classic engineering problem. We have the best product, but people just don't want to adapt and don't see the business case. Or it's just, "Hey, this is not my focus. Now I have 50 other problems. Why am I doing this?" We do it for a selected few premium products, coffee, mooncake, stuff like that.
It's pretty much established there, but it hasn't grown humongously, right? This is like single-digit millions or something. What then happened, and that's often happening with our technology, it pivoted and it was actually a fantastic solution for battery leak detection. There you do care about the very tiny leaks. You have weird packages at times. You have pouch batteries, right? Basically, all the high-tech products that you have now, from car to mobile phones, everything else, mobile needs a battery. You need to, this is an early technology, still battery, because you try out different mixtures, you try out different formats, you make it bigger, smaller, different, whatever. In that whole world, this is actually a very good tool. It continuously is part of one of our products to test or solve this problem.
There are inline versions to it too. Can you sell the membrane as a consumer there as well? Yeah. I think it's a little bit less because it is the more possible. Yeah. It's a bit more optimized. I think it's more organized manufacturing maybe than in food. I don't fully know how much consumable we have in that space, I need to also say. It's not some significant number that I would know of. Yeah. In that world, the Contoura lives on and is a very strong product. As part of the leak detection inline testing or also in R&D, the bigger part is more outline. That's what happened with that. There are other things like this that happen all the time.
I think I don't see this as a failure or only in a failure that I would like to encourage the team to make in a sense. Because what I don't like is if we are not bold enough to go try it out because we worry this is not going to work out, then we have a problem. I want them to go there, try it out, make a product, and then we have this meter funding. We cut it off when we think we need to cut it off. You cannot cut it off at the beginning because you don't know where everything goes. Now, again, in this specific example, beautifully shows, it ended up somewhere else solving some great problem. That's all the time. This is not an exception. This is how innovation works. It's never linear.
It's always like this, some kind of weird confused path. So this energy ball, I want to maintain that produces this kind of stuff.
Starting wise, it's now quarter to 11. What do you want to do till 11?
Yeah. Maybe we do two, three more questions, and then we slowly wrap it up for the online folks.
How's that? That's the pick or?
You mean private label? Yeah. Yeah. It's about half of GenMac. So it's by $80 million or $70 million or something. Yeah. Yeah. This is important, I believe, because there is when if you especially look at vacuum technology companies, we are a pure play and we want to be that. So keep that focus. And these companies sometimes do a lot of different things, right? They pump, they have valves there, whatever, and they sell it all.
They can go and have people in cars driving around places and selling things, which is a viable channel for them. For us, it isn't. In the end, we sell to universities, three of these gauges. You can justify a person driving out there and go three times. If you sell more stuff, then that works. That is why we work with these private label partners for the broader market. As soon as we think it is strategic or it is a larger account, we try to go and work directly. Sometimes still through that channel, but then there is R&D involved. Sometimes some big science projects then morph into a direct relationship. I do not know how it exactly happened, but you could use ITER as an example.
At the beginning, they would just buy from one of the distributors some gauges, and over time, they needed a different type of gauge with certain capabilities. The industry knows where they come from, even though they have a different color. They would call us directly, and we developed this specific variant for them, and it also came in blue. That is sometimes a bit of a transition. It needs to be a strategic decision then. Because big science, we now feel it is an important market for us, at least for the big accounts. It is actually a growing market. We studied this last year as part of this exercise on growth markets, that the accelerators are particularly growing, but there are also fusion projects that are increasing. Quantum computing a little bit behaves like this too when it is about vacuum.
There's a number of these big science, some markets that actually have interesting growth rates that they're a little bit hidden away, tucked away from the rest of the world in a sense. We also needed to go there. We have a few people that come out of this space and worked. Specifically, one market manager or product manager is from CERN. He's plugged in with this space and understands what projects are there. The nature of it, we can actually, it works for us, and we have the products for it. It just needed a bit of cleaning up and organizing. That's, for instance, one of the things we did when we looked at growth markets. I think we also put it even on this slide here. Yeah, big science. You see? This is, for instance, an accelerator.
It's one of these tendencies also most recently. Many have a space program. Many of these, many countries, also of the emerging countries, have large projects like this. China has a number of them. India starts. Japan has them for a long time. This is interesting that this has an interesting dynamic. It's not big, but this is not going to be 10% of our revenue, I think. It's a good market. Also good to push R&D. Right? Any few last questions, maybe also online?
Say something about the strategy.
Yeah?
Obviously, if you want to buy technology,
roughly you have summarized it.
I mean, what we do maybe as opposed to before, maybe back to your question, is what I tried to do in 2023 onwards is to staff this differently and set up a more industrialized systematic approach to scanning all the markets equally, especially the ones where we find strategic important. Before, it was a bit more opportunistic. Then fill up the funnel, and that's why there's about 100-200 targets in this funnel. A lot of them are long-term, right? This is a long nurturing phase always, sometimes 10 years until we buy somebody. Anyway, when we look at what works for us, it is typically what you described. Because we have, it's a bit of an advantage and a disadvantage. It's not as bad as maybe with VAT, right? Our profitability, though, is relatively high, and our growth aspirations are also relatively high.
When you look at any target out there, sometimes it will dilute your financial performance. We will still do it if they have a technology that we cannot develop. The thing is, though, most technologies we can kind of develop, but we cannot develop all the stuff that we want to do. We try to balance with acquisitions also a little bit. Maybe if we buy this here, then we move the resources for R&D over here. We play with it a little bit. Out of these 150 or whatever targets, the bulk is these small companies like FabTime. We would have bought a few more. We were close a couple of times. We have at any given time about three that are hot. I do not know. Statistically, it did not happen. We were very close a month ago with one.
We found another one that is even better. We looked at it. It kind of goes back and forth sometimes. Yeah. That is roughly the strategy. I believe there is some growth in there, but you do not need to put it in your model necessarily, right? If this is just a couple of millions or maybe $10 million-$20 million, potentially. We look at big ones, but that often risk-wise and so on, it is kind of not our DNA to do something crazy. We would have to talk about it if it is really strategic fundamental to do it, but we have not found one yet that is that need. Right. You added like 50% of capacity in recent years. How much of that is in use now? That was this program from, now that I think back, maybe 2021 to 2023 or something.
I mean, a bit of stuff is down here. Not only, but one was because gauges exploded in this period of time. We have since still expanded, actually, some more. Maybe another 10-20%. It's also parties in China and parties in Finland, specifically on gauges. I think the original additional capacity is in use, but now since then we have further expanded because we tried to be ready for this ramp. The ramp that we talked about for a long time, right? It's always coming in six months. It's a little bit like ITER, the cold fusion there. It's always in 20 years. Then we have it. I don't know. I think joke aside, the ramp will come. I think it might be a little bit subdued this year because of trade confusions. We feel that it is coming. We have good signals now.
For that, we have 20-30% extra capacity to be able to digest that. Yeah. That is also reasonable if you want to not miss the boat or have very stressed customers in the middle of it. The problem is a little bit, I mentioned that before, that our cost base is a bit higher than it could be because we're kind of ready for a ramp for quite a long time now. It's probably 12 months. That is a bit unreasonable. Of course, we're smart about it. Not everything is staffed. Not everything is activated. Eventually, it's still in there. What is that? Is it all that you're talking about, the ramp? Yeah. A couple of the product lines. I mean, again, if you look at this picture on the left side, we just look at the semiconductor industry. It's almost every key account.
I mean, there's 20-30 big companies globally. All our customers, and each one has its own little story, but we had now more tangible, longer-term upwards trends. It's not fully broad yet. I am talking about signals, not about the ramp having started. I think it's a good recovery from the bottom, especially the ones that really crashed down like memory. Is it really ramping? It's always been the question. In full year for Q4, I think I remember that at the time we said we see some very narrow ramp around AI. HBM and HBC, high-performance compute and high-bandwidth memory. Since then, it has a little bit broadened. A bit more in other places. That is what gives us this moderately optimistic view.
Since then, what came as a counter-reaction was Liberation Day and the trade tension and the extreme decoupling of China and the US. That is why we feel, and we read the same reports as yours, or we even read your reports, just to calibrate here. We believe that is why maybe it is a bit slower than they maybe expected. I think we were really in a good place just before Liberation Day. There would be only optimism, I think. Now I just do not know. I mean, again, our realistic scenario, as I mentioned in the Q1 earnings call, is we have about three different scenarios, but the realistic scenario is a bit of a decoupling of US and China in any case. Maybe it could be small, but in most of the places, there is going to be some kind of a digestible deal.
That's the current assumption, I think. But it will slow down the market, right? Because everybody slows down investments. There's no magic science to this. I mean, I'm very interested also in your feedback, honestly. We are also constantly reading reports and looking at the crystal ball and trying to find out what happens next. No further questions online? No further questions. It's good. I mean, you guys can still ask questions. I'll go with you on the tour. We take a small break for everybody to air, bio break. If you want to drink or eat something, then we'll go on to the tour. For the folks online, I will go and stop at this point. I want to thank you very much for your interest, everybody, for joining. Fantastic crowd, almost 20 people online and on-premise. That's really cool.
Thanks for being such interested investors and analysts. Talk to you soon again in any other location. Have a wonderful day.
This is a very good opportunity.
Thank you.
Bye, everyone. Thank you.
Bye.