All right. Good morning, everyone. Welcome to the 2022 KLA Investor Day. It's great to have everyone here in person. Thank you so much. Also, it's great to have all the folks on the webcast, including KLA employees listening in. It's because of their collaboration, innovation, and execution that this is all possible, and that's even more true for the working team here at KLA that helped pull together this Investor Day. It really does take a village, and we appreciate all the effort. Thank you very much. We couldn't have done it without you. I'm gonna kick things off by going through our agenda. We're gonna start the day with our CEO, Rick Wallace, who's gonna talk about sustainable outperformance. If there are two words that really capture the theme for the day, it would be sustainable outperformance. You're gonna hear about that from Rick.
You're gonna hear him articulate it further, and you're gonna see throughout the rest of the presentations how we build upon that. Rick's gonna pass the stage to Ahmad Khan, who's our President of the Semi PC business. Semi PC is our largest business. It's also our most differentiated and highest margin business, and that's a lot to say when you have three businesses that are all really good margin businesses. We're really happy to have Ahmad go through the outperformance roadmap, talking more specifically and in detail around how we're gonna drive the performance that we're gonna talk to you about today. We're then gonna take a short 10-minute break. We'll come back with the next two businesses. The next two KLA businesses, led by EPC. While SPC might be our biggest business, EPC is our newest business.
It was formed in 2020, and many of you know it is the combination of Orbotech, SPTS, and a prior acquisition that we did named ICOS. Oreste today will talk more about the exciting growth drivers in EPC, more specifically automotive as well as advanced packaging. He'll then cede the stage to Brian Lorig, who heads our services business. Brian will talk about the durable revenue stream that we have within services, how we run a best-in-class service business, and some new updated long-term targets as well, which you'll hear about as well throughout the day. Then, of course, we'll take another break for 10 minutes, and Bren Higgins will come back to help us conclude before Q&A, where he'll cover the long-term target model.
Many of you have probably seen, and if you haven't, you should take a look at the press release this morning at 6:00 A.M. We announced a very large capital return plan. In addition to doing that, we also spoke about the 2026 target model, which you'll hear throughout the day. It'll be building out, and then Bren will help tie it all together. We'll then have Rick back up for some short closing remarks before we go into an extended Q&A session. We did try to design today to have more engagement around Q&A, extended Q&A sessions. Of course, for those of you that are here in person, when we're done, we'll conclude the webcast around 12:15 P.M. or so, 12:30 P.M., and we will go next door, just right outside to have an executive lunch.
Before we kick things off, though, we're gonna need a disclaimer. This is our disclaimer. It's very comprehensive. It's been updated. I will say that we will be making forward-looking statements that are subject to the safe harbor provisions. This disclaimer will be made available at the conclusion of today's presentations, as well as all the presentations that you see. We will post all of them as PDF on our IR website if you wanna refer back to anything or if you feel like not taking any notes, you'll be able to refer to everything you saw today. Last but not least, we also mentioned in our press release that we were reaffirming our guidance for this June quarter. We do have two weeks in the quarter left.
This was the original guidance we provided back at the end of April, and the quarter has progressed so far according to plan. There have been no changes made. With that, it is my honor to invite to the stage our CEO, Rick Wallace. Rick?
Thanks, Kevin. Well, good morning, and thank you all for taking the time. We know you have a lot of choices with how you spend your time, and really appreciate you spending it with us today. Hopefully, it'll be informative. There are a couple things we're gonna do. One is we'll talk about the overall semiconductor industry, which many of you know a lot about. For today, what we're gonna do is map some of the developments in the industry to the impact that it has on KLA and how KLA should perform based on those changes. We'll do some bridging of what's going on in the overall industry, plus give you an update on how we're doing across the businesses, as Kevin mentioned. Then, of course, we're gonna close with the service.
Service will be last, and then Bren will give the financials. We did preview some of that in the release that we had, including the share repurchase and the increase in the dividend, reaffirming our confidence in our business as we go forward. As all of you know, the semiconductor industry is now ubiquitous in a number of industries and is driving change in a way that we really have never seen. We've never seen semiconductors be so pervasive. The reason that's important from an investment perspective is, unlike in the past, we have an industry that's broader now, which means it is less susceptible to the ups and downs of any single particular market. For example, it's blended between consumer and enterprise. There are strategic investments as well as competitive investments that are going on, as well as capacity-based investments.
For KLA, we're gonna share our innovation process. I'm gonna spend a little more time today talking about the operating model, how we run our businesses, because I hope that that will help you understand how we've gotten to where we are and why we're confident about our position going forward. We're also gonna talk more about the overall business model resiliency, the fact that we have absorbed Orbotech into KLA, and yet returned a similar financial performance that we had prior to the acquisition. That demonstrates some of the capability that we've had as a company in terms of driving our success. What we knew in 2019, we had our Investor Day in September of 2019, and there were a number of factors we were aware of at that time, and then there's some things that, of course, have happened since.
Number one, we knew that there was a broad base of applications for semiconductors in 2019. It was very clear that there were a number of drivers, a number of industries that were interested. We saw the beginning of EUV in 2019, so we knew that Moore's Law was gonna resume. At least scaling would resume. We'll talk about Moore's Law, actually, but the scaling was coming back, and we were at the beginning of this digital transformation. That's what we anticipated. That was what the underlying basis was for our plans in 2019. We also had early indication that process intensity, process control intensity was growing. We saw it in 2019, and that was the part of the basis of our belief in our plan that we laid out at that time for 2023.
We also codified and communicated the operating model specifically in conjunction with the Orbotech acquisition. There were some questions, how could a company like KLA that hadn't been outside our core apply our operating model to other businesses? We made the case then, and we're gonna show our work now on how we did that. A couple things that changed and were very fortunate, I would say, surprises. Number one, the number of design starts in advanced nodes has reached an unprecedented level, and we'll share some of that data, and the implications of that are pretty significant. We have seen that scaling has returned and the economics of advanced nodes has returned to the industry in a way that is driving new investment by our customer. That's not an accident.
What's happening in terms of the large number of investments at the advanced node is based on the economics, and we'll share some of that data. Of course, we went through a pandemic, which really, at the beginning of it, no one was really sure what the implications were gonna be. Clearly, what happened is it accelerated digital transformation across many industries. It also highlighted concerns about supply chain resiliency, which gets to our last point, is regionalization of semiconductors became a factor and a topic around the world in terms of as people looked at supply chain shortages. We had politicians, for example, that didn't know what semiconductors were, but knew they were impacting auto production.
We had people around the world saying, "Perhaps this trend toward moving all the semiconductor manufacturing to Asia, we need to rethink some of that." There's been a fair amount of work going on, and I'll share some of that and what the implications are from an investment standpoint for the industry. This data is well communicated now. There are several estimates out there. I've confirmed this as recently as a few days ago with customers and also analysts. The expectation that the semiconductor industry gets to about $1 trillion by 2030, some estimates higher, some lower, relatively independent over that longer term of some of the macro concerns that might be happening now, for example, is the industry will grow at a higher rate because, as I mentioned earlier, there's so many additional drivers to semiconductors.
Those additional drivers, in many ways, make the industry more resilient. Because in the past, if it was PC or mobile, you had a cycle where there was an inventory buildup, there was investment, and then there was absorption and digestion of that, and that was part of what contributed to the cycles. What we're seeing now is multiple investments and multiple vectors, some that were relatively surprising. We'd heard about it for a long time, but we just started seeing it happen. We're gonna make the case that this increase will drive capital intensity and process control, and we believe that long term, this is a. We're confident of this dynamic because there's so many end demand drivers. One example is 5G. We're relatively early in 5G.
5G kept being invested in during the pandemic, and we see continued investment in 5G, and that has many years to run. A big part of what's driven this, especially in logic and foundry, is the resumption of economic scaling. What this shows is the period that you backed from 2004 to 2010, we saw this 22% decrease in the cost of a transistor annually. If you go back in time, by the way, Moore's Law was 30%, right? This is already a slowing of what was considered the traditional Moore's Law. We hit a flat point in the curve because EUV kept being delayed, and so there was multi-patterning.
The economics associated with the next generation node were not very compelling for our customers because they looked at that and they said, "It's just not worth it that much," and so there were very few devices driving the next node. We're back to scaling at this 23% rate. The other thing, and Bren will talk to this, is this turns out to be the optimal rate of scaling from an investment standpoint if you're a company like KLA. Because you can continue to invest, and those investments last longer because these nodes last longer, and yet you continue to see the next generation. If you get nothing else from my presentation, it should be this chart. This is the number of designs on advanced nodes.
As we look at from 28 to 16 to 10 on a relative basis, the number of designs, unique designs, different devices, and the volume associated with it. Why is this so counterintuitive? Well, for years, the belief was that as we went to the next generation of technology, fewer people could afford it. The cost of designs was going up. You needed to have high runners. Ultimately, the view was we were going to have fewer and fewer players at the advanced node, and that was going to be the factor that constrained the industry. What happened? Why is 7 nanometer now bigger in number of designs than 28? Well, a couple of things happened. One, scaling resumed, and so we have a cost advantage of going to the next node.
The other thing that happened is the foundry ecosystem figured out a way to make it much more affordable to design advanced semiconductors. Now it turns out, if you dig into the details of this, the first design is still expensive. What they've done is brought back the cost of the later design. If you're a year after the initial 7 nanometer or 2 years after, the costs come down significantly because the foundry ecosystem has created massive libraries, AI-driven design tools, SaaS tools to allow people to develop these. What that has done is two things. One, it's driven this number of designs up, but it's also taken out the reuse issue that we used to deal with in the industry, because when one design was done, customers would migrate a large percent of the tools to the next design. Now those are being backfilled.
The other thing is there's a whole new C-class of designers. It first occurred to me years ago when we saw Amazon started designing their own chips, that suddenly Amazon has several 7-nanometer designs, for example, and they're not alone. Other hyperscalers are doing the same thing. Part of that is the economics associated with design. If, let's say, you're running a large data center and you're paying the electricity bill, the reduction in the cost associated with reduction in the use of energy on a new class of semiconductors pays for itself very quickly. It's one thing if I lower the energy usage of my cell phone and try to sell it to you. I can't make a lot of money on that. If it's my own P&L that I'm impacting, that's gonna drive a lot of this.
Now we've seen more and more companies, and I think what you're gonna see going forward is more and more players are going to be designing their own silicon as they recognize the importance of having their own capability. Hyperscalers are already doing it. There's one car company that's been doing it. I'm quite confident there will be more, right? Because as they move into EV, there's gonna be more. It's affordable if you're not on the very end. If you're an investor and you're worried about the cycles associated with the nodes, you have a longer cycle, you have a broader number of users. The economic dynamic has changed because these are self-funding. If I'm paying my energy bill on this massive data center and I can get payback in months, it drives that competition.
The other thing that is interesting about this is if you're competing for customers, you have to keep innovating too. If you go on a hyperscaler like AWS, you can pick from different processors for your compute. It drives them to also continue to iterate this design. This is fundamentally one of the biggest changes that's happened in our industry is this inversion. The other thing, if you're in the process control business, this is wonderful because what it means is we're going higher and higher mix. If you go into fabs that have very high mix, what's your concern if you're a foundry running high mix, and I'm running a lot for a customer? I don't want to start too much silicon, and I don't want to start too little silicon.
I want to start the right amount, which means that's why process control intensity goes up because I want to make sure that I'm efficient in output so that I can match all these customers. You have more and more customers, you have more and more designs, and you're at the beginning of this competition among them to drive more and more capability so that this continues. When I talk to customers, they believe that 5 nanometer will look very much like this, as will 3. I think we're on the road of this, and we're talking about scaling continuing through 2025 before we even get to High-NA EUV, which should extend it through the rest of the 2020s. Big deal for the industry, big deal for us. I mentioned regionalization. The chart on here is where semiconductor production has been.
Those of us in the industry knew this was happening. This wasn't a surprise. Just look at our passports over the last several years, right? I used to joke there was no silicon left in Silicon Valley, and I had to have a passport to see 80% of our customers, right? That's still the case. Now because of what's happened in terms of supply chain disruption, there are 60 programs around the world that try to replant manufacturing, have capacity, whether it's in Ohio or Arizona, the work that's going on in Europe. The goal from the CHIPS Act is to get to 50/50 by 2031. If you do the math, that would be very hard to do. That would be very hard to do, right? We all acknowledge that'd be hard to do.
If I'm an investor and I want to think about intent of investment, and I'm having these regional efforts to normalize investment to make sure that I have some capacity domestic in the US and in Europe, it means a more sustained, less macroeconomic investment cycle associated with these strategic investments. We're already seeing commitments, and you all know once government approves money, they actually spend it a little bit regardless of the economy, right? They'll make those investments because they believe it's long-term strategic. We're not saying exactly how this will play out, but if you listen to some of our customers, they'll say, "If I don't get the incentive," say, from the CHIPS Act, "I'm gonna do it in Europe.
If I don't get... It's they're going to do this additional investment regionally, and that's going to be an ongoing factor for us. Again, if you think about the long term, ttat's pretty significant. Will we get to 50/50? Obviously, that's a challenge, but I think what we will see is a reversal of that curve of the continuing movement. I think we're gonna see that pretty clearly. Back to KLA. You hopefully all know KLA. Just a couple things that have changed and to point out. One, Kevin mentioned this a little bit, we're actually over 14,000 now. That was the end of 2021. We're at 14,000 employees on plan to hire again to support our growth this year. We've done a good job of hiring.
We have found that we're a pretty compelling story for employees, and we're also a compelling story for advanced degrees, and we have 65% of our professional roles are PhDs and master's. Why? Because some of the technology we work on is world-leading technology. Our challenge has always been to tell our story to talent. Once they understand it, we can get them because we have some very compelling technologies that they work on. I'll share that in a second. Who you're gonna hear from today is the top team. This is one of those rare occasions where the 5 of us were the same ones that made the commitment in 2019, and we're back to update you.
The rest of the KLA team is the same, with the exception of MaryBeth Wilkinson, who joined the company in 2020. A lot of history, a lot of experience across this team. You know, personally, I've been with KLA for quite a long time. Before that, I was a customer. For those of you that don't know, I was a customer of KLA's. I ran one of the first reticle inspection tools, a wafer inspection tool. I actually wrote software in a previous life. The quality of my software caused me to move to management. What I had and what many of us have is a long understanding of how this industry works, and that helps us operate in this environment. A lot of what we've seen today is similar to what we've seen with some nuances, which I'll talk about.
There is a push and an increased emphasis on ESG from a KLA perspective. We're excited about that. I know many of you as investors, you have funds that have an ESG mandate. We think KLA is a very compelling story. I'm not gonna go into a lot of detail today. There is detail on our website if you're interested in both in our R&D work but also our environmental work. I'd add as an emphasis, the one thing that we do, we use, Ahmad's tracked this, less than 1% of the energy in a fab, and if we do our jobs right, we increase the productivity of that fab. If we do our jobs right, we contribute greatly to increased efficiency from an environmental standpoint. On the governance, we're pretty focused on the one thing you're gonna hear a lot today, management compensation.
Our long-term equity, everybody that's gonna present, is tied to free cash flow. That's the metric that we use, relative free cash flow. When you see Bren talk about it, when we talk about it's not an accident. We didn't get told that. We believe that that's a metric, that free cash flow is a thing that we ought to be measured on as a relative basis to revenue, and that that informs a lot of our decision-making about profitable growth. I'm gonna talk a bit, little bit about the strategic planning process, and how it fits into the operating model, and I'm gonna do this because it sets up the other presentations a bit. Every company does a version of this, I would think, everybody you talk to. What's unique about KLA's? There's a couple things.
One, when we get to our financial targets and objectives, penetration, share, and adoption, these are things that I believe are unique to KLA. We look at how do we penetrate new markets, how much adoption is happening, and what market share do we have. Those are things that we drive. We do that because process control, to many degrees, is a choice by customers. If what you're doing is ramping capacity, then what you're doing is just calculating throughput. But for KLA, we have to make the case that there's return on that investment, and we're gonna show you how we do that. We also do this across the company quite a bit. The other thing that we have found when we looked at other companies that we do a particularly good job of is this red team and green team analysis.
I confess, this is in our roots. Ken Levy drove this process at the very beginning of red team analysis, of understanding what the competition was doing, to the point where we mock up competitive tools, we build competitive capability in labs, we simulate competitive tools. When we speak to you about our performance relative to others, it's not speculation. It's a process that we go through, and more often than right, history has proven that we were correct. We do red teams. We also do green teams. It's what does the market need? Not do what do customers say they want, because that's a very different thing. It's what does the market actually need? Ahmad will show, and we'll all tell you that ultimately, in this industry, what wins is economics. Economics is what wins for our customers.
The most cost-effective process control solution is what wins. They might say they want something else, but we're gonna show you more case studies that that's not what happened, because if they can solve their problems economically. The good thing, kind of for us, is our customers are big enough that they end up doing those analysis, and we feel like our solutions have demonstrated that we win every time when it comes down to overall economics. We look at the operating model. We drive this across all our businesses, consistent strategy and execution, managing by metrics, financial rigor, and discipline. I would say that we are well-schooled in this. Bren and I sit through every presentation, and we drive on this, and some of what we're looking for is these metrics for each of our divisions.
There are several in Semi PC, there are several in GSS, and in EPC, we added five. What's important about this is every division, and there are more product lines than divisions, these general managers are measured on four things. What their market share is in their segment, and we want to plan to get to number one by a significant amount in every segment. What is their gross margin percentage? We view gross margin as a differentiator. That's how we measure it. When you see KLA's gross margin, it is not an accident. It's a result of this process and our discipline around value that we create and pricing that we implement. Our operating margin, we view as a measure of how efficient we are and how much we drop to the bottom line.
Lastly, to make sure we have the talent and that we're attracting talent, developing talent, and continuing to execute. We have an advantage of running these multiple divisions in this process because we do best practices. We don't tell everybody how to get their answers, but what we do is we leverage their capability, and so we can port that to the other divisions. When we added on EPC, we had Oreste Donzella, who'd been with the KLA, he'll introduce himself, for a very long time running that with these processes. We put those divisions through these processes, and that's part of why we've seen the success that we have, and we are confident going forward. I want to talk about collaboration, and this is something, innovation and execution, that the rest of the presentations will talk about.
This has always been our model for as long as I've been with the company, is work closely with customers to understand their needs, not their wants, their needs. What is technology gonna go to? What's it gonna look like? What capability are they going to need? Innovation, investment in R&D is necessary but not sufficient for innovation. You also have to have an innovative process, and I'll talk some to that. We also blend our experienced employees with new employees when we have new skill sets. Our retention rate, our promotion rate inside the company is very high. 85% of our vice presidents were promoted from within. Every one of the management team you're gonna hear from today started as an individual contributor in KLA. We've all grown up in the process, and we drive heavily on this innovation. Lastly, execution.
Some people ask, "Why did you navigate the challenges of supply chain?" Not by accident. We do early engagement with operations, supply chain, and service personnel in our product life cycle. We don't make commitments to production unless we're confident that we have those things figured out. The service business, certainly, it has to be serviceable. These are changes that we made over time as we continued to evolve. The other thing is this hybrid approach for R&D, and I think this is again, unique, and we can afford this where some of our competitors cannot because of the size of our process control business, many of you know, is 4 times our nearest peer. They might be bigger companies, but they're not bigger in process control. We're significantly larger. We have a dedicated, highly differentiated team that works on core technologies and works with universities.
We also have engineering inside of each of these businesses that I showed you. They often work with pathfinding and understanding customer needs. Marketing, we joke, is too important to be left to marketing, so the engineers also do that. They engage with what is the roadmap gonna look like, what are the capabilities, what are the needs? We have a lot of process understanding in that group. Then we bring this together, and then we drive differentiated solutions, and that's part of the process at KLA that gives us differentiation. When I joined the company, and when I was a customer, if you looked at an inspection tool, every subsystem was from a catalog. Every subsystem was off the shelf. What KLA was an integrator. That's all we did. We integrated. I mean, not all we did. That's what we did.
At the time, that was innovative, but that's not where we are now. Every one of our subsystems now has deep, deep technical capability. Every one of our resources has that, and the reason we have that is we wanna have. In some cases, we're the only people in the world that can get the image sensor that we want. Even though we partner with a sensor company to provide the sensor, we provide all the amplification around it, all the circuitry, all the differentiated that gives us our unique performance. When we talk about laser illumination, some of the products Ahmad will talk, we make the lasers at this point 'cause we're the only ones that can make them to the specs that we need. When we talk about all these technologies, whether it's AI, algorithms, machine learning. Now, it's funny, some people talk about AI.
We have a very long history of algorithm development at the company. In fact, you would argue that was one of the first areas we differentiated on. Machine learning and AI was an extension, and we continue to invest heavily in that, and that has shown up in a number of our products. E-beam and X-ray, Ahmad will talk to this. We have very good technology and capability, and I go back to the strategic planning process. We look at where e-beam is appropriate in terms of whether it's guiding our inspectors, helping out with our overlay or in review. Or in one case, we do think there's a use for e-beam inspection, and it happens to be in reticle, and part of that is the limited area that you have to inspect to do reticle inspection and the performance that you get.
We think e-beam inspection is appropriate for reticle inspection. We've looked. I mean, let's be clear, KLA pioneered e-beam inspection. We had the first capability, and the issue was always the same, is you couldn't get the coverage you needed in order to be able to support high volume manufacturing. So it always stayed as an engineering tool. In reticle, it's different, and we'll talk to that. I'm gonna give a quick preview of what you're gonna hear from Ahmad, Oreste, and Brian. This is the Semi PC process control we gain share. We feel good about where we are in market share. You can see that over time, there's not been a lot of movement in share. In spite of many questions you've asked about how this is about to change over the years, some new inflection, some new capability, we continue.
I go back to the process that we have. I think we do a really good job of understanding the market requirements and what the market needs in implementing. I think we do and also a good job of not being seduced by new technology, which sounds great, but the economics of it aren't what our customers end up employing. A lot of times it is not what our customers will tell us. A model will tell you a story today about customers very confident that they weren't going to use Gen 5, and now very confident that we should be shipping them all of our Gen 5. So this is not an accident. We've got a great story in Semi PC, and we're going to talk about growth that we think will continue to drive us. Oreste will talk about EPC.
feel really good about where we are. We have grown the revenue as we expected, but we've actually, at this point of the operating model, we've driven more bottom line growth than top line growth. That's something a little bit we're dependent on the market and the top line, and the bottom line was a bunch of number of actions that we took. This exposes us to a couple of very important markets, obviously automotive and advanced packaging. Originally, the original acquisition of ICOS was based on our belief that eventually a lot of the value in the semiconductor would continue to have in the front end, but back end would become more and more critical. In the time that we've developed EPC, Oreste will talk to this, we found that's coming true.
Most of our major front-end customers are heavily investing in the back end now, and they're very excited to have KLA as a partner because they know they can rely on us and hold us accountable. Services business has been an amazing story. One of the things we did a few years ago, I think we were the first company to elevate service business to CEO staff with Brian Lorig in that role, and we've had tremendous success as a result. It's viewed as a critical business for our customers and for KLA. Brian's got some great data. My favorite, and I know I'm stealing his thunder, $1 billion in revenue in 40 years, $2 billion in 4. Pretty amazing growth. A lot of that, and we'll talk about the growth rate as we go forward for services.
Very good business for us and continues to be very strong. When you add it all up, this is the plan we have. We haven't announced 2022, so this is consensus estimates that we get to about $10 billion in 2022 this year. Bren will talk to some of what our expectations around the market, and I know there's a lot of noise in the market, but right now we feel good about how things have lined up for us with all these drivers that we're seeing in this current era, and I think we're well-positioned as we go forward. I mentioned free cash flow. Again, not an accident. It is a focus of our.
It's part of what informed our decision about the dividend and also our decision about the buyback is our confidence in free cash flow, and you can see it trending there. Bren, of course, will share some more data with you on that. Lastly, we do measure ourselves on differentiation in gross margin, operating margin. We think this is a result of intentionality on our part, and I can give you examples of companies we have acquired that would have been on that blue line when we started and now are on the purple line after we worked through both how they run their business, how they price, and how they execute. We think this is a core competency of the company is.
The other thing, and Bren will talk to this, is in this we absorbed Orbotech, which many of you were concerned was dilutive, and it was, but our view is we can make it better, and we're doing that, and we think there's more room to go. Let me close out with a couple slides. One, just where we were and where we are now. What you're gonna hear is a plan for 9%-11% revenue growth through 2026. Ahmad's gonna sign up to 10%-11%, which will be lower than the growth he's had in the last three years, but still outstanding as we go forward. You're gonna see EPC 11%-12% revenue growth and services up based on the success that we've had and how well we're positioned in the market.
Many of you ask about capital returns, and you saw the actions today. Our ongoing view is we'll return those, we'll increase that target of greater than 70% to greater than 85%. This is it, the get off slide. $14 billion, $38 a share. That's just math. Right? You take those prior ones, you look at the expected growth for the industry, you look at our participation in market share, our expectations around process control intensity, our growth in EPC, our growth in services. We get to $14 billion plus or minus in 2026 and $38 based on that. Of course, we'll share all the assumptions, and Bren will go through that in his presentation. To Kevin's point, sustainable outperformance driven by our operating model.
Our target model is powered by what we've been driving for years and of course, a strong focus on shareholder returns. We already have a focus on cash generation, and we think the best use of cash, of course, is to invest in our business. We do that, and then it's to return the cash that we don't use for that or M&A to our shareholders. I've touched on all these. Really, again, glad that you're here and hope you enjoy the deep dives that we're about to start. With that, very excited to announce and introduce Ahmad Khan to walk through Semiconductor Process Control. Thank you.
All right. Thank you, Rick. Hello, everyone. Good morning. My name is Ahmad Khan. I'm head of Semiconductor Process Control. I will give you two things in the presentation. The first is how did we outperform versus the 2019 investor day plan. Second is the inflections that are happening in the marketplace, some of those were articulated by Rick, which deliver outperformance between the 2022-2026 timeframe. All right, four key messages. First one, we wake up every day to ensure that we are enabling semiconductor industry with designs and everything else that our customers are trying to do. We do this by making differentiated products for metrology inspection and software. Differentiated is critically important because it really determines that we are actually adding the core value that our customers need.
You will see examples of differentiated products across my presentation. Second one, KLA has always done really well when customers spend money in research and development, but I will show you how we are scaling as customers do HVM. This is changing because R&D intensity is increasing, EUV came in, and then many other inflections which I'll talk about with transistor and memory and all aspects of that. That is scaling R&D intensity and complexity. At the same time, HVM, because process margins are reducing, is also increasing process control significantly, and this is a core area of messaging today. Third, customer engagement, this is where the magic happens. We work very, very closely with our customers to ensure that we understand their mission-critical needs and then develop systems well ahead of their needs. Rick spoke about our strategic planning process.
It is really, really important process by which we determine what are our customers' needs and when do they need them, when are the inflections coming into the marketplace, and are we going to be ready for them. Fourth, KLA operating model. Critically important to deliver sustainable outperformance every single year. Let's look at Semi PC in numbers. $5.44 billion in revenue. 19% growth from 2018 to 2021. This was an outperformance because WFE in the same period grew 15.8%. 65% gross margin and growing. This includes service. 54.4% market share versus the nearest competitor, 4x versus the nearest competitor.
Market share and gross margin are two important metrics by which I determine if we're doing a good job for our customers because they are reinvesting back in the business by giving us more share and appropriately paying for the systems because that allows us to invest back in R&D. You know, we are a metrology inspection company. You can see 5.4, 54. It's. I was thinking this is by accident. Number one in 7 out of 9. We serve all of our customers. We serve chip manufacturers, OEMs, wafer manufacturing, reticle, packaging, and now recently, also in materials through an acquisition that we just completed. 65% is wafer inspection, 35% is patterning. Okay, let's talk about this outperformance between 2018-2021. Before I go there, I will share with you a few stories.
I was here in September 2019, I think that's when the last Investor Day was, and at that time, we had only shipped 20 Gen 5s. While a few of our customers were really happy with Gen 5's performance, it was not really used by all of our customers as a mission-critical tool. We worked very closely with our customers to understand what were their critical needs and where are the areas where Gen 5 needs to improve. With about a year's worth of development, we improved Gen 5 and released a new version. I'm happy to say that almost every single customer that we engage with is now utilizing Gen 5 as a mission-critical tool. The second part was I introduced a new concept called EUV Print Check.
EUV was just getting implemented in the marketplace, and it was rising up in the number of layers. We saw that there was a need for an application for reticle requal in wafer fabs. We developed this new technology, and I'll show you later on which uses Gen 5 to determine only the defects that are coming from reticle to improve the cost of the inspection. With both of these technologies we implemented on Gen 5, we've gone from 20 systems in September 2019 to 100 by end of 2022. Pretty big change. The same goes for automotive. Our strategic plan process told us that automotive quality and reliability is going to become very important about 4 to 5 years ago. Automotive customers traditionally purchase systems from our KPC group in Brian Lorig's organization, which is really used in refurbished systems.
We felt that this is going to change, and we started getting ready for it. Oreste is gonna talk about a case study that he will show which is a combination of Semi PC and EPC of how we are seeing growth in automotive, but we see that today. I recently had a meeting with a customer which is dealing with legacy nodes and also automotive, and his message was, "I just need new tools." We developed a new tool called C205, which is Gen4 based for the automotive market, and we have seen very large growth with that system. All of these stories are the reason why we outperformed. 18-21, 19%. Industry was 15.8%. Semi PC outperformed by about 3.2%. We gained share during the same period.
We improved gross margin during the same period, really talking about customer collaboration and the customers valuing what we're developing. At this outperformance really was driven by not only R&D inflections and our growth in R&D, but really the implementation of process control in HVM. I'll share some data with you to explain that. Share is a very important metric, 54.4%. Last time when I was here for the last investor day, we made a commitment that we believe we can gain 0.5% share year on year. This would have resulted a share of 53.5% in 2023. We are happy to inform that we're at 54.4% already. This has been very good. Main gains were in optical inspection and optical metrology. Again, our plan is very, very simple.
Follow the KLA operating model, follow our strat plan process, and build unique systems that solve mission-critical problems for our customers. We build those unique systems that are differentiated, that solve mission-critical problems for our customers. Our customers come back, give us share, and give us gross margin. That allows us to reinvest back in the business. Okay, that gives you a story of 2019 Investor Day. How did we do against that? Now let's talk about all the things that are happening in the marketplace that are gonna drive sustained outperformance for Semi PC moving forward. I'll first talk about what's happening in customers' research and development segment, R&D. All of this is dependent on inflections. By inflections I mean is they're going to do something about the process that's going to change their transistor or something else on the silicon.
When that happens, all the metrology and inspection modes change, and KLA comes in with new systems to solve these problems. In logic, we see transistor completely changing from FinFET to Gate-All-Around. To power these, I believe you can put like 10,000 of these transistors, 10,000 to 15,000, depending on how you calculate the gate length, on the width of a human hair. That's the challenge. You need to find all the defects and characterize these transistors. To power these billions of transistors, you need a new power delivery system that is getting introduced called buried power rail. This is gonna bring wafer-to-wafer packaging in the front end. Memory is scaling significantly, high aspect ratios. We had talked about Axion system, which was a X-ray-based system for NAND flash.
We introduced that, and now today, every R&D customer is using that system to characterize these high aspect ratio structures. Packaging for high-speed data communication, memory, logic, all coming together in a single package, and we are providing inspection and metrology solutions for that. EUV is scaling in logic. Very clear. Its number of layers is increasing significantly, and KLA is participating in that, both in the mask shop and the wafer fab, and I'm gonna talk about that as a case study today. EUV is getting introduced in DRAM. This cycle continues as High-NA EUV comes in. It will change all the metrology inspection modes in every location where High-NA EUV gets implemented. We're using our strategic planning process to guide ourselves with all these inflections and making investments now to be ready. Let's drill down into one of these inflections, which is Gate-All-Around.
On the left side, you see FinFET. On the right side, you see Gate-All-Around. You can geometrically see the complexity between FinFET and Gate-All-Around structures by just viewing that structure. We see growth in Gen4 based on Gate-All-Around, and the reason is because Gen4 has a wavelength that allows us to penetrate 3D structures. FinFET is more 2D and then Gate-All-Around is more 3D, so you need to go further deep into the structure. I'll talk about why Gen4 and Gen5 is unique, because we have all the different wavelengths that allow us to tune the system to the need that the customer has. We will see near 50% growth in later for Gen4 as we implement Gate-All-Around.
In order to do that, we are actually releasing a new version of Gen4, and what we are doing is taking all the learnings that we had from Gen5, and we're now moving that into Gen4. We'll re-release a new version ready when Gate-All-Around goes into production. High-end films is also seeing a significant amount of growth. Now I wanna talk about customer collaboration here. I had a meeting with one of the senior VPs of R&D about 4-5 years ago that was planning to go Gate-All-Around anytime soon, hopefully. The question was that will we be able to characterize a Gate-All-Around structure, as I talked about, you know, about 15,000 of these things go on a width of a human hair, with an optical metrology solution. It was very difficult.
As Rick said, we build prototypes all the time. We built a soft X-ray prototype to characterize Gate-All-Around for metrology needs. We know these technologies, while are very exciting, they are expensive from a cost perspective. The cost of ownership of a soft X-ray system would be far higher than an optical inspection system or optical metrology system in this case. We built it to make sure that the customer feels that they can count on KLA, and it's a prototype system. With very close collaborations, we understood exactly what was required for the optical metrology system, and it was really innovation in AI and machine learning that enabled us to now characterize all Gate-All-Around layers with optical. We are ready in the future if there is need for soft X-ray. We see huge growth as these inflections happen.
These routes drive our customer spend in research and development and also drives their spend in HBM because of layer increase. Now let's talk a little bit about HBM. We always saw good growth of process control in R&D because you need to really characterize it. From process to process control is very different. In process, you buy a few chambers and you're able to make these structures. But for process control, you have very high intensity in R&D to make sure you can characterize all the different structures. But as customers would ramp the product, go to pilot production in HBM, they wouldn't scale process control at the same rate. We are seeing a change in this. Process margins are getting smaller. The number of designs are increasing, as Rick spoke about.
As you add the number of designs significantly, you never actually reach full maturity in HBM from a yield point of view. For that reason, process control intensity now in HBM is increasing. Reuse is decreasing. Reuse is decreasing also because Rick spoke about design increases in 7 nanometer. That chart showed an average of 3 years, I believe, of intensity after 7 nanometer was introduced. Customers go in and add more process control to get more yield. At the same time, number of designs are increasing. Because your number of designs are increasing, you can dial in your process for 2 or 3 designs, but as tens of designs are coming in, those process chambers are varying. For that reason, process control intensity is increasing. One other thing that is happening is backporting.
I talked about EUV Print Check. EUV Print Check was introduced. If I say today the most advanced node is N, it was introduced at N minus one. Customers looked at Print Check capability and said, "Oh, I can backport this capability now at N minus two because I have a few EUV layers." We have seen backporting of technology in previous nodes because now you're able to increase the yield by a bit more. The ROI of that is very, very high for our customers. That's a lot of explanation, but let's just go into the data and see. This is our top customers, top five customers, logic and memory, their spend during research and development, and their spend during HBM for optical inspection.
You can clearly see in R&D, intensity is increasing, and you can clearly see in HBM, the intensity is significantly increasing. R&D intensity is increasing due to complexity. HBM intensity is increasing due to complexity and process margins reducing. Okay, so now I've given you this viewpoint of R&D and HBM. Now I would like to talk about what is our customers dealing with every single day, and what is the technical challenge that they're trying to solve. This is a 5 nanometer transistor. When you scale that to a chip, you have 15 billion of them, and when you scale that to a wafer, you have 8.5 trillion of them. Our job in process control is to characterize that individual transistor, we have to characterize the chip, and we have to find all or as many defects as possible on the full wafer.
If we do any one of these things individually, that doesn't solve the full customer problem, and it's not sufficient. We have to deal with all three of these problems. The only way really to do that is with a broadband system. With broadband system, what I mean is you have to have wavelengths from as long as possible so that you can find contrast of the defects across all the different types of processes. Gen4 and Gen5 combined have wavelengths from 190 nanometers to 260-450 nanometers. That's what's unique to KLA. Of course, you can get an LED source and develop a broadband system, but you won't have the intensity and therefore not the throughput.
You have to do this across all wavelengths, and you have to do it at throughputs that really differentiate. Now, why do we need all these wavelengths? Here is a defect inspected at different wavelengths. You can clearly see that at these wavelengths, this defect does not shine, and the customer will not be able to detect it. At this wavelength, it really shines. That's what we do. Our applications engineers tune our instruments to the best wavelength that is required to find the critical defect. A different defect would shine on a different color. That's our differentiation. Very high power light source that enables to have all the different wavelengths so we can find all the critical defects. That's not sufficient. We have to do it at full wafer scale.
An e-beam inspection system could probably do this die in, like, about 3 days, but our objective is to do all the entire wafer in 1 hour, because a defect that is found here is not the defect that is here. Full wafer coverage is critically important for our customers. Full wafer coverage is critically important for our customers, and our inline systems enable us to do that. How do we do it? This is the Gen4, Gen5 system. We have 40 years of innovation. We have released 5 versions of this, Gen 1, 2, 3, 4, and 5. The magic really begins with the light source. It is a very, very high power light source, 2x brighter than the surface of the sun, and each year, we are incrementing that light source by about 1.3-1.5x.
The optics is calibrated for all the different wavelengths. It's scanner grade optics, but scanners have a single wavelength. We have to calibrate the aberrations for all the wavelengths. The sensors can take equivalent pictures of 2,000 iPhones a second. That's the speed of the sensors. Then after that, all the data that is piping out, which is in gigapixels per second, we process all of that through compute, classic algorithms, physics-based AI, and machine learning. All that is not sufficient because we have to increment these products almost every single year. As I said, customers are changing the defect types. We need to be with that at the same time. As Rick talked about, we have a very good collaboration with our CTO organization that enables us to increment these products every single year.
We have ten-year roadmaps on light sources, on stages, on optics, on sensors, on image processing to ensure that we will be ready for all those inflections, and our strategic plan process guides us to be sure that we'll be ready for those inflections. I think the last part is that our customers are extremely motivated to share with us all the information because we're the only company in the world that makes a full broadband system to find all the defects. We don't make a laser tool. We do make a laser tool for throughput, but really the broadband system really covers all of these inflections. Okay, let's look at optical inspection in summary. $5 billion of revenue since the last investor day.
We have shipped more than 400 Gen4 systems by the end of calendar 2022, more than 100 Gen5 systems end of calendar 2022. We have released 10 versions of Gen4 and Gen5 since the last investor day. Like for example, I spoke about that C205 system, specifically focused on automotive segment, so that we are able to now find reliability and critical defect issues in automotive. That's 10 different versions, Gen4 and Gen5. 31% CAGR in 2018 to 2021, far outpacing WFE, far outpacing Semi PC. While we're doing that, we gain share. Okay, now we've talked about optical inspection. There's always a question, is e-beam coming, and it's going to hurt us or not? I prepared this slide. I know the fonts are small.
I don't like small fonts, but I was told that you're comfortable with them, so I have put this slide together. A lot of detail here. All right, there's 3 segments. First one, e-beam review. Second one, e-beam inspection. Third one, e-beam metrology. Okay. E-beam review, that's this part of the chart. It scales with optical inspection. Now, we have this one thing that we constantly watch. 80% of the dollars are going to optical or not, and we've been tracking this metric for the last 15 years. We keep tracking it, how much money is going to optical, how much money is going to e-beam. That's all what this slide is. E-beam review scales with optical inspection. Why does it scale with optical inspection?
Because optical inspection does the inspection, and after that, customers want to bin these defects that we find for a particular chamber or something else, and they go do e-beam review. The more optical inspection systems we sell, the more e-beam review scales. That's the first metric. We look at 80/20, and it's 80/20. It's on this side. Okay, second is EBI, e-beam inspection. I use the Gartner data. This is official Gartner data looking at calendar 2018, 2019, 2020, and 2021, optical inspection growth and EBI growth. You see 80/20, there's some slight growth you see, but in general, you don't see much growth there. Actually, this data is not completely correct because it has e-beam metrology and e-beam inspection mixed in it. KLA decided we always split that internally, and now you're looking at pure EBI.
Pure e-beam inspection, which is really what competes with, well, perceived competition with optical inspection. In that case, the ratio goes to 95/5, 90/10, depending on, you know, how you look at it, and optical inspection is growing. We don't see it. Later on, you're gonna see a bunch of slides on e-beam from me, and then you can ask me why you're developing, and I'll explain that. That's kind of 90/10. Then there's the fourth segment, which is e-beam metrology, and there's some growth in this area, and I'll talk about it. Again, greater than 80% of the dollars go to optical. It's similar to the soft X-ray story. Yeah, you need it, you want it, but then are you gonna scale it in HBM? We don't see that today.
However, KLA is developing an e-beam platform. I talked about eSL10 in my last talk, and we have now shipped eSL10 systems to every customer out there, and we're developing a number of other systems. Why are we doing it? Because we really believe e-beam is complementary technology to optical. It's an assist technology to optical. You can always feed some critical information from e-beam to optical to assist it. That's our vision. We are uniquely positioned to do that because we have the largest optical inspection and optical metrology installed base. We bring e-beam technology and connect with it to improve it further. That's how we see it, and I'll talk about that. Two case studies. Case study number one, Gen5 plus e-beam assist. I introduced this concept in the 2019 investor day.
We spoke about it, and I said, "We believe optical assist is now coming to fruition. We're going to develop a product where we will connect our eDRX1 to Gen5. I'm happy to report we finished the development. We shipped our first system, and this is the result. Here, and again, this is only limited to very, very few critical layers. Very, very few. Most of the layers will not require this, and therefore, the coupling is limited to very few. Our vision is to support all problems for our customers, not just the ones that are on the trailing edge or the leading edge. We want to solve all the customer problems. Here, we found all the critical defects with Gen5, but we have this other area, which is called nuisance, and there's some level of defects in this area.
You can solve this problem by just running Gen5 hotter or slower, or there's many terms for that, and you can find more. The cost of that is not very good, and customers don't like to do that. We have built a proprietary connection between these two systems, and Gen5 shares critically important information about the location of defects to the EDR system. The EDR system goes there and detects the defect and sends it back to Gen5, and they communicate back to each other with a high-speed connection. We have been able to double the amount of defects and reduce the nuisance significantly. This is where we believe optical and e-beam come together. We don't believe that e-beam will come and do a full wafer inspection. There's just no physics that supports it. We see the same in metrology.
On the left side, due to the increase of EUV, we see overlay measurements going up significantly. Let me just briefly explain overlay. You're building a building, you've constructed the floor. Now you're going to build another floor, but not on top of it, on the side, and you bring it on together. All the pipelines have to connect to each other. If not, the waters will leak. So alignment of one layer to the other is critically important. Otherwise, you have defects, you have parametric problems. If this was the latest generation advanced node, and this was one before, this is one before, we see a huge growth in overlay, optical overlay measurements, huge growth because of introduction of EUV and many other factors. Just to give you a perspective, this is 200,000 points per lot that a wafer will go through in its life.
You can now do the scaling. It's significantly high. The reason is because you're packing more and more transistors together, now you have to align everything perfectly. The only way you could do this is measure more, or you could do something else, improve the accuracy of overlay. What we do is we bring e-beam technology that enables us to drop the accuracy significantly. We have done this. This was an SPIE paper published recently with a advanced logic customer, and you can look it up. They presented it together with KLA. E200 is a new technology that we're bringing in. E-beam-based, again, KLA is uniquely positioned to deal with this inflection because we can combine the two. I've now articulated optical inspection growth in e-beam.
Now let's talk about our reinvestment process and which are the important areas we're investing. I'm going to talk about reticle inspection. First, this is the chart that shows it's a result of the differentiation that we bring to the table and the gross margin that enables us to reinvest back in the business, and we continue to invest a lot in Semi PC to deliver the returns. In reticle inspection, we're also increasing spend significantly. There is a huge inflection that is happening in reticle because of EUV coating getting introduced. The pitch at the reticle is changing. Let me describe what a pitch is. A pitch is the distance between a line and a space. Why a line and a space? Because you have a wire and a separation of a wire.
If there's two wires together, it's a short, so you need always a line and a space. As you make wires smaller and thinner and thinner, you can put more wires, and that means you can scale. Reticle only scales when you change EUV wavelength. Otherwise, it doesn't scale. When EUV got introduced, reticle pitch changed. KLA decided to invest significantly in reticle. We are actually reducing our optical inspection spend, increasing our e-beam and actinic inspection up significantly to be ready for the inflection. Let me describe the market first. There's two aspects of the market. First, what I would call is an established market, which it ranges between $800 million to about $1.2 billion today in reticle. This is really about pattern inspection, reticle blank inspection, reticle metrology, print check, and IC qualification.
All of that inclusive is between $800 million and $1.2 billion. KLA participates in that. We have a share of about 70%, in this segment. There's a lot of improvements that we're doing in this segment to make sure that we continue to scale. There's a second market, we call it pathfinding, where there's a lot of information that says this market would be either between $200 million to all the way to $600 million. It's a big range, and there's three reasons why there's a big range. First, next generation pitch requirement. I just described that. What is a pitch? The pitch is the distance between a line and a space, and when you shrink that, and when do you shrink that? Number two, mass pelliculization. We have not seen mass pelliculization to date.
As you know, in memory, there is no plan to do mass pelliculization. In foundry, you can pick your number between 50%-65%, no pellicles. In logic, there seems to be a gravitation towards more pellicles. Number three, High NA EUV, which is really linked to pitch reduction. Okay, so this chart describes the availability of choices our customers have. You can use a 193-nanometer scanner, you can use EUV scanner, you can use EUV double patterning, or you can eventually use High NA EUV. On the y-axis, we have the minimum pitch. 193 really stops at around 80-nanometer pitch. KLA introduced multiple systems to meet that requirement, and we have very high share in the 193 and immersion land. This is the multiple systems.
They meet our requirement. Okay, when EUV was first introduced, the claim was that you can take that 80 nanometers that I talked about and reduce it all the way to 26 nanometers. However, due to a number of reasons, when you drive the scanner very, very hard, and due to infrastructure like resist and other things, the cost of doing that becomes high. Most customers have kind of settled in this range. I don't wanna give the exact numbers due to confidentiality, but this is where most customers have settled. Originally, we thought if the pitch will reduce that much, we need to go develop a high-resolution system.
As usual, you know, product teams, they are very innovative, and our optical inspection team said, "Oh, if the pitch is in this range, I think I can improve this system significantly and get 90% of the layers." This is what's happening in the market today. We are using our 193 system to inspect close to 90% of all reticles in the mask shop. We have been able to implement machine learning and physics-based algorithms that enabled us to scale, plus optics changes in the tool that brought us here. The remaining, which is around 5%-10%, is going to our new multi-column reticle inspector, which we have shipped to multiple customers already. We believe this is a very few number of layers.
They may or may not happen, but if they happen, we have the capability for our customer. If they don't happen, optical inspection systems will serve it. Now, next thing customers will do EUV double patterning. I don't wanna go into the technical reasons, but actually when you go to double patterning, there's a Goldilocks situation on pitch. Actually, at reticle, you relax the pitch. So we think this just moves over. In fact, it might get moved up a little bit, but I don't wanna say that. That's where it might settle. When High NA EUV comes, you can really shrink the pitch. That's why KLA says that for 5 nanometer, 3 nanometer, 2 nanometer, we think our optical inspection 193 system, along with 8xx multi-beam reticle inspection system, is sufficient.
When High NA EUV comes with severe pitch reduction and mass pelliculization, you can bring in a 10x inspection system. We could have developed that earlier, but the complexity of a 10x system is much higher, and that's why we chose this path, and therefore, we are confident that we're ready for the market at the right time. We're giving 10x the appropriate time to do the development and be ready whenever High NA EUV will be available. Our 8xx system is shipped to multiple customers, and we have this one technology called die-to-database. Die-to-database is unique to KLA. Die-to-database ensures. See, on wafer, you can miss a few defects, but they are process-induced defects. But on reticle, you cannot miss any defect because it'll print on every single die on all across the wafer. Die-to-database enables us to find all of them.
8xx system shipped to multiple customers and doing well, and we're developing our 7xx system in line with High NA EUV. I'll show you a quick video to show you the flow of how the inspection will be done. Here is a reticle with a defect going through the scanner, printing defects and printing chips on every single wafer on the die. We don't want that. Our customers don't want that. We wanna detect that defect in the mask shop and not let it come out. Our mask shop optical 193 reticle inspection system catches those defects. But as I said, mask pelliculization is not happening, so this defect can get introduced through transport, through a number of different reasons. We will catch all of them with our optical 640e system.
When this reticle, the remaining of the 5% in the mask shop will be caught with our 8xx high-resolution system, so we are saying buy 640e and buy 8xx. Customers are latching onto that. That's why we have shipped multiple 8xx systems. Now this reticle goes into the wafer fab. It may or may not have pellicle. If it doesn't have pellicle, KLA developed a 670e system that enables you to do pellicle-less reticle inspection. Or if you have reticle or you don't have reticle, you have a second defense. 670 will catch as many as possible. You have a second defense, which is Gen5. Gen5 will find the repeater defect and process-induced defects, but the unique algorithms take all those process-induced defects out, and you only see the repeater defect.
Therefore, this is our solution for the wafer fab, Gen5 plus 670. Wafer fab, established market, $800 million-$1.2 billion, 70% new market, $200 million-$600 million, and we are watching that very carefully and developing these solutions for our customers. Okay, EUV reticle in review. Greater than $2.2 billion in revenue for all EUV related to reticle, right? It includes print check and all the reticle inspection. It doesn't include we have implemented Surfscan systems to determine chemical issues related to EUV and resist issues. That's not in this number. This is just related to reticle and that. Greater than 3,640 640e systems have been shipped and are using. This is 640e systems, only EUV systems, not counting 640 optical systems, EUV systems that are implemented.
When I say 90% of all reticles are inspected through this, yeah, because we have shipped 30 systems and we're continuing to scale. Greater than 50 670e plus Gen5 systems implemented in HVM to do reticle re-qual. If you combine the established market and the new market that I talked about, greater than 85% of all reticles go through KLA, and we have more than 300 die-to-database systems running from the inception of when die-to-database was developed till today. Huge amount of experience to develop this technology. It's not easy to develop. This is our entire roadmap for reticle. One more thing. There's petabytes of data being generated at every fab. Wafer fabs are getting larger and larger.
Metrology and inspection systems are generating a lot of data, and our customers need support to organize this data and make it actionable so they can control process tools. We've been doing process control for many years, but about 5-7 years ago, we decided to ramp this up. With organic growth and a number of acquisitions that we did, that is growing this segment. Our customers are delighted that we're doing it. This is not FDC. This is process control. All the data from defect, all the data from metrology, all the data related to alignment of wafers, all of those together, we're able to put intelligence with algorithms and turn it into actionable results. This is now a $200 million business and growing. With that, this is our commitment, $5.4 billion today, going to $9 billion.
This assumes a WFE intensity of 7%-8% and Semi-PC CAGR 10%-11%. We'll be growing share, driving margin, and technology. Those are the four messages. Again, we wake up every day to make differentiated systems to help our customers. Our relevance in R&D is increasing. Our relevance in HVM is increasing further. We have deep customer collaborations. This is where all the fun happens, and it's all the collaboration that is critically important for us to understand what they need, and our strategic plan process guides us the roadmap, and then the KLA operating model delivers the returns. That's all I have. Thank you.
Thank you, Ahmad. Great. We're gonna go to break now. 10-minute break. We are running about 10 or so minutes behind schedule, but we're still gonna stick with the same time break, 10 minutes. Control room's gonna keep us honest. They're gonna put it up there. We'll do our best to make an announcement at the 5-minute mark before, and you should, when you hear that, you should think about maybe coming back. Thank you. With the second part of the day focusing now on EPC. I'm very happy to have with us, as I mentioned earlier, Oreste Donzella, who is the head of EPC. Please welcome Oreste to the stage. Oreste.
Thank you, Kevin. Happy to be here after 3 years in New York City, live. Today I'm going to share with you the progress versus plan that we made in EPC, starting from the Orbotech acquisition in 2019. Also giving you an idea about how we implemented the operating model that we introduced officially to the investors and to the rest of the community during that particular investor day. My main message here, first of all, starting from operating model. Operating model is essential to the successful integration of the newly acquired company like Orbotech and SPTS came along with Orbotech. We were able to have a much larger participation in the markets that we serve. We were able also to make sure that we are operating efficiently in terms of our spending, in particular in R&D, but not only.
This can be confirmed by the numbers I'm going to share later. I'm going to base the presentation, the three differentiators that Rick mentioned before, collaboration, innovation, and execution. How the operating model is guiding us to leverage these differentiators on a much broader market base that we serve, versus the original semiconductor process control that KLA has been serving for 46+ years. Rick showed this slide. I think this is extremely important. I would like to say the most important point of this slide is the 2 numbers together. Because it's one thing to show revenue growth at 15% CAGR in the last 3 years.
If you remember, we said double-digit in the Investor Day in 2019, we were able to exceed our plan with 15%. Also, this needs to be in correlation with what we are showing in the operating margin CAGR at 40%. These two numbers don't match easily together because sometimes in order to achieve revenue growth, you are sacrificing the R&D and the ability to grow the share, or you are going to sacrifice R&D because you want to gain operating margins, but in that case, you don't deliver the top line revenue numbers that you are projecting. The two numbers together are extremely important. The single source of this is again, the operating model.
It gave us the possibility to overachieve in the top line, but also show an incredible operating margin growth in the last three years. Rick introduced how we are exposed to many different markets. In particular, the most exciting that I would like to share are automotive and, in particular, advanced packaging. This is our journey. We started a long time ago, actually much earlier than the Orbotech integration. We did the acquisition of ICOS. ICOS was our first attempt to enter the packaging arena with a differentiated supplier in a final inspection, what we call component inspection, final package inspection and metrology. We kept ICOS standalone until we found the house, a new home in EPC after we acquired Orbotech and SPTS that came along.
Now the EPC comprises these companies, ICOS, SPTS, and Orbotech. We are in the stage of integration, and eventually we are taking advantage of the operating model at KLA. Numbers. We closed twenty-one close to $1 billion revenue. As I said, the 15% revenue CAGR, including what we estimate this year to close, and the 40% operating margin CAGR, including 2022 estimate. Also, from gross margin, we are seeing a 400 basis points expansion in our gross margin. We have 2,500 employees spread all across the globe. One important point is all these revenue that I'm mentioning in my presentation are system revenue.
All the service numbers associated to these new business and new divisions are reported by Brian in the next presentation. One thing that is unique about EPC, the reason why the title of my speech is EPC Business Expansion, is because, again, after playing and serving the semiconductor front-end market for many, many years, KLA, with the acquisition of Orbotech, SPTS, was able to expand also the reach into the electronic supply chain with new markets like PCB, display, specialty semiconductor, and of course strengthen our position in packaging, not only component, but also wafer level packaging, wafer level assembly and IC substrate. The expansion of KLA in these markets will offer a significant opportunity for growth of the company outside our core business in semiconductor process control. This is represented by this chart.
This chart that some of you may have already seen in other presentations represent the entire electronics ecosystem. From the silicon, from substrate, from reticle, up to the end product, a smartphone or a board in the car or in a data center. You can see these circles are where the three business groups are represented here. If you look at Semi PC that Ahmad presented earlier, so this is in the front-end side of the fabrication of this device. You see EPC overlapping with the semiconductor process control, a little bit in semiconductor front-end, in the packaging, then moving forward, leaning towards the final product, and of course, the services across the board.
In this case, we have seen an opportunity, while creating EPC and developing new collaboration with the customers, not only to collaborate with our direct customers, but with the entire ecosystem, with car manufacturers, with mobile OEM, with data center hyperscalers, companies and so on. We see more and more exposure to the end products by having EPC in our portfolio. Of course, when you look at two of the end products, smartphone and the electronics in the car, you see that EPC really touch all the electronics components of these products. Staying in automotive, I have a few more details later in my presentation. Automotive is really the secular shift.
It's an incredible inflection that we have perceived the change in the last chip shortage in the last couple of years, how electronics, how semiconductor, how boards are becoming absolutely essential to the production of the cars and the vehicles, because many more components are inside the car and the vehicle, and also because of the complexity of the qualification of semiconductor or electronics inside the car. I'll touch base a little bit later in my speech. It's not only diversification in terms of end markets, it's not only that EPC is touching the final products in the previous slides and saying how we are diversified, how our business is driven by very, very different markets. Also inside EPC, we are very diversified.
If you look at the four markets we serve, with our five divisions, specialty semiconductor is mostly driven by automotive and, smartphones. If you look at, for example, components and printed circuit board that are mostly in the PCB and the packaging space, you can see a big exposure, very significant exposure to smartphone and data. When I say data, it means, data center, but also PC and laptop. Final and flat panel display is mostly exposed to consumers. This diversification of the end markets with multiple, end drivers and also diversification internal inside EPC, give us a sense of stability because these markets sometimes are not in sync and give the possibility to play, pretty much any single market that we see is going to grow, more in the next maybe a couple of years.
It's a good place to be because we have this incredible diversification, not only in the market, but also inside EPC division by division. Now, let me switch gears and talk about the core of my presentation. That is how the operating model is really driving EPC to be best in class in the differentiators historically at the foundation of KLA success that are collaboration, innovation, and execution. Let me start from collaboration. There is no surprise when we bought Orbotech and we got SPTS, and also with the previous acquisition of ICOS, that we wanted to expand the presence of these companies across the board in semiconductor also. At that point, we were thinking about packaging is going to be essential to the semiconductor technology roadmap.
The fact that we were seeing at that time, 3-5 years ago, the biggest semiconductor manufacturers investing in packaging gave us an opportunity to open the door of these customers to products from Orbotech, SPTS, and ICOS. I remember 3 years ago I went to talk to a top customer in packaging of one of these top 5 semiconductor manufacturers, and he told me, "You know what? I don't believe you have anything to offer to us." That's it, period. The same person I met last month, and he told me how happy he was that KLA now is a player in packaging with all the product portfolio that I'm going to talk a little bit later.
The fact that we were able to open the door for Orbotech or SPTS solution to these top semiconductor manufacturers in packaging was extremely important. This is just an example. I can have more also in automotive, for example. The possibility to collaborate earlier with our customers and also with their customers. That again is another part of the playbook of KLA. We operate across the electronic ecosystem in a broad way. We don't limit our collaboration to the direct customer. We talk to their customers, and we create this big ecosystem of value that we can provide to every single specific industry we serve.
This collaboration, again, give us the possibility to develop a differentiated product portfolio or portfolio of solutions that is important for us to stay at the top of the supply chain, at the top of the market share in the next few years. Let me start from this example, and this is something that I care personally a lot because this was a pull from the automotive industry. 4 or 5 years ago, as Ahmad said, we got pulled from car manufacturers, and I show a little bit 3 years ago in a separate presentation of automotives, what we were doing at that time. The automotive market was asking KLA to provide ideas or methodologies, best practices to screen chips for reliability in the fab.
Because the biggest issue the automotive market is facing right now, maybe it's only second to chip shortage, but chip shortage is temporary. The reliability problem is not temporary, will stay there forever, is to make sure they can find reliability failures before the chip is mounted on a board and eventually installed in a car. Because whenever the chip fails in a car, it uses a huge liability and a huge recall cost. They said, "Okay, KLA, you are the best in class for process control. It's great that we have the possibility to increase the yield, but let's go further.
Let us know if you can do anything for reliability," because reliability is driven by same defects that can impact the yield, but they go to the next level of being subtle, and sometimes they can escape the functional electrical test and become a huge liability because they fail in the car. Since then, we introduced the so-called IPAT. That is our in-line part coverage testing methodology, machine learning based. The IPAT now is written in some of the critical specs for some of the automotive makers. Not only open a very good growth opportunity for semiconductor process control, like Ahmad reported before, because together with IPAT, we created also a portfolio of products tailored to the automotive market that, Ahmad has been developing in the last three years.
On top of that also, because we are EPC, we would like to talk about SPTS as well. Our specialty semiconductor market grew quite a bit in automotive. Why? Because of the inflection that we are seeing now in electric vehicles, because of silicon carbide. I think most of you have known what silicon carbide is. It's a different type of material than silicon. It's very hard to etch. It's very hard to treat. It's immature. It has a very, very low yield. Perfect for KLA. We can improve the yield, we can improve the quality, and we can even etch with plasma etching technology in SPTS. Now it's all in, and you can see in the chart also huge improvement on the system revenue of SPTS. That is the lighter purple in the chart in the last couple of years in automotive. Packaging. What is happening?
Packaging is absolutely unprecedented. Packaging has been like this low-end industry for many, many years. You wrap a box on it, around the chip, you put in a package, you protect it from environmental damage, and then you ship to the board integrator, and then eventually gets in any product. It's no longer the case. Now, packaging is an enabler of the semiconductor technology roadmap. Moore's Law is resuming, and you saw the presentation from Rick and Ahmad. I'm so happy. I've been in the industry for 30 years. 23 years in KLA, before I was a KLA customer working in semiconductor fab. I see this flourishing of new technology coming in and new innovation in semiconductor make me happy every day. I love it. I keep my passion and my love, I spread to my kids and so on.
I'm a big ambassador of this semiconductor industry. I will be forever. Now we need something more. The semiconductor front end cannot be sufficient to deal with this appetite for data. This is happening right now. When you have data, of course, the speed of communication between memory and processor gotta be simultaneous. Having all this function on a silicon chip is becoming expensive. What is happening? There is disaggregation. What happens is that the packaging, in particular. There are many different packaging types available right now in the market. I'm focusing only on this. This is the high-end, the highest-end packaging type supporting data centers, PC, high-end laptop and so on. There are two things I would like to remark on this slide.
First of all, the logos in the left of this slide. These are the people driving the technology roadmap in packaging right now. These are the same people driving the technology roadmap in the front end. It's a great play for us. We know all these people. We know because KLA has been in front end for 46+ years, and now we have the opportunity to work with the same companies, as I said in my previous remark, and then becoming a relevant player in packaging as well. The second thing I want to say is the complexity. At the same time, the packaging industry is trying to do three things. It's trying to put together a heterogeneous integration type of packaging. What does it mean? More chips with different functions inside one single package.
We have DRAM, we have GPU, we have controller logic, we have CPU and so on the same package. The second thing is the geometry is shrinking. In the next 5 years, the geometry of packaging will shrink much more than the previous 50 years, because now everybody's trying to apply the same lateral scaling that we have seen in front end, of course, relatively, in packaging. The third thing is now this integration is getting more and more vertical, and we see all these 3D packaging. I would like to start a video right now to give you just an idea how this integration works.
For example, you have a memory logic chip, and then you combine together the famous chiplet type of architecture, and then all the integration needs to be done as seamlessly and aggregate as many function as possible in a single package on a piece of substrate, either interposer or organic substrate. Of course, all this process is increasing complexity. Not only complexity, also a lot of risk. Because what happens if one of these chip is going to fail? All the packaging is gone. This is a huge impact to the economics of our customers. Rick said it's all about economics, and packaging is all about economics too. You want to make sure that you have as clean as possible wafer or chip when you integrate these multiple chips on a single package, because otherwise the cost is enormous.
KLA playbook, the page of our playbooks in this case. We went to talk to customers and start collaboration, in particular with the top 5 semiconductor customers that we know very well, and they are the biggest influencers right now in the packaging roadmap. We have 30-plus joint projects. When I say joint projects, it means on-site evaluations with the top 5 semiconductor manufacturers in packaging and IC substrates. To make stuff more complicated, when you try to connect all these chips, generally today is done via bumps, and this is the mechanism that you use to make these multi-chip packaging. In the future, our customers, because the latency problem between chip to chip becomes so important, so essential, they want to do direct bonding.
That means they want to get rid of the bumps for some of these interconnects, and they want to do die-to-die or die-to-wafer or wafer-to-wafer direct bonding connection. This is extremely complicated and it opens up a huge opportunity for KLA because of the complexity of this direct connection. For example, the wafer that is going to welcome all the other chips on top needs to be extremely clean, because even a small seed, a small defect, when you connect the chip on top of the wafer, you can create a void. When you create a void, you will never see the void. You need to make sure that your wafer is spotless. No particles, no defect. The other thing is the wafer needs to be very, very flat, so the geometry needs to be very well-controlled.
Of course, the KLA has tools from front-end that Ahmad has been developing for years, that now we are deploying in the back-end, in the packaging. This is just one example. Of course, we have also process tools from SPTS that have been developed to serve this particular inflection in the market. Okay, let me give a couple of examples. First of all, I want to talk about ICOS. Again, this is our final inspection. This is not to be taken for granted. Doing a final inspection today at the speed that the packaging industry needs, and now the complexity of the package itself is extremely difficult. We are developing new products.
We are developing a tool that can, first of all, handle larger form factor package, because the packages are becoming bigger because they need to integrate many chips. The second thing, we are developing new optics to find the smaller defects, and we are developing new metrology capability to make sure that the package is very flat. It doesn't really warp when you place on a board and eventually in the end product. ICOS, you can see the revenue only for this particular segment, that is what we call 2.5D and 3D, that is pretty much a data segment. The revenue is increasing quite a bit. We expect this growth to follow.
Talking about the process tools, although I believe plasma dicing, although when you think about etch, this is not inspection, this is metrology, it's a process tool. I would say this is a process control tool. Because today, the way the industry cuts wafer, because of course, we don't put wafer on a card, we put a chip in the package in the card, so the wafer needs to be cut or singulated or diced. Today, the industry is using mechanical saw dicing or laser dicing. All these processes are extremely dirty because when you put a saw and you try to cut the wafer, what happens? You are cracking the wafer, you are creating delamination, chips, particles. The same is with laser. The cleanest way to cut the wafer is through plasma, but it's not easy.
What we deliver over in SPTS is the ability to have the entire process integration, not only the etch process step that we develop in SPTS, but also the other process steps to make sure that we have a lab where our customers are sending samples, and we can give them the full solution. We say, "Okay, give me a tape frame wafer and I'll send you pretty much diced already." We have the possibility to show how this plasma dicing process works. Eventually, you see the difference. I hope you can see there, but on the right, these rough die edge that are created with blade dicing, the saw, and on green, super clean edge die that you create with plasma at SPTS. Of course, what happens? The plasma dicing booking is going up quite significantly.
Silicon carbide dicing is going to be an incredible challenge that we are taking right now in SPTS. We would like to provide the automotive makers in the next few years also the ability to do plasma dicing on silicon carbide. The only player in the world that can do is SPTS. Of course, leading into my innovation part, innovation doesn't happen without collaboration. The SPTS plasma dicing tool called Mosaic just received the Queen's Award in the U.K. Of course, SPTS a U.K. organization. I was very honored and humbled to receive this award that the Lord Lieutenant of Wales handed to me. I was explaining how dicing work to the Lord Lieutenant of Wales, that it's kind of interesting challenge there.
I hope I got him on pretty much on what we're trying to do, but he gave us award, so it seems that he understood what we are for. Innovation. Second topic, innovation. We didn't talk much three years ago. We were focusing on how we can organize these companies, how can successfully integrate from an efficient point of view, from cost point of view. We didn't say much what we can do from technology innovation integration. Now we are doing it. I'll give you a couple of examples. First of all, let me talk about the long history of KLA innovation that, of course, Ahmad and Rick talked about, and it's something that makes me proud every single day of my life.
We have a lot of innovations and technologies that we can leverage in EPC that came pretty much free. I'll give you an example. PCB. One reason why we bought Orbotech was because we knew at that time that PCB was going high-end. We knew that PCB eventually is going to deliver the so-called IC substrates that are so essential for the packaging technology roadmap. They're so essential to the semiconductor technology roadmap. We knew also that the industry was behind. Who is behind to find defects, who is behind to do photolithography. What we have done, this is just an example. There are plenty of examples. I'll give you only one.
We create a product coming from the convergence of three different technologies, the Orbotech platform, the stage that is built in the Central Engineering in another organization, the optics designed by our CTO office. We put all this together, and now we are creating a tool that our customers will deploy in the next few years to see defects that the PCB industry has never seen before. This is thanks to the operating model, thanks to how we can deploy technology innovations from KLA into EPC and to new markets. Finally, execution. Of course, I'm going to talk about where we are versus what we presented as ideal thesis of cost synergy. Very happy to say that, at that time we said $50 million, and now we are at $92 million cumulative synergies here.
You see all the buckets there. It came from, of course, public company costs, corporate overhead, but also business rationalization. We were able to use rationalization of our spending to target only the growing segments, the growing opportunities where KLA, and in this case EPC, can play a unique, relevant, differentiated role. That's at the core of KLA. How we target opportunities is paramount when we needed to grow our companies. We don't want to go and shoot for every single opportunity. We want just to serve the applications where we know we are providing a unique role and a unique value. This is from cost point of view, and from, as I said, this is just the last chart from packaging. From revenue point of view, this is the expansion in packages you can see.
We are talking about generating double CAGR versus the previous history at KLA. We expected to have 24% CAGR from 2019, where we bought Orbotech, up to 2026. It goes all the way into wafer level package, IC substrate, but also assembly and test. Finally, this is the chart showing our revenue forecast for 2026 in our revenue plan that Brian is going to talk more in detail later, from $0.95 billion to $1.65 billion. Adding another $700 million to the top line for system-only revenue, and it comes from 290, how the market is going to grow. We expect the aggregation of this market to grow 5-6%.
Then you can see the other buckets where we expect to outperform the market, even in packaging, automotive, and even display. This is my last slide. Thank you very much for the attention. Again, everything I want to pass is operating model, essential, and we validate it. We validate in markets that are not the KLA core markets, and this is quite important. The second thing is how we are applying the differentiators from the KLA playbook in many, many different markets. The differentiators are, of course, collaboration, innovation, and execution. Thank you. Now I'm transitioning the stage to my good friend, Brian, who is going to give us a lot of exciting news about our service business. Thank you very much.
Thank you, Oreste. All right. Good morning. I'm Brian Lorig, responsible for KLA Services. I've been with KLA for 24 years. Great time to be part of KLA, great time to be in the semi industry. You've heard a lot about the operating model today. I'm gonna talk about how we apply that in our services business to enable the growth of a durable revenue stream. The five key messages for today. First, we are a customer-focused organization. Over the last three years, which have been challenging in a lot of ways, our partnership and collaboration with customers have only strengthened, primarily because we have goal alignment with our customers to maximize the value of their fabs.
Next, we have a stable and growing install base, and we are uniquely positioned to support the high complexity, high mix, and low volume nature of that install base, leveraging the KLA Services network to aggregate that install base and ensure that we're meeting customer expectations. Additionally, we have an evolving data and analytics platform, which is moving more and more of our service from reactive to predictive. Again, the operating model institutionalized in everything that we're doing, we're gonna talk about the way that we collaborate, innovate, and execute. Finally, that leads to an improvement in our overall growth rate. We've plussed it up from our 2019 Analyst Day 9%-11% to 12%-14%. Here is service by the numbers.
First, we're nearly a quarter of KLA's overall revenue, 22%. That was $1.8 billion in 2021. It's about a year ahead of our 2019 plan. 99% or nearly all of that $1.8 billion is pure service revenue. What we mean by pure service revenue is this is service contracts and break-fix maintenance. This does not include things like refurbished equipment. It doesn't include things like upgrade or other WFE-centric like revenue. This is just service contract and break-fix maintenance. That speaks to the durability and resilience of this revenue stream as we report it. A few other critical metrics for us. First, how much of our revenue, of that $1.8 billion is coming from, subscription-like service contracts? Greater than 75%.
That's up about 5% from 2019, and more than 95% of that is renewed, also up from 2019. The length of our service contracts are also extending. They're now averaging greater than 3 years. That's an increase of 40% and really speaks to what our customers are. Our customers' desire to provide more visibility to us so that we can better plan to support their needs. Finally, we grow faster than our install base. Our revenue grows faster than our install base, 4.5 times faster, and that's driven by not only new install base growth, but also new revenue from our existing install base, and we'll talk about that in a little bit. Again, go forward, CAGR, 12%-14%.
Among the reasons that we're confident in that increased growth rate is we believe we're well aligned to benefit from industry megatrends. You've heard about these in previous discussions, but these megatrends are not just supported by leading-edge technology. They're also supported by legacy node technology. When we think about our service business, we think about it across these three market types. First, leading-edge development. That makes up about 20% of our business. High volume leading-edge manufacturing, that accounts for 45% of our business. And finally, high volume legacy node manufacturing, which accounts for about 35% of our business.
You saw this chart in Ahmad's presentation, independent of where the customer is in their product life cycle, whether they're introducing new technology in R&D, ramping that technology, and ultimately fanning it out to their high volume leading edge manufacturing or sustaining legacy node manufacturing. We're very well positioned in KLA Services to provide customized offerings to support whatever need that customer has wherever they are in that product life cycle. We talked about. We have a very stable and growing installed base, and we have a very unique signature. It's different than other equipment manufacturers in our space in that it's very high mix, it's high complexity, and it's relatively lower volume. Let's talk about each of these in a little bit more detail. First, on the high mix.
We have more than 16 product families across KLA, and each one of those product families has something like 12-15 different products. We have more than 200 product lines or products rather, in our install base, each serving a very unique customer requirement. That's kind of the power of the KLA portfolio. If you pick one of those products and you talk about the complexity, you got a sense for this in the videos that Oreste and Ahmad showed, but any one of those given products has more than 16,000 parts in the bill of material, more than 300 field replaceable units. Those are across those critical technologies that we've talked about throughout the morning, whether it's light sources, optics, precision control, precision stages, sensors, image compute.
When you look at each one of those particular technologies, we have exclusive relationships with our suppliers. Ahmad talked a little bit about that. Bren's gonna talk more about that, but it's a very long time to development on those key technologies, which leads a lot of times to a sole source situation with that supplier. Ten years in the making, relatively low volume, so KLA is the primary buyer or the only buyer. When you look at that high mix, at any given customer fab, we have a relatively smaller number of systems installed.
Again, because of that, high mix, the ability for our customers to be able to source parts and/or provide training and keep people proficient in training is very difficult to do. That's where the aggregator power of KLA's service network comes into play. We look at that installed base across all the fabs, not just at a single fab in a single country. This all creates very high barriers to entry for our service. In addition to high mix, high complexity, low volume, our tools have relatively long life. If you look at the chart here on the right-hand side, CY 2021, you see the stacked bar totaling close to 50,000 tools. The stacked bars are what we call classes.
You see class of 1990, class of 2000, class of 2010. Those are the years that we introduced the product and began shipping it. Let's just pick one class here, the class of 2000 in light pink. We first introduced those products and started shipping them in the years 2000, 2001, 2002, 2003, 2004. You see over the next 15 years, we started to ramp that install base, ultimately reaching its peak in 2015. Then it's relatively flat as you move across to CY 2021. In fact, more than 95% of the install base is still in existence. That's more than...
Greater than 50% of that install base is more than 18 years old, which really speaks to the enduring value of the products that we ship. It also is long after our customers have depreciated that asset, and so they are demanding that KLA continue to maintain this equipment, which is both a great responsibility but also a great opportunity. Because of that, over the life of a tool, service revenue exceeds initial tool price. Now I'm gonna talk a little bit about how the service business again or leverages the KLA operating model across collaboration, innovation, and execution. For collaboration. We think about our collaboration across three constituents. First, customer. The KLA services team is embedded with our customers. We are a true extension of their manufacturing teams, supporting their customer goals. Next, we have deep supplier relationships.
Again, I talked about that a little bit, and it's not just about that beginning and product inception, but it's also being able to maintain that continuity of supply throughout 20 or 30 years of the lifetime. Finally, we're very closely coordinated with our KLA design and manufacturing teams. Rick talked about how we've inserted our services team very early in the product life cycle, and we're present throughout the life cycle again in ensuring that we're gonna be able to support our customers' goals. Just a little bit more about our collaboration with our customers.
I can tell you that we do a good job of this, but ultimately, the way we measure our collaboration with our customers is how much revenue or how much of their spend is done with KLA, and secondly, of that spend, how much is on a KLA Services contract. You see the chart here on the left from 2001 to 2021. Over the last 20 years, we've steadily increased both the amount that our customers are spending with KLA as well as the amount of service revenue that's coming from service contracts. That's the CY 2021 greater than 75% number.
If you were to double-click on that CY 2021, you'd find that products like our latest generation BBP, our latest generation reticle inspection, our latest generation PCB tools, all those have contract penetration well north of 90%. Why is this happening? There's a few environmental factors. First, our customers are experiencing record-level factory utilization. Utilization rates are running greater than 95% for more than 24 months. This is really sustained significant high utilization rates. Next, we all know that our customers are expanding geographically. When our customers expand geographically, it not only puts pressure on the home country, but it also puts pressure on that new ramping factory. This growing importance of specialized talent, our customers are focused on their core competencies.
They're depending on KLA to be able to ramp up their new factory, as well as depending on us even more in the home country as they've had to ship resources abroad to help support that new factory. Finally, this renewed emphasis on reliable supply chain. If you think about our service business, we are a critical supplier for our customers. This renewed emphasis, you know, we talked about this enduring value of our tools in the long life, very difficult to sustain that lifetime. Our customers are coming to us, giving us much more insight into their ramp plans, into their long-term plans, extending service contract life to ensure that we're able to support them over the long term. This increased complexity usage and urgency is driving increased service contract penetration.
We expect to continue to drive this. This is not just on leading-edge. This is also legacy node customers. This chart is a very important trend for us, for both our service business as well as our product business. On the right-hand side, you see this is a legacy node customer and how much money they spend with KLA. While it's a customer, this is representative of many of the legacy node customers that we're working with today. Over this period of time, this is revenue that you're seeing. This is basically the same installed base. This growth is not related to install base growth. It's relatively the same install base. You see from CY 13 to CY 18, relatively flat spend with KLA service. Why is that?
Well, our customers had factory utilization rates at 70% or 80%, pretty stable manufacturing process, and their customers weren't changing requirements for them. They were running a fairly stable manufacturing process. What did they do? Well, our customers did everything they could to save a dollar. They were harvesting parts from underutilized assets. They were leveraging secondary tool market to buy tools. They were utilizing their underutilized resources and/or trying to get third parties to come in because they had this flexibility given that they were running at lower utilization rates. You see this starting to change in 2019, 2020, 2021. If I had the 2022 bar on here, you'd see an additional uptick.
What's happened with our customers is their factory utilization in this legacy node has increased just as it has on leading edge. Now you're above 95%, and the end customer requirement, think about the automotive charts that Ahmad and Oreste talked about, the zero defect. The end customer requirements have increased significantly. Our customers' manufacturing environment has changed. They need higher reliability, quality, faster delivery, more performance. This positions us very well to be able to support our customers and drive this additional revenue. What did we do? Well, we created a dedicated team to focus on these customers, and that improved our overall collaboration so that we could truly understand what was the problem that our customers were having and begin to collaborate on potential solution sets.
That led us to invest in engineering resources to further sustain the life of the product. As well as hiring new resources, training them on some of this older technology so that we can provide both the labor and the increased inventory investment required to now support this growing demand environment. Hopefully this gives you a little bit of a sense for the environment that we're operating in. Because of this environment, we have to innovate in every aspect of our business, people, parts, infrastructure, knowledge. We have a multifaceted service innovation strategy. It starts with people. We talked about this growing importance of specialized talent. We work very hard to attract and retain not just our service engineer talent, but also the new talent that we need to drive this innovation strategy.
We also recognize that as our customers expand geographically, we have to be there to support them. More and more, we're putting more and more resources closer to our customers. Of course, we've always had service engineers, but also now driving engineering resources, more apps resources closer to our customers in order to accelerate delivery and installation. Next, I'm gonna talk about some of the innovation that we're doing to drive our fleet or improve overall install base. This digital transformation is more about what we're doing to improve the transactional aspect of our business. For KLA, we're a high transaction business, and so we look at every aspect of our value delivery chain and look to innovate to improve productivity of the business.
We're hiring faster than we've ever hired before, and that requires us to change the way that we're training our resources. It's a long time to proficiency, so we've had to completely reinvent the way that we do that to get people shorten time to contribution as well as time to proficiency. Finally, we've invested in mixed reality support, developing more augmented reality-based solutions for remote service, and I'll talk a little bit more about that. Here are four specific initiatives that are aimed at driving customer value and again, aimed more at our installed base management. First, digital twin. This is evolution of the KLA digital twin. We've been doing this for a long time. It started many years ago manually, with lots of spreadsheets.
We moved to ERP and ECO control, and then we're now moving into a much more digital thread, strengthening digital thread so that we have better visibility into how our tools are performing and independent of where they are in manufacturing or in our customer sites. This is really important for KLA and has been for a long time because our customers expect that when we ship a tool, it's going to match exactly, which means if we ship them a tool today, it's gotta match the tool that we shipped two weeks ago. It's gotta match the tool that we shipped two months ago, this tool that we shipped two years ago, because they need it to seamlessly intersect in their manufacturing process and produce exactly the same result.
We also need to make our equipment smarter, and part of that means sensorizing more and more of our tools to give us the insights of how our tools are performing and whether or not we see any drift to specification and can better predict when a failure is going to occur. Now, we recognize that as smart as we make our equipment, we're never going to predict every single failure, so we have to drive an increase in overall digital services. We've seen a 7x increase in AR/VR-related mixed reality remote support. This means we can bring experts from around the world directly in with a CSE in a fab, connect them and allow them to troubleshoot and problem solve live to drive faster time to resolution and ultimately higher availability for our customers' install base.
Finally, once you know that you've got a healthy tool, Ahmad and Oreste have talked about this, virtual process development, we wanna try to minimize the time on the tool. When you have that new process that you wanna implement on the tool, you know you've got a healthy tool which you can deploy and ensure that we can meet customer requirements. Our aim here is really modernizing a toolbox for maximum tool life and performance. This is the roadmap. Again, we wanna move as much as we can from reactive to predictive, and where we can, we wanna automate some of those processes to allow for efficient service delivery. You can see on the right some of our progress from CY 2017 to current.
We've made progress moving more and more items to reactive and predictive and starting to see traction on more automated steps. We're really excited about our progress here, and we think we've got a great roadmap. We're gonna continue to invest in this 'cause we think this ultimately drives a lot of value for our customers. Okay, now moving into execution. Ultimately, this is how our customers judge us. Did we do what we say we were going to do? You can have the best operating model in the world, but if you don't execute, you're not going to be successful. The only way that you execute is if you have the right talent and the right culture. We have a very strong culture of drive to be better.
We look for problems, and that, of course, that job is never done. There are always plenty of problems to solve. Our culture is to look for those problems and then drive resolution on them and then go look for the next problem to solve. That really speaks to our talent. We think we've got the best talent in the industry. Specific to our service engineers, they're some of the most educated in the industry. More than 65% of our service engineers have a bachelor's degree or higher. That's really because our service engineers are much more systems of systems engineers. They are not technicians doing routine tasks.
It's taking all those complex subsystems that we talk about, put those together, and then ensuring that they're working together to create the tool performance that we expect and committed to. That leads to a much longer training time to get people to proficiency, 9-18 months. Again, that's part of the reason why we've reinvented our training process to get people to time to contribution as well as time to proficiency. Lastly, a proof point for sort of that extension of our customers' manufacturing operations, more than 4 million hours applied to the install base in 2021. The operating model is important not just for driving the existing business and scaling it, but also for our inorganic growth.
When we were here in 2019, we talked about how we were gonna take the operating model and apply it to some of our acquired business, acquired service businesses. The reason that this is important is many times with our acquisitions, we find that their primary focus is on new tool penetration and adoption. They don't have the infrastructure, they don't have the scale to be able to support it. Oftentimes, they're leveraging distributors or even encouraging customers to self-service by providing them all the documentation and means to be able to do it. That changes when they become part of KLA because we deploy the operating model, and there are four key parts to this. Number one is design for service.
Number two, common systems and metrics, and then finally, leveraging KLA service infrastructure and our go-to-market strategy. I just wanna drill into one of these items, and that's the design for service, because this is where it all starts and is the most important. Our service is embedded in our product design or product development life cycle. From time of inception, we talk about what are the service requirements across reliability, availability, maintainability. In addition to that, we also look to understand what are the financial returns we expect from service on this product. Throughout a product life cycle, we maintain that discipline, that reliability, availability, maintainability, as well as the financial returns.
We look at that at every part of the product life cycle to ensure that we are doing what we need to support our customer and also driving the expected returns that we have. You see the results on the right, our ADE acquisition. You see over ten years, we were able to double the service revenue per tool on that business. Additionally, embedded in our 12%-14% CAGR is an improvement of the EPC service revenue by 1.5x. Okay, this all leads to the results. We call this our 44-4 plan. As Rick said, took us 40 years to achieve our first $1 billion in service revenue.
We will exceed $2 billion this year in 2022, 4 years later, and we will exceed $3 billion 4 years later in 2026. Of course, this is really important because it's a resilient and predictable revenue stream as well as a strong generator of free cash flow. Here's the bridge, $1.8-$3.3 billion in CY 2026, across our SEMI-PC as well as our EPC businesses. Here's the summary of the points that we talked about today. I'd just like to close by acknowledging all the great work from our global KLA service team.
You know, they were dealt a lot of challenges over the last three years, and every step of the way, they were able to overcome those in support of our customers, and ultimately enabled the great results that we were able to present and also give us the confidence that we can continue to grow this business. Thank you.
Thank you, Brian. We're gonna head to a break, but we are gonna make an adjustment because we are running behind schedule. Rather than a 10-minute break, it'll be a 5-minute break. If you plan to exit the room, please do so now so that you have enough time to get back. We will then resume with our chief financial officer, Bren Higgins, and then Q&A. Thank you.
Over the last six years, I have also oversee the global operations and supply chain organizations, and had a few questions about supply chain. I'm gonna cover supply chain a little bit today. When I'm not designing big buyback programs, I'm spending an inordinate amount of my time on supply chain issues. I think it's important to understand about our business and how we think about supply chain management. It is far from transactional supply chain. You've heard a little bit about it already today from Rick and from Ahmad, and I'll dig into some of the details about it. I think it's important to understand what we're optimizing for, and then what we accept as a result of that. I'll get into that.
I'm gonna start initially with some financial perspective on our performance, both near-term and also over the long run. We'll dig into a couple of issues that have been impacting our financial performance, both in terms of what we've seen over the last number of years, but also we believe will be factors moving forward. Provide a little bit more context on the capital structure actions that we announced today. Then I'll conclude by building up the 2026 model that Rick referenced earlier, which is $14 billion in revenue and $38 in earnings per share. If you think about our key messages, I'm gonna talk about sustainable market outperformance, give you some perspective on that. Obviously, you've heard earlier about what our growth objectives are for the next five years.
All of our reference points are 2021 to 2026. When you see compounded annual growth rates and so on, that's what we're really referring to. That the portfolio strategy of the company is very important in terms of how we go to market and how we compete. Most of our competitors are point product competitors. They come to our customers with one solution for an inflection or a situation. One of the real competitive advantages of KLA is being able to deliver a variety of products that help our customers work through their process maturity, meet technical requirements, but also meet economic requirements, and it's an important part of how we invest and how we go to market. I mentioned supply chain. One of the other dynamics, and we talked a little bit about the pace of Moore's Law and what that means.
I'm gonna talk a little bit about that, plus the trends in end demand and what it means to the product longevity, the extendibility of our products, and therefore what it means in terms of revenue, cost optimization, new capability for customers, which means cost of ownership improvement, and extendibility of platforms, which affects ultimately our R&D intensity, which allows us to shift resources around more freely. Leveraging the balance sheet. If you've noticed our performance over the last few years, we've been very public with our leverage targets of 1.5-2 times gross leverage. With the growth of the business, we've been operating at about 1 time.
We wanna use the balance sheet, and I'll walk through the construct of how we're thinking about the balance sheet, particularly as it relates to the 2026 plan, to, of course, support our growth, look at M&A considerations, but also fuel what we call assertive capital allocation. If you look at the growth prospects of our organic businesses, what you heard today, there's nothing better that we can do with the capacity of the company and the go-forward free cash flows to augment our EPS growth and accrete our shareholders. You saw that in the announcement today. We'll spend more time on it. Takeaway, 9%-11% revenue growth, 1.5 times our revenue growth rate in terms of earnings growth, and our capital return target, given our focus, that we're raising from 70%-85%.
Where were we? September 2019, we were talking about a growth rate of 7%-9% from 2019 to 2023. Embedded in that was an industry assumption. I'll show you some data later, but for the first time, we were starting to see, because of the introduction of EUV and design starts starting to pick up, investments in the logic foundry part of WFE. It was really the first year after, frankly, a negative CAGR for a number of years. While we were optimistic, we really wanted our plan to be focused on what we were expecting to see from our customers and that the industry would do what it would do. Of course, we would participate if it were stronger than that. But that was really what drove our assumption.
I'll talk a little bit more about where we are today and where we're going with that. We were also talking about new products that for the first time, given the spend dynamics, were going to drive process control intensity and KLA share higher. That was, for the first time in a while, we were pretty excited about that. Integrating the acquisition of Orbotech, but also a number of smaller transactions that we've been involved in. Ahmad mentioned it earlier about some of the things we've done in terms of our software businesses, for example. Orbotech was important. It needed to expose us to what we call More than Moore, the beyond-the-wafer-fab core wafer fab equipment markets in terms of new SAM opportunities.
We wanted to take a market-leading position, drive synergies through it to pay for our control premiums, and position it to scale at corporate-level incremental operating margins. Codifying the operating model, you heard a lot about that. Part of that was taking it to the integration, but part of it was also pushing it down within KLA to support what we expected to be structural growth going forward. A capital returns target that at a minimum we thought would be about 70%. Where we are today, as Ahmad talked about the market share gains we've achieved, we feel very good about where we're at from that perspective. Our share of WFE has gone up. It was in the high 50s in 2018. 2021 looks to be somewhere in the low 60s.
I think in 2022, based on the guidance we've provided on the year, should put us somewhere in the, I'll call it, in the mid to high sixes. We're seeing share of WFE increasing as well. Recurring revenue from services, mid-teens% growth, exceeded the synergy targets. Oreste talked about that. A big chunk, normal synergies, but also a big portion about business rationalization. How do we take the operating model and go into the businesses, fully allocate the costs, and say, "This is the profitability picture. These are expectations." There's no moving numbers around to give people a false impression of performance. A transparency to profitability, very important aspect of the operating model. We feel very good about where we're positioned from that point of view. We've invested in talent. Rick talked about 13,000 employees at the end of 2021.
We'll go to about 15,000 at the end of 2022. Go back to 2019, it was about 10, a little over 10,000. Investing in people really across the organization, service, operations, of course, to support the growth we've seen, but also revenue-dependent resources across our product development teams and even some in our corporate functions to make sure we can support the activities and growth that we're seeing. We've invested a lot in CapEx. We opened the site in Ann Arbor, Michigan. We're very excited about our HQ2, another significant presence in the U.S. We spent about $600 million in 2019 to 2020, 2021, and I'll spend another $300 million in 2022 as we expand our footprint around the world.
All of our factories, we're increasing space in a significant way to support not just the demand we're seeing today. This takes a while to put facilities in place, but really what we expect going forward. We returned about 80%. We thought 70% was probably. Couldn't conceive of a situation where it'd be any less than 70%, as we looked at opportunities out there to deploy our cash, and it was actually greater than 80%. Where are we going? A continuation. As you heard from Ahmad today, you heard from Oreste today about how we continue to optimize the business, how we go to market, continuing to lead and innovate.
The 9-11% growth rate from 2021 to 2026, which implies a 7.5% WFE growth rate, which translates into WFE from 2021 to 2026 at about $125 billion. I think everyone can do the math on that, but I just thought I'd throw that out there so people understood how we're thinking about it. Share of WFE. We've gone about 75-100 basis points, I think, by the time we get to 2023, which was our original plan. I think there's another 75 basis points plus of opportunity over the course of our 2026 target model.
Building on our gains over the last number of years in terms of operating profit improvement that we can continue to sustain our 40%-50% incremental margin target, and delivering, as I said earlier, you know, the 1.5 times EPS leverage through our capital structure actions and then our new capital return target. If you look at the revenue, Rick showed a chart earlier, and it talked about declining capital intensity, and that was really driven by logic foundry having a negative CAGR from about 2012 to 2018, a negative 4%. Through that period, KLA was fairly range-bound between $2.8 billion-$3.2 billion, most of the growth coming from some modest acquisition activity, but also growth in our service business.
We're now in our seventh year of growth as a company, so we saw the business continue to accelerate here. You can see the introduction of the Orbotech revenue in 2019. As you saw from Oreste today, from 2019 to 2021, 13% CAGR on that. 19% CAGR in our Semi PC business through 2021. Of course, service supporting that very robust demand environment, but also customers using and leveraging the install base in different ways given the growth of the legacy markets, contributing significant growth there. That's about a 14% or so CAGR as you get to 2022. On the right side of this chart, you see some of the assumptions about how we think moving forward.
If you look at the from 2021 to 2026, 10%+ growth rates in our system business, 12%-14% in our service business, that will translate into about $14 billion and 11% growth from 2021 forward. A lot of these metrics I think are important to understand about KLA. You can see that, as Rick mentioned earlier, we acquired Orbotech, added about $1 billion of revenue that had diluted gross margins to the company average in 2019. A lot of work and focus there. Not every business carries margins like our Semi PC business, but we were looking for markets that had market-leading positions or companies that had market-leading positions and markets where complexity was increasing, that we could leverage operational competencies, but also technology, customer engagement, and drive improvement over time, as I suggested.
We've seen a nice trajectory of growth here. Certainly, the mix has been fairly favorable in terms of our Semi-PC contributions, and they're not quite back to where we were, but at 63%, in terms of our 2022 expectations. As everyone knows, there's a lot of pressure on cost in the near term, and this reflects our perspective of the headwinds and tailwinds. A lot of headwinds out there in terms of cost and work for us to do here going forward, both in terms of how we manage our suppliers, how we engage ultimately with our customers, and how we think about pricing in the long run in our new product offerings. Operating margins, we spend more, so the tailwind of gross margin gives us opportunities to invest more to introduce products faster.
We also, I'm pretty pleased with the execution model. We've outperformed our target. We've been in the low-to-mid 50s% in terms of incremental operating margins over the last few years. 27% diluted EPS growth from 2018 to 2021, and then free cash flow margin consistently now above 30%, and in the 88th percentile of the S&P 500. I get a lot of questions, particularly in these kinds of times, given some of the macro backdrop, what happens to cash flow? How do you think about the resiliency of your business? There's really three factors that influence our performance if you look historically around cash flow performance. Number one is service, and service continues to grow. It grows every year. We've had one down year in service, and that was in 2009. It's a bigger part of the company.
It was only down about 11%. Service continues to be an anchor for us in terms of cash flow and cash flow predictability. Our business, despite what I said earlier about CapEx as a percent of revenue, it's fairly modest. It's not. It's fairly asset light. Most of our investments tend to be in working capital, and if you go into contraction environments, you recoup a lot of that working capital, and so you get the benefit there. As Ahmad talked about earlier, process control intensity is higher in development areas. Customers are always working on next-generation roadmaps, irrespective of what's happening with top line. We continue to sell products, our higher-end products, in those periods of time because customers are always continuing to invest. It tends to be fairly resilient.
I think we've been cash flow positive, what, 90 out of 95 quarters over the last 20 years, something like that. A very strong cash flow performance sort of through cycle and, of course, you know, we see it accelerate in these growth environments. Best-in-class margins. If you look at this 10-year view, it gives you a couple things. First of all, that's across different customer mix, spending levels, different WFE environments, different process nodes. We've been able to maintain a pretty strong lead over our peers in terms of our gross margin performance. Now, our go-to-market is a different business.
It's about demonstrating value, understanding the returns that your customers are getting, being very disciplined in understanding that. If you do a bad deal, it's a gift that keeps on giving from a pricing point of view, because the next generation or the next series of discussions are always based on the prior transaction. We've been able to manage this. As I said, we absorbed Orbotech in it, so very pleased with the gross margin trajectory of the company. Operating margin follows a similar suit. You can see KLA in purple at the top. Of course, they all march along to the different WFE scenarios.
One of the things we're really proud of is through this cycle, as you can see in the 2019-2021 range, the slope of the line is very similar, and a lot harder to go from the low 30s to the low 40s than the low 20s to the low 30s. We're pretty proud of that, and ultimately, that level of profitability drives our free cash flow margin, which is a key focus area for us. Taking a 10-year view versus the SOX, gross margin at 61%. As Rick talked about earlier, it's the sign of differentiation. Everybody thinks they have a differentiated business. The first question we ask is, "What's your gross margins?" It's why we drive our teams the way we do, given our business model. We're number 2/30 in the SOX.
The operating margin, we have to execute, we have to invest, deliver, continue to deliver value and invest, and execute our business. We're in the 90th percentile, number 4 out of 30 in operating margins, and then in free cash flow margin, the 86th percentile. We're pretty proud of this performance. Of course, free cash flow is quality of earnings. You know, how much is available for us to put to work, either back into the business or in supporting capital allocation initiatives relative to shareholder returns. Over the last four years, we've been investing more. Wanted to take a break down our investment a little bit more deeply here. You can see in 2019, about $1 billion in R&D and applications. Applications are customer-facing technical resources.
These are folks that are out working with customers to drive value out of our tools. In some ways, it's an extension of the R&D organization, and in some companies, depending on how you think about it, is it R&D really, or is it cost of sales? We've always had it in our SG&A function, but it's very technical in nature. We went from $1 billion up to about $1.6 billion in terms of investment. You can see in the light purple, you have the applications piece versus R&D. The blue line is R&D and apps as a percent of OpEx, so trending up about 70%. Most of our investment, in terms of the increases of OpEx over the last number of years, has been in these technical areas.
We're getting scale out of our corporate organizations, and so we're pretty pleased that as we spend more, we're directing it in the right places. You can see R&D and apps as a % of revenue trending down. Not as much leverage here as there is in corporate, because we need to make sure that we can support our customers, we can support all the products that are out there. There's a much heavier burden on the engineering and apps teams in a growth environment. I expect that to provide some modest leverage over time, but it's an area where we need to continue to invest. On the right, you see new products. We go to market and introduce products at a much faster cadence than our competitors.
You can see across the business that we're introducing products every 1-2 years. Some of these are iterative product introductions, some of these are more significant. We need to be able to go to market, gives us something new to sell, keeps us a moving target relative to competition. It's fundamental to our portfolio strategy as a cadence of new product introduction. We talk a lot about the operating model and how it works across the different parts of the business. From a financial point of view, I like to look at a couple areas. First of all, how we engage with suppliers in technology development, and there's a lot of engagement in terms of how we're gonna do development for next-generation products, and that starts very early in our product development process.
The other part is how we work with customers to identify unique problems. Unique problems that KLA can solve, and we can solve with a financial return. Not every problem, given the cost of some of the development of some of these technologies, makes sense for us. So a lot of times we get questions and, from investors about, "Well, what's causing that issue? Why can't you guys help?" In a lot of cases, those are markets we have decided that doesn't make sense from a financial point of view or a sustaining sort of financial point of view in terms of product introduction. Or does it translate from R&D into production to make sense from an overall financial point of view? Can we innovate? Can we deliver a differentiated solution that supports the business model, and then ultimately, can we execute?
Can we ramp the technology? Can we ramp the supply chain? Can we get the manufacturing organization scaled? Can we match performance in the field? Then can we support it over time? That's how I think about how the operating model, which is really just a series of processes about how we run the company, either operating processes or planning processes, but how it really applies to my operational/financial world. Supply chain. When you look at supply chain, what I have up here in the upper left is what's happened over the course of the last 18 months or so in terms of relative performance against semi cap peers. As the industry began to strengthen in late 2020, we saw, you know, a lot of buffer that everybody has in the system, and we continued to see demand outpace supply.
We hit this point a few quarters later, where we started to have issues out there in terms of whether the industry was positioned well to support the strong demand environment. While I think we've handled it and sort of weathered it a little bit better, we're still not delivering to where our customers ultimately want us to be. Our supply chain teams have done a great job in terms of trying to manage their way through it. To understand KLA's supply chain, you have to really look at the lower left there and see the time fence, and what that represents is when do we start receiving parts to deliver a system? The blue at the end there is what it takes to build a tool, which is about 70 days to as many as 120 days generally.
The purple there is when do we start receiving parts. If I wanna ship a tool a year from now, I was receiving parts a few months ago. We have to manage our supply chain to be able to shorten that intrinsic lead time to be able to deliver to our customer windows. Even with extended lead times today, it's still beyond what our normal time fence or inside our normal time fence in general. Now we're ordering those parts sometimes a year to two years in advance. When you look at our suppliers, there's really three categories. Ahmad showed the video and talked about some of the optics subsystems, and that's one category which is low volume, sole sourced, highly complex products.
Sets sort of the total potential in any given year of what we can do because it's very clear in terms of the structure of those relationships about how much capacity is available, and that capacity is about facilities, obviously, then there's equipment. When you think about coaters and grinders, lead time is very long. Then there's trained people, and how long does it take to train people? That sets the maximum potential that we can do in any given period. If you think about our BBP products and optical inspection, those products, customers want a lot more than we're able to deliver, and our ability to change that sort of total potential number takes multiple years to do, frankly. We're in the process, and we've been investing in that over the last year and a half.
That sets the total potential of what we can do in any given quarter. The second area, the second bucket, high volume, mostly sole sourced, critical components, mostly around volume optics. This is where our focus is. Now, these suppliers generally work with just us in our field of use, and so we don't compete with our competitors for parts here. Sometimes we compete with other KLA products, and we have to balance those capacity demands. This is where our focus, and it's a day-to-day effort to go manage these suppliers. That's what drives how we execute and deliver in any given near-term period. At the bottom is the tier two and three, four type component suppliers. Those are the electronic components, those are semiconductors, and those are in demand.
They're commodity parts and in demand across industries and across peer companies. It's a low cost of our BOM. You think about the cost of our BOM is proportional to the way I've described the three categories. Usually we can go and be creative, pay expedite fees if we have to, find gray market alternatives, qualify alternatives in our engineering teams within the time fence that's established by the other two categories. The other thing is our volumes tend to be lower. We're a higher mixed business. When you think about sets of electronics, I don't need as many as a lot of the process guys do. As a result of that tends to not be an issue that holds us up all that much. What do we optimize for and what do we accept?
Well, I make long-term co-commitments. I need my suppliers to invest. In this case, you can see our purchase commitments over time in the top chart, and you can see that that's gone up in excess of $3 billion. I feel pretty good over the long run in terms of the demand for our systems, given the end market trends, that the systems are living longer. Also with the service business, given the nature of the complexity of the parts we have, that I'll consume those parts over time, and I always have when you go back and look at prior downturns and so on. I think long-term risk tends to be fairly light. You can also see that we do carry material. We don't debate with our suppliers about who carries the material. We'll take the material. We're well capitalized.
They're building custom parts for us. We need them to be there for multiple years, given the relationships we have. We'll carry long lead time parts and ensure that we can deliver in a more predictable way. We can be more flexible in terms of how we manage the business. Decade-long relationships. 96% of our key suppliers under contract. Product development, engineering, and supply chain are closely linked. There is no throwing it over the wall to manufacturing that happens in some industries at KLA. Because we have to be linked from the beginning to ensure we've got the right suppliers, our platforms are living longer, so we have to keep going back and redesigning parts. If there are issues, sometimes we have to qualify alternatives. Very closely linked process between our development teams and our manufacturing organization. Multimillion-dollar commitments.
I've talked about executive-level engagement. Everybody you've seen today manages a supplier, manages our key suppliers. So there's a high level of engagement at the highest levels of this company, but also our supplier organizations. We'll also make investments. Over the last couple of years, we've invested over $150 million in different structures to ensure we have dedicated capacity to support our growth. Something else that's happening, you heard earlier, is this pace of Moore's Law. Now that we've got a pace of Moore's Law that's a little bit more measured. Historically, given that rapid cadence is you spent a lot to develop new platforms. That's the biggest decision you make in product development, is when to go to a new platform.
The time frame in which you would develop that would drive your R&D intensity up. Products are living longer today. You can see in the top part that this is our R&D as a percent of Semi PC revenue, that it's actually about 20% lower than this six- or seven-year period that's cited, 2011 to 2016. At the same time, the ROI on our products has actually increased about 40%. Now, this is our return on investment assessments when we look at systems business and long-term service streams. We're selling more for longer. It's interesting when you look at the BBP business. Gen5. Historically, we introduced Gen5, Gen4 would fall off. That's not what's happened.
Gen4 is outselling Gen5, frankly, today because of some of the dynamics in the roadmap and our ability to deliver incremental value out of that product. That value is also driving cost of ownership improvement for our customers. R&D intensity isn't the only factor in returns, but we are selling the products and we're also able to do a lot of cost work and introduce new capability over time. As an example, when you look at the Gen4 product line, what you have here, the bars are our spending. The blue line is the revenue. This is a product serving the patterned inspection market. We serve this market with three different products.
What you can see in this case is we started developing the product, spending a significant amount of money about three years before, in terms of major development, three years before we started to revenue. Subsystems, as Ahmad talked about, can go back seven years or more in terms of getting those in the right place. Then we see the sustainability of revenue over time. We have five product iterations here. You can see at the top the GM improvement. From the time we launch a new platform, it's about use case validation, it's about technical feasibility. Then over time, we introduce new capability or productivity, drives customer cost of ownership down, and at the same time we can also do cost optimization, so we get ASP improvement, we get cost improvement.
In this particular product line, we've driven about 10 points in gross margin over the 10 years. It works for our customers, and it works for us. That R&D intensity either is a leverage to our model or allows us to divert some of that spending into other areas. Here's an example in metrology. This is SpectraShape, which is a CD, optical CD product, and you can see a similar trend. Now, this is a more competitive market, but you see in this case we've improved gross margins about 750 basis points, and the return is about 10 to one in terms of the R&D investments versus the contribution that comes out of that product. Previous was 15 to one. You can see Gen 4, which is obviously a very strong example within the company.
We have plenty of examples like this across our product lines. One of the other factors driving our business is the incremental margin of service. Now service has a, we believe, a dilutive gross margin. Now it has an accretive operating margin stream because there aren't many fixed costs there, but everything shows up in gross margin. Because service has grown faster, it tends to be a dilutive force. What you see here is you see a 15% CAGR in revenue. You can see that the incremental margins on the service growth is less of a headwind to the company's overall gross margin, as it's been consistently above 60%, if you look at the linear line. Now what you see in the actual line is you see what happens in service when you have to support customers.
You can see in 2016 and 2017, as China started to ramp in terms of its geographic footprint, the duration of time it takes to hire service people to get them trained and competent. You have to make investments before you start to see the revenue streams and those fabs start to scale. You make those investments, and then you see nice returns, and you see the accelerated returns. If you look at where we are today, given the strength of the overall industry, we're making a lot of those investments again, particularly starting to make investments supporting some of the regional investment strategies that Rick talked about at the beginning. I'll call it less of a headwind from service driving the service incremental margins driving gross margin on the growth.
Transitioning to discussion of capital structure, you can see our free cash flow performance in over four years about doubled to about $2.5 billion. As I look at 2022, and Rick showed a chart earlier, we get closer to $3 billion in terms of free cash flow expectations this year. On the left, you see generally how we go to market and what the business model is. Given our views of what you heard, we feel very comfortable with our ability to raise our long-term target. When you look at the balance sheet and where we are today and where our targets are in terms of how it supports the 2026 plan, we had about $2.6 billion in cash at the end of March.
We believe the right level of cash for the company is about $2.5 billion-$3 billion in terms of our working capital requirements, our reserves for dividends, our operating or strategic buffer and so on. We look at that. We think that through the period, we think we can operate at that level. It's important to establish these levels because in the end of the day, that shows that, hey, every dollar gets allocated. Cash doesn't get valued unless it's allocated productively, and I use these words a lot. We upsized our revolver to give us an additional backstop. This was announced last week. It's sustainability linked at $1.5 billion now. Our leverage ratio, as I said earlier, it's been about 1x.
Our long-standing corporate target has been about 1.5-2 times. When you look at the announcement today, the $6 billion share repurchase, roughly half in the form of an ASR that we'll execute over the next 3-6 months, the other half funded out of ongoing cash flow, given our expectations for the business over the next 12-18 months. A 24% increase in the dividend. I'll talk about the dividend a little bit more, but part of our usual cadence, 13 years in a row. We would expect we're gonna add new debt into the capital structure to support these activities, and I'll have more to say about that over the next several weeks. Effectively taking us back into our target range.
Not extending any beyond where we've said that makes sense to operate where we think it makes sense given the characteristics of our business. Our share repurchases, we've been fairly consistent over time in terms of how we do share repurchases. It's programmatic. We can be opportunistic when we need to be to ensure that we're buying below VWAP, which is usually our goal, and it starts with the principles I talked about before in terms of how much cash you need, what you're gonna generate, and then putting every dollar to work. Then the dividend, thirteenth consecutive, as I said, really governed by the growth rate in our free cash flow, which has been about 15%.
If you think about our earnings model, if you look at 9%-11% revenue growth, an earnings model that gets you 1.5 times that revenue growth puts you in the mid-teens. We should be able to grow our dividend, can continue growing it in the mid-teens. The increase this year was larger. We take a long-term perspective. Certainly, the business was stronger over the last couple years than we expected. A continuation in the overall strategy here.
Theoretical capital allocation, the biggest differences between the two periods, and they're different periods in terms of the amount of time. What's really different is you can see the 22% acquisition allocation versus the go-forward, which is fairly modest and really a placeholder in terms of how we're looking at the future and an effective doubling of the share repurchases in the go-forward period. Expect to get a little bit of scale out of SG&A. Expect to continue to spend in R&D, and would expect to see CapEx be generally around 3% or 4% of revenue. Our priorities haven't changed, right? We're gonna fund our R&D. We're gonna invest in working capital.
We're gonna enable organic and inorganic opportunities when they present themselves, and we're gonna invest the necessary CapEx to support our growth, and we're gonna return everything else. The 2026 target model, I'll spend the next few slides talking about what that model looks like. You can see that, you know, it's predicated, and Rick showed the chart, about 1 trillion or above out to 2030. When you think about semiconductor revenues as a % of GDP, for this broader period of 2004 to 2016, fairly flat. You can see a CAGR that's about in line with global GDP when you think back to our original assumptions in the 2019 plan. Our view is, hey, look, it's gonna be semiconductor revenue. Capital intensity is gonna be flat to rising.
I think what's pretty clear over the last number of years, and I'll talk about it in a moment, but capital intensity is rising. It's not flat. That's pretty clear about where the industry's at. Of course, over the last number of years, we've seen an acceleration of that. There've been reasons for it. Obviously, scaling has resumed, which has driven the design start environment that we talked about earlier. Moore's Law has enabled more benefits to fuel that. New products, the pandemic's effect, I think, in the long run, but also very disciplined pricing behavior by our customers.
We've seen very public announcements there about leveraging those positions in terms of pricing overall and with a slowing Moore's Law, not as much a requirement for discipline because you don't get as much cost reduction, particularly as customers are designing around certain attributes of Moore's Law for some of these end markets. You saw this chart earlier, and this represents the design starts and wafer starts over the first three years of a node introduction because you're seeing today, right, you're seeing 28 nanometer being invested in as customers are investing across multiple nodes. It also highlights the capacity reuse that was happening as the overall industry the scaling in the industry was delayed.
Again, we feel very good about where 5 and 3 nanometer are, and you can see that dynamic drove that -4% CAGR that I referenced earlier. Yes, we've seen meaningful growth driven by the overall market demand, but also what it means in terms of capital efficiency when you don't have a robust demand or design environment. This is capital intensity, and this is nothing new to everyone in this room. I think I'm talking to the most informed audience on some of these dynamics. You can see capital intensity back in the year 2000 or so you go back 20 years, was in the mid-30s. It went down all the way down to about 20%, and it's been rising ever since about 2013, 2014, in that timeframe.
The biggest drivers of it was the wafer transition, right? 200- to 300-millimeter was a big one. The transition from IDMs to foundry was another. Industry consolidation. All important factors here. There is no wafer transition to come. It was always the big driver of enabling that improvement in overall capacity because due to competitive dynamics, maybe discipline, the process guys never were able to get the full value of the incremental area that was being added, but that's not coming anytime soon. That's putting pressure on capital intensity to rise. Our model is it's better. It grows. Doesn't grow a lot, but grows over semiconductor revenue. Semiconductor revenue is 6%-7% in that 2021-2026 timeframe. That capital intensity will drive WFE to be 7%-8%.
One of the questions I get asked a lot is, well, they're spending so much more. What's happening? How does it happen? Well, if you look in logic and foundry, and we looked at it across 17 advanced node foundry and logic manufacturers, you can see that WFE as a percent of EBITDA has actually been fairly flat for a lot of the reasons we've talked about, and their financial performance has continued to grow. Again, now I don't think it's news to this audience, but one of the things that's been a factor in this has been the growth rate in legacy node versus leading edge and how that mix has changed. What you see in the bottom table is you see that the mix used to be about 65/35 leading edge to legacy. Today, it's about 50/50.
If you look at the growth rates, it's about 1.75x the growth rate in that 17-21 period. 20% growth in revenue from legacy versus 11% growth in revenue from leading edge. What's important about legacy is that all the equipment's fully depreciated. Our customers' profitability in this area, particularly given and also with their pricing strength, has driven very high profits out of this part of their revenue mix. It's not the only factor for what's driving the top two charts, but it's certainly a factor that is being driven by the end market dynamics today, given the broad-based demand in semiconductor for semiconductor revenue. If you look at our model, I talked about the 4-5 and where it's going and where the 7-8 is.
We still expect with our systems businesses to grow greater than 10%. That'll contribute 1%-2%. And then service, even with the adjustment up, will only contribute about 1% because the systems business is growing faster in terms of how we're modeling. If you take the $14 billion from where we were in 2021 at $8.2 billion, you end up at about 11%. If you take where consensus estimates were at in 2022 and go to $14 billion, you end up at about 9%.
This is kind of how we tried to bracket given where we are in 2022, and there's still some uncertainty about what will happen this year, particularly in the second half. I think more uncertainty in other parts of the industry than we have given the comments that we've made about our views of the second half and that we made at earnings, which is still our perspective, but gives you some sense about how we're thinking about the overall model. This is what you saw earlier aggregated for the company in terms of contribution. We have an M&A placeholder of about $250 million. Our focus is on our organic businesses. We think that there's nothing more compelling than driving those.
We'll allocate the capital either way, but there are opportunities from time to time for us to augment our offerings, typically make versus buy smaller transactions that we tuck in, some SAM augmentation from time to time in terms of new market opportunities. We just did something recently in materials process control, or verticalization to the extent that we think that there's something from a supply point of view that we need to take in. But de minimis in terms of the overall contribution and focus in terms of our 2026 plan. More detail on the overall model, 63% gross margins. Given our incremental gross margins is at about 60%-65% overall mix assumptions, I think gross margins stay kind of about where they are right now.
We've got work to do from a cost point of view, as I said, in terms of some of the more current cost pressures we're seeing and some of the pressures we're seeing from our suppliers, in terms of cost increase and other inflation that's out there. We think we can maintain our 63% gross margins. R&D, a little bit of leverage, but still gonna spend, to ensure we've got the right portfolio fundamentals to the company, compared to our 2023 model, 8% for SG&A. Operating margins between 41%-43%, and incremental 40%-50%.
With that incremental, you can certainly influence as we grow, but probably doesn't move a lot given that we're kind of already there, and I think we can sustain the gains we've seen from, you know, the rapid growth over the last number of years in terms of how we're gonna operate. $38 in EPS, ±$1.50 and the 85%, and then on the right, you see our assumptions in terms of overall mix for the industry WFE. In closing, before we move on to some closing comments from Rick and then on to our Q&A, hopefully I've talked about the sustainable market outperformance we expect across the business. The portfolio strategy matters to how we go to market and compete.
We're gonna continue to make the investments and optimize for our supply chain to continue to deliver and meet our customer expectations and have the flexibility that's required given the inherent lead times of our products. Pace of Moore's Law right now is good. It drives a healthy demand environment and allows us to get full value out of our offerings, while at the same time improving our customers' cost of ownership and continuing to be proactive about allocating every dollar productively with the company, either go forward cash flow or the capacity that we have. With that, I will turn it over to Rick for some closing remarks.
Thank you, Brian. I actually just have three stories to close, and then we'll come back for Q&A. First of all. Well, thank you for keeping your cameras on, all of you. It's nice to see everybody in person and have live audience to interact with. I think our stories hopefully have explained KLA. You understand it better. Hopefully, you get a sense of how we do business, and I wanna give you a couple other kind of stories, anecdotes to let you know what informs our thinking. Years ago, I was the product manager on the KLA 2110, which was the high speed. First time, it was the highest speed inspection system in the market. I came to New York with our co-founder at the time to introduce it. This was like early 1990s.
Product was 100 times faster than the previous generation. It was amazing. I couldn't believe it. Nobody could believe it. We came out here and gave this presentation to a group of investors, and the stock was at $7. After the meeting, the stock was at $7. Nothing happened. I talked to some investors after, 'cause I'm new to this world, and I said, "I don't understand." They said, "We hear 50 of these a week, of these great breakthrough technology stories. You need to show us the numbers." Part of KLA's belief is we need to show you the numbers. We'll talk about our forecast, but it's always backed up with what we've done. That was all along when we talk about these forecasts, like can we deliver?
We hold ourselves to a very high standard in terms of setting expectations and setting business expectations. The other thing about the 2110, later I was the general manager, and our goal was to beat Tencor. In fact, my mantra in all hands meetings was, "Ramp, beat Tencor. Ramp 2135, beat Tencor." That was. I told everybody in the division, "If you're not working on one of those two things, talk to your manager 'cause you're not working on the right stuff. Make ramp, beat Tencor." That was my mantra. Let's keep it simple. I go out, and we have this new product, and I go out to talk to customers, and they're like, "We love your product. It's awesome." I said, "That's great.
Why are you still buying the Tencor tools?" They're like, "Oh, 'cause the economics are much better for these layers." When we bought Tencor, we talked to those guys, and they said they were telling them the same thing, make the AIT better 'cause it can take out the KLA systems. The message that they're sending to everybody is if you do these things, you'll get your business. What we found out is that's not how they made those decisions. When Bren talks about our customers and our competitors being told they're gonna take a market, that's the incentive of all the players. The power of this portfolio is once we had both the Tencor and the KLA offerings, we didn't have to tell them what to buy.
They made the most cost-effective solution, and that's kind of the point of the power of the portfolio, is those two factors, is when we can offer them choices, trade-offs to make, they will optimize the economics. The third story I wanna tell you that I think is fascinating as you think about, look, we know there's a macro storm going on right now, right? There's a lot of concerns about inflation and we understand that. Why would this industry be different? What if part of the industry is that we're part of the solution? I'll give you an example. There was a 45 nanometer node years ago, one of our customers, and I was talking to them, and they had. This is years ago. They had the game manufacturers were their big customers, and they were moving to 40 nanometer.
45 to 40, to those of you who remember, there was no performance improvement. It was the same technology, just less power. I was saying, "I don't understand. Why are you guys going to 40?" They said, "Well, we're a game manufacturer. We have a $60 fan in every one of these game consoles, and if we go to 40 nanometers, we can take it out." That node transition was not about their performance, it was about the cost savings of the power reduction. Hopefully, what you understand is a lot of what we're talking about now that we have resumption of scaling is I firmly believe semiconductors, in many cases, will be part of the answer to lowering cost for our customers, especially when it's enterprise-based. Especially when you're talking about running data centers, when you're talking about automobiles.
I would just suggest you think more broadly about the impact that semiconductors have had and can have. When it comes to KLA, think about our track record of execution and the power of our portfolio. With that, thank you all for this time. We're gonna take 10 minutes?
It'll take about 5 minutes, maybe.
Oh, we're gonna set up chairs. We can do that pretty quickly. Okay. Then we'll come back, and we'll take on all your questions. All right. Thanks. Typical round-robin type launch. Okay?
Keep an eye on where we are, right? We don't.
Where's the other one?
Run all the way to the-
Come up with me right now. No, come up with me in the front. Like, yeah, you can go to the other half, Christine.
Me? Okay.
Yeah. All right. I think we're all set. Let's go ahead and open up Q&A. There we go. Okay. We got the mic runners here, and hopefully you guys can. Right here, straight in front of you.
Please state your name and.
Yeah, name and firm, please, for the webcast, so everyone knows who's asking the question.
Thank you. Vivek Arya from Bank of America Securities. Thank you so much for a very informative day. I think you've made a very, you know, interesting and compelling case for the long term. My questions are unfortunately more about the short term and medium term, you know, given what we are seeing in the market. First thing is, what is your baseline assumption of WFE for this and next year? Let's say if we do go into a downturn, what is the maintenance WFE? What is the minimum that you think your customers are gonna spend on process control? And in that situation, kind of part B, what happens to your margin structure?
In terms of what we said about 2022, the baseline for 2022 is what we said at earnings, that we see the industry at about $100 billion. It's predicated on a second half ramp, and we'll see what the process guys are able to do in the second half. Our views are unchanged from earnings, from that point of view. If you look at the demand we see from customers today, we continue to see our customers push us pretty hard for deliveries. Lead times are long. As we see this ramp into the second half of the year, I don't see any indications of anything changing in terms of the trajectory as we sort of exit Q4 and move into 2023.
We haven't provided a perspective on 2023 WFE other than, you know, as I look at it, as we run the factories, we're continuing to drive the capacity to deliver more capacity, and I have plans to be able to sustain the business moving from the December quarter. It would be a pretty dramatic change from conversations we've had in the last few weeks with customers, because customers have been making the rounds, and we have obviously asked that question. I know Ahmad and I have both met with a number of customers, and you've seen a major one just reaffirm their CapEx for next year, right? TSMC just came out. Of course it can change, but we're not seeing it changing right now.
I think when you get to the what would happen if it does, obviously the resiliency of the model, we feel pretty good about where we are in our operating margin. You could imagine that we could absorb quite a reduction. Look, how do you decide on a dividend policy, for example, on a buyback policy? You run Monte Carlo simulations of every downturn you've had in history, and you make sure you can maintain it through it. That's what we do.
I mean, we're a pretty analytical group, so when we put this in front of the board and we map this out, we say, "What are the dynamics? What's it been? If you ran it through, can you support that dividend?" 'Cause our view is dividends are only valued by investors if they grow and if they're maintained, not if they're optional. That's how we go and do that, and we talked about the service business essentially funds the dividend, right? The debt structure is not that expensive right now, right? Yeah, look, at the end of the day, our focus is on looking for signals from our customers. If we start to see signals, we have variable elements of our compensation structure. We have open reqs as we're driving to
You know, we're very focused, as we always have been, on delivering and managing. Yeah. Our balance sheet, our profitability enables us to invest in our roadmaps to make sure that we're in position to continue to achieve the growth objectives and gain share as we're moving through. Historically, it's, you know, you look at the investments we're making in supply chain and all that, I feel very good about. Look, there's always near-term sort of excess demand or no demand type of exposure. Over the long run, if you look at prior downturns, even in 2009, we brought every inventory reserve we took to the P&L within two years.
I think that making sure we're making the investments to support our strong view of growth structurally, we're gonna continue to do. The operating model is about managing through the environment. There's a number of operating processes there, and so we'll react as we need to react, balancing sort of near-term with our longer-term objectives.
Got it. If I could just squeeze in a quick follow-up. Rick, if you go back to your 2019 Analyst Day, you had expectations of share gains. How did they play out, and what specific areas did they play out? Which are the areas do you think that are ripe for share gains from here? Are there areas where you think the business might be actually subject to movements on the other side as well?
Sure. I think we, you know, had an expectation. I'll let Ahmad talk to it since he follows it very closely. We had an expectation that optical inspection based on our positioning with Gen 5 was gonna gain. We actually exceeded the model that we had or that we committed there. I think the other one is metrology had some gains.
Yeah. Very much what Rick said. Optical inspection grew quite a bit, and that's where we gained share. In optical metrology specifically and overlay, we gained share. Film thickness, which is the high-end film thickness area, a lot of customers are moving to on-product measurements because these gate structures are very difficult to control on monocrystalline wafers, and we have a unique technology that enables us to do that. We grew there significantly in share. In macro inspection, Oreste spoke about hybrid bonding. That's an area where the dimensions have gotten really critical and macro inspection grew quite a bit, and we gained share. I think future opportunities are in reticle inspection.
Reticle, as Ahmad talked about, the 8xx isn't in the Gartner numbers right now, and that's upside. Of course, after that is the 7xx. I mean, film metrology is also a business within the company. That's the biggest business that no one ever asks me about. But it scales like a process business because for every process tool or whatever the ratio is in terms of how customers add, whether it's overlay, as Ahmad talked about earlier and how that's scaling with EUV, or in film metrology, we've seen better than market growth in those areas. That's been a driver for growth on the metrology side.
Okay. I think the next question's over here.
Yeah. Good morning. Thank you for today. CJ Muse with Evercore ISI. I guess two questions. The first one, you guided process control intensity growing again, and I guess would love to hear your thoughts around the interplay in terms of what I imagine would be a tailwind from Intel, and then perhaps a headwind from lagging edge. The second question is around reticle inspection for EUV, and I think you highlighted greater than 85% market share to date. I guess would love to hear your thoughts on specifically 3 nanometer and then as your confidence in terms of having your tool available for the insertion with High NA EUV . Thanks so much.
Ahmad, why don't you go ahead?
Yeah. I think on the first question, I won't name customers, but process control intensity in logic is increasing, and one of the reasons for that increase in intensity is because with EUV introduction and with smaller process designs, the process margins are getting smaller and smaller. For that reason, we have seen optical inspection grow significantly in logic. It has always grown in foundry, but it is now significantly growing in logic. We see that. The second part of your question related to 85%. Eighty-five percent is the overall layer share, meaning those are the total number of layers that are getting inspected on KLA. When I talked about share, I said two things. One of them has to do with the established market.
Established market, our share is close to 70%, which is primarily in the mask shop and wafer fab, including print check. Then, as you go into the wafer fab, we have good share today because mask pelliculization is not implemented. As mask pelliculization comes in and pitches get reduced, we will have our 8xx system and 7xx system that comes out, and that drives share. I think that covers it. Did I miss something?
You asked a question about legacy. If you look at process, about 80% of our revenue is below. When we think about legacy, it's 28 nm and above, right? Process control intensity tends to be more like memory in the legacy markets for KLA, which I'll call it sort of 10-ish%. Of course, that includes a lot of activity that's happening in native China also. Most of our we can look at those customers in terms of how they're adding capacity, typically, if they're adding capacity to existing lines, or adding new capability, it's, there's no real change. It's fairly mature. There's no real change in terms of process control intensity. Like in automotive, for example, where you're seeing reliability requirements change, you can see a change in behavior from customers.
A lot of these customers have sort of talked about the ability to not be able to rely on foundry for excess capacity and the need to insource a lot of that. They were able to leverage the used market in a lot of cases. All that is not available anymore. The conversations with those customers is that it's a different CapEx pattern moving forward, but it really would require a change in what they're doing or product specifications for our customers to behave in a fundamentally different way. As I said, it's probably 80% or so of our business tends to be the leading edge. Process control intensity is very different, even in mature logic foundry, which is not in the mid-teens like it is at leading edge.
On reticle, my comments were related to 3-nanometer. You did ask that question, so it's related to 3-nanometer. I think in the mask shop, we have a very high presence at that node. Really, it's because, as I talked about earlier, the pitch reduction is not that severe for our optical inspection systems, and we can cover most of the reticles, probably around 90%-95%. Whatever that's left over, one or two critical layers can go to 8xx system. You do see in the marketplace, there is some initial investment on actinic inspection, and that's the reason I had put that $200-$600. You know, it is.
These systems are long lead time, so customers have to place orders initially, and they place them, and that's why you see some rise in the actinic inspection orders. But all of this settles down as the technology is decided for HVM. We see HVM reticles all going to KLA.
All right, there, end of the second row.
Harlan Sur, JP Morgan. Thanks for hosting this event. Ahmad, you talked about some of the big inflections that are coming up. You specifically talked about the move to Gate-All-Around. You just talked about intersection of actinic with High NA EUV. The one inflection that you had on your presentation but you didn't really touch upon is the move to backside power distribution, lower interconnect resistance, smaller cell size. If you think about the architecture, it's a huge. I feel like it's a huge overlay challenge. Is that an opportunity for KLA? What are the other opportunities from either inspection or metrology perspective as backside distribution gets adopted by your customers? One question for Bren is, off of the 41%-43% EBIT margin targets, how should we think about the free cash flow margins?
Okay, I'll take the first, the buried power rail. Yes, it's a pretty significant change. Traditionally, as you add more transistors to the front side of the silicon, you added more metal layers to do the interconnect. But the number of transistors are increasing significantly, and customers have to come up with unique ways to interconnect all these transistors. You can do it from the front side of the wafer, but it's not as productive. Most customers are gonna go to wafer-to-wafer bonding, which is essentially you make the entire front end, and then you do some interconnect with the second wafer. Now, it's very important because now these are two very valuable wafers that are completed, and now you need to interconnect them to make sure that this happens, and overlay is a huge challenge. I go back to broadband.
KLA, all of my presentation and comments were really proxy, right? When I talk about Gen4 and I talk about broadband, it's proxy for Gen5. It's also proxy for optical. We have a broadband system that enables us to choose wavelengths that will penetrate through silicon. We have released a new product for overlay that goes to 1 micron wavelength that enables us to go through the silicon and make overlay measurements between one wafer to the other. That is a critical aspect of the solution, and we have now brought that tool out, and all of our early development partners for wafer-to-wafer bonding are using that technology.
Wafer shape is very critical because these wafers are deformed in some ways, and many years back, we acquired a company called ADE, and from that, we developed a technology called PWG that measures shape. For years, it was not moving into HVM. Memory decided to adopt it first, and now we see logic adopting shape metrology. If you have a defect on the backside or the front side of this wafer and you do wafer-to-wafer bonding, you will have voids and other issues. Macro inspection is gonna grow in this area and also edge inspection. To summarize, I think overlay has the biggest growth potential in this area. We have the system available. It is image-based, which is a KLA unique technology. It's broadband that enables it.
We see many other tools that will enable wafer-to-wafer bonding. We are fully engaged with every customer. First, we were engaged in memory because memory went to wafer-to-wafer bonding first, and now we're engaged with logic customers.
Let me add an additional piece of information. Whenever you do the backside power distribution network, it's opening also opportunities for SPTS because all the interconnects will come through nano TSV.
Right.
We are developing a specific project with imec to make sure that we can etch through this TSV.
On free cash flow margin, the biggest variable in that is the growth rate that we're dealing with, right? If you look at for now, the investments we're having to make in supply chain, the inventory we're carrying, given the way our business model works. What was on the chart, as you saw, our days of inventory has been very flat for us and it'll never perform really in line with where the process guys are just because of how we generally manage it. Our service businesses are different, so we've got to make sure, given the useful life, that we're doing a lot of end-of-life buys and so on. If you're in an accelerating environment, the working capital commits will be higher.
When you look at that plan, I think that, you know, we're low 30s% today because of the demand environment we've been supporting. I think low-to-mid 30s% is about where it'll be given the expectations around operating margin targets.
Okay. Thank you.
That's great then.
Thanks. Sidney with Deutsche Bank. Question on the remaining performance obligation. I know you didn't talk about today, but it has expanded substantially over to $10 billion, and that also seems to be a significant chunk tied to shipment beyond 12 months. My question is, do you think this RPO is sustainable once these supply shortages go away? In other words, are you seeing customer behavior changing in terms of how they work with you and how they order with you in the context of what you guys talk about today?
Yes. I'll let Ahmad take the question about, you know, just the customer engagement. He runs our sales organization focused on this part of the business. As we said at earnings, you're right, it has accelerated, and given the lead times now, it's over $10 billion. I said at earnings that I thought that we'd continue to see growth, that demand would outpace supply or book to bill would be greater than one through the year. I don't see it changing in the near term in terms of where customer demand is and haven't seen any fluctuation within it. Most of the fluctuation would come from generally from us in terms of our ability to deliver, given some of the dynamics I talked about earlier.
Customers are getting in the queue because for certainly around certain products, the lead times are even beyond what we're slotting out into 2024 and so on. No change so far in that behavior. Look, we take the orders, and we make commitments, and we're gonna, you know, build the systems to support. You know, that's one of the signals that if you do get a signal that you feel like things are fundamentally changing, then you react to it. We're not seeing anything like that right now.
Yeah. Can you clarify the customer question one more time?
Yeah. Is there changes to the way the customers work with you guys putting longer lead time stuff or just longer-
Yeah.
order times?
Yeah. Yeah. Customers are really engaging with us positively in this area. You know, the example I give you is, legacy customers did really well in the beginning of this pandemic because we had trained them for years that you have to give us, you know, 9 months to 12 months lead time. When lead time started to get longer and longer, we clearly saw that customers that gave us long lead times benefited. Now, everybody, top five customers clearly understand that we need to provide forecasting till 2023, 2024, in some cases, 2025.
I mean, you know, I'm really reacting to the earlier question also, and Rick attends some of the meetings with me, and then I write passionate emails afterwards about how good the meeting was, which is all about when are you gonna get my systems delivered as soon as possible because they want them. While we have done really well from an operations perspective and ramped up, there is no slowdown from our customers demanding those systems. To your specific question, I would say the top five customers have, you know, stepped back and realized that giving, you know, same year same month order and PO and delivery is not gonna work.
We're working with them to have a very good understanding on their near-term, midterm, and long-term needs for systems. I'll give you one story specifically related to next generation inspection system. Here, we haven't developed a system completely yet. The node is not fully developed yet, and the customer and KLA needs to make a decision on how we are gonna ramp the product. We both have come up with methods by which we're going to agree to specifications that we will take mutually risks on, but we believe we'll be able to ramp. This is the second node we're gonna do this with the customer. The first node, we agreed to some very simpler specs, and then we decided we're gonna take the risk and ramp the product. We did really well.
Of course, we always are improving, and that's part of our customer collaboration culture. You know, just because you agreed to this and we agreed to this, we're not gonna leave each other high and dry, and we close it. Now we're doing it for the second generation systems. Lead times are long. We all have to take risks. They're taking risks. We're taking risks. I think it's all working out really well.
Yeah. You know, this earlier question on the WFE for the year, we say 100, I think it's gated. It's clearly supply gated, right? It's gated by overall and mainly, I think at this point, the process equipment. I don't think. The other thing that's happening, I talked about regionalization, is part of the confusion customers have when they have these regionalization plans, they're surprised, even though some of them aren't breaking ground, haven't broken ground or they're about to, that they're struggling getting slots or what their long-term planning is right now. We have a customer that's, you know, Intel announced they're breaking ground in July in Ohio. How big that is and how fast that is depends on other factors.
I think when we get to these obligations, part of it is because they've got these regionalization plans in place, and there's some expectation. I think what would vary over time, depending on the magnitude of whatever industry dynamic there is not when they started, but how many wafer starts they have at the beginning and then how much they ramp. There's still this initial investment. Because also often process control tools are front-end loaded, you know, not only do we ramp now, as Ahmad said, with production, but they need us at the front end because what they're trying to do is ramp these new facilities. I think in many ways, capacity has elongated this current environment for WFE because there's still all these strategic plans, but they're struggling with getting the capacity.
Back to Bren's point, part of our investment with our supply chain is to make sure that we have the capacity over the long term to support these plans, and that's how we're making these investments. As Bren says, we have, you know, we have ways to manage that should things change, but that's the dynamic right now. It's not just conversations we're having about deliveries in the next
We can't believe you guys are 2+ years on some of these, because we are. In some cases.
Yeah
We're at two years, we're navigating through that right now.
Yeah, given our focus on leading edge, right? Strategic customers obviously are very important in terms of how we think about slotting. I think the other thing is trying to support the ecosystem, right? To the extent that they're semi-providers in terms of how we're prioritizing, we're juggling a lot of different things, but we wanna try to do what we can to keep the industry moving, right? You know, to the extent that there's, you know, we can prioritize or help with some of our slots to help our customers that are supporting the rest of the industry, then it's in our best interest to do that. There's a lot of juggling that we do to try to manage around some of these dynamics.
Great. Maybe a quick one for Oreste. On advanced packaging, if I think about chiplet designs, I think about smaller dies and simple implementation. That would likely mean yields are better and larger than the large monolithic dies. I assume there are a number of new steps that require more process control, but can you talk about why advanced packaging is a net positive for KLA?
As I said, first of all, because of the complexity that comes into the packaging roadmap right now. This is number one reason. When packaging was not shrinking, you didn't need to find the very, very small defects, so the relevance of inspection metrology was much, much lower. Also with the heterogeneous integration, as I said, then the verticalization of the bonding of one die on top of each other, on top of the wafer, creates a lot of opportunities for us to provide unique value. As I explained in the previous presentation, people are really, really worried about voids. Whenever you bond one surface on another, you create voids. We also wrote a technical paper saying that these voids are created on a seed of a particle.
Every single defect on a wafer can create a void when you are stacking a die on top of that wafer. The customers are coming to me and say, "In the past, we need a few micron sensitivity on the inspection tool. Now we need 100 nanometer." Nobody can provide 100 nanometer inspection sensitivity backend right now. Call KLA. KLA has a lot of tools in front end that can be redeployed for the back end, and this is the first net positive for packaging. The second net positive for packaging comes from process. The process tools are becoming very, very customized for advanced packaging. Of course, we have an organization that is mastering customizing tool, that is SPTS.
If you look back at what SPTS was really shining in the industry was creating applications, creating processes that are tailored for the needs of specific markets, either power, RF filters or MEMS and now packaging. We have been partnering with the top five semiconductor manufacturers, not only on the plasma dicing use cases that I showed before, but also on deposition, film deposition that are very, very unique for packaging right now, in particular hybrid bonding. SPTS can provide a low temperature, and this is the only provider in low temperature of this kind of bonding, keeping the same film strength, the same film quality, but done in a low, low temperature that creates a much less impact to the chip and the wafer itself.
Anyway, there is really multitude of reason why KLA is shining in packaging, starting from relevance of process control, inspection metrology, the ability to get some of the tools that Ahmad is developing from front end and moving to the back end, but also the ability to create customized process solution in SPTS.
Yeah. Just to build on what Oreste said, one of the fastest growing business in optical inspection is our 89xx macro inspection system. You can look at the inflection in bond size reduction and the growth in that product line. It is just very, very clear, and it's more than doubled the business because people are just buying that system to make sure these wafers are clean before they do the wafer-to-wafer bonding. Also for all the RDL layers, we see it there too.
The importance also of the channel. We decided 3 years ago, 2 years ago, to form a dedicated packaging channel. All in organization working for me, directly for me, but represent every single product KLA, including Ahmad's products, and we go to the customer with a single voice. This is something that can look trivial, but has been a differentiator of KLA versus other big companies that go there with a bunch of voices, a bunch of people negotiating with the same customer. For us, it is, we have a dedicated channel for packaging, dedicated to packaging. Don't get distracted by the TSMC 3 nanometer technology. It's only dedicated to that. Create a solution collaborating with the customers that their customers can use in the next 2, 3, 5 years.
In even in packaging or PCB, some of the products that we have there, they're not inspection and measurement, they're production equivalent, like direct imaging tools, which came out of originally understanding a lot of the optics and then applying it. Some of our core competency that we're applying in those is toward production. It's not necessarily. Because it's true, process control intensity in packaging is never gonna approach what it is in the front end, that's part of why we have those production tools in there.
Yep.
I think we have in the back.
Hi, it's Krish Sankar from Cowen and Company. I have two questions, both product related. First for Ahmad. You mentioned the term broadband a couple of times. Your competition also has a broadband product. Arguably, it's more bright field or gray field optics. I'm kind of curious how you differentiate with them. And along the path, you know, you have a strong in optical inspection, they're strong in e-beam review. So does bundling actually work in today's world? Do customers accept bundled products? And then a follow-up for Oreste Donzella on the hybrid bonding side. Looks like one of the issues today is on the metrology front is the scanning acoustic microscopy, which you don't do. But is KLA doing anything on that front to help with hybrid bonding?
Do you think you need to have a partnership similar to what AMAT did with BESI in order to accelerate hybrid bonding? Thank you.
Okay. I think the first question was that we have, I've spoken a few times about optical inspection and broadband, and one of our competitor has a broadband system. And/or if it's bright field or dark field. I'm not aware of that. If you can elaborate which competitor you're speaking about. My understanding is that the competitor that you're outlining makes a laser-based system that would make it a single wavelength system. It's a small band ± a couple of nanometers. That's not a broadband system. When I say broadband, I mean every single wavelength from 190 nanometers to 400+ nanometers is available to us, and we can pick different bands during that wavelength and the size of the band and all of those things.
We have amazing flexibility in our system, and our optics is calibrated to be able to deal with all these wavelengths. The system I think you're referring to has a bright field channel and a dark field channel. It's a combined field channel. It doesn't have the same capability. Therefore, I think they're apples and oranges at this point. That's why, for all high-end inspection, KLA does a really good job serving our customers. We do also make a laser based system for cost of ownership. It's a scattering system, very similar to our competitor, and we compete on lower end, high throughput layers. KLA commands a lead in that segment as well. All this is conjecture, it's really about share.
As I said earlier, we've signed up to 0.5% year-on-year share gain since the investor day of 2019. We're well ahead of that plan. Optical inspection was a major aspect of it. When I say optical inspection, in this context, I mean bright field inspection from Gen4 and Gen5 and laser scanning combined. We grew share in that segment. I think that was your first question. There was a second question for me, and I can't remember.
That was.
Okay.
No. It was about review and bundling.
Oh.
Bundling.
Bundling. Yeah. You know, I really believe that our customers are looking for solutions that are really best in class. KLA works really, really hard to build best in class solutions. Meaning that I'm not gonna be able to take a good system and couple it with a not so good system, and that will work really well. It doesn't really work. When I showed you our Gen5 system coupled with our eDRX1, we first did an evaluation on the eDRX1 separately from Gen5. The customer said, "Yeah, this system is performing to our expectations." Then we said, "Okay, you use that for your classic use case, and we will compete in that segment.
On top of it, now that you think that this system is performing appropriately, let's combine between Gen5 and eDRX1 through this proprietary connection and improve the performance further. The same goes for the overlay example I said. We have image-based overlay, scatterometry-based overlay. We're growing share in overlay. The number of measurements are going up significantly. We go to our logic partner in this case because it was Intel and the SPIE paper was written, so it's public. They did this evaluation with us over four years. We started the development of the tool with a really old electron beam system.
After that, we brought in our first generation system, second generation system, and then that system had to perform to the best of its ability, and then we found ways to combine the two. I think that's the best way to do things, is to build the best systems and then differentiate them.
Yeah.
Regarding the scanning acoustic microscope that is a technology that can be used to find voids in a fully bonded packaging. We are not investing in this kind of technology, and the reason is it's extremely slow. We cannot scale to production. It's extremely unreliable, too. Actually, in the last couple of years, also our customers are redirecting the focus on, as I said, finding the defects at the source, finding the small particle seed on top of the wafer, drive the sensitivity down of these inspection tools instead of buying a tool that is not usable for production and is not reliable. I can find maybe one void in a very, very small area of the wafer.
That's the reason why instead of developing a Scanning Acoustic Microscope to find the voids, we are providing the customer with the inspection sensitivity they need to avoid these voids at the source. That is finding the defects that eventually can create the voids. There is a technical paper that says one void can be like more than 20x the dimension, but even 100x the dimension of a small defect. The small defect is what today the customer cares about it. That's the reason why we are developing this solution for providing the sensitivity they need so to find these defects at the source.
Mike's over there.
Yeah, thanks. It's Joe Quatrochi from Wells Fargo. I wanted to kind of double-click on, you know, the phenomenon that we've seen in terms of the lack of tool reuse at the leading edge. How do you think about that? Does that continue into 2026, as part of your 2026 model? You know, if it does, how do you think about the profile of the business maybe changing in terms of just leading edge versus trailing edge? Maybe does that mix change over time?
I'll start and Ahmad can talk a little bit about the dynamics of the leading edge. I think as I said earlier on the legacy nodes, there hasn't been reuse in terms of, you know, buying used equipment or, and so on in that part of the market. It really comes down, given the fact that there aren't any used tools out there, it really comes down to what parts are they building and all those parts changing. Brian talked a lot about that, and in some cases that could accelerate the service business as well.
I think it was something that if you just go back and look at the design starts and then how does that translate into wafer starts, is you had the leaders that were doing the development that were benefiting from a lot of volume on, at those nodes. As they moved forward, the follow-on designs didn't happen because the attributes of Moore's Law weren't all that compelling. There was no technical driver generally. I mean, there were some small changes in architecture, and there was very small changes with multi-patterning in terms of scaling, but you didn't really have the technical drivers which enabled the customers to get a lot of efficiency out of that CapEx. Moving forward in a robust demand environment, as they ramp for 5-nanometer, 7-nanometer is being consumed and they're still adding.
You also have a technical driver with EUV adding, you know, this technical requirement above and beyond the capacity requirement. As long as we have a robust environment, I think it will be good for the business from a process control intensity point of view. The technical drivers are increasing to a lot of the inflections that Ahmad talked about.
I think Bren covered it really well. The way I really think about it is that complexity in HVM has gone up significantly. Because this complexity has gone up and wafers, and design starts have increased quite a bit in legacy and in the leading edge, that HVM is not able to maintain high levels of yield without doing intense process control. That's what we see. We see it in our strip plan process guided us that this is gonna happen in the automotive and legacy nodes. That's why customers per stop significantly reduced used tool purchases, and they asked for new models. We are making new models for optical inspection. We are making them for metrology products, for overlay products, all in that legacy segment, and that's driving.
That is really because of what Oreste talked about, the cost of that failure is just too high, and customers want to maintain it. In leading edge, the same thing goes on because complexity is high, design starts are very, very high. They're not able to tune the yield because you know, the process chambers are dealing with different processes after you know, a few lots. You process a few lots through the process chambers, you bring in a whole different process in the next minute, and that really drives complexity quite a bit. For that reason, people are buying more Surfscan systems, a lot more Surfscan systems or bare wafer systems to make sure the chambers are clean, optical inspection systems, metrology systems. It's really about complexity that is driving.
Now, if that changes, meaning wafer starts come down, will customers do a little bit more reuse? That is possible. We also have upgrade plans from node to node. When customers do that, we leverage our upgrade ability. But I wouldn't be having meetings with customers who plan to do reuses, and then they're giving me forecasts till 2025, 2060 timeframe and demanding that, you know, we need that. We have a very good model, top-down model that estimates the capacity of customers based on wafer starts that they put out, and we don't see a heavy reuse currently planned in the future forecasts up till 2025.
If you go back in time, there was one main customer that probably drove the most reuse, and it was a lower level than what we saw more recently. That was because they were driving microprocessor technology. They'd finish with one node and go to the next. They would try to reuse a fair amount of the capital. Then we saw it at foundries during this period where there was no scaling happening. That was actually the heaviest period of reuse for us, and that has kind of gone away as scaling has resumed. I actually think the folks that drove the microprocessor roadmap with reuse won't be doing that either. I think there's likely to be less going forward than historically there was.
It was really this period we showed on that curve where it was flat, where, you know, you're going literally one technology for one phone to the next phone to the next phone, and that was the, where the reuse was, and that's kind of already diminished.
Yeah. My comment on legacy nodes was the status quo design rules, right? To the extent you have migration of design rule to more advanced design rules, then to Ahmad's point, it creates some of those complexity dynamics. You are seeing more of that in the lagging edge today in those legacy markets. It depends. If it's status quo, less so. If they're migrating design rules, then more so in terms of the benefit we would see.
Okay. While you get the microphone back, we're gonna have time for one more question here on the webcast before we break for lunch. Of course, we can continue this Q&A at lunch for those that are here. You have the microphone?
Hi. I wonder if you could talk to export controls a little bit, both, you know, do you anticipate the sort of surgical entity list actions that we've seen in the past? Do you think there could be broader export controls that affect your shipments in China? I guess, do you have a voice in this? Are you? I know some of your device customers are lobbying heavily for some of these things. You know, do you guys have efforts to be part of that conversation?
Of course, we're in conversations, and I think the primary focus, as stated by Commerce and I think DOD, has been around advanced nodes, and that's a very small part of our business for that region. Bren has quantified that before, and we've not had that for some period of time. In terms of will there be additional ones, there's different conversations, but you know, our assumption is that we'll be able to continue to support customers at the nodes that are above 14 nanometers, which is similar in what there is in terms of the memory markets. The concern is more about foundry and logic than memory. You know, we don't view that as being a major factor.
We are in conversations about that, and we'll continue to, of course, give our perspective and, you know, work with the appropriate entities for that.
All the demand, virtually all the demand we have is considerably above that design rule. Not an issue with the current business levels.
Okay, great. Thank you, everybody, for being here. We're gonna conclude the webcast now. I'd ask that you take your belongings when you guys go to lunch, just 'cause I know that they may be changing this room up, and so I don't want anybody to have anything misplaced. Management will be joining you shortly, and you'll see once you're out there, everything should be organized and we'll see you in a moment. Thank you.
The guests of Jack Yu are meeting in the Hudson Room.