Okay. All right, I think we're on. Thank you all for coming. My name is Toshiya Hari. I cover the semis and semi-cap equipment space at Goldman Sachs. Very honored, very pleased to have Bren Higgins, CFO and EVP of Global Operations from KLA with us this morning. I have a bunch of questions, but would like to keep it pretty interactive. To the extent you've got questions, please be prepared. I'll pause here and there. Bren, first of all, thank you for coming and supporting the conference.
Yeah. Thank you for having me. It's good to be here.
Yep. So in terms of, I guess my opening question, it's a pretty open-ended one. It's been about a month since you reported results and talked about the forward outlook. There's a bunch going on, you know, at the macro level, perhaps a little bit less so at the industry level, but still a couple of your customers pretty public about their spending intentions. Can you give us an update in terms of what you're seeing in your business today?
Sure. It's an interesting time, and we've been doing the conference thing the last few days, so had an opportunity to interact with investors and talk a little bit about what's going on. I think the interesting part about it is that I think it highlights how much the industry has been under-supplying demand. Right? This year has been an incredibly challenging year from a process or just from a shipment point of view, supply chain challenges and so on. While we've done a pretty good job over the course of the year, and I think, you know, the June quarter was a sequential uptick, we talked about sequential increases through the year. We guided September up sequentially at $2.6 billion in revenue at the midpoint.
We see the quarter shaping up as expected, and expect the December quarter to also grow sequentially. At earnings, I talked about the output demand profile as we look at March today, and I see a sustainability in output. I'm still under-supplying our customers. Our customers are putting significant pressure on us. It doesn't feel like it's getting any better or the pressure is getting easier. On the supply chain issues, some areas getting better, some areas, but a lot of other areas are still very challenging. We're trying to get a lot more out the door than we are. But at the same time, I think it's been a great year for the company.
We're gonna be, if you take what I just said and translate that into year-over-year growth, we're probably gonna be up somewhere in the low 20s, mid-20s%, in terms of growth on top of 46% or 35% growth for the total company last year, somewhere in the mid-30s, and in the mid-teens in 2020. Significant growth overall. In fact, the company will be almost twice as big in 2022 as we were in 2019. It just gives a sense of the magnitude of where we're at. We have strong backlog, and you've got to remember that KLA is a technology enabler. More of our business happens in development and in the ramping of facilities.
Less so in production, although production's getting better as the design environment has become more robust, and you're seeing more of our customers having to monitor more activity in their fabs. That's been good from a high-volume production point of view in terms of the impact of our business. More of it happens in working through the process technology, adding the element of volume, scaling that volume, and that's where KLA mostly participates. As we look at where we are and our customers' desires to pursue their roadmaps, particularly at the leading edge, where 80% of our foundry logic business is at the leading edge, you've got a robust demand environment. You've got, you know, 5G units are still strong. You've got weakness in some of the lower-end and mid-end phones.
You have a number of players driving technology roadmaps. It isn't just about mobile and one player, but you have multiple players, particularly as it relates to high-performance compute and data center. We feel very good about the leading edge part of the business, our role in kind of enabling technology. Our customers in memory have indicated into next year that we'll see some reductions in their business levels. It's still very strong today. Sometimes in memory, particularly today, where you have to add wafer starts, where in the past you could do shrinks and effectively address most of the supply requirements, that today there's incremental capacity, so you have to add wafers to meet demand. We'll see.
I mean, our memory customers, since this dynamic started to happen in 2018, and we'll see our memory customers sometimes have to make quick turns and quick decisions because the capacity investments are bigger bets. We'll see as the year plays out based on the pricing environment and everything else and how that plays in terms of their spending next year. So far we see that, but we see a pretty competitive environment at the leading edge. I think on the legacy parts of the market, particularly as it relates to automotive, industrial, we're just starting to ship to those customers. We haven't shipped that much, and would expect those to continue to be pretty solid.
We'll see about the other parts of legacy as it relates to consumers and second sourcing relative to, you know, the TSMCs of the world, where maybe if utilizations were to come down, do some of these customers push orders or not? We haven't seen any of that yet.
People have been asking me a lot over the last few days to handicap what's going on in the overall market. What some of the factors might be as it affects 2023. While too early to guide a specific number, I think these are the factors that are gonna play out in terms of what happens in the industry. I think specific to KLA, as long as they're pursuing technology roadmaps, we're seeing those competitive environments we're talking about, I think, you know, we're pretty well positioned as we move into next year. As I said, demand's been very, very strong.
Great. That's an awesome overview. Since you touched on it, Bren, I wanted to start off with some supply-related questions. I know you personally have been spending quite a bit of time on that dynamic, perhaps more so than you'd wish otherwise. Are the key pain points as of today primarily external to KLA? Is it components and, you know, other things, if you will, that are outside of the company and outside of your control? Are they, you know, internal in nature as well? Most importantly, when do you expect supply to catch up to demand?
Yeah, it's a great question. The issues are different, right? The issues are evolving. I said at one point, you had this Whac-A-Mole, where you have an issue, and then you kind of beat it down, and then another issue pops up. As it relates to the industry was impacted a lot by commodity electronic components. I think in that area, it's gotten better. Part of it getting better has been, I think, we've all escalated, and I think we've all gotten creative in terms of qualifying new parts and finding other sources of inventory. Our customers have become more sensitive to our needs to be able to supply equipment for them. There's been some facilitating effects there. I think in general, that part of it has gotten better.
If you look at KLA's supply chain, and it's very much of a collaborative supply chain strategy. It's far from transactional supply chain. It isn't so much about trying to leverage our heft and spend to drive suppliers to do what we want. It's much more about how do we find suppliers that can meet our technology requirements. We're a differentiated supplier. Finding ones that will work with us and that we can partner with in a way that over the long term, we can meet those technical requirements. Those tend to supersede some of the usual commercial dynamics that happen in supply chain. Because so much of it is there are very few people that can meet our needs.
You can look at one bucket, which tends to be high-end optical components, and there's a fundamental gap in the demand versus supply, and it takes a very long time to add capacity in that area. If you think about anyone who supplies optics, lithography, we're all struggling in that particular area, and it's literally years to be able to add capacity. You know exactly how much volume you can get. Tends to be fairly predictable, and if there's that gap, it takes a couple years to catch up.
In those areas, those customers, we have very long lead times. Customers are in line to try and clamoring really strongly to try to get those products. The electronic stuff I talked about earlier, you usually can manage your way through that part's gotten better. Then there's what we'll call volume components, typically around optics, stages, light sources. In those areas, the challenge we have is we have sole source relationships generally. We have multiple products that are using those suppliers, so we're trading off across products within the company. Those are interesting general manager discussions when I'm allocating capacity.
In those areas, you have just strain in the system, where they're running 24/7, they're adding capacity, but anytime you have a yield issue or you have a COVID disruption, people aren't being affected in a meaningful way from a health point of view with COVID, but if you're running 24/7 and you lose people for a week, it's very hard to make up. In that area, that's our biggest challenge, and we're working our way through it. As I said earlier, I'm trying to get more out than we're able to. At the same time, I think our supply chain teams are doing a great job.
We are investing in those suppliers to provide investment capital, get dedicated capacity to do the training and so on that's required to try to get requirements or to get our needs met in the short run, but also to build capability for our structural growth thesis over the next few years. We wanna make sure that we have these suppliers positioned to support our expectations that we outlined in our Investor Day plan a couple months ago.
Got it. I guess to your earlier point, even in the March quarter of next year, you'd expect to be short of higher customer demand from a supply perspective?
I think it's getting better from a supply point of view, and there's new capacity coming online. I would expect, as I said, a sustainability of December kind of output levels as we move into March. I mean, that's what it feels like today.
Got it. Okay. A WFE question. You kind of answered it in your first response. Thoughts into 2023, again, I don't expect you to give us a number today, but when you think about the puts and takes, you talked about memory being a little bit more squishy. I don't mean to put words in your mouth, but then your confidence around leading edge and perhaps some of the even specialty technology nodes. As of today, based on your backlog, based on customer conversations, which areas could be the potential swing factors? Both to the upside and the downside.
Yeah. As I said, I feel like the leading edge competitive dynamics and the roadmap opportunities out there, I think, look pretty good. I think those will hold together fairly well. You mentioned memory. In the legacy area, process control intensity tends to be lower. If you think about it, if you're an automotive supplier, you've been making the same part. Now, this is an EV automotive, which we'll call general automotive. You've been making the same part for a decade and you wanna add capacity. You're gonna add capacity. That'll impact the capacity, the process guys more. It's fairly stable in terms of your maturity of your process, maturity of your output, your yield, and so on. The process control part of that, there is a piece, but it tends to be fairly small.
If that customer is changing specs, and you're seeing that in some cases in automotive, as semiconductor content rises because of reliability issues and cost of recall and so on. Specs are changing, design rules are changing. You're seeing some dynamics change. But in that particular part of the market, process control intensity is more like a memory. When we talk about process control intensity, it's process control as a percent of wafer fab equipment. Memory tends to be 9%, 10% as a percent of wafer fab equipment. High-mix leading edge foundry logic tends to be mid-teens to high teens. I would take legacy foundry logic to be more like memory, so about 10% or so. We're less exposed to that area.
You are seeing in that. You know, you still look at car lots, and you're still not seeing a lot of cars out there. I think some of the challenges are there, and we would expect to ship to those customers over the remainder of this year and into next year. I think that's, that part of the market, seems like it's pretty good. I talked about the second sourcing. China logic and foundry, I would expect that to be kind of flattish based on what's in the plan for next year. We've seen a lot of demand, and lead times have been long, but I think we'll do about $1.5 billion or so in systems revenue in native China this year.
I would expect it to be kind of roughly consistent as we move into next year. I think the memory wildcard is interesting. This year's been a good year for memory for KLA. With the introduction of EUV into DRAM, and the infrastructure requirements to support DRAM, it's been, we've seen a nice, uptick in investment to support that, technology transition in that part of the industry. We'll see. We'll have more to say about it as we get, we report out for the December quarter. For now, and then we'll have to see how everything we hear, how does that translate.
Juxtaposing that against everything I said and the strength of the backlog, which in semi PC is in excess of the Semiconductor Process Control business with KLA in excess of $12 billion, it feels like it's gonna be, you know, fairly resilient given that and given our technology focus in the company in terms of the role we play.
Got it. Bren, when you think about the CHIPS and Science Act and government funding across the globe more broadly, you know, curious how you're thinking about the potential implications, both direct and indirect, you know, in terms of customers perhaps being a little bit more aggressive with their projects. Internally, perhaps, you know, you need to make investments to support your customers more on a global basis as that sort of manufacturing footprint globalizes, if you will.
Right.
How are you thinking about those things?
Yeah. I think first of all I think it's very good for the United States having more semiconductor manufacturing in the U.S. is a good thing for industrial policy for the country. At the end of the day I don't think it creates this influx of incremental supply. It's really a cost offset to drive incremental investment here versus alternatives. While it could provide a bit of a cushion depending on the timing given that there's dollars available and sort of how that plays out in terms of how planning happens, and people might spend money when they otherwise might decide to pull back, you could get some benefit from that. The industry has achieved significant efficiency and capital utilization over the last couple of decades, mostly with this transition from IDMs to hyperscale foundries.
Some of that efficiency, as you're spread across a broader footprint, I think you'll lose some of that, particularly as teams are spread more thinly, customers are operating multiple sites. There's probably a benefit to the equipment industry from that. For a scaled supplier and being able to support a broader footprint, and you got to think that, you know, if a customer has their team spread more in a confined way and now has to support more facilities, being able to rely on suppliers to help, both in terms of our applications people that help people run tools and get full value out of KLA equipment or our service infrastructure, all of that, I think creates a share opportunity for sure for the company, and probably does for any of the larger players out there.
I see the benefits more in those areas. I think these customers have always done a pretty good job of balancing supply and demand in the long run. As you're ramping up a facility, it's at an unproductive state. Depending on the time of that, there could be some short-term benefits. But I think over the long run, it gets worked out fairly rationally. But I do think there's an efficiency gain that the equipment companies get kind of at the expense of our customers. And then I think there's a share opportunity as well to support it. Now, to your point, we have to put the infrastructure in place. We have to hire the people to be able to support, and we've been doing that in Arizona already, just starting to do that in Ohio.
There's training to get to competency from our teams in terms of being able to support our products there. We're making those investments. We'll have to get depots to put parts in and so on. In the long run, we see those investments, and it puts a little bit of pressure on the incremental gross margins, but over time, as you get that to scale and you start to see the revenue come in in terms of systems but also the service streams, we tend to get fairly good efficiency on those investments. We saw it play out like that in China, where you had a very broad geographic footprint that we had to invest in.
There was pressure for a period of time, but now as we're seeing more and more service revenue opportunities, we're seeing some nice leverage on that. Talked about that a little bit at Investor Day as a driver for our service incremental margin. I think it plays out like that.
Got it. Makes sense. The geopolitical backdrop, specifically, you know, U.S. versus China is a tricky one, to say the least. I know you guys, along with everyone else, have kind of invested in government-facing positions, if you will, over the past couple of years. One of the most common questions I get from investors is, what if a headline hits, you know, tomorrow and your $1.5 billion native China business is cut in half or what have you? How do you think about that risk? I know there were some headlines yesterday morning. How are you digesting all of this?
Yeah. I think first I'm gonna just back up and look at the China business we have. It's really in three buckets. You have the foundry logic part of the business, which is mostly legacy, great majority legacy, which is 20 nm and above technology. Very limited capability at what's 14 nm. 14 nm was an architecture change to FinFET architectures from, which was a transistor architecture change. That happened at 14 nm. You look at the memory part of the business, which I think is, you know, not necessarily caught up in some of these discussions. And then there's infrastructure investment that's also happening in China around wafers, but also reticles.
New projects and customers being able to get allocations of wafers has been challenging over the last couple of years, so there's a desire to build wafer capacity to be able to meet the domestic requirements there. On the reticle side, a lot of the reticles in the trailing edge coming from the merchant mask suppliers haven't invested a lot in those nodes over the years. Pricing's increased a lot, but being able to get those parts has also been a challenge. You're seeing investment in reticle infrastructure as well. If you look at KLA's business, we're exposed to the bare wafer and the reticle more than some of our process tool peers who don't have exposure in those areas.
If you look at the mix of the business, I'd say maybe 30-ish% or so, more or less, is the memory part of it. I'd say 20%-25% tends to be this infrastructure, and the remainder is logic foundry. The great majority of that is 28 nm and above. What's 14 nm and below, which is what you referenced, and we were notified prior to earnings, and we made some comments at earnings. I don't think I have much that's incremental to say. What we're hearing and what we saw in the news recently was I think a codification of what we heard was that anything that's 14 nm and below would require licenses.
We can get the material to our business expectations remainder of this year and into next year, which was our comments, and we'll continue to work very proactively with the government to try to understand their goals and what we can do with our systems to try to support our customers, but at the same time, be fully compliant with what's being asked of us.
Mm-hmm. Is it fair to say you guys have a pretty, you know, constant, healthy kind of back and forth or-
Very.
Okay.
In fact, we're working very closely also with our peer companies. One area of agreement, I think that Lam, Applied, and KLA, we all can get on the same page in this particular area just to make sure that we're collaborating in a way that supports the, you know, us as suppliers, as U.S. suppliers, but at the same time allows us to collaboratively work with the government to accomplish the government's objective.
Got it. In terms of technology inflections over the next, you know, call it three years plus or minus, you know, process control intensity, you know, has grown over time, memory more recently to your prior point about EUV, but as a company, what are you focused on in terms of inflections, and how are you thinking about process control intensity going forward across logic and DRAM and NAND?
Yeah, it's a great time, particularly in the logic foundry space. One of the stories we were trying to tell at Investor Day was to try to explain what had happened in the middle of the last decade, we'll call it 2012 to 2018, and what changed to where we are today. I was saying earlier, we went from, like, severe drought in logic and foundry WFE investment, it was a -3.5% or so CAGR over that timeframe, to the perfect storm of goodness in some ways in that part of the market. Given what I said about process control intensity earlier, our strongest part of the market for KLA. When I say perfect storm of goodness, what do I mean? Well, you had a much more diversified demand environment. This is great.
It isn't just about PCs or mobile phones, but much broader than that. Also rising semiconductor content in some of the more legacy markets, but those are markets where we're able to get more life out of our tools, so the platforms we're selling, and the incremental engineering requirements are lower, so the profitability from an operating margin point of view is very good. It's also very good for the service business. We've seen technology inflection. That's what changed, was the introduction of EUV, which created stronger Moore's Law attributes that drove a more robust design environment. The technology inflections and the introduction of EUV has been great for KLA, and we're seeing rising layer count.
We're still kind of in the mid-teens right now in terms of layer count, and so you're gonna see that continue to progress, and that's an opportunity for a number of our businesses. Finally, we had this design environment, right? Where the design environment's picked up, and so that's been very good in that customers have to monitor more designs that are testing design rules in different ways, more customer process flows, having to deliver in tighter windows, and so that's been good for general process control intensity that they're having to monitor more in high volume production. They also have this follow-on demand that's consuming the capacity, so their ability to take that capacity and move it to the next node has been compromised.
For a number of years, with the delays in EUV and very few designs, there was a lot of capital efficiency by taking tools from 1 node and moving them to the next. Much harder to do today because of the demand that they have, and they're having to invest across multiple nodes, and they did that this year. That being the biggest driver, but also that they have this technology change, and the technology change makes it even harder because they have to find different smaller defects. They have different defect mechanisms and so on. If you look at what's in front of us, changes in architecture again at the 2 nm node and beyond, you start to look at gate all around, buried power rail and power distribution and how that's changing.
Ultimately, as you start to get in the middle of the decade, High-NA EUV, which changes probably the pitch, and therefore you get more density in reticles, you get more EUV reticles generally or more EUV layers. All that's gonna be, I think, very good for our part of the business.
Got it. The E-beam versus optical debate is less of a debate, I feel like, these days. It used to be a pretty significant pushback, if you will, on the KLA story. I think you guys have demonstrated that optical continues to improve in its technology, and you guys have obviously done a lot of work on the E-beam side as well. Internally, when you have that debate, if there is a debate, what are the thoughts around that potential transition? I know you guys think about it as two complementary technologies as opposed to one taking over the other, but what's your current thought process on that?
Yeah, they really are complementary, and we've spent a lot of time talking about it over the years, and it does get back to some of the fundamental physics that when you talk about inspection really limit the ability to use e-beam in a production environment and why optical is the flagship. People talk a lot about the extendibility of optical from a wavelength point of view. We're figuring out ways to extend it. We're using AI techniques and so on to get better signal from the noise, which is our biggest challenge. Our biggest challenge is more about finding what's relevant, less about do we have enough light for it. We find a lot of stuff. It's finding what matters from what doesn't.
If you think about E-beam, there's really, and I think sometimes investors are, and people confuse, there's really kind of three use cases out there. The first is using review capability, and what that is that your inspectors find defects, and you use review tools to go take a look at them. The optical inspection market has grown significantly, maybe the fastest-growing market of scale in WFE over the last few years. If it grows, E-beam review grows. They scale almost 1:1 in some ways because you have to look at those defects. We have a review system. We have a competitor that's the number one in the marketplace for review.
What we're trying to do is how do we take our review system, put connections between the tools, and be able to do fast review for fast classification, faster time to results, and how to do that more effectively with our inspectors so they complement each other that way. More of assist technology. You have inspection. Inspection, we use that to point our inspectors, but you have huge gaps, like thousand times difference in terms of production speeds. You can't do one thing about in the fab, to find random defectivity, you gotta have full wafer coverage. You may have defects that show up in one part of a die on one side of the wafer and not on the other side. This whole process uniformity challenge has to be managed.
You have to be able to use optical inspections to be able to see that. In development and in defect discovery, you'll use E-beam tools to try to work your process through. When you scale to production, given those limitations on speed, you have to have different solutions. You're seeing more application of E-beam in metrology, where you can actually do in-die measurements to point your optical tools because a lot of the optical tools in metrology is model based. There are metrology applications, particularly around overlay, where you're seeing more E-beam capability being deployed. We're investing in all those areas. We see it as an assist technology.
If you back up and look at all the dollars being spent, tends to be about 80% optical, 20% E-beam, and we're still seeing it there. I think that it continues there for all the reasons I talked about. Now, given we have the optical portfolio, and I think it's important part of KLA's competitive advantages to be able to compete with our competitors using the portfolio to allow customers to meet technical requirements but also economic ones. I think it's a very important distinction of how we go to market versus our competitors. A point I think it is, it's the data really demonstrates, right? Maybe it took a while for that to happen.
It took a while for Gen5 to get traction and to kind of prove out the growth that we've seen over the last several years to demonstrate that this is why customers are behaving the way they are, why they're making the purchase decisions they are. I think you'll see it play out that way over time.
Got it. Super helpful. We have a little over 10 minutes left. I wanted to pause here to see if we had any questions. Can we get a mic, please? Second row. Thank you. In the front.
Thank you very much. I actually have a follow-up question on this topic on EUV. How do you see a Teron inspection tool sort of come into play? Is it also kind of a complementary role like e-beam with optical, or is there a more significant role in this technology like?
Yeah, sure. No, it's a great question. In Reticle Inspection, particularly as it relates to EUV, there's what's called an actinic inspection, which is an EUV wavelength inspector, and it's used to inspect the reticles. We think, you know, over time, it'll be used more and more to inspect reticles. Today, not really. It's being used today more in some very limited, what we call pathfinding or defect kind of R&D applications. It's an area of our future product portfolio we're investing in. Back to, as I said earlier, what we offer to the Reticle Inspection market is we offer a portfolio of products that allows our customers to leverage multiple technologies to meet their needs.
For example, 95% or so of all the EUV production reticles are running through KLA 193 nm 640e systems. That's what's happening in the mask shop. We also have for those pathfinding examples, a new product in the market that's an e-beam based multi-column system. It is a technology that makes sense for e-beam given what we're doing with reticles, and the size of reticles, and we have multiple columns, so we can address some of the throughput challenges there. It's a very high-end inspection. If you have a defect on the reticle, it prints on every die. You have to have 100% defect detection, so this technology is very valuable.
There's also supporting capabilities to inspect against what we call design databases and so on, and be able to render those images to compare against the physical images you're getting. You have that very high-end application in the mask shop. You have a couple of tools that are addressing that market. We have a couple of tools on the e-beam capability that's in the market for that pathfinding, and we think that'll meet the needs over time. We'll start to see revenue from that product next year.
In the wafer fab, customers will do what's called a requalification, wanna make sure that as they're using reticles, they're verifying that they're good before they go and start to run wafers. We sell two different technologies there that customers are using. One is in reticles that don't have basically particle prevention pellicles, which is a membrane on top of a reticle to protect it against particles. In those situations where you don't have that membrane there, you use a version of a Reticle Inspection System called 670e that customers are using.
They're also using where they have that membrane, that pellicle-protected reticle, then they'll use a wafer inspection system where they can print the wafer and validate the fidelity of the reticle based on the information we get off the wafer. You can see that you know customers are able to mix and match, and that's what they're doing to meet their production requirements today. I think this market and this opportunity around Reticle Inspection and the evolution of EUV reticle, which is actinic, it coincides more with the introduction of high NA for a couple reasons. High-NA EUV will allow for a lot more reticles, but it will also drive a pitch change in the reticle.
A pitch is a geometry shrink or design shift, and at that point, you start to see more relevancy of actinic solutions to be able to find those smaller defects. Because you have more reticles, you'll have more pellicles because you'll put more production at risk, so customers are likely to do that. Although there are a lot of challenges in all of this. A, that the actinic schedule holds. B, pellicles tend to absorb energy, so that affects the throughput and productivity of EUV scanners. There's a lot of complexity in all this. I think when you back up and look at the way we're addressing the market, A, we're getting the lion's share of the production dollars that are being spent in both areas.
We have investments, in fact, it's the most expensive R&D projects in the company to develop capability going forward. I think it's an example of where we allow customers an opportunity to leverage our technical solutions, but also to meet their cost of ownership requirements. When you're running a facility, the last thing you wanna do is bottleneck your facility and have unproductive wafers when you're starting 60,000 or 70,000 a month. They take 90 days to build. There's a lot of economic dynamics there that also drive how they mix and match across the portfolio to try to meet their requirements. I think that's how that market plays out. We spent some time talking about it in Investor Day. There's some slides out there, and if you wanna go, our Head of Semiconductor Process Control did that presentation.
A little more technical than I am, but that's how we see that working.
Bren, hoping to talk a little bit about the services business. It's a beautiful business, really. I was hoping you could kind of level set the audience by explaining kind of the fundamental differences between your services business relative to the process tool companies, the stickiness that's really inherent to your install base.
Sure. When we report service revenue, it is all service revenue. It doesn't include upgrades. It doesn't include systems or anything like that, so it's all classic service revenue, and 80% or so of it, certainly in the Semiconductor Process Control, maybe it's a total company when we factor in our EPC business, about 75%, all contracts. Service contract that can be either tool contract or can be fleet-wide or fab-wide type contracts. I think the way our business behaves, a couple reasons why customers and why it's sticky is part of it with process control, because at the end of the day, we're a monitoring company. We're monitoring their process, so we're not contributing to output, contributing to good output, but...
Customers tend to buy what they think they need and then run the tools very, very hard. They're very complex systems. They have to match performance. If you're a customer, it's hard to build internal capability to provide your own self-service when you just don't have that many. There might be an ROI if you've got, you know, 100 etchers to do that, particularly if the complexity is not that high and you can get access to the parts. But if you've got 20 broadband plasma inspectors, that takes us a year and a half to get somebody proficient, and they only service that system, really hard for our customers to do. They buy what they need, they run them hard, they're very complex, so it's very hard to do the service.
Our supply chain tends to be very custom. There were years ago we had off-the-shelf parts. We don't have any off-the-shelf parts anymore. The ability to get those parts, we did generally control the supply chain. All that drives our customers to buy a certain level of service and support and to have a contract structure that provides the predictability to keeping the tools up and so on. That includes things like response times and inventory levels, and there's a fair amount of detail to how we put these together. That tends to drive this service stream that's very predictable, and allows us to optimize underneath it the cost structure of support we deliver to that support model.
We have clarity about what that looks like, and so we can get a fair amount of efficiency about how we do it. As we've seen the industry consolidate, we've been able to get. It's been a nice tailwind also for the service business overall. I think that's why it behaves a little bit differently. It doesn't have the consumable element that a lot of the peers do. If you back up and look at how are customers running our systems? They're using them for a long period of time. In fact, you can go back to 80%+ of all the systems that KLA and Tencor shipped are still in service in some way today and are still being used by customers.
As you're seeing semiconductor content rise in these legacy markets, you're seeing the useful life of our systems increase. You're seeing inflections that are driving our customers to run at higher utilizations and to try to use these tools more. Then you've had just this very nice ramp in the business. All these tools with higher complexity and the dynamics I talked about are now starting to go from warranty and into contract, and we're getting leverage on that too. Pricing tends to be as a percent of ASP. To the extent that we're adding value and we get paid for that value, then it also translates into the service pricing structure as well.
It's very good, and I think over time, we've seen, you know, how much of sort of tool system revenue do you get in service revenue over time. If you would've asked me that question five, seven years ago, I would've said maybe 40%. It's probably closer to 60% today.
It's on its way, certainly. It was a big part of our plan on a go forward. It's a bigger part of the company. It's a stronger anchor in terms of profitability and cash flow in any WFE environment, 'cause customers always run their install base. We've been able to drive a fair amount of efficiency out of it over time too. It's certainly an aspect of how we've gotten more comfortable over the years with paying a meaningful dividend, having more leverage in the capital structure, is having the reliability of this service stream. We think we can take a lot of the principles in how we do it to acquire businesses. Acquired businesses struggle with the footprint, struggle with the investments, struggle with customers beating them up on trying to monetize service.
We do think that there are opportunities for us to get more value out of the service businesses of the companies we acquired, and that's part of our long-term plan as well. We went from a 9%-11% long-term growth rate to 12%-14%.
In terms of what our expectations are moving forward.
Makes sense. Last question on capital return. Again, to your point, you guys generate very strong free cash flow irrespective of the environment. You were early in embracing capital return, a capital return program relative to your peers. Extremely consistent dividend growth, you know, consistent buybacks. Given the macro, does that, you know, make you kinda rethink the near-term cadence of capital return, or are you just gonna keep going at it?
I think our capital return strategy, and it goes back, you know, a little over a decade or more, was it became very obvious to KLA, A, the dynamics that we're talking about and how that affects our ultimate cash flow performance. We've had three down quarters in 25 years, only one in 2009. How our business behaves in any CapEx environment has been very resilient from a cash flow point of view. I think a lot of companies are sometimes slow to realize how much value gets trapped in how you finance a business and how you allocate the capital that it generates.
At KLA, we've been very aware of that, and we spend a lot of time looking and doing analysis every year as we go through it with our board, and saying, "Okay, what are the capital cash return requirements? What are the simulations?" I made a joke the other day. I don't know how many boards are looking at different Monte Carlo simulations about what happens to cash in different revenue environments. We go through this, I think, fairly rigorous exercise, and we have the amount of cash we think makes sense to have on the balance sheet. Then after that, it doesn't get valued unless it's deployed productively. What's the right level of cash? What's the right level of leverage?
1.5x-2x , we believe, is the right level of leverage for a business like ours, given its business characteristics. Our cash target, which is, we adjusted up, it's about $2.5 billion-$3 billion, and I can go through the pieces of how we get there to that. We deploy the cash flows to get a value, to either back into the business to meet strategic opportunities, and if we look at M&A, the alternative is buying back our stock. It's a very clear hurdle. We do a risk-adjusted assessment of that. At the end of the day, what we're really trying to do is make sure that we get the sort of the total company value.
If you look at the plan we put forward in Investor Day at $14 billion in revenue and $38 in earnings, we don't think that there's anything more compelling to support that, and we thought the reaffirming actions would be clear there. The best thing we could do for our shareholders is drive earnings per share growth by putting the go-forward capital and the capacity to work. That's how we thought about it. I think it's a regular exercise at KLA. We think to get it valued, you have to be explicit about it, so we're very open, and we think that as you take a look at it, you can come away with a longer-term perspective to how you think about it.
Environments come and go, but we've done the exercise. We know why we have the amount of cash we have. We have it because it's the right amount given the dynamics in our business. You know, the leverage effects and all that kind of stuff kinda go into that overall assessment.
Awesome. On that excellent note, I'd like to close. Bren, thank you for all the insights, and good luck.
Absolutely. Thank you.
Thank you all.