Good morning. Welcome to day one of Citi Global TMT Conference. My name is Atif Malik. I cover U.S. Semiconductors, Semiconductor Equipment, and Networking Equipment Stocks here at Citi. It's my pleasure to welcome Bren Higgins, EVP, CFO KLA, as well as Oreste Donzella, Chief Strategy Officer. Now, Bren is going to open up with a few remarks, then I'll ask my questions first. I'll open it up to the audience questions a bit later, and if you have a question, please raise your hand, and the mic will come to you. Bren?
Sure. Thanks, Atif, for having us. Good to be here in New York. Just want to take a minute and introduce Oreste Donzella, who a lot of you know. Oreste is transitioning into this Chief Strategy role for KLA, from running our EPC Group, which is our More Than Moore part of KLA. One thing that's happened, and I know investors like to talk a lot about it, but we've certainly seen a transition for our customers, where the back end starts to look a lot like the front end, in terms of how our customers are managing their packaging activities. Obviously, there are parts of the end markets that are driving that, and it's an area of growth for the company.
But it's also driving organizations in terms of how we go to market, how we interface with customers, how customers want to interface with KLA. So as we're starting to see that transition and how it affects our specialty semiconductor business, which has a lot of exposure to advanced packaging from a process point of view, it made sense for us to do some reorganizing. So Oreste is an expert on KLA, a long-standing KLA leader, so brings a lot of insight to not only KLA products, technology, roadmap, customer process, and customers' customer. So I'll rely on Oreste a little bit today to provide some of his insights, and I'm happy to have him with us today.
In terms of just where things are, just level setting, some messages out of earnings is 2024 is, you know, a transition year with WFE levels in the mid-$90 billion range, which is, you know, maybe up a little bit from where we ended up last year. We think the company is performing at a pretty high level. We're excited in terms of corporate priorities around not only supporting customers, which is what we always do, but also driving the product roadmaps and then preparing for ramping of the leading-edge.
And so I think we've done a pretty good job over the last couple of years, where you've seen a lot of legacy and mature investment, which tends to be a little less Process Control intensive, but leading-edge tends to be the biggest driver of KLA's business. And we're seeing some strength in the second half of this year related to some of our customers' investments around, well, with Intel and with TSMC N3. And would expect as we move into next year to see a ramp in that investment. So that's a big driver. China's continued to be strong for us, and would expect that half to half overall, that's about flattish for us, and seems to be reasonably stable as we move forward into next year.
But most of, you know, certainly the growth coming from investments at the leading-edge. I talked about Logic Foundry, but also would expect to see, you know, a recovery in memory spend as we move into next year as well. So we think that overall the industry this year is mostly Foundry Logic, probably 70%-75%. Next year, you'll see some memory growth, which tends to be a little less Process Control intensive, but at the same time, we think we're pretty well-positioned in terms of inflecting products, particularly around the wafer inspection, where new supplies come online, and we think that will drive a continuation in relative performance for KLA in terms of share of market as we move into next year.
Our service business continues to grow solidly. We'll be up at the high end of our target range of 12%-14%. We've had 48 quarters in a row of year-over-year growth in our service business. Our service business is a little unique in the industry in terms of its subscription-like contracts. About 75%+ of the revenue stream is subscription. And we're seeing a lot of dynamics in terms of extendibility of platforms and useful or average life of tools increasing and total revenue capture on tools growing over time. And so all those are really good drivers for our service business as well. So profitability model is working, gross margins in the mid-60s. Would expect operating margins based on you know consensus views into the fourth quarter.
We talked about sequential growth, we talked about in March and reiterated again, in July, about sequential growth through this year, which should translate into high end of our target model in terms of incremental operating margin leverage on the revenue growth that we expect this year.
We raised the dividend after market yesterday, which is our 15th consecutive dividend increase, annual dividend increase, about 17% increase to $1.70 a share per quarter, which is just a continuation of not only our confidence in the business, our view of assertive capital allocation, and in a business like ours that has the service business I mentioned, but also generally asset light in terms of how we're able to run it, that we can pay out a meaningful dividend, that we can grow in line with our long-term growth rate expectations for earnings and cash of the company. So I think that's where we stand right now, and why don't I shift and happy to take your questions.
Yep, that's great. Bren, a lot of our investors try to simplify things by looking at end markets and associating a particular equipment maker with that end market and its growth. And so when we look at your outperformance this year, I feel like you guys have been, like, a silent outperformer. And can you walk us through the drivers in terms of end markets? You know, DRAM has been a strong growth here this year, when Foundry Logic was probably down from last year. Can you help us understand what has driven your outperformance? And as we look into second half, you have indicated that, you know, you're still going to outperform WFE, and also into next year. Like, what are the drivers of outperformance historically, and what... How do they look like?
Mm-hmm
... over the next six months to next year?
... Yeah, so this year, I think Logic Foundry overall is probably, you know, flattish for the year, although there is a switch between, you know, lower legacy investment and more leading-edge investment. And so that tends to be more Process Control intensive, as I said earlier, and so that's a good driver for the company. Most of that is N3-centric, but would expect to start to see some N2 investment as you move into particularly from our largest customer in the fourth quarter and into next year. So that certainly is an area of strength for the company and a driver, and you know, the end market drivers around high-performance computers is driving the incremental strength that we're seeing there.
On the memory side, I've been pretty consistent that I thought that this year was much more a year of device market device company improvement, right? That the pricing was improving, that would improve profitability and ultimately cash flow. So, you know, our customers' utilization rates were lower. They're able to utilize more capacity, now we're seeing utilization rates increase across all our customer segments. So that's the first thing you see. You'll see tech investment, too, in terms of incremental shrink investment. But that memory recovery is more of a bigger driver into next year. Yeah, as I said, China has been relatively consistent year to year, so I think that continues, you know, to be a, you know...
When I mention legacy, it's more the non-China legacy dropping off this year, given some of the corrections in industrial and in automotive. So, I think from the, you know, just a broader point of view, I think the leading-edge strength that we're seeing is probably the biggest driver and certainly the strongest part of our market. Process Control intensity tends to be very high in that part of the market. And so, that's good for us. And then I think the service strength, both from higher utilization rates, but also from a lot of tools coming off of warranty and going into contract that we shipped in 2021 and 2022, early 2023, is also driving. And we've had very good visibility to that, driving the incremental growth in service.
So I think that has been pretty predictable, as it generally is. And the leading-edge opportunities are encouraging, particularly what we're seeing now, but also as that moves into more N2-centric investments moving into next year.
Just staying on the leading-edge in 2-nanometer, I feel like, you know, historically, when there's a wavelength change in the industry, for example, EUV adoption, you guys, you know, kind of benefit, you know, along with that wave. But in the last, you know, three, four quarters, I would say, you know, you guys are starting to separate yourself from the lithography company in terms of, maybe you, you're shipping more equipment for 2-nanometer qualification, but it's not tracking as closely as it has in the obviously last couple of years in terms of lithography adoption, and you had Print Check and all that. Are there timing differences in terms of your equipment going into the 2-nanometer foundries right now for troubleshooting the Gate-All-Around and you guys being kind of early beneficiaries of spending?
Mm-hmm. Yeah, so I'll weigh in and then I'll let Oreste add as we go here. But it doesn't always line up perfectly, right? I think one of the drivers that we've seen in terms of incremental growth this year has been, frankly, adding additional supply around wafer inspection, which has been a nice driver for us this year, where we just had a lot of pent-up demand. So that's been a factor.
Most of what we've seen so far has been more incremental. N3, of course, packaging has been a big driver, as I mentioned earlier, where we've actually upticked our business expectations for advanced packaging from about $300 million earlier in the year to about $500 million, an expectation of, you know, I'll call it, you know, a multiple point, sort of long-term growth rate expectation versus WFE, over the long run. We'll see N2 pilot line investment that will happen towards the end of this year. But the bulk of the N2 investment will happen as we move into next year.
So typically, early in a development cycle, they buy the—you know, customers manage and sample very tightly, manage their process, and so they buy the fleet of systems that we, as a portfolio company, we offer, you know, a comprehensive set of solutions across inspection and across metrology. So customers will buy everything where they're, you know, really trying to qualify the process. So we tend to participate at higher sampling levels earlier in the process, and then as they go to then introduce the element of volume and ramp, then we also participate pretty heavily. As they go into production and mature their processes, they tend to historically have invested less, but we're seeing that start to change in that, because of the design environment, there is less reuse of old capacity, so that tends to be a driver.
And you've got technical drivers of more advanced design rules requiring more capability. So that happens. And then they're also managing a very robust design environment, and those designs can test design rules in different ways. And so, you tend to see higher sampling rates in production today. So while we're absolutely leveraged to kind of development and ramp, we are seeing our participation in volume production with higher level systems, more capable systems, increasing over time. N2 is unique in that you have a transistor architecture change, and so that will drive new defect mechanisms. Lithography could be relatively constant in terms of EUV layers, but those are complicated layers, and smaller defects are continuing to be a challenge for our customers.
But the architecture change will drive different materials, different processing steps, and that will create new defect mechanisms, certainly new metrology requirements, which should be an incremental driver for the architecture. And if you go back to the last local high that we achieved in terms of Process Control intensity, and the FinFET transition was pretty high as well. So we think that there's a pattern recognition relative to what's expected at N2, and early indications are pretty positive.
Great. Um-
Yeah.
Oreste, let me come back to you. Yeah.
Let me chime in on this, because this is an important observation between KLA and the process equipment companies. So when you have a big inflection, like for example, a change in wavelength going to EUV, the photolithography company leverages this inflection. When you have a Gate-All-Around, probably we are not scaling lithography almost at all, and who is going to leverage the most is the process tool, the process tools like deposition and etch. KLA can leverage them all. Because we leverage either if the inflection is in lithography, the inflection is a transistor architecture, the inflection is in backside power distribution network or even packaging. So the beauty of KLA, probably the diversification of KLA versus all the other companies in the world supplying equipment for semiconductor, is how much we are exposed to every single inflection in the semiconductor industry.
That is quite unique, and that is why maybe you see a mismatch between lithography and KLA, but now probably we are developing tools and selling tools not for lithography in the 2-nanometer, but for Gate-All-Around, and then you see KLA participates in all.
Yeah, that makes a lot of sense. Oreste, just sticking with you, first of all, congratulations on your new role. It makes a lot of sense. You've seen both sides, EPC and like Bren pointed out, back end is a new front end. There's definitely a re-rating going on in terms of packaging. And when we talk to some of your peers, the small cap metrology inspection companies, they, you know, they have a view that there are a lot of opportunities on the back end. And so can you just help us understand where are you going to be focused, near term, longer term, in terms of taking advantage of some of these, you know, challenges on the back end using your front-end portfolio?
Absolutely. So, let me say, first of all, that we are very happy, as Bren has said, to see now we estimate the revenue of this year for packaging business to be pretty like $500 million plus. And this is a big, big step up from last year, but also is a big step up where we started. So if you look at five years ago, before the Orbotech SPTS acquisition, our presence in the packaging market was only limited to component inspection. It was probably in the $100 million range. So in five years, you can say 5x this number.
And the reason is because of the complexity that the packaging technology roadmap is bringing to really this heterogeneous integration, 2.5D or 3D integration on multiple devices, and also the way how some other packaging architectures, like fan-out, is shrinking the critical geometry. So, this is mostly driven, of course, by AI and data. So we are in the AI and data years right now, and everybody's trying to accelerate the scaling of the semiconductor roadmap, and front end can bring you to a certain level. I mean, people are telling me, I've been in this industry for 32 years, and we are telling, is Moore's Law dead? Of course, it's not.
Whenever we can go to the next transistor architecture to deliver the power from the backside of the wafer, or going to the next level of NA in the EUV lithography, we are scaling the technology based on Moore's Law, but it's not enough anymore, so packaging is complementing the advancement in semiconductor roadmap in a very, very cost-effective way. I explained one time, I said, "We are moving from SoC System-on-a-Chip to SoC System of Chips." Because everything in advanced packaging is starting from disaggregation of the functions that in the past were developed on a single chip. Now, it's not cost-effective anymore, so you disaggregate the function, and then you aggregate the function in packaging, through either 2.5D, 2D, or 3D heterogeneous integration.
This is adding a huge complexity to the packaging industry that the customers in packaging couldn't really bear in the past. Without the investment of the top semiconductor companies in packaging, the packaging roadmap couldn't exist. This is how we see the TSMC of the world, Intel, Samsung, and the other memories heavily investing in packaging, because we need this kind of complexity. We need this kind of budget in packaging now that the old OSATs couldn't really afford. Where do we play? We play everywhere that there is complexity. I'll give you an example. When you look at the 2.5D CoWoS, for example, now, of course, almost all the GPU are made on CoWoS, that is based on silicon interposer, that provides a very stable platform for the further integration of the chips.
On the silicon interposer is an extremely, extremely precious part of the final package of a GPU. In this case, you need the most advanced inspection capability and highest possible sampling. This is important because generally, we have not seen 100% sampling on inspection. That means every single unit is inspected since a long, long time, but now we see in the silicon interposer people are, because this part of the package is so expensive, so important that our customer need to be 100% sure there is no defect. Okay? We see every time there is a shrink in geometry or complexity in integration of the packages, we see KLA participating. With what? With the semiconductor front-end equipment, because we have inspection metrology tools already serving the front-end wafer fabs.
We need to customize them because there is no, like, 100% transition from a tool in front-end to package, so we need to do some customization. But at the end of the day, you get the basic technology that works in front-end at three nanometer, and then you customize it for the most aggressive challenges in packaging. So another example is hybrid bonding, 3D hybrid bonding. So when you stack die on top of each other without bumps, so the complexity of inspection metrology and even process, because with SPTS now we also have a process division, we see opportunities for us to provide unique value.
For example, other area where we don't participate is bump inspection, because we believe bump inspection is not high-end and is very well served anyway, is not high-end enough for us to provide unique value. Remember, KLA always search for opportunities where we can be highly differentiated. Otherwise, if the market is already served by other people and we don't provide unique value, there is no point for us to enter. And actually, going back to my new role, that's exactly where I'm going to focus the most of my time, thinking about what are the strategic opportunities for us to provide extra growth to the company and have Bren, Rick, and my colleagues in the executive staff focus on this.
I understand it's early days, but have you done an exercise in terms of incremental TAM that the back end can open up for you, given your leadership on the front end?
As you know, I mean, putting numbers is not easy all the time. But I would expect, for example, we see this year a continued demand of tools in terms of our bookings is very, very strong in packaging. So I expect, for example, next year to have other big growth in the packaging because the bookings is, are very, very solid this year. For example, hybrid bonding, I cannot quantify to you, but hybrid bonding is right now only pretty much on a couple of devices. But when hybrid bonding is going to be established, so you will see more and more inspection metrology from our tool, more and more plasma dicing and silicon etching from SPTS, and the market is going to open up even bigger.
Yeah, that's some work that we have to do. I mean, we expect the market to grow faster. And to Oreste's earlier point, you know, as you shrink the lines and spaces, so the pitch starts to shrink in these connecting layers, it drives the need for more capability. And there's always a trade-off in our business between the sensitivity of a system and the speed in which it runs. So if you're running it at very high sampling rates, then there's a driver for more capability, which tends to be a higher-end system, tends to carry higher prices, higher margins. But also to maintain the same sampling rate, then you need more systems.
So, you know, right now, our quantification has been more about just, you know, where we see the trends today in terms of overall momentum and our views of the long-term growth rate. But I think there's more work we can do here.
Great. When a segment of memory, we didn't talk about NAND, which has been fairly dormant for the last few quarters. I mean, what are your expectations in terms of NAND recovery?
You're right. It's very low, right? Very low, you know, mid-single digit type WFE levels this year, and so I would expect we'll start, you know, as you move into next year off the low base and the improvement, I think, in the business and financial fundamentals of those companies that we'll see more investment into next year. Off a low base, but still, I think that the mix of investment next year is probably more DRAM-weighted, or incremental investment, but I think we'd expect NAND to grow as well.
And then on China, it was more than 40% for the first few quarters of the year, and you're expecting it to come down in the back of this year. Can you just talk about what you're seeing in China and how your, you know, China customers are looking at investments, second half into next year?
Yeah, it's been pretty broad-based and stable. We made some comments about, you know, half to half, and I've been looking at China in halves generally because you have a lot of greenfield projects, and so greenfield projects can be impacted by construction issues and so on. So, you know, trying to identify exactly which quarter that you'll revenue, particularly if you have a new customer and if you look at a revenue recognition policy with new customers, you know, we tend to have a longer cycle of acceptance for rev rec. So half to half, roughly on an absolute value basis, about even. I think that as you look at right now, it feels like Q3 is probably a little bit stronger, and then it drops off into Q4. But we'll see how that actually plays out.
So what's driving the percent coming down is growth from our non-China customers in the second half of this year. And then I think as we move into next year, I expect it to be. I'm using the word stable. It you know could be up a little bit, it could be down a little bit. I don't think it falls off all that much. So likely then, given expectations of growth from other non-China customers, that we'd see the percentage drop into the you know high twenties or so. Again, across a number of legacy projects around the country, and you know you're seeing just you know. Again, back to my earlier comment about how customers buy typically, that they'll you know buy the whole portfolio.
They'll work a process, they'll try to do R&D, try to be credible in demonstrating that ability to customers. And so you see this sort of continual sort of cycle of investment and maybe smaller increments versus some of our process peers, which it could be more lumpy, big investments, and then a back off. That provides, I think, some consistency to our business expectations from China. We also have wafer and reticle infrastructure investments for wafer houses where they're manufacturing wafers or manufacturing reticles in the country. And so those were choke points in terms of supply back in the 2021, 2022 , and so there's been domestic investment in those areas, and that's an area we participate that a lot of our process peers do not. So you've got a few vectors that are there that are driving that outlook.
Great.
Yeah.
Let me stop here to see if there are any questions in the audience. If you have a question, please raise your hand. It's one in that seat.
I'm just curious, should tax rates go up significantly in a new administration, would that affect your capital allocation policies in any material way?
I don't think so. I mean, certainly, you know, you may be paying higher taxes, but I think in terms of how we think about allocation of capital, it's about what's required to execute our business and have, you know, in terms of the process we run through, in terms of working capital requirements, strategic capital, and so on. Our philosophy is around allocating every dollar productively. We don't have excess cash at KLA in terms of how we think about it. So we'll continue to be assertive in that way. So, you know, I'd have to understand all the details of the ins and outs, but generally, our policy should be consistent. The fundamentals of our business are not changing in that environment, and so we'll continue to deploy capital aggressively.
We like the fact that, look, our business generates cash that we don't have to reinvest, and so we'll put every dollar to work.
Hi, thank you so much. A question regarding, you said that N2 is in the pilot phase. Just wondering qualitatively or quantitatively, how that this process compares to your prior experience at the prior nodes at N3 and N5 and N7? Just trying to get a gauge of the magnitude, the materiality, just any insight to call it out.
Yeah, so we're very early. I mean, the general sense from our customer is, you know, our primary customer is a lot of customer interest and an expectation, perhaps, of a faster ramp over the first few years of a node. And so that's typically how we look at it. So, you know, I don't want to comment on progress yet, but at the same time, our expectations are pretty strong, and we've seen those expectations increase over the course of this year in terms of the pace of that ramp and more products driving it, where it isn't necessarily a you know, a mobile device that debugs the process, but you could have multiple devices that are driving the node earlier in the node ramp, which can create opportunities from a yield point of view, which translates into opportunities for KLA.
Hi, apologies if you already answered this, please discard. I was about five minutes late because it took me forever to get an elevator. But, there's been, you know, some of the foundries recently have announced various, like, CapEx cuts or at least intention to maybe shift how they're doing their spending. How would you expect that to impact your business?
You know, most of what I've said today, you know, we stay very close to our customers, and so, you know, I think it'll—I think by the time sometimes things make announcement, announcements happen, we, you know, we tend to, I think, because of the engagement, have a pretty good view of our expectations, given our lead times. So, I don't have any changes to the outlook I provided today. I haven't quantified, you know, next year's WFE levels other than our expectations of growth with the leading-edge, so...
Let me just go back to the advanced packaging side. You mentioned bump inspection or metrology is probably not something that you think is high value. Can you just maybe help us understand what you consider higher value? Is it like the RDL layer?
Yeah.
And maybe why that is more high value than, say, measuring the bumps?
First of all, the bump inspection is a very, very important application, no question about it, especially in memory. When you do HBM, all the HBM stacking now is through micro bumps in the thermal compression bonding. But the bump's geometry is not scaling to the extent where we can produce unique value. So today, the most advanced bump pitch size is around the 25 micron, probably. And this is not even today, it's the next generation, HBM4, for example, for DRAM. So in this range of sizing, we believe the market is already well-served. When you look at what is happening in silicon interposer, silicon interposer is providing interconnect at the size of 0.5 micron, 0.4 micron.
So you have a 0.4-micron line of space in interposer, where the challenge to find very, very small defects are way, way higher than the bump inspection to measure coplanarity or bump defects. So this is a classic example. The fan-out wafer level packaging, for example, is moving into 2-micron line of space or even 1.4, 1.4. The heterogeneous integration also is providing us opportunities to differentiate not only in defect inspection, but also in metrology. I'll give you an example. Whenever you do die-to-wafer bonding, in hybrid bonding, you need to make sure that you place the die in the correct pad size, because otherwise you have misalignment of the chip on the pad, and they create, of course, a short or other electrical failures.
This is an opportunity for us to leverage our overlay metrology from front end to become really the tool to use in packaging as well. That's what I feel. Whatever is inspection or metrology application is already established, and we don't see technology roadmap in terms of sizing or complexity of the process integration, we prefer not to participate. We prefer to participate in other areas where we can provide extra value. Did I answer your question?
Atif, if you had a question earlier, you were about patterning versus inspection, just in terms of just performance overall. And I wanted to make sure I covered that. I mean, certainly, you know, what we've seen, our inspection business has had more strength over the course of this year. And understanding how we break up the business, I think it's important to understand some of the drivers. Inspection, when you look at patterning, you have patterning tends to be metrology. It tends to be film metrology, critical dimension metrology, overlay metrology, and reticle inspection. So the metrology tools tend to scale with process tools because you need to measure the films that are being deposited or etched away by our customers. So it's a business that tends to carry a little bit more cyclicality.
It ramps and grows very fast. It's also a shorter lead time business in expansion environments, and then it corrects a little bit more in more contraction or stable environments, and so inspection has been a bigger driver over the course of this year, and I would expect to start to see the patterning part as it relates to metrology grow more into next year. Reticle inspection is also part of that reporting and covers from the mask shop, where you manufacture reticles all the way to the fab, where you're validating quality and doing quality control steps, and different products serve different parts of that market.
Parts of that market can have very high integers in terms of the ASPs of the system, so it can create some lumpiness. In the EUV world, you most of reticles in EUV are moving through our Teron optical system, but as you haven't seen a lot of expansion so far, it's been a little bit slower part of the business. You know, you're seeing plans around the fabs, and that should drive more of the fab investment, both for Print Check but also reticle systems that are for the fab. There's also part of the market that's what we call the pathfinding part of the market, which is what, well, customers will ultimately need a higher-end capability for inspection post-patterning, but after you write the pattern in the mask shop.
And so we've talked a lot about that market. We call it the reticle market, which is an EUV-based inspector, which we are targeting to intercept the market. We've been investing for a long time to intercept the market opportunity there, which we don't think really becomes a production requirement until you get to High-NA. So we've been investing there. We've recently you know solidified a you know a formal relationship with Carl Zeiss, who's a strategic supplier of KLA in the wafer inspection side for many decades, but also now to support our reticle efforts moving forward. In terms of aligning the roadmap for inspection and the NA requirements within lithography and having sort of a synergistic alignment between what's going on in terms of litho roadmap relative to the inspection roadmap.
So, you know, this is a more formal relationship than what we've had, and we're looking forward to really leverage their clear expertise in optics as the supplier of optics for lithography, and also our inspection, or our expertise around the inspection. We think that that roadmap will provide our customers, you know, a clear plan and view, not just through High-NA, but also Hyper-NA and beyond. We think that, you know, that there's an opportunity in this market. We believe it happens with High-NA, which is likely, you know, in and that feels like it's shifting a bit in maybe 2027-2028 in terms of high-volume production.
You know, the ability to ensure that we have a differentiated system that can solve, you know, our customers' problems.
Yeah, that was super helpful, putting the context on metrology as we move to Gate-All-Around, as well as on a Teron inspection longer term. A couple of areas we didn't talk about services. What's driving it towards the high end of the 12%-14% growth, how sustainable that is?
I think it's pretty sustainable. There's a couple things that drive. I mentioned utilization rates improving, that's a driver. Tools coming off a warranty is another driver. The average life of a system, if you go back to 20 y ears ago, five-ish years, right? 50% of the install base today has an average life of over 20 . We're seeing, as you see more demand in the legacy nodes, that our tools are living longer and our customers are continuing to use them. There are a number of tools we shipped in the nineties that are still in service, and so that creates a unique opportunity for us to capture more by providing that service expertise, to capture more value.
If you would've asked me when I first became CFO back in 2013, 2014, how much of an ASP value of a system we capture in service stream, I would've said 40%-50%. Today, it's over 100%. So it shows. It gives you a sense of as the lifetime of the tools is increasing and the percent a customer pays as a percent of ASP is increasing, that it's creating this very strong revenue stream. Because we have contracts, we can optimize the cost structure underneath to deliver to the entitlement that we establish with our customers, and so that helps us be pretty efficient in terms of how we deliver. And so, you know, over time, got the question earlier about capital structure.
Over time, we thought, you know, when you have a service business like ours, it can give you the confidence to very predictably pay a consistent dividend over time and be pretty explicit about it, given the fact that it's had one down year in 2024 or 2025. But now you can look at it and say, it can pay a dividend, it can pay the debt service of the company, too. So in terms of the profitability, now accounting can be a little bit arbitrary in terms of how you do accounting, but in terms of how we look at it, that's generally a factor in how we think about some of those decisions.
So I think that these drivers are pretty consistent here moving forward, and, you know, we don't have anything to announce in terms of a long-term growth rate, but we think that we're trending towards the higher end. And if we can continue to deliver value to our customers here, I think that there's opportunities for us to continue to drive that growth. We're also trying to then take some of the capability to some of the other acquired businesses, and that will take longer. We can leverage infrastructure first. Go-to-markets can always be a little bit different, but, you know, to try to bring. I think one of the challenges of small companies always have a hard time of optimizing service, 'cause you just don't have the infrastructure for it.
So as we have acquired companies, we've been able to bring our service expertise and infrastructure to play, and it's, I think it's another factor of growth moving forward.
Great. We'll wrap it up here, Brian and Oreste. Really appreciate your insights today.
Thank you.
Thank you for having us. Thank you.