KLA Corporation (KLAC) Earnings Call Transcript & Summary

Great. Let's go ahead and get started. So I'm Joe Quatrochi, the semi cap analyst at Wells Fargo. Today, we're pleased to feature KLA, CFO, Bren Higgins, as well as the Head of Global -- I'm sorry, Head of Global Support and Services, Brian Lorig as well. Maybe first, Bren will tell a little bit of news out over the last 24 hours. Can you just -- can we knock out this China question and any sort of there any comments?

Yes, a little bit. Yes, so happy to talk about this a little bit. I just wanted to thank Brian for being here. So Brian runs our service organization. And I think sometimes the service business of KLA is not as understood as I'd like it to be with investors. I think it's a unique business. It's a huge value contributor to the company. And hopefully, as we move forward today, Brian will share some insights about how that business is growing, some of the drivers of that and how service overall is becoming more and more strategic to our customers as they support what is what we believe a pretty compelling leading edge road map, but also dynamics within the mainstream, I'll call it, part of the business, but also our mainstream volume but also legacy. So as it relates to China, so yesterday we received some new export controls. They've been talked about for a while. So it created some noise in the market over the last 12 months or so. We're still digesting the document. It's about 200 pages, and it's incredibly complex as we work our way through it. We did an 8-K this morning that reiterated our guidance for the December quarter. So we feel that with all the moving parts in the December quarter, we're still -- we're comfortable with our guidance that we provided. As it relates to a broader picture into 2025, we've been looking at it. And our preliminary estimates is that in 2025, it probably has an incremental impact of somewhere in the neighborhood of about $500 million, plus or minus, I'll call it, $100 million. Now that's preliminary and is based on not only backlog and some expectations from customers in the near term, but also forecast dynamics as you move into the second half. So there's some fluidity in terms of how you think about that. About 70% of that would be systems, roughly 30% of it would be service. If you look at our overall China business for 2025, that would imply our overall China business is down somewhere in the low 20 percentile or so as we look at it. So back at earnings, we saw it somewhere down between 15% and 20%. Now I think it's somewhere in the low 20th percentile inclusive of these actions. Now we'll see how licensing goes. There's always clarifications of what the intent is and whether we can get licenses and to still support these customers in line with the intent of the regulations. The licensing process has been somewhat inefficient over the last couple of years here in the U.S. And so we'll have to see how that goes. I'm not planning for efficiency there, but if we do have a more efficient approach and that creates some additional opportunities to mitigate some of that exposure, that would be a positive. There's also some systems that we sell that have long lead times where we've had supply constraints. And so the ability of perhaps some of that business that might be out into '26 pulling into '25 could also be a factor that would mitigate that as well. But overall, that's more or less how we see the impact, and we'll have more to say about it at earnings, we'll provide an update. But I think in terms of where we're sitting today, our preliminary assessment is in line with that. So as we look at 2025, though, we're pretty excited about the opportunities that exist with investments at the leading edge. The N2 ramp is compelling and is driving our business into next year. But not just as it relates to logic, but also in high-bandwidth memory, which is driving memory intensity and also advanced packaging, which is also a good driver for the company as well. We've seen that inflect from what was about $300 million in business in 2023. That looks like it's in excess of $500 million in 2024, it's probably north of $750 million in 2025. So we're pretty excited about that opportunity. I think the business is pretty concentrated around those customers that are exposed high-performance compute and the N2 node. But we're pretty excited about what's in front of us there. Over the long run, if you look at the company's plans, and we've always expected that the industry would rationalize and normalize over time, and then were some unique factors, I think, that drove China to an elevated percent over the last couple of years, driven by supply constraints back in 2022 and then some of that business pulling in and raising the level of the last couple of years. So we've already I think, been fairly cautious about expecting some normalization of that spend moving forward. And as you look at the long-term plan of the company, our long-term view is that semiconductor revenue is really end market driven, not capacity-driven, but end market-driven, and that's going to drive a growth rate in semiconductor revenue of somewhere in excess of 7%. We think capital intensity from a planning point of view, likely is somewhere that's ahead. I mean, I think capital intensity is growing. So capital intensity climbs, I think, slowly. So it enables WFE to grow modestly faster than semi revenue. Turns out that there's more inefficiency and capital intensity rises faster, I think we've proven in the company that we can ramp our business to support that. But that's generally how we see it. I think given the growth rate in logic and foundry, we would expect the mix of that WFE to be more logic and foundry-centric. That tends to be good for process control. This process control intensity is higher in logic and foundry. If you look at within process control, the fastest-growing markets in process control our markets that KLA has a very strong position in. And so I think that creates an opportunity for KLA to grow even faster than that. So we think that, that, coupled with service, which is growing at 12% to 14%, which we'll talk about more. It's a pretty compelling story over the long run. So I think what's happened recently doesn't affect our long-term view. I think we expected that you can only invest ahead of revenue and then there were some unique dynamics associated with that investment for so long. And that, over time, you would expect some normalization of it. So I wanted to provide some view of how we're seeing the long term because I think it's driving not only how we're investing in the company, we're sizing, but also thinking about the business model and the capital return/capital allocation strategy underneath it. I think I'll stop there and...

You give me a lot to digest there. Maybe just -- maybe -- can we revisit some of the stuff that you said about China, just I want to make sure I understand it correctly here. So you're saying the preliminary estimate from what happened yesterday is $500 million, plus or minus $100 million for 2025 just looking at like kind of your backlog that split 70-30 systems versus services. And then can you repeat that one in terms of how you're thinking about the percentage of revenue from China next year? Because I think in the earnings call, you said -- I want to -- you said 20%, right? Now you're thinking low...

That's the down decline year-to-year...

30% of total.

So the decline year-to-year is low [ 20th percentile ] from '20, the decline in '25 versus '24. If you go do the math on the mix of business, we're going to go from roughly 40% [indiscernible] this year yet, and there was some impact this year of the restrictions, although there's a lot of moving parts in the quarter, which is why we are comfortable with the guidance that -- from around 40% to somewhere, I think, probably in the high 20%. So before we [ thought it was ] about 30. Now I think it's somewhere in the high 20s. We said sort of low 30-ish percent or somewhere in the high 20.

Okay. Okay. That's clear. Maybe just a couple of more questions in terms of just kind of the bigger picture and then I promise we'll dive in on the services. You talked about 2-nanometer and thinking about just the positioning there. Can you talk about like the increase in capital intensity for 2-nanometer for process control, specifically around the gate-all-around? Like how do we think about sample rates? Is it too early still to understand what that looks like? But just kind of how do we think about that? And then also I'll sneak one more in there. You're also positioned in the mask shop to benefit from 2-nanometer. So kind of can we dissect that a little bit?

Sure. So 2-nanometer, the biggest change in 2-nanometer is an architecture change, going to gate-all-around architecture from a FinFET architecture. And architecture changes are good for process control, good for KLA. If you go back to when the industry transitioned from planar devices to FinFET that, that was the last local high in process control intensity. So we're very encouraged by what we're seeing there. If you look at our engagement with our primary customer there, we're seeing the process control intensity higher, and we're encouraged by that. I think KLA share of WFE higher. And so I think that's reflective of the challenges associated with the architecture change. The litho scaling is -- or the number of litho layers is relatively flat, but still driving significantly challenging layers and challenging defects that we're going trying to debug and to help our customers. So I think overall, from an architecture point of view, and in node point of view, we're pretty compelled by what we see. Our customers are excited about the number of design starts. And in a similar state of the ramp versus 3-nanometer, there are more designs. And high-performance computing is driving some of that design activity. More designs driving leading-edge process integration is good because they're managing a much more fluid environment and these designs, test design rules in different ways. So that tends to be good also for process control versus maybe one design driving the debugging process of a new node. So the other factor is the 3-nanometer node has also been fairly tight in terms of capacity. We've seen some incremental business this year from 3-nanometer. And so that makes it very hard for our customers to take capacity and try to reuse it, to try to be efficient with it as they go to think about the 2-nanometer ramp. So I think all those factors are driving a pretty compelling environment around process control intensity and KLA share of the opportunity there. So that's 2, and then, of course, what's driving the biggest port of memory is high bandwidth memory. And high-bandwidth memory carries higher process control intensity relative to baseline DRAM because of the nature of the devices, the dies are bigger, drives more capacity, there's logic circuitry that adds the need for more sampling. Your stacking dies or higher reliability challenges. And then you have to then take those die and integrate them with the GPU and the interposer substrate, to drive the advanced package and package now carries a tremendous amount of value -- there's complexity in how it is all linked together in terms of the interconnects. And so all of that tends to drive opportunities not only in terms of the packaging opportunity with it, but also the value of the device and failing one of these devices because of some defectivity issue starts to be pretty costly for our customers. And so that translates into higher sampling rates. And more complexity moving into these packages over time, creates a need for more technology in the package inspection environment as well. So I think there are a number of factors that are driving this 2-nanometer node. And the memory -- the leading edge memory and packaging, which are big drivers for our business into next year. There was another question that reticle... So reticle inspection is not really changing node to node. It tends to drive like a process -- it behaves very similar to a process tool and that's really driven by designs. Every design has a reticle set. So as you have the need for incremental capacity to produce those reticles, then that drives the reticle inspection business. KLA tends to get in the mask shop as you're writing those reticles. Most of those EUV reticles actually flow through KLA systems, probably greater than 90% of them. And then you have to then monitor and requalify those reticles in a fab environment. And so there's multiple products that KLA sells that also addresses that part of the market where there are specific inspections that happen after a certain number of exposures to ensure that the reticles are -- there's been no contamination or some other defectivity that's materialized in the production process. So it's part of the overall view of strength of N2. I think it's -- reticle fidelity is also important. The pitch sizes are not shrinking, which is driving higher sensitivity requirements, but the overall demand environment from a capacity point of view is pretty strong. And I would expect that part of the business to also grow probably faster than overall market into next year.

And you think about like I think in the past, like just one more question on the reticle space. In the past, like you've seen where there's been some level of reuse, just given that they haven't seen that trickle down like products moving from 7 to 5, et cetera, et cetera. Like do you -- as you get back into 2, do you see that kind of dynamic where they're going to have to -- like your customers need to add a lot of capacity. Because I mean clearly talking about a lot of good design starts or reacceleration designs starts at 2-nanometers. Like do we think about the reuse at 2 as increasing relative to prior nodes? Or how do you think about that?

No, I think it's less. We tend to see less on process control generally anyway because there isn't redundancy in process control as much. And so our customers tend to monitor processes for yield because they only want to start so many wafer starts to address the demand that is out there. So we usually don't see a much, but typically, the big driver for the size of the ramp is how robust is the design environment from the more mainstream designs as they follow the leading edge design into [ a node ] right? And I think it's early, but the expectation that the N2 node in the first few years of the node, it's likely to be probably the largest node in terms of design starts. Now we'll see how that plays out. But if you just go back, was 28-nanometer than 7-nanometer was -- had more design starts in the first few years than 28. And I think the potential for N2 is that it could be bigger than 7. So a lot of excitement about it. N2 is a very power-friendly node. So if you're a data center and paying your own power bill, the migration to N2, even if it is perhaps a more costly node, the benefits from power consumption can be a huge driver. And I think that's why we're seeing not only the competitive environment as it relates to high-performance compute driving faster product cycles. But N2, in particular, is more attractive because of its power-friendly architecture.

Okay. Maybe let's shift gears over to services. Maybe just to start, Brian, like can you give us kind of just like A little bit of an overview of the services business, like what the portfolio looks like in terms of the offerings and why services is important to your customers and then any sort of metrics you're willing to kind of share in terms of just some of the KPIs that you focus on?

Yes. Thanks for having us, Joe. And excited to talk a little bit about the services business. So our service business makes up about 25% of overall KLA revenue, and it's a very durable revenue stream, as evidenced by our 49th consecutive quarter of growth on a year-over-year basis. And that durability is really a function of a few things. One, it's a customer-focused organization provides a lot of value to our customers. And so there -- we service a lot of different customers, 1,400 or so facilities across the world, so it's fairly diversified in that regard. We also have a very large and growing installed base of more than 50,000 tools. And finally, about 75% of our service revenue is on subscription contracts, long-term subscription contracts. And those renew at about a 95% rate. So it's a business that provides a lot of value to our customers and in turn, is a really good business for KLA.

Maybe like double clicking on that. Your services relative to some of your peers, right, that's a little bit more like brake fix, like talk about like the things that you do for your customers relative to that, that drives that kind of stability in terms of the contract, 75% on contract?

Yes. And it is industry-leading 75%, and so it is a differentiated service model. Ultimately, when our customers select a KLA product, they're selecting it based on sensitivity, the ability for the product to find or measure the defect and then also how fast can it do that throughput. And so those 2 things from a service perspective, we call performance. And then they want that performance at a very high availability. Those tools have to be reliably and predictably available. And so we call that availability at performance. And ultimately, that is the outcome that we sell with our service business. And as you look at the complexity of the products that we're introducing, the only way that you can achieve the ROI on that investment that you're making in a KLA system is to have that availability of performance at a very high level. And the only way to do that is through a KLA service contract, comprehensive service contract. The other thing that's a little bit different with respect to the KLA service businesses, is related to our installed base makeup. And we have a high mix, high complex, low redundancy installed base. And by that, I mean, if you go into a customer fab, you'll see a lot of KLA tools in there but there are a lot of different product types inside of those number of tools. And so that's that high mix. It also leads to lower redundancy. So we're a little bit less susceptible to some of these changes that you see in utilization rates because yield always pays. And so even when you see utilization rates decline in a fab, customers still want to leverage their inspection metrology equipment, they might just sample higher. So they're not idling any of that equipment and that creates, again, a lot less vulnerability to large utilization fluctuations. The last thing is we have a very pure service model. We don't have a lot of capital or any capital related investment in our service revenue profile it's more around operating budgets. And so when you see WFE fluctuations, it doesn't impact our service business in the same way because we don't have refurbished equipment or upgrades in our model. It's more around servicing the existing installed base. So that purity leads to a lot more stability, and that leads to this growth year-over-year. Again, as I said, 49 consecutive quarters of growth on a year-over-year basis. And those are certainly some of the factors that drive it.

It grows every year, right? I think we've had one down year in the last 25 years, and that was in 2009. And it was down about 10% or 12%, we'll call it low double digits. So because of those factors, you get the predictability of the contract stream at a pretty high ratio and you get it consistently with growth year-over-year. And of course, in downturns, you do see some idling and utilization rates drop. So that tends to drop the growth rate down below as we saw in 2023 below our long-term target. And then you end up in a year like this year, where you're seeing that tighten and new tool that are coming off of warranty that we shipped over the last couple of years that go into contract with a very high attach rate. And so that tends to drive growth that's more in line or even above the long-term target. So it's -- as Brian said, it's a high mix business and it isn't just about fixing tools, right? We're part -- spread. There's a lot of services that we provide in terms of preventative maintenance to ensure that the tools can deliver the performance and deliver performance across the fleet in a particular fab at very high uptimes, right, 90% plus uptimes. And it's one of the reasons that as our customers buy our products, they run them at high levels, they expect a lot of information from them. And it makes it because of the complexity of it, very hard to, and frankly, a very low ROI to go and do self-service. And so they rely on us to keep the systems up to drive that availability of performance. And so it behaves very uniquely, I think, relative to some of the other service businesses that are out there for those reasons.

Yes. I mean it's very impressive that -- Bren, I think you said 49 quarters of year-over-year growth, and you still also had to comp like export restrictions, right? And over the last -- right? And so it definitely speaks to the consistency of that business.

Yes. So you get -- you have some headwinds there, right? And you also get some headwinds sometimes from foreign exchange. Your costs are denominated in local currencies. But given the strength of the dollar over the last couple of years, it's been a bit of a headwind from a revenue point of view. So you do have those dynamics. Obviously, we invested a lot of inventory to support these very long lives of these systems and the custom nature of the parts. So we're optimizing for the business dynamics that we just talked about, which are very compelling. And we accepted that we're going to have to make some investments in working capital to be able to support the business over time.

Okay. Thinking about -- you guys have talked about like a services venue growth, I think the CAGR that you gave at the last Analyst Day was 12% to 14%. Well, I think you're going to have another Analyst Day here next year, you're operating above it. I mean talk about like maybe what's been the surprise that's driven the upside to that growth rate? Is it the installed base or just the service intensity? Like how do you think about, I guess, like that dynamic of what's driving the upside?

Yes. I think a few things. I mean, the biggest driver, of course, is new installed base growth. And that certainly, the KLA lineup of products has been very strong even in these nonhyper growth WFE years, we've still continued to drive market share and intensity improvement. That's driven new products, which has driven our new installed base, which has driven new opportunities for service. And again, that intensity is high at the leading edge for sure. The other thing is existing installed base growth. And so unlike a lot of companies, when you introduce a new product, oftentimes it cannibalizes the existing products. That is not the case with inspection and metrology equipment. When we introduced new equipment, it's incremental installed base growth. Our existing installed base, there are tools that have been out there for 10 years, 20 years, even 30 years and customers routinely ask for the tools that are 30 years old to extend those for another 10 years. So you've got a very, very robust existing installed base, which allows us to drive value-added services there because maintaining some of that equipment that's that old, it's a very big challenge in terms of maintaining supply chain and driving obsolescence projects, engineering projects to support it. But it's long been depreciated assets for our customers. So those are important and still very much providing value to them in their production environment. So certainly, new installed base growth, coupled with a very stable and opportunity to drive additional value-added services to that existing installed base has really helped drive the growth rate. And as you said, in 2022, we increased the growth rate from -- we were at 9% to 11% in our 2022 Analyst Day. We increased to 12% to 14%. And the recently completed September quarter, we were at about 15%. So as you said, top end of that range. We're not updating that model today. As you said, we've got an Analyst Day coming. But when we look at the fundamentals of the business and long range, we still feel very good about the growth prospects.

I think one other thing is there's an opportunity for us, and we've seen a little bit of it. But just when you acquire companies, and we've had some -- we did the Orbotech transaction, but we've also done some other transactions over time. What you find is a lot of companies, particularly small ones struggle to monetize service. We certainly optimize it. And they don't have the infrastructure necessarily to support a global footprint. And so one of the drivers for us, we think both near term and long term, is the ability to get more out of the service businesses of acquired companies. You go back a long way and you look at some of the deals we did back 15 years ago when we acquired ADE and we acquired Therma-Wave and the other companies, we've been able to drive significant service revenue streams out of those businesses, much more than we ever contemplated where it turns out from a deal thesis point of view, you might have been able to justify the deal just on the service stream. Let alone the system contributions and IP contributions come from it. So there's an opportunity there for us. Each business behaves certainly. So we have to deal with different -- particularly in different markets. But still, there's clearly an opportunity there for top line, but also for efficiency in terms of the service delivery system itself.

Yes. And I know you guys like past the Analyst Days, I think you've talked about like a lot of work that you guys were doing like on the Orbotech side, I'm just trying to kind of pull that service model to be more like core KLA? I mean, where do you stand on that? And like how do you think about the -- just kind of the go forward there?

Yes. I think on some of those businesses. FPD, of course, we're exiting that business, and PCB is certainly more of a capacity-related business. And so -- that market has been a little softer. So from a top line perspective, a little challenged. But from an operational efficiency perspective and driving the KLA services operating model into those businesses, I think we're right on plan or maybe slightly ahead of plan. And so we do see some of the efficiencies that Bren talked about. And those companies can't necessarily invest in the same infrastructure as well as balance sheet. They don't want to necessarily make the investment in inventory. And our services business is an inventory heavy business because we got to have parts available on -- at a moment's notice. I'd say that the place we're doing both reasonably well, top line and operational efficiency is that SPTS business. We've made some headway there. We've driven some improved contract penetration there. And so we remain pretty confident and optimistic about our opportunity to drive that business. And the last area that I think is a real opportunity for us. Bren talked about advanced packaging for the company. If you look at services intensity and advanced packaging versus historical service intensity and packaging, it's certainly increasing. And for all the reasons that Bren outlined the stakes are much higher when you're putting these packages together. And therefore, you've got to make sure that you've got the same availability of performance lined up on those products. and thereby, you're increasing some of your service attach rate.

Yes. And that's a good segue into trying to think about like the services intensity, how that has increased over time. I mean what are those like discussions with your customers like? I mean when they're -- you're saying, look, like things are becoming more difficult? Like how does that like work into the contract negotiation of services and just like the overall handholding or help that you're providing them in those situations as your customers are moving to new technologies or next-generation nodes?

Yes. As Bren outlined, there's a lot of complexity at moving to the next node. And therefore, the products that KLA is delivering the complexity is increasing, and therefore, the need for KLA service is even greater. And time to market and time to yield is still king. And so they really need those products available all the time. So you get more -- they then want higher availability, for example, Bren talked about 90%, but for -- in a technology development area, they might want 95%, which means additional coverage for us infrastructure investment. They also drive the tools very, very hard from a performance perspective. And so one way to think about services intensity is contract penetration. And so if you were to look at contract penetration at leading edge versus our sort of 75% baseline, it's up 10 or 15 points, 20 points on some product lines where they're attaching at a very, very high rate on leading-edge products with respect to service contracts.

I guess like in a situation where someone doesn't want to sign a contract, I mean, what's the driving factor of that?

Usually, it's a little bit of -- it's a budgeting constraint. And so it's usually more an environmental issue where if they think that they might have lower utilization in some period where they can maybe reduce some period costs for a particular quarter. They might not want to do a full service contract. And so unfortunately, over that period of time, it's going to cost us more money to go into a time and material but given the environment that they're in, that's usually the driver is that they've got an environment that they're unsure of -- and so they don't want to make a long-term commitment, They'd rather go day-to-day, month-to-month.

And you have that, but you also have, I think, the ability of us to customize our offerings whether it's per tool, whether it's fab wide, whether it's uptime commitments, over specific time frames, response times, stocking levels of parts, all of these things go into our calculation in terms of how we think about how we price these. And so we can be pretty flexible with our customers to meet some of their some of their challenges with their budgets. And even to the point where there's -- we allow them to do some idling and some other things. So I think fairly customer-friendly and it's customized and then we want to deliver to that entitlement. And so then we build the system underneath it. The visibility helps us, I think, optimize from an efficiency point of view but also we can help them manage a very fluid environment in terms of how we offer what we can do with our service business.

Okay. Maybe as you think about a lot of the discussions still happening around like regionalization of chip production, the geopolitical kind of climate of just making sure that we've got leading edge production in the U.S. So I guess like is that an opportunity for the services business in terms of kind of moving away from more regionally large fabs to smaller scale fabs that are in different regions. How do I think about that from services?

Yes. I'd say, it's tailwind and a headwind. I mean, the headwind is, of course, when you have mega fabs, you can drive some operational efficiency that you can share with customers. From -- but from a regionalization perspective in driving a tailwind for the service business is that when our customers are trying to expand geographically, it puts a lot of pressure not just on the new region but also on the home region because they're moving a lot of those same resources to the new location to help ramp up that new facility. And so they need more help in the home region, they need certainly more help in the new regions. So they really lean on their strategic partners like KLA because they've got to deal with hiring people. They've got to deal with the governments in terms of getting things lined up. They've got to do drive the facility. And so we really have an opportunity. And it comes down to very detailed level work, which adds a lot of value for our customers, which is looking at every single slot they have. We have a very robust ramp process that we work with our customers on, where again, we look at line by line, slot by slot, making sure that we prepositioned people. We've hired all the people that we need. We've repositioned parts. And then we accelerate the installation to production-ready time. And our commitment to our customers is that we will improve that lead time each ramp. That's our goal with our customers.

Now when they change their schedules, then there's a headwind there, right? So both in terms of -- you think about the -- we talked about the complexities, right? So the training lead times are pretty long. You have to hire people and regionalization is driving those investments. We have to position parts. And then if the ramp plan changes or the fab delays, then we have to figure out how to absorb and management, still keep optionality to support in the long term. The other thing the regionalization also drives is, I think, from a market share point of view, given the reliance model of managing a broader footprint, I think it help systems market share also because I think they want our customers generally want to rely on us for more support more applications, where we have people stationed in the fabs to help optimize and get value out of the systems. So I think having the broader footprint allows them to rely on us and it translates as a scale provider, I think it translates into market share for systems as well as driving a service penetration that's favorable.

Perfect. And I think we're out of time. So let's end it there. Thanks guys.

Thank you for having us.