KLA Corporation (KLAC) Earnings Call Transcript & Summary

Good morning. Welcome to day 2 of Citi Global TMT Conference. My name is Atif Malik. I cover U.S. semiconductors, semiconductor equipment and networking equipment stocks at Citi. It's my pleasure to welcome Rick Wallace, President and CEO, KLA; as well as Bren Higgins, EVP and CFO of KLA. We also have Kevin, our friendly neighborhood IR, in the audience as well. Bren is going to make a few opening comments, and then we'll dive into the Q&A. Bren?

Atif, thanks for having us. I appreciate your interest in KLA. So just to level set here with some comments from some things from earnings as we had -- June quarter was a good quarter, a raise on the September quarter. Obviously, '25 is setting up more or less as we expected through the year. My business is really being driven by what's happening with the high-performance compute market. On the logic side, the N2 build-out, which has been very solid for KLA from a share and an intensity point of view. High-bandwidth memory is creating opportunities for process control inflection. We'll talk more about that later today, but we're pretty excited about what's happening in that part of the market. And advanced packaging is growing as an opportunity as the complexity of the package increases and the ability for KLA to differentiate also plays along with that. So we feel pretty good about the trajectory of the business. Our service business, despite some restrictions and access to fabs that have come from export controls, continues to grow in double digits, consistent with its long-term model. And we're pretty excited about what's happening there in terms of the lifetime of our tools, the customers' desire for availability and performance in the service installed base. So a good situation there. Overall, the financial model, inclusive of tariffs, we provided some context around 2025 at 62.5% gross margins and operating margins in the 43%-plus range and outperforming our long-term incremental operating margin target of 40% to 50%. So '25, I think as we continue here, we provided some soft guidance in terms of expectations of revenue levels being more or less around current business levels as we exit the year. Early views on '26. We'll see how it plays out. Qualitatively, we think the investments continue at the leading edge, and that will be a positive for WFE, same with the high-bandwidth memory part of the market. Flash market is reasonable and at low levels. So I don't see things falling off there too much. And I think the legacy business is overall about -- I don't think it's getting worse, but we'll see if it gets better around it. So in general, that's where that is. And I think China, after a couple of years of significant investment in legacy we would expect, and we're seeing a down year in our China business this year and would expect next year to probably correct some more as we move forward. So we're constructive on a growth year overall for WFE and for KLA's relative performance within it. So why don't I just stop there, and we'll go on to your questions, Atif.

Great. Rick, let me start off with you. I've been doing sell-side for 15 years, and I know you've been at KLA for longer than that. There is something different about KLA. Your SPC systems grew 15% last year, outperforming WFE. And again, this year, you're outperforming WFE. Understand the process control intensity was around 13% per the Gartner data. So if you can walk us through -- and your stock multiple has expanded, what is different this cycle versus the prior lithography-driven cycles?

Thanks for having us, Atif. Well, I think if you look at the dynamics in the industry and where the investment is going and the technology that's being deployed, Bren mentioned it a bit. But if you think about the hyperscalers and the work that's going on high-performance compute, there's really 3 things that are happening simultaneously, which are opportunities for KLA and then we can talk about our execution against those opportunities. Number one, in terms of the number of designs that are happening at the leading edge has historically been a concern of the industry that with each subsequent node, because of the cost, you would see designs go down. Now that inflected at 7-nanometer and partially because the design tools became more usable and the number of people that could do designs in the foundries went up. But now there's another dynamic which is driving a lot of that high-performance compute. The need for silicon proprietary technology by a number of end users is driving the number of people that are doing designs up. So if I go back even a couple of years ago, and I remember when the first hyperscaler started doing their own design, Apple, of course, had been doing it for a while, but suddenly, you see design starts going up. And then more recently, I think we count up to 10 hyperscalers between the 4 big ones and others doing custom silicon. And that's driving -- why is that important for KLA? Well, it's important for KLA because a couple of things happen if you have multiple design starts. One of them is the fab complexity goes up dramatically. So you have a much higher mix fab, which means the ability -- if you're running 100 process flows and multiple designs for multiple people, you've got to make sure you don't start too many wafers and you don't start too few and you've got to ramp them quickly to yield. All that's good for process control because in order to do that, they want better insight, our customers want better insight of what's going on in the process. So just start with the number of design starts inflecting, and it looks like 2-nanometer is going to be even more design starts because more people are doing them. The other thing that's happened due to the work on high-performance compute specifically is AI is the die have gotten larger. So very large die for defectivity meaning the same number of defect across a wafer will have lower yield. So the result of that is customers end up inspecting more to protect and drive down that deep activity to increase the yield. So bigger die, more design starts. And then the other thing that happens at the same time is in the reticle and the manufacturing reticles, you have more reticles and you have more single-die reticles. So our reticle business is going to have a record -- all-time record year in 2025. And we have a number of opportunities to continue to grow that based on our product road map and based on the needs. That's just the foundry logic side of things. And this is in a market where really only 1 player is at the leading edge. And I'm sure we'll talk about if there's opportunity for others to get in. But right now, that's 1 player doing it. The other thing that's changed with high-bandwidth memory is the memory for many years was the intensity, in other words, the amount you had to inspect to produce memory was not a very high percent of the budget. And this went back to years ago when customers started building in redundancy into their memory and they would repair all the memory, and it was mainly consumer application. When high-bandwidth memory started being used, it's for -- it's always-on, it's for -- and so customers have been demanding less redundancy, more reliability in these chips because I can't just reboot the PC; and more complex design architecture around periphery, all of which is driving inspection intensity up. And then the -- so you have logic driving it, you have memory driving it and then the third part, which we'll talk about, is packaging. So advanced packaging I'll admit, we thought advanced packaging was going to be important when we bought ICOS almost 20 years ago. But now it's finally happening. And it's happening and what's happening is the demands of high -- of the advanced packaging look more like what was in the fab. And so a lot of our equipment is being redeployed to address the packaging. So you have all these dynamics at once. Process control intensity goes up in foundry logic because of design starts, bigger die, more advanced nodes, new architectures for transistors, goes up in memory and it goes up in packaging. Overlay that with the execution we've had on our product road map and our ability to address those, that's what you're seeing right now. And I would say we're in the relatively early build-out of, and we'll talk more at our next Investor Day about what we see as this progresses, but we think there's more to go.

That's great because just on the last point, we appreciate you guys walk us through process control, intensity assumptions underlying your target model. But why is there a ceiling at this number when you're talking about more design tape-outs and AI is getting more complicated, you have 12 and 8 HBM stacks and all needs to be tested. Otherwise, the Blackwell fails. So help us understand what determines the ceiling on the process control intensity for you guys? Because the impression that some investors have that you guys are early beneficiaries when a new technology like Gate-All-Around or something comes around and it might not be sustainable. So why would it be sustainable this time?

So we're still -- I mean, you mentioned 13%, right? The question is always a combination of what drives the need and do you have the capability. And there are more problems to be solved that we're not solving right now that if we solve them would drive it up. If you look at reticle manufacturing, the making of reticles, probably 45% process control intensity in the making of reticles, right? It's probably -- every reticle gets inspected multiple times. I mean, if you look at the intensity in that, now that's a pretty unique area because the cost of a reticle being bad is so high that you get 100% inspection, and you have to have 100% capture rate of critical defects. But there's still room to run in terms of the process control intensity. We don't really know and our customers don't know. We know that it doesn't move that fast. It takes each cycle for it to increment, which is why we talk about a trend line as opposed to an ultimate destination for this. But it is a combination of need and capability. And of course, customers are doing everything they can to reduce the need for more process control in terms of making more stable processes. But it's been creeping up. And the other dynamic is, as I said, there's only one real player at the leading edge in volume, right? So that would also change the dynamic.

Yes. Atif, I think when you back up and you go, you look at where we were in 2020, 2021, KLA's share of WFE was somewhere in the low 6 percents. This year, we're approaching about 8%. You look at share of advanced packaging, it was 1% or 2%, it's about 5%, 6% today. So we're seeing some sustainability to the relevance of KLA over this time frame and our relative performance, I think, is reflective of that. These drivers that are, in some ways, uniquely benefiting KLA as it relates to high-performance compute and as that becomes a bigger percentage of the overall market certainly provides a tailwind of opportunity moving forward, as it relates to those metrics. So those are metrics we focus a lot on. It's not just our share within our market, which is obviously very strong, but also our relevancy within WFE. And it's a couple of points -- in the last few years, but that's a couple of points on $100-plus billion of opportunity. So we're pretty encouraged by what we're seeing and we think that these tailwinds as it relates to high-performance compute will have some sustainability to them going forward.

And again, the most significant thing that's happened is increased number of designs. That's the most significant for -- I think, for the industry because it speaks to how many people have now viewed having advanced semiconductor as part of their differentiation. And obviously, the world years ago, the belief was every node, you'd have fewer players capable of competing. That may be true in the manufacturing of it, but it's no longer true in the design and the economics of that design are many customers', customers are all determining that they need to have custom silicon.

Great. Rick, going back to advanced packaging. Historically, you guys have been tied to kind of lithography-driven cycles where you have a new wavelength comes along and you have more inspection need for it. But advanced packaging is a market where lithography is not driving any demand, it's mostly process tools, inspection tools. And you talked about 85% growth this year in advanced packaging. So can you just touch on what's unique about your advantage in advanced packaging? How sustainable it is there? There have been some incumbents in the market as you come in.

So I'll do part 1 and let Bren do part 2. So part one, what's different? Why -- so my historical view of this, I think as a company, has always viewed this as, this is, in the end, all about economics. Everything in process control is about the economics. How much is it worth it to inspect something or measure something has to do with what do you do with that information and the relevancy of that information to manufacturing process. So the reason -- like if you overperform a spec and somebody else can essentially do the job at a lower price point, they're going to win because it's economics. So in the reticle case, back to my very high percent of inspection on reticles, the cost of an advanced reticle is so high that you can add a lot of costs in the inspection. Take the other end of the continuum, we were briefly in a market through acquisition where they were inspecting solar wafers. And my joke was you could step on the wafer and they would yield. So there is no inherent value in that inspection on solar. So eventually, it went away because the cost of solar was so cheap of those wafers. It's economics in the packaging. The value of these packages is so high and the cost of failure is so high that that's what's driving the increased need. And in the initial phase, they're going to largely inspect everywhere, and they're going to look to try to debug this process. But because the cost of the package is so high, the end product, it's going to have a relatively high intensity over time, the sampling will change as they debug the process, but the view of how much is that loss at risk is what's higher. And for a long time, like in the chiplet era, it wasn't -- it still wasn't valuable enough, right? If you're making chips for a mobile phone, you can't add that much to the cost. That's not the case. That's not what we're seeing in high-performance compute. That's what's changed. And so a little bit for us because we get asked this question, would you go after these other lower-end markets, it's like we don't want to go after a market where differentiation, you can't get paid for differentiation because what that will be, there's a lot of people playing in it and the margins will be -- won't support our business model.

I think you covered most of it, Rick. So I think when you back up and just look at the growth and what we said at earnings is we were slightly over $500 million last year. We think and we've upticked the number over the course of this year, about $925 million across the portfolio of KLA. About 70% of that is in process control. The other 30% is in our process business, which is -- has historically participated in a number of these markets designed for these markets and now those markets have been reflected. And so we've got a pretty good competitive position there. On the process control side, you have most of the optical inspection business, which is the bulk of that, but there's also other businesses within KLA that are also benefiting. You have to inspect the interposer. So we have patterned inspection. So you inspect the substrate. That's an opportunity. Customers are also starting to use more advanced capability, and we'll likely do more of this going forward to point to have a reference for some of the higher volume optical inspectors. And so we have that capability in the company. So as the customers start to deal with a shrinking design rules, it creates an opportunity for us to leverage the existing portfolio. And we have a number of platforms from kind of the base level where we're operating today. So we think competitively, we're in a pretty good place as the technology profile of packaging starts to change and increases over time.

All right. So just on the competitive side, AI was ramping very quickly and the incumbents benefited in year 1, but now things have kind of settled out with OSATs and they're looking at your tools, which are definitely far more capable of being employed in the front end first and now they're looking at the back end. So is it safe to assume that those dynamics will continue into the future where they'd like to invest in a tool that is 3, 4, 5 years advantage.

Well, that -- it's a very cost-sensitive sector. So you have to win on cost of ownership and that, as you know, is a combination of things. But it starts with you have to have capability for the job. And then the intangibles that you get, you can win if you're tied if you have a road map that the other side doesn't have. You can win if you're tied if they're familiar with your product because they've used it. I mean it's less true in some of the OSATs, but it is more so the case that people that went from the front end to the back end and packaging wanted KLA there because they had used it in the back end. And then a lot of that gets dictated to their supply chain, too, of what they're using. So -- but we still have to win cost of ownership. You're not going to charge you might have a higher price, but the cost of ownership needs to be competitive because these are all very sensitive markets, right? And so you might have an initial build-out, but eventually, you've got to get competitive. So we don't chase every opportunity because some of them, there's just not enough differentiation to make sense for us.

Great. And let's talk China. One of your U.S. peers provided a weak October quarter guidance, driven somewhat by China and your outlook on China is up half-over-half. Can you talk about, at least on your side, what you're seeing in China? And then also if you can touch on any initial thoughts on the recent news that the Commerce Department is revoking the waivers for the international customers in China.

Yes, sure. As I talk about earnings, I think our view on China has been pretty consistent over the course of the year. We were, as a percent of business, top 40%, 41% in 2024. I think this year is going to be somewhere in the neighborhood of about 30% plus or minus. I talked about that perspective back in January when we lead into the year and it has been pretty consistent in each earnings call with that perspective. We've seen some strengthening as we've seen some strengthening in the overall business, but the ratio of China has been more or less the same. So that translates, I think, into -- given that I think we've had consistent outperformance each quarter relative to consensus, where we originally thought 15% to 20% overall, that the China business will be down at somewhere between 10% and 15% today. It can be lumpy quarter-to-quarter. If you look at the first quarter, it was 26%. Second quarter, I think, was over 30%. So I would expect the second half to be stronger. We'll see how it plays out over the course of the year. Right now, I would say, of course, there's revenue recognition dynamics and shipment dynamics, but September is probably stronger in December. But the overall view for the year is more or less consistent. And as I said in the opening remarks, as we look at next year, you're seeing investments change. We've seen more investment from some of the more mature players in China as we think of the future outlook. But I would expect that just continue to kind of normalize as we move forward. We'll see in terms of guidance into '26, but I think there's an adjustment down off of the pretty elevated levels over the last few years. Correct. And then on the last question, I don't think it changes. You just move to license-specific environment. We'll be supportive of our customers and the license requirements to get the necessary approvals to ship to those facilities. We have service business in those facilities that's not affected by anything in the current ruling. So I think in terms of impact on business, there's nothing there.