KLA Corporation (KLAC) Earnings Call Transcript & Summary

Great. We're going to get started. I'm Tim Arcuri. I'm the semiconductor analyst here at UBS and very pleased to have KLA with us next and Bren Higgins, who's the CFO, I've known Bren for a long time. So very, very pleased to have you, Bren. Thank you.

Yes. Thank you for having me.

Great. Well, a lot of things to discuss. But the first thing that I wanted to lead off with is there are some very interesting dynamics in your business. I mean, this is my 25th year covering the stock. So -- which doesn't make me any smarter than anybody else, but I have seen a lot of things. And what's interesting is that we've never seen -- you have so much backlog. You've got basically 14 months' worth of backlog. It's been coming down a bit, but not much. But because of this book-to-bill has been low, it's been below 1 now for 4 quarters in a row and you're just basically shipping off backlog. And this has also been a strange year from a WFE perspective, because I know you're saying $80 billion, I'm going to ask you about that. It seems to me like it has to be a little higher than that this year. But you have all these different dynamics where customers booked out with you so long in the future, and you're now shipping off that backlog, and you've always relied on backlog, but it's much more extreme right now. So can you talk about just how confident you are, given all the backlog you have? And maybe when do you see the book-to-bill coming back? The customer is actually placing bookings again or just they don't have to because they've already placed the orders and you have so much backlog.

No, it's -- a lot of moving parts in that question, and I've been with KLA 24 years. I think when I started, you -- I was in our corporate FP&A group, and you were a new analyst in the space, and I was new to the company. And I think in some ways, we learned some of it together. So good to be here again. So over the last number of quarters, we've been working through about 500 million of backlog per quarter. So the orders have been below the shipment revenue levels, not much difference between the 2 and it's been that way for about 4 quarters. Now the orders have -- or the backlog has been unprecedentedly high levels. I mean normally, we would run with 7, 8 months of backlog, somewhere in that ballpark. And so it's been much longer than that. And we went through a period of significant growth in the industry, right, in '21 and '22. And if you look at what KLA's own semiconductor process control systems performance in '21, we were up 46% year-over-year and '22, we were up 36% year-over-year and dealing with a lot of supply chain challenges through all that. I think customers were trying to get in the queue and to try to lock up positions. If you look at what happened in '21 and '22, most of our shipments were going to our biggest, most strategic customers. We're also prioritizing the semiconductor supply chain because we were all short certain semiconductors for our equipment, and so we wanted to make sure we were supporting those customers that we're gaining our ability to ship and others' ability to ship. And so -- and that -- so we had a lot of customers that were coming in, a lot of the current business we're seeing today in China, but also in non-China in the legacy parts of the market that, frankly, were further down the list in terms of prioritization given the strong demand we were facing. So these were orders that we took over the couple of years and this year, as we moved into this year, when we've seen the leading edge parts of our business weaken, we've seen memory on the margin weaker as well. We've seen those customers now sort of step up and take those tools that they ordered. And so we've seen about 500 million. We started the year about $12 billion of backlog, a little over $10 billion today. So it's been about 500 million a quarter or so of shipping out of backlog. And then obviously, new business has been also a portion of that. So I think it probably continues. Orders have always been lumpy, and there -- we were having massive orders as customers were trying to get in line as you go back into that time frame I was referring to. So we'll see how it goes as we move into next year. I mean if you can just take a step back and say, okay, well, what does it mean in terms of business levels. If you look at overall KLA semiconductor process control systems this year after last year's 36%, WFE was up about 8% or 9%. And if you go to this year, we'll see what WFE ends up, somewhere maybe down 10% to 15%. And I think our systems business is going to be probably down somewhere around 10% if you take the midpoint of the guidance that we provided. So I think that's how it sort of plays through this year. And as we look at next year, at least through the first half of the year, we talked about a stabilizing demand environment. I think that process control system shipments and revenue likely be sort of stabilizing, sort of flattish around the current levels that we're seeing.

So you think next year's -- whatever WFE turns out to be this year, which we can talk about in a moment. But whatever it turns out to be this year, you think next year is characterized, and I'm not asking you for 2024 WFE guidance, but do you think next year is kind of a flat year, plus or minus, for your business? At least through the first half.

Yes. I think in the second half, we'll have to see if we start to see some recovery in our customers' business. If you just look at their current profitability levels, cash flow, you'd like to see -- I think our customers are going to want, particularly given the cost of financing today and so on are going to need to see a sustainable improvement in their business before you start to see big capacity investments. And so as I look at the first half of the year, I don't really see a lot of that. We'll see how it translates in the second half as we move forward.

So in 2021, your WFE share was about 6.3% roughly. 2022 was a huge year for you. You gained 160 basis points worth of WFE share. Of course, it depends on what WFE turns out to be this year, but you're going to -- at -- I mean maybe if WFE is 90 billion, flattish, maybe you lose a smidge, but it's sort of maintaining at this high level. So can you go back and like deconstruct why you've gained so much WFE share over the past 2 years? I -- from my perspective, it's easy to say, well, it's mixed because foundry has been so good and memory has been so bad. But it's not just mix, there's been other things happening. So can you talk about that?

Yes. No, it was an important part of our message in our Investor Day back in the middle of '22. We -- as we have moved into an environment where scaling is happening again and we've got a pace of Moore's Law that's driving enough compelling attributes to the end markets to drive leading-edge designs and we've seen design activity pick up. And so that's been good for investment overall in logic and foundry, where you went through a period of time where logic/foundry WFE was a negative growth rate, right, from 2012 to 2018 or so. You saw this reacceleration of investment. Logic/foundry tends to be stronger for process control. The dynamics of -- in memory, you have more commodity products, more redundancy, more repair in that environment. So process control intensity tends to meaningfully higher in logic and foundry. So that's been a nice driver. But we've also seen the investments in -- because of design starts, we're seeing more investments in production, and that's all been good for process control. We've also had good overall share of process control growth in the time frame that we're talking about. And I think a lot of that is driven by our exposure to certain markets that have been inflecting as a result of scaling, things like optical pattern inspection, which I think, is one of the fastest-growing markets in WFE. Even this year, our Gen 4 product is likely to be flat this year, even potentially down WFE environment. So there's a lot of -- I think the drivers of scaling that have enabled our customers to invest more in supporting a broader design environment, less reuse because they're supporting more designs. All those have been factors in the overall growth of KLA share in the market. This year, look, '22 was -- there were a lot of supply chain challenges. I think a lot of people struggled to get shipments out the door. And so I think it distorted a little bit, back to your question about what '23 WFE looks like and you probably had, I don't know, $5 billion or so that was really '22 kind of carryover into '23. And I can remember a year ago, you and I were at this, how can KLA have this 5x the industry growth rate but still kind of perform in line with the market. And I felt that because of mix, because of our exposure to these parts of the market that some of the export control impact and some of the other things that I felt like we were in a pretty good place to at least maintain the share of the overall. So I think because of that carryover effect into this year, could it be -- we said 80, could it be 85. Now that everybody's guided and I have seen the results, I can see it a little bit higher. I don't think it -- back to the point of our share of the overall, I think we're probably flattish to where we were last year. And I'm pretty encouraged by given the adjustments we've seen at the leading edge and how well we've done in a year where most of the investment in logic/foundry has been lagging edge that normally what drives our business and drives process control intensity at higher levels is at the leading edge. And so to, I think, be able to maintain that, I think, pretty good relative performance overall in an environment where leading edge has been weaker gives me some pretty good confidence about go forward as we see the leading edge start to invest again.

So let's talk about China for a moment because I think China -- there's this perception that it's a bunch of capacity and it's low end, and so you shouldn't participate in that. But actually, you're participating as much, if not more, than anybody is. And part of it, I feel like is because you've got this surge in China DRAM orders and their yields are still very low. They're just trying to get off the ground. And so you seemed to be punching above your weight in China relative to what we would typically expect. Can you just talk about how China is maybe helping your share?

Yes. So the DRAM piece has been a nice upside factor this year, and it was -- earlier, it was -- we didn't think we'd see it this year. And then after some clarifications about where the technology lines were, it did enable us to ship some business this year. So it's been -- we don't talk specifically about yields or customer mix, that's again kind of 20-ish percent of the overall China. There's been a lot of investments in China in infrastructure, and we call it infrastructure, where it's around silicon wafer capacity and reticle capacity that these were choke points in '20, '21, '22 for our Chinese customers to get reticles where they didn't have captive mask shops so they would have to buy from the merchants and so there was limited capacity investment there to support the demand for -- that they were trying to drive silicon wafers, again, long-term supply contracts from a lot of the silicon wafer providers. And so as a result, we've seen investments try to build up that capability for self-sufficiency. Overall, there were choke points before and also to support the overall export market for electronics in China. So that's been a nice upside factor for us this year as well. We're seeing that investment. And that's WFE that we're exposed to that a lot of the peers aren't, meaningful parts of the CapEx investments that are there in those -- in that part of the market. We have our traditional customers in China, and there's a portion of the WFE that have been long-standing customers and are subjected to a lot of the general same supply-demand dynamics that any of our semiconductor customers are exposed to. And so we've seen that business move around a little bit, but it's been pretty resilient. And then there's a bunch of new projects. There's a lot of new projects where they're subscale, they're just starting out, where if you're doing a limited set of wafers or new wafer starts, you're trying to prove out technology. You're trying to be credible with customers. You have a lot of customers that are investing. And it's subscale. They tend to have to buy the whole portfolio from KLA. They're doing a lot of sampling. They're trying to drive development. R&D never goes down. And so it's sort of R&D kind of risk production investment. And that's spread out across multiple legacy fabs around the country. And so that's the part of the business where we've had those orders that we took over the last couple of years, and it's been a nice contributor over the last 12 months as we've seen some of our larger customers move around. And it's been the driver, I think, to some of the puts and takes, the upside part of WFE. As we move into next year, I've got a lot of backlog from those customers. I had deposits from those customers. So we'll see how it plays out over time, but I feel like as I look at investments in China into next year, I think it's more or less flattish overall.

And how do you sort of think about all these -- there's a host of new customers you're shipping to that you've, in many cases, never heard of before, so you're taking more of a down payment from them. We don't really know, I mean, in some cases, who they're affiliated with. But for now, the U.S. government is allowing shipments to those companies until they can be proven otherwise. And there's a question that I get because there seems to be a different sort of burden of proof as a U.S.-based company that the government has long-arm controls, meaning that they can -- have you have to report back to them in terms of what the tool is actually being used for in the field. Whereas it sounds like the Japanese and the Dutch don't have long arm control. So it's improved upon shipment. But then when it's in the field, they're not required to then be responsible for how it's being used. So can you talk a bit about that, about how much of a burden it is on you to then, on an ongoing basis, from a service perspective due to this long arm control dynamic for you to then validate how the tool is being used? And at what process nodes it's actually being used at, right?

So everything we're shipping to, and it's been pretty clear since October of last year, where there are very clear technology thresholds and end user controls. And so whether it's advanced logic at 14-nanometer and below, whether it's DRAM at 18-nanometer half pitch or whether it's 128 layer flash. So that's clear, and that business is taken off the table, and we had talked about, we sized that impact for 2023, and it's largely played out closer to the bottom end of that range because of some of this DRAM upside that we -- we spent an inordinate amount of time, effort and resources working very closely with the government to ensure that we're in compliance with all export requirements. You get licenses for systems, you have to get licenses for spares and you validate on an ongoing basis how your equipment is being used. It's going, in a lot of cases, into specific fabs at specific design rules. I think there's pretty good clarity about what they're using those tools for in terms of process technology. And it's an active effort, and we spend a lot of time on it, and we're very, I think, generally pretty conservative about how we manage our way through it.

Right. So you -- so the point being that on an ongoing basis, you can't even service the tool unless you can ensure that it's being used at a process node that's not banned.

Right. Yes, because you have to get licenses for systems and a separate license -- set of license requirements to get the spares for those systems. So it's -- there's an administrative effort to work your way through that. And so yes, that's true.

Got it. Let's talk about gross margins. So you guided gross margin down. It looks like you think it stays pretty flat. Can you talk about what's actually going on there? I would think that China would be helping your gross margin. So can you just talk about the dynamics for December and then looking into next year?

Well, product mix is the biggest factor in our gross margin. And our business model is really predicated on our ability to differentiate, to understand the value that we provide to our customers. I think it's the clearest indication of a differentiated business is just to look at the gross margin performance. Given the elasticity of demand for process control, we have to have a pretty firm understanding, I think, of the returns that our customers get. And we have to be disciplined about it because you can't undermine that value prop from share increment, the share increment or even from node to node because the baseline for the future negotiation is what they paid for the last generation of tools. So it's an important aspect of our business in terms of how we run it. And the product mix across products, we also have service, which has a dilutive gross margin. And then you have our EPC business, which tends to have lower gross margins as well. So those factors, the puts and takes in a given quarter will drive our variability of our margins overall. So slightly higher last quarter. We had some upside in the quarter in terms of outperformance relative to guide. Most of it came from our semi PC business, and so that drove the upside. But even within semi PC, it's not all the same. So you have different types of tools, different configurations. In a lot of cases, you may sell a version of a tool that have a higher COGS, but has an upgrade path over time. The upgrades have a higher margin because the incremental cost of goods is less. So there's a lot of those kind of moving parts within our product mix. But in general, we're operating today right around 61.5%, it's consistent with how we guided. It was a little bit higher last quarter. And it can be plus or minus depending on these factors overall. From an incremental point of view, incremental gross margins tend to be 60%, 65% for KLA. And I think as we start to see a recovery overall in industry investment, we start to see growth. I have a lot of confidence in our ability, I think, to continue to drive the kind of gross margin performance we have historically. I think the market positioning is good, the market positioning around the products, I think, that will grow as a percent of the overall is good and carries higher margin. And we made a lot of investments that are kind of a headwind today at these business levels to support our '26 plan of getting up to $14 billion in revenue, plus or minus or even if you were to extrapolate, take that further out. So we've done a lot to position ourselves for growth moving forward. We start to get a resumption of growth. I feel pretty good about our ability to scale it over time. And then the drop-through is 40% to 50%, very consistent with what we've done over time and how we run the company.

And then do you think all things equal, if China continues to be and maybe even be a growing portion of WFE, ought that to be a positive effect on your margin because you probably get higher gross margins as a general rule for domestic Chinese customers versus outside of China? Is that fair?

My margin profile, given the nature of our go-to-market, is pretty consistent across different markets and different customers. Obviously, volume is a factor and what they're buying is a factor. I would say from an operating margin point of view, it tends to have a more attractive operating margin because the incremental investment to support the legacy investment by those customers is less. So the R&D work that I have to do, the support on an ongoing basis is less. So I think accretive operating margins versus other regions. But gross margin, not a lot of different regions and customers.

Got it. Let's talk about your non-process control business, your EPC business. It had its ups and downs since you bought Orbotech, but it's about the same size actually as it was in 2020, whereas your process control systems business is much larger than it was back in 2020. Now I get different dynamics. But can you talk about what's going on in EPC? And maybe what are some of the growth drivers that could maybe help EPC start to grow a little faster?

Yes. EPC is really a tale of 2 businesses, right? On one hand, you have specialty semiconductor, which is being driven by advanced packaging and power semis. And that's a business that has had nice growth over the last couple of years. We expect even this year to be flattish. And you had a switch from -- had an RF percentage of the mix and then mobile slowed down, but we've seen because of automotive, the power part of the business accelerates. Advanced packaging is also a factor there. So that business continues to perform well, we said in our report. You can get a view of the revenue and the gross margin profile of that business overall. And we're engaging with customers. I think one aspect of our thesis was that given our -- the size of the company, we could engage at a much higher level with customers strategically. I mean obviously, we have more at stake. The customer is more comfortable. Sometimes smaller suppliers struggle with having the engagement at the higher levels. And so we've seen that, and we're pretty excited about some new product offerings that are coming there. The PCB, Display and Component Inspection of -- or part of -- segment of EPC has been weaker. Now it tends to be more mobile-centric, and so -- and it's more capacity, more consumer-driven. And so that business has fallen off a fair amount. I think we're excited about what exists in PCB, particularly related to advanced packaging and how the substrate and you start to shrink the lines and spaces on the devices and the substrate integrates into the heterogeneous package. And so there are going to be new product offerings to serve that part of the market. So we're encouraged about the advanced packaging opportunity, and it was part of the thesis of us acquiring Orbotech was that evolving opportunity there. And our ability to take the KLA operating model and differentiate in process markets of our choosing, and I think we've demonstrated that overall. The gross margin profile of those businesses is different than semi PC. But part of our thesis was that on growth that we could drive corporate level incremental operating margin, so we could scale the profitability of the business at least 1.5x faster than the revenue growth rate from a growth rate point of view. And from the time we bought it in '19 through '22, we were able to do that. Obviously '23, it's corrected. But I think over the long term, we feel pretty good about the long-term opportunities to see high single-digit kind of growth from EPC overall, driven by some of the things I've talked about. But we've done a lot of cost work, too. So I think our ability to scale the profitability of that business in line with the overall company averages is pretty strong.

Great. Can we talk a little bit about the bare wafer business? You alluded to it before. China is building out their supply chain that includes bare wafer. It's an idiosyncratic driver for you that others don't have. But you don't tell us that much about it. You don't tell us how big it is. It sort of gets embedded in the numbers that you put out. Can you just -- I'm not asking you to tell us how big it is here, but it's definitely becoming a much bigger piece of the business. Maybe help us understand a bit more about that business.

So it's unpatterned wafer inspection and metrology. So you have the unpatterned wafer inspection where you're basically inspecting bare wafers. Customers -- I see customers also use these tools to monitor for cleanliness and run monitor wafers through process tools. So there's -- those are the applications. So in some ways, it behaves like a -- in an IC fab, it behaves very similar to a process tool. And so you have that part of it. The other part is that each of our wafer customers has to qualify the wafers before they ship them out to their customers. And so there's defect inspection, and then there's also metrology for flatness and wafer geometry and so on. So generally, about 50-50, so if you take the overall unpatterned wafer inspection and metrology business, about $1.5 billion this year. If you look at the Gartner data, last year, it was like [ 1.3 ] it's expected a little bit. We have a very strong market position there. Generally, it's been about 50 -- over multiple years, tends to be about 50-50 to wafer customers versus IC customers. This year, given the inflections that we've been talking about, both with China, but also non-China, it's been a higher percentage than that. Not a lot higher, but higher than that. So it gives you a sense for the sizing. But over time, it's been generally about 50-50, but it varies because it does -- those customers do invest a little bit on a different cycle. It takes a long time to add that capacity. And so there's a lot of preparation that's going on for demand recovery. It was a shortage point. So customers are not doing just in time for wafer supply anymore. A lot of our customers are signing long-term supply contracts with the wafer providers. You have drivers for more wafers, packaging, wafer-to-wafer bonding that's coming with packaging. It's going to come into the front end with buried power rail as an example. So between inventory stocking, between drivers for positioning for demand, memory consumes a lot of wafers, so recovery in memory at some point. So that investment is happening today and probably continues next year, although I don't think it grows next year. And then that part of it will fall off, right? And it tends to have -- that part of the business underperformed in 2020. So it just has a little bit different cycle relatively -- but it's a good period, and it's a very strong market position and the financial performance of that business is commensurate with its market position.

Can we talk also about your attach rate to litho? Going back to my comment about WFE. This is just such a strange time because WFE -- personally, I think this year is going to turn out to be closer to 90 than even the mid-80s largely because of litho because you and Applied and Lam and everyone is down in a range that would convince you that WFE is 80, but really because of litho it's actually probably closer to 90. You tend to get -- I have always mentally thought of you as having a fairly high attach rate to litho. So in a year where litho is over shipping, you get not completely pulled along with that, but you get partially pulled with that. So can you talk a bit about your attach rate to litho and maybe how you've been carried along with all those litho shipments this year?

Well, and the timing is never perfect, right? Getting back to last year, I think litho is a market underperformed last year because of some of these challenges we talked about earlier in terms of rev rec and shipments. I think if you actually looked at the overall, certainly underperformed relative to like where we were last year as an example. So you could argue that some of what's happening in '23 is a carryover effect from '22, as we talked about. Generally, though, the scaling effect of litho and EUV is a driver for our high-end business. And so as customers are driving to smaller feature sizes, it drives our high-end inspection capabilities to support all that. So it's one of the big things that has driven, I think, more compelling Moore's Law type attributes for our -- for end markets, which is driving customers to adopt the next node, which when in the time of limited scaling, a multi-patterning from the 28-nanometer node down to the 10-nanometer node, you had high-volume parts that were migrating, but a lot that were staying at previous design rules because it just didn't make a lot of sense to move forward. Now with chiplets and also the scaling effect of EUV, that's creating a more robust design environment, and that's been good for KLA overall. DRAM, we've had the introduction of EUV into DRAM. That's been a factor in our DRAM process, control intensity. I think that continues. Memory never looks like logic/foundry, but I think that we can sustain some of the improvements we've seen there. So scaling is very good for KLA, smaller feature sizes is good for KLA.

Let's just talk in the last minute here on China. So one thing that's unique about KLA is unlike for some of the films companies where there are local competitors in China that are becoming a fairly big factor within China, there's not really anyone that I see that really competes with you that's a domestic Chinese supplier. I mean at the end of the day, you're more of a software company than really anything else. And it's been challenging for China to compete in what you do. So is there a competitor that you're worried about in China on the horizon? Or do you feel like you're fairly insulated from getting picked off from those dollars?

I think we feel pretty good about our competitive position. And a fundamental aspect of our competitive position is the portfolio that KLA offers. Most of our competitors, China, non-China, tend to be point product competitors. And that forces them to go to market and have to try to come up with various use cases that sort of support that one technology. And we can go to customers with a portfolio approach with a network effect across the tools. We generate more data than anyone in the fab. And so we can aggregate that data, we can use AI to try to leverage data analytic technique, simulation techniques to try to interpret and leverage that data in a meaningful way. But we allow our customers to address technical issues or economic ones. And we're somewhat agnostic between what they use, what type of tool, for what application. And that -- depending on what their needs are. So I think from a competitive point of view, it's very hard to come in and try to compete with KLA, where I have one type of tool when you've got this whole breadth. Look at wafer inspection. We have 5 different platforms we're serving the wafer inspection market with. All having different sensitivities and throughputs. And so there's a cost element that sort of drives a lot of those decisions. And customers sample and are different stages of maturity in different ways. And so as a result of that, I think it enables us competitively to position in the right way. And I think that we've demonstrated over time that as long as our products are relevant and solving the problems that are scalable and scalable into production that we tend to be able to share in the value that we help create.

For sure. Well, we are out of time, but thank you, Bren.

Yes. Thank you for having me.

Appreciate it.