KLA Corporation (KLAC) Earnings Call Transcript & Summary

Good morning, guys. Why don't we go ahead and get started. I'd like to welcome everyone to this afternoon's session with KLA. It's my pleasure to introduce both Rick Wallace, President and Chief Executive Officer of KLA; and Bren Higgins, Executive Vice President and CFO. We've got a fireside chat format for about the next 30 minutes. If you do have any questions, feel free to e-mail me and I'll try to incorporate them into my prepared questions. But with that, I'd first like to thank both Rick and Bren for being here. I wish it was in person down in the desert, but virtual is the new norm in 2020.

Well, Rick, I wanted to kind of start off this fireside discussion, just helping -- allowing you to kind of position the company, I always like to say that of all the major semi cap equipment companies out there, KLA is perhaps the most unique and as a result, the least well understood by the general investment community. To be clear that the tech specific investors know you well but I'm wondering if you could just spend a few minutes kind of positioning your company to the broader investment community, your core IP and the way you've leveraged that IP into your core markets?

Yes. First of all, John, thanks for having us. And I share your desire, next year we'll do it in person, but this seems to be working pretty well for this year. I appreciate the opportunity. And the question is a good one. I think that we often get lumped in with companies, which we're actually quite different from. And I think there's 2 reasons. When you talk about what we do as a company, in process control, specifically inspection and measurement, I think what differentiates KLA from our peers in this overall semiconductor capital equipment is actually structural. It's structural in 2 ways. It's structural in the equipment that we that we produce, the complexity of that equipment is structurally different than process equipment. And it's also -- the industry is different structurally, and I'll get to both of those things. First of all, inspection and measurement has evolved over the years from when I joined the company many years ago, to now has multi disciplines at which you have to be an expert in whether it's optics, electronics, metal, all these things, algorithms come together to create very complex tools that it's very hard for people that aren't in the space to be able to afford to compete, and we'll get to the IP part of that in a minute. So we have this advantage that the tools are incredibly complex, requiring lots of different disciplines. As a result, 40% of our workforce has advanced degrees. We hire a ton of PhDs in areas that you simply have to have the challenging work that we have to be able to attract these people. And that creates the advantages, and it's in areas that are very esoteric, whether it's in optics or plasma physics in just a part of it, not like the edge guys, but in the part that we do, whether it's in algorithms for defect detection. And even though there are companies that seem to do what we do, maybe lithography being the closest example, it's really quite different because of another reason is that we have a very large number of products, so very high mix, low volume. And so we have to be very good at producing offshoots of tools that look similar, but in fact, do very different jobs for inspection and measurement. So those 2 create what effectively becomes differentiation and every peer that we get that tries to show up will often come to market, and they'll come with something that might work but won't perform as well and won't have nearly the margin profile because they have to buy things that are more off the shelf. At this point, we've vertically integrated development in optics and electronics and lasers where we have custom capability in every subsystem that allows us to perform better and ironically at lower cost. So we end up with higher margins than the rest of the space because of the way the industry has evolved and structured. So back to the IP point, we end up having tremendous knowledge about how to drive capability in pretty subtle subsystems. So I'll just give you one example. We acquired a company -- and I know we'll get to this, Orbotech. It turns out we're able to custom design optics because we have an optics design team in Germany that can allow us to get optics that are much lower-priced than if you go to the optics providers and buy them off the shelf. So we're able to bring a lower cost design that has higher performance because we can afford to have a design team. That's just something that our competitors who have point products can't do. So over time, that has allowed us to have this differentiation. And the other thing is it's very low volume. So you have to really -- in the end, it's harder on our operation side. But in the end, it's very hard for people to get a footing. And we'll talk about countries that have tried to compete, but they -- in this business, they've never had a big enough market to be able to sustain the R&D that's needed. Does that help?

No, that's extremely helpful. I wonder if you can elaborate a little bit, though, Rick, because I think that we on Wall Street spend most of our time when it comes to process control thinking about the technology you need to find a defect. You talked a little bit about algorithms and the importance of algorithms in kind of the moats that you create. So how should we think about kind of what you need to be able to do once you actually find a defect effectively?

Yes. So the biggest challenge has always been in inspection. And if you just back up a little and think about how inspection works, you really -- the way it mostly works is you're looking for a difference between one semiconductor chip and another. So you're taking images and you're analyzing the difference and you're looking for subtleties. That has always been an algorithmically intense operation. So years before people were calling it AI, we were effectively doing very high-speed advanced processing of that imagery to find anomalies or differences. We also do it to the design. So we do it to the database, and we're looking at -- and we pioneered that technology years ago in our reticle inspection, so we're doing algorithms to understand what should be there and what is actually there. It creates for us an opportunity, but it -- there's tremendous amount of data that we have to process. So we were one of the earliest users, for example, of the graphics chips that are now famous for other applications because we were looking for very high-speed ways to do that. The other thing is image shaping. So making sure that you have -- it's not just the optics to detect, it's the image that you have to illuminate the substrate with. So the laser, in our most advanced systems, we have custom-design lasers because they're none commercially available that provides the capability. We have a capability where we'll explode a plasma and evolve to create the photons that we need. So it's all these things that come together, and then you have to process that data. Nobody else really does that in our space. So everybody else will do an approximation, which may produce a result but won't result in the same fidelity of the image, the sensitivity and ultimately, the value to our customers.

No, that's helpful. I think the other big existential concern out there is you've been so dominant for so long in the optical space, but as Moore's law progresses and feature sizes get smaller and smaller, the industry is going to be forced to go to new sort of detection technologies like e-beam, like actinic. And I think you guys always live with that -- the investors always fear that when those transition points happen, you won't be ready. Can you kind of sort of explain to us how you think about market opportunity, timing and intersection of your R&D dollars with real market opportunities around things like e-beam and actinic?

Of course. Yes. The -- this is because it's so complicated, it might be a little hard to simplify it. And what I mean by that is, in the case of inspection, what you're not doing is resolving an image. And I say that because resolution is what happens in lithography, you have to resolve the image to print the image, which is why there's this clear delineation when you go to the next design role, and EUV is the greatest example where you had to skip wavelengths to get to a smaller one. Because defect inspection is about detecting contrast, you don't actually have to resolve the image in order to find the defects. So you have to just be able to detect differences. And what we found is optics, if you give enough light into this substrate, you can produce differences that are well below what is you could optically resolve, which is why waiver inspection has lasted much longer optical than you can imagine it would have if you'd gone by that strict definition. And I'll give you the most sufficient -- we talked about Gen 5 being our most advanced technology. We're in the early innings of Gen 5. We're seeing great adoption for it. But when we talk about the technology of gate all around, the transistor that's coming for gate all around, that is going to be a Gen 4 solution. And the reason for that is because the wavelengths of light necessary to detect the defects of gate all around, it's not about resolution, it's about contrast. And in that case, as you go up in wavelength, so we're actually extending Gen 4 for gate all around. That's not an e-beam application, and that's not a Gen 5, that's a Gen 4 application. So what happens is they're only a very small number of defects that are at the most critical design rule, which will require, let's say, e-beam will have the ultimate sensitivity. But most of the process is not -- no one would design a process where every single point of failure would be at the most advanced node. In fact, it's a few layers, which is why when we talk about the adoption of EUV, it's only a few layers that get EUV initially and then it fans out. And of those layers, most of those defects are detectable with Gen 5. And so what we found is as we pour more light in, we can detect more with more algorithms. We probably got a whole another generation using AI of the optical. Now at the same time, we also have developed an e-beam technology. And we've announced the eSL10, and it's on the market. And it's -- we didn't want to do it until we had a differentiated design, and it works in complement with our optical systems so that customers can pick. The most compelling marketing we've ever done was after we bought Tencor, and I realized early on, customers want a choice in their portfolio. So you have the best opportunity of solving customer problems if you give them a portfolio of solutions and don't tell them, this is your answer to everything. It's multi-beam e-beam because that isn't the ultimate answer to everything, it's a segment of the answer. So e-beam will be a part of the market. But our analysis shows it has been 20% of the market for 10 years. And because optical advances, e-beam advances, we could talk about actinic, which is a special technology used for reticle. But again, it's a subset of the answer, and we're still developing it. So we're not saying that you don't need it. We're saying that customers will default to the lowest cost capability to provide them the information that they need. And once we figure that out, which is back to my Gen 4 extension example, customers will always want to default to the lowest cost solution, and our job is to provide them a portfolio, and that's what we've done. And by doing that, we allow them to mix and match these technologies to satisfy the latest problems, and they're going to be different as we go to 3 nanometer. And as I said, the architecture around the transistor impacts that, the materials impact that. And so all those things require us to have an understanding of what's coming and, therefore, to be able to handle it. Our view -- and this is part of why, I think last year, we reported we gained market share overall in process control. When we laid out our 2023 plan, we think with the things that we've got in place, we're well positioned to continue to marginally gain share, we're not talking about a huge amount of share, 0.5 point per year over time because our customers are always going to seek alternatives. We're well aware of that. And there are a lot of competitors spending a lot of money to provide those alternatives. So we're aware of that. In contrast to how much we spend, though, it's not even close. We spend more in R&D than the revenue of our biggest competitor because of our large portfolio market position overall. And as I mentioned earlier, that R&D that we have scales across all our products. So we have that additional benefit. We'll develop sensors that can be used in multiple products whereas our competitors are stuck trying to develop those capabilities for maybe one.

But given that some of our more recent channel checks suggest that a large logic customer converted to Gen 5 recently, it sounds like you've got a pretty good track record of figuring out what the market needs and when it needs it.

Well -- and John, as we discussed last year at the Analyst Day, I think you might have asked that question, what about e-beam? And why did the world believe it was going to e-beam? Our customers thought it was going to e-beam, right? So it took them Gen 5 in use, and in some cases it wasn't the first implementation. I have to tell you, the other thing that we go back to is algorithms, the subsequent layer or generations of Gen 5 will get stronger and more capable because it will add more light and we'll add more algorithms. And those things allow us to solve more problems. And our customers, we had one that dedicated years ago told us, don't do Gen 5, we're going all e-beam. And now they're yelling at us for Gen 5 capacity, and Bren is being beaten up because operations report to him about how you can get that. And that's kind of a good problem for us, but nonetheless, I think we'll have the biggest year for optical inspection ever this year, Bren, I think that's right, right? Bren if you can...

Yes, year-over-year growth in excess of 40% in that business, overall.

And if you look, this is the biggest single year of optical, in spite of it also being the implementation of the e-beam, right? So the idea that e-beam was going to take over is simply not playing out. And it continues to be the challenge, e-beam is a point solution and optical keeps getting better.

Well, Rick, that's a good segue into my next question on kind of EUV and the impact on your business. I like to say that after ASML, you're perhaps the most levered semi cap equipment player to the introduction of EUV into the marketplace. And you guys have talked about the difference between shrinking and scaling, what that means. Maybe you can elaborate a little bit about how EUV has impacted your business?

Of course. Yes. I think that we were -- we looked at it from a couple of fronts. One, there's always an advantage -- a benefit to process control business if defects that are smaller matter. And that happens when design roll shrink and suddenly a defect that before was benign now becomes critical. So what happens is that drives adoption. You have to inspect more to find those defects, those wafers are cleaner, they have to be. So you have to look at more, and you have to look for smaller defects because those defects then become more critical over time. So that just drives adoption. But the other thing is -- EUV, so we have that traditional scaling for years. It kind of slowed for a few years, frankly, waiting for EUV once we are kind of stuck in this. And I think we saw a foundry investment kind of go flat for a while, and we also saw process control intensity not increase. But now with EUV, what we have is additional change based on just EUV, and it happens. It starts in the mass shop where we need different capabilities to detect the defects. And we've extended our optical platform. We're looking at our 6xx and it had -- this is an example where even we didn't think could be as useful as it's been for as long as EUV on EUV, but we kept adding capability to it to provide a bridge for customers until they get to actinic with high-end EUV. So that's driven a mass business. And then you print the wafer and you have to validate that the design is what you thought it was, that's driven Gen 5, what we call the mass qualification or print check application, that's driven adoption, and then just a smaller defectivity. So in net, the tailwind from EUV is slightly bigger than we thought. The other thing that's happened that maybe is a surprise to everyone, is the number of designs have gone up as we've gone to advanced nodes, and that certainly isn't something that people were predicting. In fact, not you, John, but some of your peers were saying there'll only be a couple of companies that can afford a 7-nanometer designs, the number of design is going to go to 2, 1 and then we'll have none. It's gone the other way. And so that complexity has actually driven more process control. So EUV drive scaling, it drives smaller defects and it drives a technology change that has caused customers to drive up adoption of process control both in defect inspection, but also we're seeing now increasingly in overlay because the overlay challenges of EUV as they implement are getting higher, too.

Rick, you mentioned earlier in your comments, both Orbotech and Tencor. And I'm old enough to remember when you guys bought Tencor in 1997, which is why despite all the money you spent on rebranding, I still want to call you guys, KLA Tencor instead of KLA Corporation, and I apologize for that. But I wondered if you could talk a little bit about the Orbotech M&A because it's a little bit different than some of the M&A you've done in the past, which was really broadening out your footprint in your core markets. This was really about also driving additional core markets to go after. Talk about the strategy and kind of grade yourself on how you think you've done?

Sure. So I think from 2 standpoints, we're looking at with Orbotech. One was, are there adjacent markets, which we think will have higher growth where we can leverage our core inspection and measurement capability as a way to continue to expand? And the one that we've always felt was going to get more important was going to be the more than more play, the packaging, the things that are going on. And that, I think it's undeniable, even though it's taken a long time, I remember questions about TSV inspection years ago, but packaging is becoming more important, and we're definitely seeing with 5G, the benefits of having printed circuit board inspection and not just inspection, but also process capability for PCB. So those -- that was one that we always felt like the more than more and we had this ICOS business, and we felt that if we had more capability in packaging, what surprised us a little bit is the number of big semiconductor guys that are pulling on packaging there was one thesis that said we're going to have all these new customers, which there are some of that, but it's actually our old customers that are doing more in packaging and maybe driving more ultimately of the upside opportunity and we're getting a seat at the table where it was much harder for Orbotech to get that seat at the table because those guys are wanting to make sure that it's a big enough player that they can hold them into account. The other business that we truly didn't fully appreciate until we got in it was a specialty semiconductor process business, which happens to be very well positioned. It's a very well-run business. It's very well positioned for power, some of the dynamics that are going on in the automotive, but also 5G, these other areas, and that's a business that -- it turns out, we think we can add a fair amount of value to also. And that's been a great pleasant surprise, we didn't fully appreciate what a great business that was when we originally did the deal. Flat panel was the other part of the Orbotech. We never expected a lot out of flat panel. I would say that flat panel is at a point now of maybe bottoming, and we're going to hope to see some recovery but that was never the thesis of the deal. It was much more the PCB business and the specialty semi as augmenting. We're impressed with what we've seen. I think we feel good about where we are in synergies and we've been very impressed with the technological capability of those teams. So I think so far, so good. We're tough graders. So I look at it, the comparison would have been buying back our stock or making that investment. I think at this point, it's been slightly positive if we can see some of these realized synergies and revenue, which we're getting early indications, but we haven't seen yet, then I think it will be definitely a major plus for us. And we're well on target for the 2023 plans that we laid out. Bren, I know you spend a lot of time on Orbotech, so maybe you want to weigh in here as well?

Well, Rick, I think you covered it pretty well, overall. I mean it does expose us to -- I mean, we've always had strong exposure to front-end semiconductor manufacturing, and that's clearly something our customers are driving. And driving today with the introduction of EUV as strong as they ever have. So this does expose us to the other things that they're doing to drive performance or power consumption or just overall cost. And so as you look at the integration of the printed circuit board and how it integrates into a more complex package over time, certainly, that exposure for us, the broader channel across the specialty markets and packaging, just gives us, I think, a few more vectors in terms of where our customers are focusing and where more value is being created in terms of long-term performance. On the synergy front, we feel very pleased from a cost point of view, and we've really gone after it in a couple of ways: first, traditional synergies, which I think everybody generally understands around corporate and supply chain and so on. Channel, there really wasn't much here. I think one of the things that we're pretty happy with is that we've been able to drive an overall level of efficiency out of the business over time. And we're on our way there, and we're not there yet. But I think as we look at the next couple of years as we introduce systems and can deploy what we call the operating model or the KLA operating system into the business, that we ought to be able to drive higher leverage on the expected growth of that business. We expect $400 million to $500 million in our 2023 plan. At Investor Day, we talked about $400 million to $500 million of growth between now or between '19 and '23. And then on that growth, we felt that even though the margins, the margin ratios are lower in those businesses that we could drive incremental operating margins or the drop-through on the revenue growth at 40% to 50%, consistent with what we've seen and what our model is for the semiconductor process control part of the business. So, so far so good. I think as Rick said, the top line opportunities are really what we think will add that incremental value and we got to make some progress there. But overall, I think we're in a pretty good place with where we're going directionally and are pretty excited about the opportunities here.

And maybe, John -- sorry, but just one last point on that. I think the validation of the KLA operating model with the business as different as what Orbotech, in many ways, is the most encouraging for the long-term success. That's been the part that I think we've all come to realize the benefit in that because there was the question, could we run these very different kind of businesses with this operating model. And I think we're proving that we can, and it's making the overall model stronger.

Well, to that point, Rick, and Bren, I'm glad you brought up margins because it gives me the opportunity to ask the question that I ask almost every quarter. When you guys bought Orbotech and then had your Analyst Day, I guess, a year ago or so, you set out kind of a long-term gross and op margin target for 2023 and you've been executing above those targets, at least of late. How much of that is transitory relative to the mix and what's driving the business today? How much of that was just conservative guide? How much of that was better Orbotech integration? And when are you going to raise those targets?

Well, certainly, the growth that we've experienced over the last year has been a driver to that, right? The leverage that we've gotten on the infrastructure of the company has been really good this year. But it's not just that, I think that the introduction -- and we talked about Gen 5, but we've got a number of new platforms in the company. And usually, when you introduce a new platform, it comes in, the margins are a little bit lower, the design isn't as mature, sometimes supply chain doesn't perform and you have to retrofit. And there's a lot of things you have to do that just create some underlying in efficiency and a new platform. One of the things that I think that we've been really pleased with over the last year or so is that as we have rolled into the next iterations, both from a pricing but also from a cost point of view, is the engineering execution has been pretty good. And so the introduction of the platforms, the value that we're able to drive with the platforms has been pretty solid. I mentioned the -- just general operating model deployment and overall margin improvement. And I think we're seeing some of that in the acquired businesses. So there were pieces of it, and it was a different revenue levels and there was a mix assumption built in. But I do believe that a lot of this is structural and that we'll be able to drive this -- I don't think it's going anywhere anytime soon in terms of -- that we're seeing some short-term inflection. I do think we're operating probably where we talked about 61% plus type gross margin profile. I think we're at the higher end of that range. And I'm not introducing a new model today, but at the same time, I'm pretty pleased overall with where the trajectory of performance is right now on the margin line, and I don't think it goes anywhere. I mean quarter-to-quarter, we always have mix issues that will drive underperformance or outperformance across the business. But whether it's the platform dynamic, whether it's general growth of the business, we haven't talked about service, but certainly, the leverage that we're driving in our service business are all factors that are leading to above model performance that I believe is sustainable.

Yes, I could spend the next, I think, hour talking about company-specific drivers and still not exhaust my questions, but I want to spend a little bit of time just talking about kind of your views of overall WFE, both kind of in a near-term perspective, but probably more importantly, on a longer-term perspective. Rick, you and I have talked about the fact that I think one of the things that holds back multiples -- public multiples for all the companies in this space is investors' unwillingness to underwrite the fact that this industry is growth cyclical again, and not just cyclical. As you look out over the next couple of quarters, but more importantly, over the next couple of years, how are you thinking about WFE process control intensity, capital intensity, cost of capacity?

So maybe I'll take part one and then let Bren add on. I think that -- fortunately, what we laid out in our thesis for last year is being borne out in some degree. I mean, this is the first year I can remember where there's been an economic recession. And yet, the semiconductor industry and capital equipment has grown significantly. Now obviously, this is a unique, hopefully never to be repeated scenario. But what is clear is the digital transformation that is going on in the world has been, if anything, accelerated through the events of the last year. And the multiple drivers for the semiconductor industry, as John, as you pointed out in your work in the past, are pretty clear on display now because we've gone through a year where it's been the second year of handset decline. Now we think 2021, they're going to grow again. And yet, that would have been the past spell downturn for the industry. And yet, the other factors which we always believed were coming, have contributed to overall industry growth and anticipated continued growth. So the diversification of end demand is playing out. The broad-based nature of the ubiquity of semiconductors is playing out. The fact that there are multiple designs at advanced nodes if you will, that we're back on a cadence where there's more applications, more use cases. And as more are being discovered, more are being created. I think this industry can take some credit in a lot of the analysis that's going on that's produced some of the results we've seen in terms of vaccines because there's just so much more computing capacity that people have to analyze data and information. So AI is definitely playing. So then you look at overall and you say that the drivers are all in place, there seems to be less cyclicality and then to Bren's point, the argument -- and we'd love to share our deep dive on this, the industry is less cyclical and KLA is less cyclical than the industry. So that we're actually in a point where we're behaving more and more like companies, which have enjoyed higher margins because we have growth cyclical and we're less cyclical than we were and more growth than we were. And we're driving this bottom line performance, which we've talked about. And reasonably for us, we've demonstrated the ability to expand beyond our traditional markets and drive increased performance across a variety of kind of businesses and scenarios. And then the service piece, which is driving also its own growth vector for the company. So I think all those factors, I think '21, if we had gone back to the beginning of the year or a year ago, and you told me some of the macro events for '20, I don't think I would have guessed this industry performance. I'm sure I would not have. But it only sets up well for the future because a lot of the things we're learning societally from all the digital transformation is only going to build on as we go forward. And so I think those dynamics are good. Bren could speak a little bit to the drop-through. I mean our capital requirements are incredibly low, and we feel really good about, as we said, the model, and we're going to keep driving. And Bren can speak more to the specifics.

Well, you've got the end market adoption part of just what's -- it's not all about PCs and PC infrastructure like it used to be. And then 2 things are true here that you've got capital intensity flat overall for the industry or slightly going up now and our customers on a through-cycle basis are more profitable than they've ever been. And I think if you look at all the efficiency that came from -- that drove declining capital intensity, whether it was wafer transitions or whether it was consolidation, the movement to foundries, there's lots of issues that drove efficiency that kind of kept WFE from growing even though semiconductor base growth was growing. All that has been played out generally now. And so the opportunities are there, and customers have to invest in them. And I think it's been good overall. I think for discipline overall from a spending and pricing point of view. And it's changed the industry from a modest growth or very low growth to a structural growth that we believe at a minimum is at least in line with semiconductor revenue growth rates on a go-forward basis for the broader industry. And that with process control intensity related to a lot of the dynamics we talked about and some incremental share opportunities we think are out there, that we ought to be able to grow a couple of points faster than the overall industry and the systems business. And another point to 2 points that comes from just service at a higher growth rate overall. And then drop-through on that revenue growth rate of 1.5x earnings growth -- 1.5x revenue growth rate in terms of earnings growth, which is our long-term 2023 model and how we run the company and how we make sizing and portfolio decisions underneath the top line.

And Bren, if it's fair to say, as was reflected on your most recent conference call that if you look at near-term order patterns, backlog, all of these dynamics seem to be playing out in the bottoms-up drivers of the business.

They do. I think one of the changes we saw as we moved through last quarter was certainly strengthened not only in the order profile in the current quarter and what we saw in December, but also how things were filling out in the first part of the year. So I think what was important was sustainability of the business levels that we're seeing today as they continue in the first part of '21. We haven't put a point out there around what we think WFE. We think it's a growth year for next year. We don't think -- see any change really in foundry logic investment, we think that's very sustainable into next year. And we think memory is better overall. And so that should drive growth into the industry. And we'll have more to say about that when we get to our earnings call in late January. But I think the setup is pretty good, and the margin discussion and all the other things that we had earlier, I think, will sustain itself as we move into next year and see another growth year for the industry.

Perfect. I think that's a great way to end this session. We've run out of time for this fireside chat, but I want to thank both Rick and Bren and everybody virtually tuning in for this. And I want to pass along our well wishes. Rick, Bren, I hope you, your immediate families and the broader KLA family continue to be safe and healthy in what's been an interesting and challenging 2020, excuse me.

Yes. Thank you, John. Thank you for this opportunity.

Thank you. Yes, likewise.