Global TMT Conference. My name is Atif Malik. I cover U.S. semiconductors and semiconductor equipment stocks. It's my pleasure to welcome Bren Higgins, EVP and Chief Financial Officer, KLA, as well as Kevin Kessel, VP IR and Market Analytics at KLA. Welcome, Bren.
Thanks for having me. You know, you almost got through it.
I know. Rick—
It's been seven years.
Yeah. Rick was not happy. I remember a couple of years ago when I almost said that.
Tencor didn't exactly just flow off the tongue, but thank you for having us.
Yeah.
Giving you a hard time.
All right. Definitely, KLA. All right. Bren, you lifted your Wafer Fab Equipment expectations this year to mid $150 billion and talked about upside potential $290 billion WFE next year. You mentioned you're having conversations around deliveries to second half of next year. Can you just help us understand how you're seeing the market shape—
Sure.
—in terms of Wafer Fab Equipment this year, next year, and even to 2028, if you have any commentary?
2028. Yeah, I will just level set a little bit on where we are at right now and some of the themes that came out of our earnings results back in July. It has been a pretty exciting time at KLA. If you just go back and look at in 2024, we grew 12%, 2025, we grew 17%, 2026, if you take our performance plus some of the guidance we gave, you end up in the low 20%.
So, over that timeframe, the financial model has moved. We have been at the upper end of our incremental operating margin model, and gained about 400 basis points over that timeframe in terms of operating profits. So, one of the strongest models in the industry, both in terms of opportunity and share of market, but also in terms of leverage and operating performance has played through, and we are pretty proud of that performance.
2026 has shaped up to be a year of increasing momentum, as you said, right? We started the year, went back to our Investor Day. We thought the industry would be somewhere around $135 billion. That was just in March, $135 billion- $140 billion, and we did talk about greater than $150 billion expectation. So while our customers are struggling with space, they are figuring out ways to take equipment. First half of the year, a little slower in terms of sequential performance growth into the second half of the year, half to half, 20% higher in the second half versus the first half. So, we are excited about that we are starting to see the momentum building, not only in what is happening with our customer engagement, but also in our ability to supply and get the needed components that we need for our systems.
2027 looks to be a pretty favorable environment. I think growth rates into 2027 are probably in a similar range, maybe higher. We will see as we get closer to it. Some different views on that. Our orders are certainly our backlog is very strong. We just disclosed our backlog just short of $13 billion in our 10-K that we filed back at the beginning of August. And the order funnel has been very strong, and I would expect that backlog to continue to grow. So the visibility that we have from our customers into shipment expectations through 2027 and in some cases into early 2028 is pretty good. So, our customers' profitability levels are extremely high, and so they certainly can support the level of investments that they are making. And the priorities for the company in this environment for us is just to, first of all, the execution.
I talked about supply chain and trying to get systems out the door, aligning to our customer expectations, delivering on next generation products. I talked about the operating model earlier. That comes from the innovation that we do at KLA, the differentiation in our products that is reflected in our gross margins. Our gross margins, I think, are the best indicator of differentiation in the industry. It is typically what we point people to. If you have a differentiated business or not, it typically is reflected in the gross margins. We need to be able to deliver those programs that allows us to do some adjustments for pricing, both in terms of delivering new capability, but also dealing with costs. It is pretty important effort for us not to lose sight of our development requirements at the same time we are ramping to execution.
Driving our ability to support our customers through service applications, and so on, so we can keep these fabs running at high utilization levels. Given the value of the chips that they are producing, both in logic but in memory, yield is really critical. Printing money, if you will, in terms of incremental yield performance, and how that translates to their financial results. So that is important for us to make sure that we are in position to be able to support them.
Great. Bren, let me double-click on two things you mentioned, backlog and the supply situation. On the backlog, in your filing, your backlog grew 60% versus FY 2025 to $12.6 billion. How should we think about the quality and the duration, or what portion of this backlog is expected to be converted to revenue in the next 12 months or so?
We did not disclose that, but we are taking orders basically for the next, in most cases, it is 12 to 18-month time frames. In an environment like this, you might be further on the end of that, and it can be product specific. Certain products, for example, our Gen 4 and Gen 5 tools will have longer lead times than our metrology tools, for example. But it is typically in a delivery window that is more or less in that time frame. You can have facility schedules that will drive a customer to place an order to ensure that they are getting tools when that facility opens. Those schedules are tied to construction. In some cases, you will see orders from customers to ensure that they are slotted in the appropriate timeframe for that construction schedule. Our engagement with our customers is very high.
I mean, look, our top five customers drive a lot of our business. But getting the orders tends to solidify, particularly in a constrained environment, both timing, commitment, and pricing in. So certainly, getting the orders is really, I think, important to just lining those customers up the way we'd like to see it. So quality is always good, right? I think in this environment, customers, we behave differently when we get an order. So we want to make sure we get the order, and I think it helps set up how we're thinking about how we're building our own capacity, how we're driving capacity into our supply chain.
On the supply side, Bren, where are the biggest constraints for KLA today? You have talked about optics components being tight and sensors compute, and just help us understand from a supply perspective. And some investors are kind of extrapolating peer commentary of doubling manufacturing to some sort of doubling WFE in 2028 to $300 billion type numbers. But from your perspective, where do we stand on supply constraints?
Well, typically where we see the most constraint typically is around optical components. Optics have an intrinsic lead time that is just very long. Calcium fluoride lead times can be 18- 24 months or more. The equipment that's used to make optics has a long lead time. So it's something that you have to plan for, not just what you're doing in the short run, but you also have to think about long-term. We are having conversations with our key optics suppliers around capacity requirements out into the 2029 and 2030 timeframe. So you definitely have to plan with a long-term horizon. And we think that generally ensuring that we've got the capacity will take inventory. It's important for us to ensure that we can meet where our customer lead times are not necessarily going to match that lead time. So that's something we have to take a long-term.
We take a very partnership-driven approach to how we work with those suppliers, how we invest in both development but also in capacity, and it can take different forms. But they're pretty critical to us and if I could get more, I could probably ship more today. But we feel like we've got a pretty good trajectory in terms of our plans and what we're planning for as it relates to the different scenarios of growth over the next year and a half or so. Part of the reason why we've seen the second half start to improve and our construct on 2027 is new capacity that's come online that supported the second half. The industry pivoted very quickly. If you go back to this time last year, we weren't talking about $150+ billion .
We weren't talking about growth rates into 2027 that are similar to what I mentioned earlier. Anytime you have a quick turn like that, it does put a little bit of pressure in the short run. But in the long run, and our customers understand this, and we work with them in terms of slotting out how we can deliver capability, we're seeing more come online here in the second half, and we'll see that continuous move into next year.
We're fair to conclude, like you said, your outlook on second half is indicating that some of those constraints are improving.
Yes, for sure.
All right. Let's talk about the end markets. DRAM is expected to lead the growth this year, and even in next year. You guys benefit the most from foundry logic growth. There are many more steps, many more types of chips and mass sets and all that. That kind of plays into your strength on your portfolio of different inspection and metrology. So where are we in terms of the foundry logic growth? There's been this transition to gate-all-around for the last couple of years. We're seeing somewhat more demand on 3 nm and customers are moving to 2 nm. So help understand, how do you look at the foundry logic market this year and next year?
Well, foundry logic tends to have higher intensity for a lot of the reasons you talked about than memory. Foundry logic, over the last year to two years, has had a broader level of participation. It is more than just one player investing, and so that is good for the efficiency. The industry was very efficient at the leading edge. You could argue maybe a little less efficient going forward as more players are investing to meet what is a pretty compelling opportunity that is out there over the next several years. The technology roadmap, both from a scaling point of view but also architecture, has been good for our business. The design environment tends to drive process control intensity, and that is a good thing. More designs moving through fabs means our customers have to manage a much more dynamic environment.
They have to deliver to tight market windows, different designs, test design rules in different ways. More designs, I think, benefits KLA in a unique way. The fact that you have more designs moving through fab does not necessarily change a process opportunity, but changes process control a lot. The value of those chips that come from those designs also is a factor that just if you have more at risk, you tend to invest more to ensure that it is functioning, not just that it works and yields, but it also bins consistent with specs and performance. So those are all good factors for process control intensity. We have seen intensity in the 2 nm node is higher than what we saw in 3 nm, and I think it is driven by a lot of these factors.
Big die also carry defect density challenges, which is also where you have the same number of defects on a wafer. The yield impact is more significant when you have much larger die, fewer die on a wafer than in mobile, for example, where you would have more. So all those are unique opportunities for process control, and we have seen it translate over the years in terms of higher intensity, more relevance, not just in accelerating yield cycles and ramping a fab, but also in high volume production. So that has been, I think, a change that has been good for us. Memory is changing though, and it has been good. High-Bandwidth Memory carries some unique attributes, both in terms of performance, less redundancy, larger die, more logic-like attributes, more logic in the base die, for example.
In a High-Bandwidth Memory stack, for example, the actual device only performs as good as the weakest DRAM in the device. So that has been a good thing for process control intensity. In some products, we have seen intensity in High-Bandwidth Memory that has been consistent with what we see in advanced logic. So overall, we think High-Bandwidth Memory has been a positive. It is not the same memory. Conventional memory has been a driver this year, and there has been more tech upgrades in terms of upgrades of existing capacity that historically has been a bit of a headwind, although you have more EUV layers and conventional DRAM devices today than what you had in the past.
Conventional memory is better, High-Bandwidth Memory is better than that, and advanced logic is carrying some of these unique opportunities that we think creates an environment not just for more KLA relevance in the early phases of a ramp cycle, but also in production. We are seeing backporting of capability into N3 today, in addition to some of the N2 investment. As we move into 2027, we are going to see the next investment in the next node from multiple players. I am pretty excited about that.
Great. When you mentioned HBM is logic like there is higher process control intensity, one of the concerns clients have is that with this kind of memory de-speccing aspect, some of the AI servers are putting less memory into their servers and maybe coming down the stack from 12 high to 8 high or something. Can you talk about what impact, if any, that will have on your overall equipment demand if the industry is in this transition period where they are stacking less dies versus historically?
I think the roadmap dynamics I talked about earlier, those are generally unchanged in the HBM environment. Obviously, the reaction from customers is to the pricing environment that exists today in memory with the supply-demand imbalance. I think these are techniques that I think customers are looking at how they can become more efficient, and because as they even see with our own memory costs, I do not like paying the prices that I am paying for memory today. I do not think they like it either. I think these are adjustments they are making to try to get more efficient with the memory that they are using. But the roadmap dynamics around redundancy and I/O count, size of die, performance specs, you cannot bin into a lower end device. You cannot bin an HBM device. All those things we think are pretty positive in terms of overall intensity.
I think that is more about the pricing environment and what is happening in terms of supply relative to demand that is driving that behavior.
Let's talk about new customers. Are you seeing a broadening in foundry logic customers? There's a new fab or a new project in the U.S. The NPU maker is talking about significant increase in their CapEx next year. It's not really a new customer, but they're definitely ratcheting up their plans and spending. Just in terms of the kind of greenfield opportunities, you guys do benefit from that a lot more initially, versus brownfield. Just talk about how you see the broadening of foundry logic customers kind of expanding your TAM.
Well, like I said earlier, one thing about the last several years, there was a lot of investment, if you go back a few years ago, a lot of investment in logic that was happening in the legacy nodes. But at the leading edge, it was incredibly efficient, effectively one player investing. As you look moving forward, you have multiple players, and so we're excited about the opportunity. Obviously, they feel that it's appropriate to invest given the opportunities they're facing. But that inefficiency that comes from multiple players investing will be good for KLA. I think we've gotten really good traction for the reasons I talked about in terms of the kinds of devices that they're making. But also in our collaboration model, working with customers to deploy our capabilities in ways that accelerate those learning cycles, that show KLA value.
I'm optimistic that because that's the way it translates into high volume production capacity adds. So if we can demonstrate that value earlier on in the investment cycle, then it does translate into more investment in capacity. So we're excited about those opportunities. We're investing to ensure that we have the resources to support these fabs in different locations, and also ensuring that we have not just people to service and to do applications, but also inventory to be able to support. So there's a lot of activity there that more customers is better. Excited about that.
Let's talk about advanced packaging. I think one thing that has kind of bifurcated your stock performance as well your fundamentals from lithography in the last few years has been your participation in the advanced packaging market. You're talking about 70% year-over-year growth, which is 2x the rate of the market growth in the advanced packaging market. The idea is that you guys have these front-end tools that are very much capable of looking at larger geometries and all that, and you've been able to enter that market and take share. Can you help us understand where are we in that kind of adoption cycle of your kind of front-end tools being increasingly adopted on back-end advanced packaging side, how much more room there is to expand share and—
Yeah, no, it's a great opportunity. If you just go back a few years ago, our share of the advanced packaging market was a couple percent, just a few years ago, 2023 timeframe. Today, I think this year in 2026, likely to be somewhere between 7% and 8%. So it's been a great opportunity, and as you've moved to more front-end like processing in advanced packaging. Historic packaging was fan-out packaging. It wasn't particularly complex from a process control or an inspection point of view, and sampling rates weren't particularly high. The need for advanced capability was much lower. It was a very competitive market, multiple players investing, certainly on the logic front. But as we've moved to high-performance compute packages, that has changed. You've seen more front-end like processing, and that processing has driven more front-end tools.
Our CEO tells a story that he got pulled into a meeting just a few years ago about the need for our front-end tools in advanced packaging. Of course, the reaction from us was, "Hey, these are higher-end tools, you might not like the price." The response from the customer was that, "No, we're moving to front-end processing in packaging, and so we're going to need front-end capability." We had to do some engineering, and we did some engineering in our macro products, which had the highest market share in the segmented part of macro inspection. There was the front-end part, and then there was the packaging part, and there's 2.5D and 3D. We had the front-end share, so it was in some ways kind of logical that when they moved to more front-end requirements, that the front-end share leader would get pulled into that.
It's been very good for our business. More success or more momentum on logic side. OSATs are an opportunity. Memory, less complex, but we think the share opportunities there will grow over time as new technology is introduced. So it's been a great opportunity. Very high sampling rates because you assemble this package, you have to integrate the HBM with the GPU or the custom ASICs. There's a lot of surrounding peripheral die. You have to integrate it, you have to make sure it works and hits specs. So the sampling rates are very, very high, and that's been really good for that business. There's new technology related to hybrid bonding, which is more exposure and moves the sensitivity requirements down in terms of sub sort of one micron sensitivity.
We think that you're going to be even, in some cases, down to even more advanced sensitivity requirements. It's driving more of the front-end portfolio. We've been able to leverage the R&D we did in the macro product in the other parts of the front-end portfolio. So we're getting some leverage on the R&D investment, mostly for handling and handling the different substrates because they're different in packaging than the standard 300 mm hard wafer that you could use in the front end. We're starting to see now this year, for the first time, we're actually seeing with the introduction of this new technology, that we're starting to see some of the higher-end tools adopted. It's a pretty small percentage today of the overall. Say it's probably somewhere on the order of maybe 20% or so.
I think over time, of the total number that we talked about, that 70% growth number is $1.1 billion. As we move into next year with what's happening with agentic and CPU growth rates, and some of the inference chips, it is going to create even more opportunities. We are pretty excited about the die stacking opportunities that will happen there in advanced logic in the packages.
Let us talk about gross margins. It has been a topical topic with investors. We were a bit surprised by some of your systems gross margins being 70%. Those are top class. Any room for improvement in the gross margins to, let us say, to reach your target level or even beyond that? What knobs do you have in kind of value-based pricing, operational efficiencies? Obviously, memory pricing coming down will be a huge tailwind whenever they start to normalize.
Yeah, no, memory has had an impact and when this. I thought it was we were pretty smart early on securing supply very early in the process, but then the momentum in the industry has driven more demand, so we have had to buy more memory at current prices, which we do not love. The impact on our gross margins, which we thought was closer to 75 basis points, is probably slightly ahead of 100 basis points today. So it is certainly more of a headwind. Our target model for gross margin is, if you look at our business, it is generally to drive 60%- 65% incremental gross margins. Target model we presented was between 63% and 64%. We are operating in the low 62% range today. We do have the headwind that you mentioned. There are puts and takes. Value pricing is nothing new for KLA.
Our approach to our customers is a deep understanding. We think it is a core competency of the company, of the value that our products deliver in terms of faster learning rates, yield improvement. As we introduce new capability, we improve our customers' cost of ownership, and we get to share in that value that we create for them. That is typically where we adjust our pricing is when you introduce new capability, improve cost of ownership. That is an opportunity to deal with your costs if you do your engineering right, and also to meet their targets. So customer benefits, but we also benefit. We are able to usually adjust, and we have been, but things like memory have escalated at a pretty rapid rate, so it will take time for that value pricing opportunity to play out. There are puts and takes in mix.
Packaging, we just talked about it, but packaging generally carries a lower gross margin on great gross margin dollar opportunity, which is, I think, really what we are here for. Not every market carries the same margin profile. Growth in our EPC businesses, our PCB businesses, our semiconductor process business, those margins are lower. That business is now growing, supporting the high-performance compute environment. Those carry lower margins, too. I think there is some puts and takes. We factored that in as we thought about the long-term gross margin model that we should be able to drive 63%-64% over time, and that is our plan here moving forward. We will drive leverage out of our manufacturing operations on the scale of volume. There is a ton of investments that are happening really worldwide. I run facilities also inside the company. In every site, you have a project plan.
When you are dealing with structural growth the way we think, the way we laid it out at Investor Day, it is as soon as you get to a place where you think you are done, you are moving on to the next decision. For example, I am going to Singapore later this year to do a grand opening and a groundbreaking on the same trip. I am going to try to be efficient with the trip, so it will be clip and scoop and try to get it all done at once. These are things that we are going to invest in that will support the growth moving forward. But I think that there is leverage in the model. Ultimately, it is going to translate in that 60%, 65% how we run the company and 40%-50% incremental operating margins.
Great. Let me pause here and see if there are any questions in the audience. If you have a question, please raise your hand. All right. We will just keep moving. Bren, KLA uses a lot of AI and software in your tools, and I believe that is one of your key moat. Do you view that AI models becoming more capable to be a net positive for KLA?
It is interesting, we started investing in AI back in 2015 or so. We have had it in basically in all of our products. First product was 2018, 2019, an electron beam inspection product. It is deployed across the portfolio today, physics-based AI. First of all, no one generates more data in a fab than KLA. The ability to process that data, move from a raw image to a processed image that factors in both hardware dynamics, context, applications, and then the algos that ultimately allow us to understand what is going on in terms of signal-to-noise and get ultimately speed time to results is a core competency of the company. We are able to then take some of that capability. For example, I mentioned the electron beam tools where you can use AI to train your inspectors, to point your inspectors more effectively, as an example.
We're using it in a number of ways. We're using it in recipe development and applications. We use it in our systems. We're doing things inside the company to drive productivity, both from an R&D point of view, but also in our business operations. We're pretty excited about it. The architecture benefit is clear. We see in our own image computing where we're able to drive more capability at lower cost. We're big believers in this transition in high-performance compute. What it means, we think it's a pretty good microcosm of what an AI data center is facing in terms of the compute transition. We're seeing it in our own systems, and we think it translates. We're pretty excited about the benefits we see in our own systems, and we're able to use the capability in a lot of different ways.
We didn't talk about services business with all these increased shipments this year, next year to reflect WFE. Should we be thinking differently about the services growth rate beyond the mid-teens range with these higher tool shipments?
Our service business is pure service, right? There's no systems in there. We have an updated target growth model of 13%-15% moving forward. 80% is contract-based, so subscription-like in a lot of ways. The service business, we believe, carries an operating margin profile in terms of profitability that's accretive to the overall. It's a very resilient business for the company. It provides a nice anchor through any sort of cyclical dynamic where it's only had one down year, I think, in 25 years, right? It's a very unique service business. As we ship more tools, it creates more opportunities. Our business is a very relatively low volume, very high mix, high complexity service business. It's very hard for our customers, given the fleets they have. They don't have redundancy.
They tend to run them at very high up times to do their own service. They rely on us to do service, and to have a predictable level of support that shows up in that contract that we can customize to their needs. We can deliver uptime, we can deliver matching performance, and so on. It carries a resiliency, and it will grow as new systems are shipped. If we end up at a higher growth rate overall in terms of our system expectations, certainly when we laid out that target of 13%-15% back in March, we had a view of the industry over the next five years. I would say that that view looks conservative today. I think it does create an opportunity for us to drive more growth out of service.
It will move slowly, but it tends to be pretty resilient over time, and would expect it to be pretty predictable as well.
All right. One question on the E-beam product and the attach rate. You have talked about expecting 80% to 20% optical to E-beam split. When do you see that reaching that level?
Yeah. I mean, historically, it has been about 80/20, and there has been periods of time where it has been higher. I think today it is probably a little bit higher in the higher end of the maybe 85%+ . A lot of it has to do with the relative growth rates of both parts of the market. Customers who use E-beam inherently is a slow technology, so its applicability in production is not as high as an optical solution. So a customer is really trying to, what is the most advanced capability to meet the problem at the most effective cost?
And so typically, what you see is optical technologies in production. But there are certain electron beam opportunities where, again, you will use it around certain defect types where you can only find it with E-beam, and so customers will deploy it in production because they are inspecting for a very specific type of defect.
There are a number of those buried defects in a gate-all-around structure that we have participate in, or use it to point the inspectors. So, I would expect that ratio to hold more or less consistent. You are using multi-beam technologies, which are, you sacrifice a little bit of resolution relative to a single beam, but there are some electrical defects and buried defect use cases that make sense. So the tools will continue to be complementary.
I would expect that one of the fastest-growing markets, certainly in process control, maybe in the overall industry, will be in optical inspection, optical high-end optical pattern inspection. And so you will see that continue. But E-beam has its place and will continue to grow I think along with that. So I do not think the ratio changes all that much moving forward. And I think it is just because of the nature of the physics.
It's the optical solution meets the requirements in high volume production. You'll see that in the vast majority of the use cases.
Great. We're almost out of time. Bren, thank you.
I have a question.
Oh, sorry. You have a question? Can we have the mic here?
I'll bring it forward.
Yeah. Oh, right here.
Bren, I had a question on market share.
Yes.
Applied has been pretty vocal in describing process control as one of the best growth opportunities. That business for them is going to grow 50% this year. Their approach is this integrated bundling strategy. How do you think about that threat versus your position as best in class in process control?
We've gained share the last year. Over the last several years, we've continued to gain share within process control. We're the only provider that comes to in this industry with a portfolio that we have that allows our customers to mix and match, depending on where they are in a development or in a yield cycle, mix and match the kind of capability they need to meet their needs. We feel pretty good about our competitive position. Our business model and market attracts competition. I think that they like the profitability, but we feel pretty good about our strategies. We feel pretty good about the portfolio. If I had a nickel for every time one of our competitors talked about gaining share in the industry, I'd have a lot of nickels.
I feel pretty good about our position and I think we'll continue to gain share here moving forward. One thing that is clear is that if we can deliver more capability, there's room for us to have more share. We've seen as we've worked with customers in the most critical areas, but even in less critical areas, that we can work in a much more collaborative way to gain share. We feel the packaging we talked about earlier, E-beam is another opportunity, and then some of the core markets, we continue to maintain the momentum we've seen.
Thanks.
Thank you, Bren, for coming to Citi Conference.
Thanks for having us.