KLA Corporation (NASDAQ:KLAC) BofA Securities 2024 Global Technology Conference June 6, 2024 1:40 PM ET
Company Participants
Bren Higgins – EVP & CFO
Conference Call Participants
Vivek Arya – Bank of America
Vivek Arya
Alright. Good morning, everyone. Welcome to this session. I’m Vivek Arya from BofA Semiconductor team. Really delighted to have Bren Higgins, Executive Vice President and CFO of KLA joined us, KLA is the most profitable company in semi-cap equipment and Bren will give us all the secrets as to how to make company the most profitable in their sector, but really delight to have you.
Bren Higgins
Yeah, thank you for having me. Good to be here.
Question-and-Answer Session
Q – Vivek Arya
What — I thought we could start with Bren is maybe at the high level, in ’22, you had an Analyst Day where you laid out the target of 9% to 11% topline growth, right, getting to $14 billion in sales and $38 in earnings. I was just hoping that you could maybe take a step back and tell us where KLA is in that journey and what needs to happen from an industry perspective, what needs to happen from a KLA perspective to kind of make progress towards that goal?
Bren Higgins
Sure. And it was an interesting time back then where we spent a lot of time talking about supply-chain challenges and shortages, the ability to ship to demand which was hard for all of us in the industry. I think we navigated it pretty well and then, of course, went through 2023 with a correction from our major customers, but cushioned by some of the dynamics in China.
But the plan we laid out to your point was 9% to 11% expectation for topline growth and we get there with a view of semiconductor revenue growth of roughly we’ll call it 6% to 7% this plan of our view of 2030 at about approximately $1 trillion in semi revenue that we would see a continuation of rising capital intensity. So for the past decade or so, we’ve seen capital intensity rising in the industry and that would drive the WFE part of the market, which is the front-end of investment for our customers and equipment that would grow a little bit faster than semiconductor revenues, let’s call it, 7% to 8%.
If you look at the mix and some of the drivers, particularly given what we’ve seen in the last couple of nodes in terms of design activity and the robustness of the design — designs that are moving through these nodes is that, that would create an opportunity to see a reacceleration of investment in logic and foundry, which went through a period of time where it was really kind of negative, right, from about 2013 to about 2018 with very limited design activity outside of some of the high-volume parts and limited scaling benefits from a Moore’s Law attribute point-of-view. 2019, that all started to change with the introduction of EUV, it’s 7 nanometer node and then we’ve seen that escalate as we move forward.
So given that, that you’d see roughly a 60% plus kind of mix of logic/foundry WFE and that’s an important — the mix is important for KLA because of the relative process control intensity of the logic/foundry business. We have lots of designs. We have customers that are designing for specific nodes. In contrast to memory, where you have more commodity-like parts and redundancy and repair and in the process. And so as a result of that, the process control intensity can be mid to high teens kind of leading-edge process control intensity for leading-edge logic/foundry and memory is a little bit different between NAND and DRAM, but we’ll call it somewhere closer to 10% plus or minus.
So that mix was going to — would ultimately drive the process control part of the market to grow a little faster than WFE. And then if you look at KLA’s position in process control, which is approximately five to six times more than our nearest competitor, that in the markets that we’re in and we’re in a number of segments within process control that the relative growth of those markets, particularly like optical pattern inspection and then also some share opportunities with some new development programs that we thought that we would see that an opportunity for KLA to grow a little faster than that.
So you take that, that’s our systems business gets up somewhere closer to 10% as a result of that. And then you take our service business where we updated our long-term service model to be 12% to 14% through cycle growth. That was up from 9% to 11% and we’re seeing a number of factors that are driving this. The biggest being obviously the growth in systems, which you have tools that go out of warranty and come on to contract. And our attach rate is very-high in excess of 90%. And about 75% of the revenue is subscription-like contracts. So we get a contract stream with an average of about three years. So pretty predictable service revenue stream overtime.
You also have the extending life of our systems in the field that are servicing a number of markets where you’re seeing rising semiconductor content. So automotive, industrial, communication, infrastructure, IoT and so on. So all that is positive for overall service growth. And if you look at where we are today, we generally think we’re probably towards the higher-end of that range than the lower end of that range.
Then we have our EPC, which is the more than more platform for KLA, which is electronics, packaging and components, has specialty semiconductor, which has had nice growth over the last few years, better than market as you’ve seen [Technical Difficulty] collections and it’s a diversified business.
Initially, you saw a lot of drivers related. We saw mobile investment in 5G related to RF capability, then power, semiconductors for automotive and now seeing demand in advanced packaging. So that business has done pretty well. It’s been offset by some weakness in more consumer-centric markets like our component inspection market and our PCB market. And then the flat-panel part of the business, which we made a decision last quarter to exit that business.
So when you take that, I think we had an expectation of somewhere around 10% growth from a systems point-of-view in that part of the business and we think that is likely somewhere less than 10%, just given the growth that we have to grow from where we are today over the next couple of years. Fast forward to where we are today, looks like given WFE levels somewhere in the low to mid-90s today, our view of the ’26 plan got into the 120s. And so I think if you look at where we are today and expectations over the next two years and drivers, I’m sure we’ll talk about, that implies that a kind of a mid-double-digit, mid-teen kind of growth rates over the next couple of years to enable the plan.
If you drop through the plan, gross margins of 63% plus, pretty confident in our ability to continue to drive scale. We’ve made a lot of investments to — that are in some ways headwinds today to be able to position us for improvement in margin as we approach a $14 billion level from the low 10s where we are today in terms of expectations for 2024. And with our incremental operating margin target in terms of how we size the business overall that will deliver 40% to 50% incremental operating margin. So that will translate into operating margins somewhere 42% plus in earnings per share the $38 plus or minus.
Underneath that plan, we’ll continue to be assertive in capital structure and in capital allocation for the company, we plan to — we like the businesses we’re in. We think we can continue to return about 85% of the free cash flow we’ll generate. And so you’ll have that shared between the dividend, which is growing over the last 14 years consecutively and growing at about 15% CAGR and then the remainder in deployment to buybacks.
And so we also did some — did a large debt-financed ASR to sort of reinforce our view of the strength of the model and the relative valuation at the time versus what the implied value of the plan. So I think in general, we feel pretty good about this trajectory. We’ll talk about some of the drivers, I’m sure. But that’s — in a short summary, that’s the plan.
Vivek Arya
That’s great, very comprehensive. We’ll talk about the product side of the business, but the services business of KLA, I find fascinating because it is not really cyclical per se, because with a lot of peers, what I find is what is classified in services, right, is a lot of cyclical usage-based spares, right, as refurbishment of old equipment, right. Of course, there is a maintenance part of it, right, that is more subscription. So talk to us about — maybe help us unpack what is in your services business and what is helping it to grow 12% to 14% this year even though overall tool spending is really not growing that much this year?
Bren Higgins
Yeah. And not really correlated to WFE, more correlated to semiconductor revenue because ultimately our customers running their installed-base is what supporting their businesses is what drives service. You’re right, it is different. And our service revenue stream is all associated with service. There’s no equipment in that number. There are no upgrades or things like that. And there’s limited consumables. And so if you think about the business, it is a high mix, high complexity, relatively low-volume business if you look at what we sell. And then even within each product family, you have versions and variants of certain types of tools.
Our customers buy process control and tend to — want to spend only what they feel like they need to because ultimately you’re really trying to drive productivity and cost performance out of the fab. So we can contribute to that, but they usually start with, hey, we want to spend a certain budget. And then we will have high uptime expectations, 90% plus generally across our systems. They’ll run the tools pretty hard. They like and need matching performance across those systems.
So we’re able to go in and provide an entitlement stream that the customer wants in terms of support levels across different tool types or broadly across fabs. And we can customize these offerings and these offerings can vary around response times and can also include assets like inventory and inventory in the fab, inventory and region and all these things we can try to monetize and then get clarity in terms of what we have to deliver to you over an extended period of time and then do a lot of cost work underneath that to make sure that we deliver to those commitments.
Given the nature of the complexity of the systems, there’s a lot of custom parts. Given the breadth of the company and our engineering teams, most of what we do is custom. And so the ability to procure those parts can only come through KLA. And our engineers are trained such that they can generally service one tool type. So there’s a high level of capability that’s required in our field service engineers to be able to support it.
Even in downturn environments, our customers will always — they run their installed-base and they tend to run process control at a higher level. They’ll take offline some of their process tools, excuse me, as utilization rates come down. But in a capital constrained environment, higher yields and only starting as many wafers as you really need is something that our customers in a finite sort of capital environment are very careful with.
So our utilization rates are always a little bit higher. Contracts, even though they’ll idle some capacity, the contract model stays intact. And so as a result of that, even last year where a number of the service businesses were down because some of the lower utilization rates across most segments, we still saw mid-high single-digit growth last year. And so you’ve got those dynamics that drive service and our service business has grown every year except for in 2009, it was down about 12%. So you have this, I think a fairly predictable stream overtime of growth expectation that you can plan around and I think you can optimize for from an efficiency and profitability point-of-view.
So as utilization rates improve, that’s been a driver. I talked about the extending life and all the metrics we look at for service, whether it’s contract revenue, whether it’s how much of a tool sales price we capture overtime in terms of revenue. We’ve seen all these metrics improve over the last five years, but the useful life of the tools or the service line. We are still servicing tools that we shipped in the 90s that are making parts for a lot of the industrial markets and appliance markets and things like that.
So I think there are unique attributes to our service stream that creates this opportunity. And if you look at where we are today in this increasing growth; A, we shipped a bunch of systems and so we can take the growth rate higher that — and those tools will come off a warranty and go into contract and will drive $250 million to $300 million of incremental revenue this year irrespective generally of WFE and unless WFE levels come way down because our customers take utilization rates down or something like that, but in terms of where forecasts are.
And then we’re also seeing utilization rates improve, which is also a driver to that. So we feel pretty comfortable about that target range. And I think we’re probably trending more towards the top-end of that range than the bottom. And everyone does accounting a little bit differently in terms of how you look at the cost structure, but we believe it carries a stream that’s in line or better than our corporate averages. So it does have an effect on gross margin.
Gross margins are lower, but in terms of operating margins, it’s more aligned with company average. So it’s — it certainly is part of the business that has gotten us comfortable with some level of predictability, gotten us comfortable with how we think about how cash flow behaves overtime and through cycles, which has then led and fed not only our capital structure decisions at a certain level, but then also how we think about returns to shareholders.
Vivek Arya
Got it. Maybe one kind of near to medium-term question. As we look at the demand for wafer fab equipment this year, how is second half looking versus the first half, right? What are the moving pieces in terms of whether it’s by end-market or whether it’s by geography, China demand second half versus both?
Bren Higgins
Sure. So part of our messaging at earnings, there were a couple of messages. The first is in some ways over the last 12 weeks from the beginning, from sort of March until — or from the January reporting out and expectations for ’24 and where we ended up in reporting in late April that there wasn’t a lot of change in terms of how we’re seeing the overall market in WFE levels. And our messaging around that the March quarter was likely the bottom for us and that we expected sequential growth moving forward.
So we are down in the 2, 3 range and then we guided in the 2, 5 range from a revenue point-of-view for June. And expectation of sequential growth through the year, which implies second-half somewhere in kind of mid-high single-digits of expectation that translates more into the semi, kind of a flattish WFE environment. Obviously, WFE is a little bit stronger, I think we’re in a position to continue to grow with the market perhaps a little bit faster, frankly, given some of the incremental demand we’ve seen.
Certainly, the leading-edge investment is a big driver for KLA. One of our objectives for this year is preparing for growth at the leading-edge because leading-edge has a very different process control intensity profile than legacy investment like you’ve seen over the last couple of — last year or so. So, we need to be able to support our customers. The competitive dynamics are pretty strong and the investment plans in the second half and feeding into 2025 look pretty encouraging.
We made some comments, I think some constructive comments around expectations for the 2-nanometer node as we move into next year and beyond, both in terms of our customer expectations, the design environment and expectation of design, certainly at the front-end of the node and potentially the size of the node moving forward. The HPC and AI drivers are pretty strong for M2 (ph). The architecture is very power-friendly. And so I think that that’s driving a lot of that interest.
Memory, I would say that the market — now all this growth is off at fairly low levels because ex-China, our customers adjusted pretty significantly in 2023 versus 2022. So — but I would expect to see growth in leading-edge investment, legacy investment after a couple of strong years and certainly a strong year last year would likely be down, particularly as the automotive market has corrected some.
And at the China market, the mix of it changing, but generally would be mostly flat into this year. Memory, I think we’re seeing signs of improvement in financial performance of our memory customers and cash-flow and pricing environment and inventory, all the indicators you look for that, that sort of demonstrate supply-demand parity and an expectation that I think we’ll see growth as we move into next year.
I think most of what’s happening in memory today is more about utilization rate improvements, some technology investment and — but I think that as you start to see some correction obviously in DRAM, we’ve seen some of the AI investment and some of that capacity conversion and that’s having an effect on this utilization and I think plans for the future. So I think those markets generally lining up the way we would expect to see in a recovering environment with some optimism as we move into next year.
Vivek Arya
Got it. Does HPM require higher process control intensity than DRAM? I would assume so, right, because yields are still low, but it’s consuming a lot of wafers. So is it still is — I assume it’s not as close to what logic process control intensity would be, but does that become a driver for higher process control intensity?
Bren Higgins
No, it’s a great question. And it’s not I think getting to where logic is, is at very different level. But if you just back up and look at the die-size, the die or bigger, you’ve got the introduction of EUV, which has been a real positive for process control intensity. You’ve got higher reliability requirements in those devices as you’re stacking and you have to connect these devices. And so you want to make sure that they’re all working and functioning as designed. You’ve got a silicon trade in terms of bit supply, right, so it’s consuming some capacity.
So in general — and then, of course, then you’ve got to take the device and then you’ve got to connect them together, but then you also have to integrate it into the GPU package, which our logic customer is doing. So I think all those are drivers that are positive for us. And I think as you start to see continuation of that growth, not only will it — I think not only is it driving profitability, which is good for our customers, it’s consuming capacity which is driving up utilization rates. And because of the capacity trade and the go-forward technology roadmaps, it does create perhaps an increment moving forward for process control intensity that we’re encouraged by.
Vivek Arya
Got it. It’s — you mentioned that — so historically the mix of foundry/logic to memory has kind of been this 55 — 45 or so in WFE. I think you’re suggesting something in the 60s. We heard from Applied earlier today, they also suggested something in the 60s, right. So it’s more like 65, 35, shouldn’t that be favorable to KLA because it doesn’t — it kind of mix-up process control intensity. My only kind of Part B of that question, how much of that is going to be taken up more by litho? So ex that part of the wallet, does the — like what drives process control intensity higher from here?
Bren Higgins
Yeah. No, it’s an interesting question and I said 60% because I — but I do think if you look at just where the trends have been that there’s potential kind of upside to that number. I think that was just in terms of how it supports the ’26 plan and the revenue associated with it. We’ve seen a nice move in process control intensity and KLA share of WFE over the last few years that have been driven by the incremental scaling benefits that we’ve seen with more-and-more EV layers. The design environment that’s consuming capacity, it’s driving our customers to do more sampling and production because they’re managing a more robust design environment and making sure that they can deliver to their customer requirements and that tends to be a positive.
We’re getting ready for some significant changes to transistor architecture and power distribution that are coming and so we think those are drivers as well. I think if you back up and look at the overall markets over the last five years, I mean, litho intensity has increased and process control intensity has increased. I think in most of the other markets and maybe there’s some smaller parts of markets that haven’t, but in general, they’ve been — the players have changed a little bit in terms of share, but overall the markets as a percent of the total haven’t changed all that much.
So we think that there — because of these challenges that our customers are facing, the ROI is pretty easy if you can improve yields sustainably in this sort of rich design environment or speed the time through a ramp cycle and get to volume production sooner. So as long as we can continue to demonstrate our relevancy, I think we’re in a pretty good position to be able to not only help with R&D, speed ramp, but also increase our exposure in production. We’ve been supply-constrained around some of the most important products that the company sells, our broadband plasma high-end optical inspection systems, we have more capacity coming online.
I think it affected our sort of share and revenue performance in ’23 because we are limiting capacity. As that capacity comes online, I have an increment that improves 2024 and expectation around that business and even more into the future. So I think that’s reflective of the benefits that this product can provide and the constraints that we’ve been under to be able to ship to market requirements even despite what has been the correction period for the industry.
So we feel pretty good about our relevancy if we can continue to execute on our roadmaps and introduce our products. And I think that the value case, if we can do that is pretty easy to demonstrate to our customers, which is why it’s important for us to really understand and collaborate closely with our customers to deliver and innovate and invest to be able to deliver products and new capability at a faster cadence than our competitors and then to understand that value and have a shared relationship in terms of how we share the value of that in terms of how we price those products. So we feel pretty good about the roadmap and the opportunity moving forward to continue to see improvement in KLA’s share of the market there.
Vivek Arya
And I wanted to touch on advanced packaging. I think about $400 million this year, half in specialty markets. Maybe tell us where exactly does KLA play in advanced packaging and how correlated is it to just the growth in the accelerator market?
Bren Higgins
Yeah. No, it’s an interesting thing and we come at-the-market with multiple products in the portfolio. And we are even seeing some strengthening as our customers are investing more with some of the improvement in the AI market that we talked about $400 million growing from $300 million last year. And I think it’s probably $450 million to $500 million at this point, right, as we just look at expectations for the second half and some of the incremental demand that we’re hearing from our customers.
As you said, we have a split. We participate in process where we do plasma dicing, we do barrier seed layers, we do RDL and TSV reveal steps. So a number of process steps through specialty — through our specialty semiconductor business, which has I think overtime uniquely designed for some of these market opportunities and has a fairly differentiated position and I think that’s reflected in the gross margins. We do segment reporting, so you can see kind of the gross margins of our process business relative to some of the others that are out there.
We also monitor chemical properties and when you’re doing depositing metal layers to do like TSV formation and so on and so we have a business that participates there. So that’s roughly about half of it. And then there’s actually, I left out, there’s a piece that’s tied to IC substrates, right, as you think about CoWoS and wafer on substrate and how the substrate integrates into the board. Where there’s increasing opportunities, I think overtime as lines and spaces as future sizes shrink to drive more demand in that part of the market too. And that’s part of the reason why we bought Orbotech was this exposure and this evolution of the substrate into the package. So you have that piece in it too. So that’s roughly half.
And then the remainder is in our core inspection and metrology businesses. There are metrology steps as you think about bonding steps and you have to do overlay and there’s feature size measurement and [indiscernible] weight shape and so on. And then inspection, both in terms of inspecting the connecting interposer layers as you’re connecting the various sort of tiles or chips into the overall device.
So we come at it in a number of different ways. And we think that overtime, as you start to see more complexity and shrinking densities in these layers that it will drive the need for higher-level inspection. And the challenge in inspection is that if you increase sensitivity, you have an effect on the speed, right, so very sensitive tools are very slow and less sensitive tools are fast. So right now, they’re using lower end tools that run pretty fast and you can get pretty high sampling levels. But overtime, if you — the sensitivity requirements are increasing, you’re going to need more capability and that will have a throughput hit, right, in terms of what you can actually — how many wafers an hour you can process.
So in that environment, so if customers are going to maintain similar sampling strategies, then there’s an ASP element, but there’s also a unit element as well. So I think those are the sort of the bigger drivers as it relates to the packaging part. Obviously, what is driving the need for more packaging is much more relevant to KLA. And if you look at today, just with all the focus on, high performance compute devices with larger die with a similar defect density, you have more yield challenges, same number of particles fall on larger die, your yields are lower, higher-value die and more pushing of leading-edge design rules earlier in node ramps are all good drivers for — ultimately for our business.
And I think an increment versus historic where you had more mobility driving most of the leading-edge progression and process debugging, which smaller die and a lot of volume around singular sort of device design versus a more robust design environment. So all these things are good. And then, of course, it translates into this packaging opportunity, which likely grows faster than overall WFE. And with increasing technology requirements within that overtime. So I think we’re encouraged by a lot of it, it’s — excited about it.
Vivek Arya
Are there parts of advanced packaging where you don’t play yet? Because we look at a number of other smaller companies who are seeing 30%, 40% kind of growth rates. So are there parts of the market where you don’t play yet where KLA can play, but in meeting your business model, of course?
Bren Higgins
Yeah. I know that there are — there is some segmentation within it. I mean, we’re seeing within our business those kinds of growth rates. But if you look at in certain parts of HBM, how it integrates through TSV formation into the package, that’s more of a CoWoS step and I think there’s participation there. But if you look at how the DRAM devices are stacked within an HBM device that you do have some traditional packaging steps like bump. It’s packaging which drives bump inspection, which is a not as much of an advanced packaging step, more of a traditional packaging structure that is a requirement where we do have some competition and is a kind of a lower-end of requirement in the market.
And so we’re — we don’t really serve that part of the market I think given some of the dynamics relative to price and cost of ownership and the needs that our customers have, right. At the end of the day, our customers want as much capability as they need to drive the highest-level of productivity. And — so you see that in a lot of markets where if you come in with more capabilities more than you need, then there isn’t necessarily as much value in paying for that. So there are opportunities there. I think that we think overtime as you move to different hybrid bonding techniques that you start to up level the technology requirements there, so create some opportunities in the future. But for now, we’re less exposed in that sort of HBM formation of HBM stacks than the other parts of the packaging market.
Vivek Arya
Got it. I know it seems a little early for ’25, but given the lead times for your tools, maybe you could help quantify what the lead times are? What is kind of the early preview of what ’25 might look like?
Bren Higgins
Yeah. Well, we’re in a very good position to be able to supply. I talked about some of the challenges in some of the product types that we have. It’s interesting. I said this in the call about 40% of the backlog of the company is tied to supply-constrained products where you either have the biggest part being the Gen4 optical inspection that I talked about earlier, but we even have products that we’re selling into some of the N minus 1 and N minus 2 market opportunities where you’re going from very low-volume levels to much higher and so spinning that supply-chain to be able to supply has been somewhat of a challenge.
So I think we’re in a pretty good place. I mean, it’s one of the things that we did in the last cycle was in preparation for this plan that we’re pretty convicted about, was making sure we had the investments in our supply-chain to deliver the capacity. I’ve continued to take inventory even though I haven’t necessarily had the demand for it to keep my suppliers committed to the supply that they’re providing for us in anticipation of the future.
So I think we’re in a pretty good place from a company capacity, from a supply-chain capacity point-of-view as we think about next year. Our lead times are generally — they vary across products, but generally somewhere around the nine months plus or minus range. That [Multiple Speakers]
Vivek Arya
What next year.
Bren Higgins
Well, that being said, we’re willing to also do what we need to make sure we don’t lose business because we can’t deliver. So I think given some of the leading-edge commentary we talked about earlier, I would expect that foundry/logic investment next year will be pretty good. And I think the memory dynamics portend I think a pretty good environment for DRAM. And Flash is not really growing this year. We’ll have to see. I think it’s going to be more consumer market-centric as we see that next year. But I think in anticipation, based on what we’re hearing from customers and the improvement we’re seeing in their businesses, I think we anticipate a pretty strong demand environment into ’25.
Vivek Arya
On that positive note, thank you so much, Bren. Thank you for having the time.
Bren Higgins
Thanks, everyone.