MKS Inc. (MKSI) Earnings Call Transcript & Summary

So thanks for joining this morning for the first session for the Wells Fargo TMT Summit. I'm Joe Quatrochi, [indiscernible] cap equipment analyst and excited to have the MKS team here, John Lee, CEO; as well Paretosh from IR. Lee [indiscernible] started a little bit of news this week already and kind of some updated China export restriction so let's just dive right in, like any 200 pages of reading [indiscernible] read at all.

[indiscernible] memorized.

Yes. Any initial kind of thoughts in terms of [indiscernible] restrictions?

Yes. Well, certainly, as a reminder, most of the impact MKS was -- occurred in October 2022 when the first fiscal came out. And so many of our revenue that went directly to Chinese equipment OEMs disappeared there. We had publicly said the impact of $200 million to $250 million of that 2022 number revenue was going to disappear. So that has to happen. We still ship a little bit whatever we can. But these new rules don't really impact that direct sales that we have to Chinese equipment OEMs. Of course, we ship to all the big OEMs outside of China that then ship to China. And so we are exposed to that. But as you probably heard from several press releases from certainly the 4 big U.S. OEMs, they all kind of reaffirm guidance, like, good, it's early days, but they didn't see a big impact. And so that's pretty much the guidance we would give as well. And we'll see over the next several weeks, we'll dive into that and see if there's any other impact. But right now, I think very minimal impact to MKS.

Okay. And then remind us, the $200 million to $250 million, that was the indirect impact?

No, that was just direct.

Just direct, okay.

Just direct. And so that's gone. Most of it's gone.

Okay. That's perfect. Maybe let's kind of -- before we dive into some of the [indiscernible] services parts of the business, if you take a step back and a lot has happened in the MKS story over the last couple of years. Talk us through like when you kind of approach the company and you think about the opportunities over the next few years, like what do you think investors are missing? Or what's underappreciated? And like I guess, where should we spend some time maybe doing some better on MKS story?

Yes. I would say MKS started as a critical subsystem supply in semi CapEx, right? And so that's been our history. It's still one of our biggest markets. So people understood that story pretty clearly. But 4 or 5 years ago, we started seeing that the continued shrink in semiconductor chips was no longer sufficient to kind of extend more than Moore's Law. And you needed to package multiple chips together. And we saw that 4 or 5 years ago, and we actually did something about it. We said packaging is going to matter. We want to be foundational to this next large growth segment for more than Moore's Law, enabling more than Moore's Law. And so we -- that's why we acquired Atotech. And that was new. That's a little more different, a little more complex story. But I think what's underappreciated by many investors -- some is that we are so foundational to that entire ecosystem. So when you look at just semi CapEx , 85% of every piece of equipment in every fab in the world has multiple MKS subsystems in it, 85%. We can come up with that number very easily because we just add the market share -- market share of Applied Lam, KLA, ASML, Tokyo Electron. Just those 5 is 85%, and we're on every one of their tools with multiple subsystems. When you shift over to this new leg of growth for enabling advanced electronics and that's packaging. And when you look at how complex these packages are, when you look at the cross-section of NVIDIA server, you see the chips at the top, you see HBM, you see GPU, CPUs. And you see RDL layer, a CoWoS layer. But if you actually look at a real picture to scale, there's now 20 layers of advanced package substrates under that 20. And under that, there might be 40 more layers of HDI or MLB package substrates -- sorry, PCB. And so when you look at the amount of content that there is underneath the chip, the just -- connecting the chips together in the package, we see that as a great driver for future growth and critical. So semiconductors are necessary. Advances in semiconductor processing technology is necessary but no longer sufficient you need that packaging. And I think that's a little underappreciated. The story is getting out there with the help of people like you, Joe, telling them the story. But that is the ecosystem we're in. And we're in -- 70% of all the steps that make that package substrate. So think about that. When you have an NVIDIA server board, 85% of all the steps to make the chips, we're in. 70% of all the steps to make the package, we're in. No one else can say that. That's how foundational we are and how integrated we are into that ecosystem. I think that's still underappreciated.

I think like back when you guys acquired Atotech, one of the things you said was do you think that PCBs and semis are basically there's like a decade difference in time. Do you think that like that time difference is shrinking, like given what's kind of happening on the AI side and everything you're just kind of talked about in terms of packaging and the importance of packaging?

Yes. I would say -- when I said that it's probably 2 decades difference in terms of semi and packaging. But you can see that our thesis is accelerated. So that's even more true now. You can see that -- you've talked about some of these MLB boards are 40 layers now with HDI on top and the package substrates are going from 20 layers to 25 layers. The features on them are smaller. The number of layers is bigger. The size of the board is bigger. That is just like semi, right? Remember semi, wafer has got bigger, more layers and features got smaller, exactly the same story. And I agree that it is accelerated. So that is really catching up fast.

Maybe kind of shift gears a little bit, right, like more near-term cycles are still kind of cycling across both your end markets. Can you kind of talk about maybe each one and then we'll tie them a little bit more?

Yes. Certainly, semi CapEx has cycles for sure. And right now, NAND is kind of in the doldrums, if you will. One of the other things that -- going back to your first question that I think is underappreciated about MKS is we do have a lot of position in NAND. And so when NAND has been dormant, we've seen a larger degradation in our semi revenue, but NAND will come back. I think the investors have appreciated are running the company during muted market cycle in a downturn of semi and packaging, we're 47% gross margin, generating 16% free cash flow in Q3 and lowering our interest payments and paying down -- prepaying the debt. And this is with a NAND, a very low market demand for NAND. DRAM is coming back a little bit, and logic has been what's held us up. So we really look forward to demonstrating the financial performance during an upturn.

Yes. And I think like last quarter, you guys are, I guess, technically this quarter, you made the largest prepayment on your debt that you've done since increasing your leverage quite a bit post the acquisition. Talk about like the confidence in recovery of like forward demand and free cash flow as we look into the first half of the calendar '25? And like -- I mean, it seems like you're pretty confident that making that such a large payment. And I guess like how do we think about the cadence of those payments going forward, given the kind of step up there?

Yes. The way we look at prepayments, we'll prepay any excess cash towards that debt. And just as a reminder, in 2024, if we hit the guidance -- midpoint of the guidance of Q4, that's $3.5 billion in revenue, and we are prepaying over $400 million of the debt. And that's just a testimony to our ability to generate free cash flow during a relatively very low muted market cycle. And our OpEx is under control. Our gross margins are already close to what our 5-year model target is. And I think quarter-on-quarter, there can be variations in how much we can prepay. I think Q1 is always a little lower because you have variable comps and all that. We also have CapEx investments that historically have been 3% to 5% of revenue in terms of our CapEx needs, maybe a little higher in the next year or so because we're building a Malaysia factory for BCP reasons. But our CapEx needs are quite low. And so we'll always look at each quarter and see what we can do. But any excess free cash, the first priority, of course, is the debt prepayment.

Okay. Maybe jump into -- let's talk a little bit about the semi business end market. Can you talk to your visibility recovery? It seems like, at least from an investor standpoint, like sentiment as we look into calendar '25 is kind of continue to maybe deteriorate a little bit. We kind of -- you go back to the middle of the year, people kind of were thinking maybe we could get a WFE of 120-plus and now we're probably kind of skirting, I don't know, low 100s, maybe for next year. Like kind of talk about your visibility and how we think about like order lead times with your customers?

Yes. I think you're right. Beginning of '24, everybody thought '25 was going to be a lot higher and every quarter that '25 estimate has gone down. We worry about that. Obviously, we care about that, but we also know that we can't really do too much about it. We can only control the things we can control, like profitability, gross margin, prepayment of loans. I would say this, our lead times are back to normal. Lead times for a big portion of our portfolio can be 6 weeks to 12 weeks. That has been the reason why our visibility is a little lower now because our customers know they'll get the stuff, right, within 6 weeks to 12 weeks. And that's also been the reason for a couple of slight upside surprises the last 3 quarters, which is in quarter turns -- booked in turn within the quarter because we can. So I think our visibility is we'll read the same stuff you guys read. I think we're prepared to run at this level with this profitability is 2025 were flat. We have also demonstrated over time that when that ramp occurs, it can often occur very quickly. And we have our factories, they're scaled to do that. We have enough capacity for that $125 billion WFE run rate. And so I think we're well positioned for that.

Okay. And like the inter-quarter kind of demand that you talked about, I mean, what is driving that? Is that your customers that are just kind of trying to drive their working capital turns just a little bit harder? Or is it that they're seeing maybe pull in their order of equipment orders from their customers, okay? How do you kind of parse through that?

Yes. I think it's a combination of both. I think it's hard to tell, right? We know what our customers are shipping. We know that in NAND, there's still our inventory in the supply chain. So we won't get too much help there until that burns off. But it seems like everything else has really kind of burned off. I would also say that our customers are holding probably a little more inventory than historic averages, but I think that's by design, right, because of the past supply chain constraints. And so these incremental in-quarter turns have been a good sign, but they're also not changing our revenue by 20% within the quarter. It's incremental relative to our guidance.

Yes. Okay. As you look at your customer kind of base, you said you think inventory is kind of largely normalized so maybe there's still some of the end kind of pockets of inventory. But I guess as you think about like looking into '25, like your ability to outperform your peers or WFE just given that you kind of have that digestion that's negatively impacted that relative performance in '24. Is that a fair statement we should think about that kind of return maybe what we normally expected outperformance of WFE next year?

Yes. I think it's one of two things, right? If WFE is flat, let's say, in 2025, then we'll probably run flat relative to WFE just because of the inventories have burned out. I think when we outperform is on the upturn. On the downturn, obviously, we and many of our peers underperformed because of the inventory digestion. But we always look at it long term, how are we doing relative to WFE in the long term. And as we've said in the past, we've been able to demonstrate our performance by 200 basis points over a decade. And I think some of the things that we're doing now will put that in that position. One is world-class optics. This is our effort to gain share in lithography, metrology and inspection customers and applications. Our power still remains one of the areas where we think continued growth will occur as things get more vertical. So things aren't getting more -- less vertical. That's for sure. Things will become more vertical, and that requires more high asset ratio etching, more RF power. And so these are the kinds of things that allow us to outgrow WFE because we have the broadest portfolio and no one can predict which subsystem is going to be really critical for that next inflection. No one would have predicted VNAND and how RF power would grow. Very few people would predict EUV, that was going to take off. And so having that broad portfolio allows us to move the R&D to wherever the highest growth opportunities are.

Okay. And maybe kind of sticking with NAND a little bit and talking about that vertical is kind of increase. And I think your main customers talked a lot about '25 being more driven by just upgrades of the existing systems or existing footprint, not necessarily increasing net new wafer starts. How -- like how does MKS participate in that? And then how do I think about like -- I think a piece of that is services maybe for them. But like how do we kind of -- how does it trickle down to MKS?

Yes. So the idea of cryo etch, right? There is an upgrade to NAND [textures] today. So cryo is a way to cool the wafer and you can use different chemistries so that you can etch more higher aspect ratio. And I think it really depends on how much of an upgrade the end user is going to want. Do you want a slight incremental 10%, 20%, maybe you don't need a new whole set of RF power supplies. But if you're going to increase the high aspect ratio of etch from 100 layers to 150 layers to 175 layers. I don't see how you can get around that without new -- entire new RF power set. And the reason I say that is because we're on G9, Generation 9 of VNAND, right? Started at 32, 48, 64, 96 on and on. We're in the ninth generation. Every generation, you've needed new or power, either all 3 are new or 1 is new, this generation and then the other 2 new the next generation because you need a higher aspect ratio etching, therefore, more RF power. So I think it depends on what form these upgrades take, whether they're incremental, slightly incremental, in which case it's not as beneficial to MKS, maybe slight. Or if they're really big increments, then, of course, it would almost look like a greenfield in terms of RF power.

Okay. That's helpful. So I mean, it's really about -- I mean, I think what Lam has talked about is 2/3, I think, of the wafer start capacity out there, still sub-200 layers. So moving above that threshold, I think, probably drives a fair amount of RF power requirements.

I would imagine so.

Yes. You mentioned cryo etch. Like can you just kind of maybe help us understand like the position there? I think there's maybe a couple of customers that are using cryo etch or shipping cryo etch or trying to?

Yes. Well, I think Lam is the one, is the market leader, shipping cryo etch. But cryo is about cooling the wafer and a different part of it. We don't necessarily play too much in that. But every cryo etcher is a -- high aspect ratio etcher has 3 RF power supplies on top, right? And so the more chambers of cryo etch or non-cryo etch, they both have RF power on top. So I think there is another customer who's talking about cryo etch. So we are certainly a company that's known RF power and pulse DC power company. This is a new type of generating power that is being tested for high-aspect ratio etching pulse DC. And so we're engaged with, obviously, all the key players. People know who we are. Obviously, leaders in the field. So they'll always be asking us for what we have in terms of technology.

So pulse DC power will be replacing RF power? Is that...

It could.

It could.

It will never replace all. So in a cryo etch -- in a high aspect ratio etcher, whether it's cryo or not, there's usually 3 RF power supplies in it. So 3 different power supplies, 2 different power levels, 2 different frequencies. That's been true for many generations. Pulse DC could replace one of them, perhaps in certain applications.

Okay. Interesting. And that's not shipping today for anyone, everyone is still using RF power.

Right. Production is all RF power.

Yes. Okay. Okay. You talked about optics. I think that's been an area of focus for you guys. And last year, we were fortunate to tour the facility just down the road, which was very impressive. Can you talk about the size of that opportunity? Where do you see that going over the next few years?

Yes. So we -- this is also another 5-year incremental -- not incremental, bet that we made 5 years ago. We said, look, we are underexposed in lithography, metrology, inspection relative to our vacuum customers in semi. And we shouldn't be, we don't have to be. But it requires investments. It requires investments in CapEx. It requires investments in process engineers to develop new processes. And so we made those investments 5 years ago, we continue those investments. And I would say if you were to look at our revenue for lithography, metrology, inspection 5 years ago, call it in the $100 million, $150 million a year range, give or take, there are cycles. Now it's over $300 million okay? So double. And that's all design wins, right? Now that can oscillate back and forth as well, but that's still a relatively small percentage of the BOM for our lithography, metrology, inspection customers versus our vacuum of customers. There's a lot more opportunity there. I think maybe the other point, Joe, there is lithography, metrology, inspection tools, they're very complex, as you know. Lots of parts that go into it. You really need size and scale to support these big customers because they have a lot riding on every component there. And I think that's an advantage for companies like us who can demonstrate we're willing to put that investment in with them even during a downturn.

Yes. That makes sense. And I think as we look into next year, right, like ASML kind of they've downticked a little bit unlike their EUV expectations. So maybe to the extent you can talk about that. And then outside of EUV, like can you talk about the drivers of the optics revenue, too?

Yes. Well, so EUV, it's a long-term play, though. And so I think EUV is going to be a healthy market for a long time. I think though that in general, whether it's EUV or DPUV or inspection of using DPUV or inspection using EUV, these are really complex optical subsystems, and there are many drivers to it. One is how fast can you go, right? Every year that every one of those tools goes faster. In order to go faster, you are developing perhaps new metrology and new parts, right, new optics to be faster. So there is a continuing evolution of these tools. They don't just design one in and just sits there because every year, it's like well, I've got the same tool, it does the same thing, but it's 50% faster, right? That doesn't just happen. That requires a lot of development of new optics, new optical subsystems, maybe integrating lasers with it, maybe integrating motion with it. So we see just a huge amount of opportunity in that whole space.

Okay. One of the things that was interesting for me Analyst Day, it was last month, they talked about basically the kind of road map for EUV is like for moving some of the mirrors or trying to kind of reduce the optics. Is that something that you guys started like that you could help with? Or how do we think about it your position like for that?

Yes. Those mirrors, we don't make. There's [indiscernible] But I think if you can -- also the question is how can they remove some mirrors?

Right.

So what are the things that allow them to do that? And there are many areas that possibly we could help. And so if we can, they'll certainly be asking us to do that. Without going into details, I think there's lots of ways to make things more efficient. Just go back to my earlier comment, that might require a lot more complex development somewhere else versus that EUV train that everybody talks about.

Right. Right. Interesting. Okay. Maybe let's just give a little bit of talk about the Photonics and Materials Solutions. I think like the recovery of PCB market still kind of influx right as we look into next year, iPhone cycle was a little bit weaker than we had everyone had hoped. I mean, how do you think about like recovery utilization rates looking into next year?

Yes. Yes, I would say this in our earnings calls in the last couple of quarters, we've had chemistry revenue or PCBs increased kind of 7% Q3 to Q3 in 2024, 2023, 19% actually Q2 to Q2. So -- but it's gone down 15% year-over-year from '22 to '23. So we're still catching up to kind of pre the downturn. But '24 was better. It's better than '23. And that might not look that way from an absolute dollar standpoint. But when you take out FX from palladium, it's actually a double-digit growth year-over-year. And so that's a good sign. That means that things have stabilized, people are coming back a little bit, still not back to an area where smartphones are really taking off. PCs are really taking off. Non-AI servers are taking off. AI servers are certainly taking off, and that's helping, but it's still only 10% of the entire PCB market. I think the industry was hoping that this year with the PC refresh, and maybe driven by AI, maybe just driven by everybody bought PCs during the first year of COVID, that didn't really happen, right? And so I think there's still -- obviously, probably because of macro issues, people have pushed that out. But the longer you push it out, eventually, it happens, right?

Yes. I mean, I guess like one thing to think about, right, you still kind of sure is still well below like where you were at previously, like at the peak, right? And I think if you kind of think about some of the structural sizes of like the different end markets, I mean, is it fair to think about like some of those is basically kind of overbuilding capacity, and so therefore, like it might take a lot longer to get back to that kind of utilization rate? Or how do you think about like that dynamic?

Yes. No. I think there are currently so many pockets where someone who's maybe got more capacity. But I think in general, many of our customers have reached utilization rates that make them think about adding CapEx. And we actually saw that right, in the last -- in this year in 2024. We saw some people adding CapEx for MLB. The lowest tech part of the PCB industry as well as HDI and certainly for package substrate. Now a lot of that we said is tied it to AI. But if they already had capacity for MLB, they wouldn't need to add capacity, right? So I think things are healthier. And I think if next year becomes even healthier, then we'll see maybe more of these customers start ordering equipment, which is, of course, the first sign.

Okay. you guys -- one of the things you have talked about, right, you talked about the strength for high-density multilayer PCB applications, some of that related to AI. Like I guess how big is that? And like, how do you think of it like the sustainability demand? Like how much capacity is like the industry adding?

Yes. Well, we talked about multiple tools, multiple customers ordering multiple tools. So that is not insignificant, but certainly not a -- we're not calling it a turnaround in the entire industry. And a lot of it is driven by customers who want to be in that AI food chain who maybe are not yet, right? They're in the PCB market, they could do it. They're not in that, they want to get into that food chain. And so they need to show that they've got capacity right to the end user there. So that's been a little bit of why, I think, we've seen that capacity increase the need for CapEx.

Okay. I guess, like as you think about within that, right, you talked about packaging as being a growth area, we talked about it a little bit earlier. But can you remind us how large is that business inside -- I think you said maybe 1/3 like how do we think about the size of the advanced packaging and then like the growth expectations of that business relative to maybe the overall packaging?

Advanced packaging is about 1/3 of our vendor packaging -- our E&P business, sorry. But if you think about the PCB market in general, we divide it up into kind of 3 sections. One is MLB, multilayer boards. That's historically been the easiest oldest technology, of which we have a lot of share. So think dishwashers and washing machines. But now maybe for AI boards, some of those MLB boards are now 40 layers. They used to be 4 for our washing -- probably still are 4 for washing machine, but they're going to 40 layers, 10x more. That's 1/3 of the market. That's growing traditionally at GDP+, right, maybe a little bit more with AI. The middle 1/3 is what we would call HDI, high-density interconnect boards. More difficult to make smaller features, think smartphones. So a lot of the smartphones have HDI, high-density interconnect PCBs. That's kind of growing at mid-single-digit CAGR. And then the most difficult 1/3 of the PCB market is what we call package substrate. This is the stuff needed for PCs, servers -- AI servers. And even some phones are moving into that. That's 1/3 of the market. It's growing at high single-digit CAGR. And so that's really the dynamics of the entire PCB industry. Atotech -- our Atotech division is a market leader in every one of those segments and a market leader in the industry electronics chemistry.

So I guess like when I think about like breaking that down, right, like for this year, how do you think about like the growth, I guess, of those different segments or the overall packaging businesses is what this year?

Yes. I think 2022 to '23 was down 15%. '23 to '24 remains to be seen, but it kind of looks like high single-digit or low double-digit kind of recovery, if you will. It's hard to parse it out in terms of package substrate, the most advanced part because while the AI portion is growing very rapidly, the non-AI portion, the non-AI servers and the PCs are not. Those are still big numbers, right? So the AI portion is single-digit percent, going to double-digit percent of the package substrate portion to the PC market. Now at some point, it might be 20%, and they'll eventually start driving a lot of the volume. But PCs and non-AI servers are still, you have a 90%.

Yes. Yes. I mean like maybe go back to something you said earlier, like you're talking about like on some of these boards going from like 4 layers that's required for like a dishwasher something that's 40. How -- like what's the difference? And is it a 10x difference in terms of also like just the amount of like chemicals that are required like [indiscernible].

Yes, it's exactly 10x. So we make 1 layer at a time and then you put them together. And so there's also chemicals to put them together to bind those layers together. And can you imagine, with 4 layers, any kind of warpage, thermal cycling, not too big a deal, keep 4 layers laminated together, [about] 40 and the boards are even bigger, right? And so while the technical challenge may not necessarily be in the smaller lines of the spaces, so a lot of technical challenges in keeping 40 big area layers together in the thermal cycling.

So when you think about like that dynamic, right, like make analogy right, it's like think about the processing time and the complexity for like HBM and DRAM, right, like should we think about it as something like that? And the fact that like it's more processing time is the yield lower or you need more capacity. Is that fair?

You need more capacity just because it's 10x more. But of course, you've got to make -- it's harder to yield. But I think it still remains to be seen what those numbers eventually will shake out to be because no one really knows. So I think -- but those are all great drivers, tailwinds for our business.

Yes. Yes. okay. One of the things when we -- you first acquired Atotech, you talked about a lot of basic cross-selling opportunities. Is there anything you can kind of update us on that today, like where do we see in terms of some of those deals turning into revenue?

Yes. We've talked about kind of a handful, maybe 5 to 10 kind of the design wins there. Some have turned into revenue, but many are really about getting designed in because you can imagine when we are talking to customers about is really talking about the next-generation development. And so there is the opportunity where we could say, can we bring in the laser tool as well or the laser guys bringing the chemistry as well. So just like in semiconductors, you get designed in, then they have to go make sure that they can get their business and then they build a whole fab around it. So it's kind of a 2 to 4 years' time frame still. So I would say some of it has turned through a decent amount of revenue, but the design wins have not turned into those kind of production volumes yet, but we have to have those design wins. And so it's the same as with RF power. We kind of talked about RF power. We talked about design wins our counterpart for about 3 years, and eventually, the market share changed, right, because it takes 3 to 5 years for those design wins to turn into real dollars that you and investors can see.

So I guess it's fair like when we sit here again, you come next year, right, as we look into '26, we should start to be thinking about some of those kind of materials?

Yes, that's our expectation as well.

Okay. And maybe like, can you help us like how many of those are MKS led versus like Atotech led? And how do you think about like some of those dynamics?

Not getting into certain particular percentages mostly, it's Atotech led. Fundamentally, it's because Atotech has had a long history with many of these customers, number one. Number two, when a customer buys an Atotech solution, oftentimes, they're buying the equipment and the chemistry. And as we've talked about, this piece of equipment is 100 meters long. It's a football field long, right? So you're building the factory around it. So when Atotech comes in, they're talking to CEOs and COOs because Atotech's tool is the factory, they call it the factory, the chemistry is a big part of the yield, a big part of the cost. And so then Atotech brings in this new brother that makes lasers as well. There are some areas where Atotech is not in, where lasers are already in, and those have happened. But majority is Atotech's relationships that are just more at a higher level, if you will, just because they're more important to the yield of that entire factory versus the laser tool, which is an engineering division, choosing the laser step, right, in the laser equipment.

Okay. Perfect. I think we're out of time.

Great. Yes. Thanks, Joe. Thanks for having us.

Yes, thank you.