Lam Research Corporation (LRCX) Earnings Call Transcript & Summary

Safe harbor to keep my attorneys happy. I don't plan to say anything that I haven't said on public record before. So don't be disappointed with that. But today's discussion may include forward-looking statements that are subject to risks and uncertainties. Actual results may differ materially. Have a look at our website for the safe harbor language to the extent that you would like to do that. So perfect. All good.

Great. Well, thanks, everyone, for being here. I'm Tom O'Malley, semi and semi-cap equipment analyst here at Barclays. Pleased to have Doug Bettinger, CFO of Lam Research with us.

Great to be here.

Thank you for being here. Why don't we start off from just a very high level. So you spent some time on your last earnings call kind of talking about where you thought WFE was this year, kind of where you thought next year would go. And I think that you also...

Yes, that's a good place to start.

Yes. Maybe a little bit more growth view because some market share gains. But why don't we talk about what's your view kind of on this year? And what are those growth drivers that are lending to your view of growth in 2025?

Yes. Just real quick. Our view for the year has largely been pretty consistent all year, frankly, it hasn't changed all that much. We think wafer fab equipment spending this year, we call it mid-90s. I don't know, let's round it off and say $95 billion this year. Within that, I guess, just to give a little bit of color, leading edge foundry and logic has been pretty good. DRAM has been actually pretty good as well. That's a combination of investments that are occurring in the country of China in addition to the fact that you've got product transitions occurring, DDR4 going to DDR5 with some of the new server CPUs coming out as well as high bandwidth memory, which everybody is excited about all of the AI compute and whatnot. So that's kind of where that's at. NAND has been very soft now for coming up on 2 years, and I'll talk a little bit about how we see that into next year, likely strengthens into next year. Trailing edge foundry and logic has been soft outside of China. In China, it's been okay this year. So I mean that's kind of the painting the picture of 2024. As we go into next year, we haven't put numbers around next year yet, but we expect it to be a growth year is the color that we've provided. I think that was important context. Everybody wanted to know, hey, how do you see next year? We'll give our normal numeric quantification of spending when we get to the earnings call in January, but we see growth is what I would describe. So let me unpack that a little bit. We think NAND spending is going to be stronger next year than it's been for the last couple of years. It's just been pretty soft for so long. And frankly, when we look at the installed base, 2/3 of it is sort of aged, I guess, is what we would describe it as sub-200-layer device capability. Our view is next year is going to largely be an upgrade year, meaning the installed base will get upgraded. I don't think there's going to be new wafer capacity put in place, at least of any meaningful amount, but we do think it's going to be an upgrade year. And there's going to begin to be a metallization change in NAND as things move from tungsten to molybdenum or let's call it, moly, easier to say moly. So that's a little bit of kind of how we see NAND strengthening a little bit. Leading edge foundry and logic, I think, is going to continue to be strong, perhaps led by one customer more than the broader set of customers. And I know probably we want to talk about that a little bit. Trailing edge foundry and logic, I think, is going to continue to be somewhat soft, right? We're in the middle of an inventory cycle in a lot of those markets, industrial, automotive and so forth. So I don't think that's a surprise to anybody. DRAM is going to continue to be strong, right? You're going to continue to see investment in high-bandwidth memory and the continued transition to DDR5 occurring. Offsetting that, everybody always wants to talk about what's going on in China. Our view is China WFE next year is going to be down from this year and likely down from the concentration that we're seeing in the December quarter. So I don't know, maybe that sets it up to lead into whatever else you'd like to talk about.

Yes. Why don't we walk through each of those just kind of one-by-one and just unpack a bit. So why don't we start on the foundry logic side. So obviously, an interesting dynamic. Two customers are really struggling. One customer is doing quite well. We've talked in the past kind of about you need to look at the demand profile from a top-down perspective to understand how this market works, where if a piece of the pie is moving away, it oftentimes goes to another. But you often get questions about efficiency of that equipment spend. So can you talk about share shifts in the foundry logic market and how that impacts your business? Do you see some of those efficiencies? Or just net-net, do you think things just move from A to B to C?

Yes. Listen, I never talk about any one or another customer. So we kind of set that aside. What I always will tell people when they ask about, hey, this customer is doing this. This one is not doing as well as that one, what have you. The most important thing in this industry and in this business, at the end of the day is the demand for the output of what's happening industry-wide, frankly, right? The demand for semiconductors is the most important thing. And if 1, 2, 3, 4 customers are out there, that's great. If it's two instead of four, again, what matters is the demand. And if it's 1 instead of 3, then the 1 is going to invest more and the other ones will invest somewhat less. Now you always want to have more customers than fewer. Of course, that kind of is pretty logical. But at the end of the day, the most important thing truly is the demand for the output of the leading-edge wafers, and that is unchanged, whether it's coming from multiple places or just one location.

Yes. Helpful. The second side is, obviously, there's share dynamics, but there's also capital intensity. And you're -- you've talked a lot about technology advancements that have increased capital intensity, particularly at the leading edge. So GAA, BSBD advanced packaging. Can you talk about, one, I think it's -- I often get confused as well is just how big of a portion of your business is that today. And then maybe could you spend some time talking about where each of those drivers is important where you would see the growth amongst those?

Yes, we've talked about kind of these big new-ish type of -- I don't know if inflections is exactly the right word, but maybe it is. But things that are driving business to grow, basically etch and deposition and specifically Lam Research. Maybe let me take you through the ones you're referencing and then you can kind of ask further questions. First, advanced packaging. We're very excited about what's going on with advanced packaging. You hear people in the industry talk about chiplet architecture, die disaggregation, high-bandwidth memory users, I call it advanced packaging, but through silicon via to put the DRAM buy together, this is a great business for Lam. If you think about to do the advanced packaging stuff, you're stacking things on top of other things, you're interconnecting things. Generally speaking, that means you're depositing conductive material to create the electron path and you're having to etch things to open up the room to put the conductive material down. That's what we do. And so when you look at this, I guess, if we started the year, we were suggesting, hey, in a couple of years' time, this will be a $1 billion business for us, the advanced packaging in total. Then as the year progressed, we said, hey, it's $1 billion this year. And then more recently, we said it's actually more than $1 billion this year. So this is a great inflection. It's going to keep growing for the foreseeable future, right? If you look at high bandwidth memory, HBM3 is going to 3E, so 8 DRAM die going to 12. 3E will go to 4. So 12 will go to 16. That's more interconnect, right? It's more equipment. Capacity is ramping for that. So there's that. And like I said, all of the chiplet architecture that you hear the CPU guys and the GPU guys talk about, we enable a lot of that with the TSV process. We are very strong in those applications. So anyway, excited about that. It's $1 billion -- it's more than $1 billion this year and growing into next year. So that's the first one. The second one that we've been talking about is gate-all-around. And again, I like billion dollar numbers. We like billion dollar numbers. And so the quantification for that for us is when you look at the gate-all-around structure, it is very etch and deposition intensive to create those -- the sheet architecture. You're depositing films, you're selectively etching them. The intensity of etch and deposition grows to the tune of $1 billion addressable market for every 100,000 wafer starts and capacity that gets put in place. So we're excited about that. Obviously, that's incremental opportunity. Now we're not going to get 100% market share there as much as we would love to do that. That just doesn't happen in this industry, but we're going to do well there. And so that's an incremental opportunity that ramps into next year. We have said actually that the investment for the whole node, the whole gate-all-around node this year, not just the creation of the sheets is $1 billion in revenue this year, Tom. So just to kind of get your head wrapped around that and growing it into next year, obviously. The next one -- and again, it's another $1 billion number is backside power. And that one likely is late next year, maybe into '26. But that also when we look at the intensity of depositing the metallization and etching the structure is another $1 billion incremental opportunity for etch and deposition for every 100,000 wafer starts of capacity that gets put in place. So obviously, we're excited about these things because when you look at the evolution of architectures, they're moving in the third dimension. They're more materials intensive. And that plays to the strength of the things we do. We deposit film on silicon wafers and we etch them. So -- and then a little bit further down the road, we've been talking about this for a while, is dry photoresist, where we are collaborating with ASML to put the resist down in a dry fashion. It enables the photons from the EUV tool to more effectively be absorbed and creates efficiency as a result of that. That's a little bit further away, but I can see in 4, 5 years' time, that also will be a $1 billion opportunity for us. So when we look at these things, we're innovating, we're evolving to grow incremental opportunities that we see because of the evolution of where the industry is going.

Yes. I mean when you hear people say, hey, these are exciting opportunities, but the size may not be substantial. I mean you're talking now $3 billion, $4 billion today moving to much more substantial market that's in the mid 90s. So it is.

That's right. And these are things, frankly, you need to have anticipated them years ago, right? You need to have the capability ready when the industry wants to ramp this stuff. And so the investments that are going to show up and bear fruit this year and into next year occurred years ago, frankly, right? And so that's the nature of this business. You need to be anticipating where things are going. And I think we've done a pretty good job here.

Super helpful. So those kind of cut across both foundry logic and the memory side. I want to pivot over to memory because we touched on foundry logic really quick. So on the NAND market, you're seeing a very interesting dynamic where the large demand drivers of NAND have not really come back. They've kind of bounced along the bottom. But you've seen this interesting kind of pickup in ESSD, which has soaked up some capacity. So I think something useful we could talk about is how does an equipment manufacturer look at ESSDs? Does it require more advanced equipment? Is that soaking up more bits. Do you see that dynamic helping you guys into next year? Or is it really just the migration of mid layer nodes into leading-edge nodes? What's the bigger driver for you guys and spend a little time on how that looks in terms of ESSDS.

It's actually a combination of both, to be honest, right now. Yes, I think if you look at where NAND bits are consumed, it's in phones. Phones are soft, right? It's in clients. Clients, I think, are going to go through a refresh next year. So I think both of those will get somewhat better as we go into 2025. There's still inventory out there in terms of NAND inventory. I think, generally speaking, if you listen to people in the industry, they believe that will largely get worked through by the middle part of next year. So that's good. Things will set up could be a reasonable year next year in terms of needing to invest in incremental capacity for those reasons. On top of that, Tom, to the -- basically the direct thing you asked, the capacity that is out there is actually becoming somewhat aged. And for the industry to continue to evolve, specifically in the enterprise SSD space, eventually, the industry wants to go to QLC, right? It's TLC today, it will eventually go into QLC. For that to actually happen technically, the installed base needs to get upgraded. And it sort of goes hand-in-hand with the fact that the upgrades that will occur to move right now, 2/3 of the industry capacity is at sub 200 layer devices. The things that you upgrade to go beyond 200 layer are also the things that enable QLC structures to show up in the architecture. So those 2 things will enable the bit growth and basically the technical specifications for QLC that the enterprise SSD market is going to need. And that is a growth area, certainly on top of the traditional markets for storage in phones and clients. So that's part of why when we look at kind of where things are this year versus where it's going to head over the next call it, 2025, next 12 months, the equipment needs to get upgraded. And when we look at what we believe is going to happen in the NAND market, it's going to largely be characterized at least from an equipment spending standpoint next year by upgrading the installed base. The good news for Lam when the installed base gets upgraded is we get a very large share of that spending because we are the constraint tools in a NAND fab. We deposit that the mold stack. We do the most critical etches down through the structure, and we also deposit all of the metallization, the tungsten. And so those are the things that constrain the output of the installed base. And when we go into an upgrade phase, those are the things that will need to get upgraded. And that's part of why when we look at next year and you think about the things we've already talked about, NAND upgrades, gate-all-around showing up, advanced packaging continuing to ramp. We believe Lam is well positioned going into next year to outperform the broad market because the things that we enable the industry to do are in growth phase.

Super helpful. Pivoting to the DRAM side, you spent some time talking about TSV, but I think there's a debate in the market right now. Clearly, there's market growth in HBM moving into '25, and I think that's well understood. But I think the debate is how much capacity has been put in place already? Is that capacity enough to serve the market? So maybe spend a few moments like talking about your thoughts on the DRAM market. Are you seeing continued upgrades to line additions? Where are you playing there? And do you see that as a strong growth trajectory into 2025.

So it's been strong this year. I think our view is it's going to continue to be strong into 2025 basically is the way to think about it. And it is characterized by, yes, there's always upgrades that are happening. There's always the installed base moving to the right, 1 alpha, 1 bega, 1 gamma and so forth, right? That's happened this year. That's happening into next year. You've got this transition, like I said, architecturally with these new server CPUs from Intel and AMD coming out, needing DDR5. It's a bigger die. On top of that, when you look at all of the parallel compute architectures, that's a DDR5 die, and it's a much larger die when you go into the TSV stack. It needs to be bigger to create the space for the through silicon via interconnect. So the die is nominally twice as big. And then you have to have equipment to put to -- I call it -- we call it the drill and fill, right? The silicon etching to create the space for the TSV. And then it's a copper interconnect put down with an electroplating process. That's our SABRE 3D tool. That's all incremental relative to the high bandwidth memory architecture. So that's also showing up. And like I said, when I look at the advanced packaging space, HBM is a key part of what's driving this year. HBM for us is tripling this year.

All right. So I made it halfway through at least without bringing up China. So can we do a little bit of China here?

Sure, of course.

All right. So can we talk about where you think equipment sales have gone into China thus far? Do you think those are being deployed? Do you think that those are maybe being saved? I think that's a debate in the market.

It's all being deployed.

It's all being deployed. And when it's being deployed, can you talk about like the sustainability of those spend trends? So you're saying next year it kind of goes back to normalized level. You spent some time on the last call talking about China as well. But over the longer term, do you see a period of digestion after a couple of very strong years? Or do you think it's like a small blip and then it comes back? Your best guess, obviously, it's not perfect.

It's hard to know. Listen, I'm not even giving numbers for '25. I'm certainly not going to go out beyond that. But when I look at the breadth of customers in China, it's a lot of customers, right? They tend to be smaller, newer customers not all of them, right? There's some well-established customers in China, but there's a lot that are new, a lot that are kind of learning how to do what they aspire to do, which is to supply the China demand, China supply. I observe that's what's happening. This all tends to be trailing edge kind of investment, right? The U.S. government has restricted the most leading edge stuff, at least from U.S. companies, our ability to sell, you can't sell the most leading stuff. And so they're investing in the trailing edge nodes, read it to be analog, industrial, automotive, power devices, things like that, that don't require the most advanced equipment capability, and it's a broad set of customers. Investment this year was pretty very strong, in fact. It's trended down through the year. And as we look into next year, we've suggested it's going to trend a little bit lower even beyond where it is in the December quarter. It's not going away, though. I want to be very clear about that. I don't view this as something that like came and went. That's not it at all, right? The aspiration to supply the China demand remains there. The evolution of the customers there. They're going to continue to get better and learn how to do this stuff and get utilizations up and yields up. And I think it's going to be around for the foreseeable future.

Super helpful. Pivoting really quickly to last week, the BIS announcement, you had a press release out essentially saying that you didn't really see your outlook changed. Some of your competitors actually had outlooks that varied slightly. Could you maybe walk through, are you seeing the restrictions touch anywhere in your business? It's relatively unchanged, but it obviously can touch certain parts, but it's not consequential. And any commentary you have just on those additional restrictions.

Yes. Listen, we've got a great government affairs team that is pretty well plugged into everything that's going on in D.C. And I would tell you the regulations that came out were largely what we expected. And so yes, I was at an investor conference last week, getting ready to talk to people, and that had just hit the tape. And I knew we were going to -- I needed a little bit of air cover. So we issued a press release saying, hey, no change to any of the guidance, no change with our view of what we've previously said. So that's, I guess, is about as much as I can say. There were a couple, 3 specifically or maybe 4 customers that were put on the end user restricted list. So there were customers. We won't be able to continue to service them as customers. And from a technology standpoint, there were some slight tweaks, but nothing of any real substance.

Super helpful. Moving to CSBG. So I think that oftentimes, from a quarter-to-quarter perspective, you have some moving parts. So you have Reliant, spares, other components. Service has obviously been fantastic for you guys as well. Can you walk through maybe sizing the portions of that business today? And into the next quarter, you've given some color on the moving parts there as well. Just remind us of what's happening.

Yes. Let me unpack it for you. So CSBG at Lam is the installed base business is how I often will describe it. CSBG is the customer support business group. So it's -- a lot of this is the money we make from the equipment that we've already sold. And so the nature of that is there's 4 components in CSBG at Lam. There's spare parts service. I've already talked about you can upgrade the equipment in the installed base. That's in CSBG. And then the last component is what we call the Reliant product line. It's the mature node, older equipment that is part of WFE, but goes into the mature node stuff. If I unpack this for you, last quarter, actually, CSBG in total was 42% of the total company revenue. So it's a big part of what we do. So unpacking it. The biggest individual component of CSBG is spare parts. Think of the equipment we sell for basically the whole industry, but us in particular, with our etch equipment, it needs -- spares need to be replaced. They wear out, things wear out, right, especially in etch, which is a destructive process. So that's just part of running an etch tool deposition as well. You have to replace things with some level of frequency. Along with replacing spares is also service, right? And in fact, one of the things we're excited about to deploy cobots to help the predictability to the repeatability of service. We issued a press release about this actually just yesterday around some of our dielectric etch tools. It's a way you can actually provide the service more predictably, more repeatably with greater less time to actually do the work, right? It can be much more predictable. So we're excited about that. We continue to innovate with service in that way, things like that, advanced service offerings and then upgrades with the installed base. So those are the components, frankly, that in almost every single year, that grows as part of the revenue stream because our equipment, the chamber count grows every single year. And so the opportunity to sell more spares and do the upgrades and provide service grows as a result of that, plus the fact that we're innovating in some of the advanced service offering. The part that goes up and down with WFE is actually the Reliant product line and specific the mature node investments that turned out to actually be pretty strong this year. So CSBG did end up growing a little bit this year. And coming into the year, I was suggesting, hey, it may be flattish, that ended up being a little bit stronger primarily because of Reliant. And so when you think about what CSBG will do into 2025, which is the question everybody has been asking me over the last couple of weeks, I feel very good about growth in spares, service and upgrades because chamber count grew this year. Utilization is improving in the memory fabs. What likely will face a little bit of headwind is the Reliant product line as some of the mature node investment, like I described earlier, it's a little bit soft. So that's the way to think about CSBG.

Yes. And correct me if I'm wrong, but it would make sense that given the exposure to China over the last year, maybe that Reliant business in terms of its exposure to China versus your total business exposure to China, maybe a little bit higher there?

It's a little bit higher because the spending in China is primarily mature node spending, right.

Okay. If I look at some of the moving pieces into your guided quarter, obviously, there is a step down in China. If I was to focus in on what areas of the business we're seeing that headwind, could you maybe walk through to the extent that you can? Is that on the foundry logic side on the memory side? Is that some of CSBG? Are you allocating that exposure to any of those end businesses?

If you look at the totality of the spending in China, it's got a concentration that skews towards foundry logic, right? That long list of customers I described are primarily mature node foundry and logic. So just to kind of get your head wrapped around how that is progressing through the year. We started the year, the China concentration, if I recall right, in the March quarter was 47%. It trended down last quarter, it was 37%. And we've suggested in the December guidance, we believe China trends down to about 30% of revenue. And again, as we look into next year, I think it likely trends somewhat lower than that as well. But it's got a heavy concentration in mature node foundry logic and yes, the Reliant product line.

Put you through the rumor on all the WFE stuff, and we talked to China. Now some CFO questions.

Sure.

All right. Operating model into next year. You already talked a little bit about gross margins moderating. Is that purely just the movement of China revenue? Or is there any other puts and takes that are influencing the margins?

Listen, I don't want anybody to walk away that it's China because it's China. It's not. It's because a lot of the customers in China are smaller customers, and they tend to pay somewhat more than a bigger customer, right? There's a negotiation back and forth. So the concentration of the smaller, and I'll always describe it as customer mix, it's not geographic mix per se, will be a bit of a headwind as we go into 2025 from a gross margin standpoint. However, offsetting that, if you've been following the company for a while, when business trended down in the beginning of '23, we restructured the company to a certain extent and pivoted to try to be closer to our customers, and that is true in R&D as well as in manufacturing. And what that meant in manufacturing was we had actually always contemplated to grow into the factory that we had in Malaysia that will continue to be the case. So as -- if we are right about it being a growth year next year, and I think we are, I wouldn't describe it if we didn't believe that, the incremental volume for the company will increasingly be built, manufactured through that Asia factory network, specifically in Malaysia. It's got a better cost structure. It's got a more cost-effective supply chain. And frankly, maybe one thing that is underappreciated is it's also closer to all the fabs in the world. And so the freight and logistics spending that we have to deploy dollars to because it's closer to where the ultimate customer is, we also will save in terms of freight and logistics. So there's a cost tailwind that, to a certain extent, depending on mix and other things, will offset some of the customer mix headwinds.

Super helpful. So the capital return story has been a really positive part of the stock. You have $6 billion in cash sitting on the balance sheet plus. Can you talk about priorities for use of cash?

Sure.

Obviously, you have the dividend side, but you have, I think, about $10 billion left on the authorization.

Roughly. Yes. So yes, obviously, the priorities for cash at the company are, first and foremost, invest what the business requires in R&D, in CapEx, in infrastructure. That's always the priority. So we comfortably are always able to do that. On top of that, we generate a lot of free cash flow. And so the priorities for the company are both the buyback, which you already mentioned how large it is. There's about $10 million remaining on the authorization as well as we plan to grow the dividend on an annual basis. And so how that shows up is the intention of the company in the medium, long term is to return 75% to 100% of free cash flow through an annually growing dividend and then topping it up with the buyback. It's meant that over the last several years, we've bought a lot of stock back, and I anticipate we'll continue to do that.

Super helpful. It's an absolute pleasure to have you here. Thank you so much for joining.

Awesome. Thanks for having us.