Lam Research Corporation (LRCX) Earnings Call Transcript & Summary

For this session, I'm Vivek Arya. I lead the semiconductor coverage at Bank of America Securities. I'm really delighted and honored to have Doug Bettinger, CFO of Lam Research, joining us this morning. What I thought we would do is I'll lead off with my questions that I have prepared. But at any time, you have a follow-up question, please feel free to raise your hand. So with that, warm welcome, Doug. Thank you for attending this kickoff of semiconductor part of our conference.

Of course, Vivek. Let me begin, first, I need to read my safe harbor. I can't see if the slide's up. But my attorneys always like me to be going with this, just in case I say something I'm not supposed to. I'm not going to do that today, but just in case. Today's discussion may include forward-looking statements that are subject to risks and uncertainties, and actual results may differ materially. Additional information concerning factors that could cause actual results to differ materially from those forward-looking statements can be found in the risk factors disclosed in our public filings with the SEC, including our most recent 10-K and 10-Q.

You read it so fast.

I read that as fast as I could.

I guess I have to exceed that level of excitement.

I could tell the buzz in the room from that, yes.

Let's start at the top. We'll go through the nitty-gritty of the quarter and then spending. This is this march in the semiconductor industry toward the $1 trillion of sales, right? [indiscernible]

Everybody is excited about it.

[indiscernible] the decade, from the $600 billion or so where we are at right now. So whether it's at the end of the decade or thereafter, what does it mean for the semi-cap equipment industry? And then what does it mean specifically for Lam Research?

Yes. No, it's a great question, Vivek. Thank you for starting with a big picture question instead of what's happening in this quarter and next quarter. Listen, this is what everybody is excited about relative to the future trajectory of the industry. And I am, too, and so is everybody at Lam Research, to be perfectly honest with you. I think of this industry as a growth cyclical industry. There is a cycle and, oftentimes, at these conferences, that's all anybody wants to talk to me about. Where are we at in the cycle? When are we going to recover? Frankly, that's important. But when you're running and managing a company in the scale and scope of Lam, you have to be thinking about what the future looks like. You got to plan for the long term. And sometimes, you need to invest through that cycle, which, frankly, we're doing this year. Because when we look into the future, we do see this trillion-dollar industry emerging. And frankly, I don't know exactly what year, Vivek, but I know it's going to happen. Everybody is all balled up, hyped about AI. AI is a real thing, and it is going to drive a ton of silicon requirements. And I know some of the questions you're going to ask, Vivek, but we'll go through that. But that is a key aspect of the growth trajectory, over and above everything else that's already here, be it analog, power PC, smartphones. None of that is going away. But the new growth vector is AI and, frankly, automotive. The unit volume in automotive isn't nearly as high, but the content growth is enormous in some of these electric vehicles and autonomous vehicles. So when I step back, and Tim and I talked about, "Hey, where are we going with the company," that's what we think about more than anything else. The company has to be well positioned for where things are going. And so we think about that. So the second thing is, frankly, then everybody wants to know, "Okay, if semi revenue is $1 trillion, what is WFE?" If I'm being honest with you, I don't know for sure. Capital intensity is going up, though. When I look at the cost of the most leading-edge wafers today in both NAND, in DRAM, in foundry and logic, it's going up per wafer. Everybody likes to triangulate around this CapEx per revenue dollar. That's an important metric. And I would encourage you, though, that is an important metric, but the most important metric is CapEx per operating profit dollar. And while CapEx per revenue dollar has ticked up in recent times, CapEx for operating profit is actually dead flat or actually ticking down a little bit. So I don't know what WFE looks like in the trillion-dollar industry. A planning metric or a thinking metric might not be unreasonable to think about 15% of revenue, might be higher, might be a little bit lower, largely doesn't matter. So let's call it $150 billion in WFE. That's an awesome segment of the business to be involved in. And like I said, there will be a cycle around moving from the $600 billion to the $1 trillion, but that's where things are going. And then if I abstract back, let's say, WFE is $150 billion, the thing that excites me most about our company is, when I look at the evolution of the architectures in most of the industry, things are going in the third dimension. That's largely what Lam Research does for the industry, right? When things go 3D, like NAND did a decade ago, our addressable market doubled for wafer. You see that beginning to happen today in areas like gate-all-around, right? You went from planar to FinFET to gate-all-around. Gate-all-around is a 3D structure in foundry and logic. Our addressable market is going to grow meaningfully. And the metric we put out is, it looks like $1 billion for every 100,000 wafer starts that we can address around things like selective etch and ALD. So we can talk, and I know you have things you're going to ask, so I won't front-run the questions. But when I look at where the industry is going, things are going in the third dimension. So semi is a great industry. Capital equipment is a great segment to be in. And frankly, the unique position Lam sits in, I wouldn't trade it for anybody in the industry. That's my big preamble.

No, look, it makes a lot of sense. I just wanted to pick up on one of the things that you mentioned about WFE intensity. That's also a topic of conversation in that, this year, if you look at wafer fab equipment, WFE spending, it's in that low, mid-90s billion or so, as you mentioned.

Low, mid-90s is how we see it, yes.

Exactly, so somewhere around mid-teens of sales. What will help it become a bigger -- what will help this intensity go up? Because -- does it stay here? Does it go up? And that question is relevant because a lot of investors look at it and say, "Well, I want to invest in cap equipment if I think they can grow faster, right, than the broader semiconductor industry." So what can help this intensity go up, i.e., cap equipment grow faster than the overall semiconductor industry?

I guess what I'd tell you, Vivek, at the end of the day, it's hard for me to decompose $1 trillion in semi revenue, how much was improvement in pricing units. The unit-based part is what's important for the equipment side of things. What I can tell you, for sure, though, is capital intensity per wafer at every subsequent process node is growing. That's the most important thing, at least, when we look at it, right? I talked about FinFET going to gate-all-around. You're getting a decent step-up in capital intensity per wafer. And like I said, fortunately, when I abstract and look at kind of how architectures are evolving, 3D materials, it benefits the things we do.

Got it. And one other interesting thing, I think, that you mentioned, and I know you will have a unique perspective, right, because before Lam, you were at Broadcom. So you have seen the semiconductor part of the industry. And some of those companies have expanded beyond just semi and into systems and software. So there are so many more levers of them generating more value and, hence, being able to pay for this leading-edge technology. And as we have seen, whether it's NVIDIA or AMD and other accelerating their road maps, right, from 2 years to 1 year, I imagine, that's also going to be very favorable, right, for capital equipment?

It's great for capital equipment. And frankly, I mean, when you look at some of these accelerator die, the GPU is at the reticle limit. And frankly, it can't get any bigger. And so how does the industry evolve in a situation like that? Advanced packaging, right? You start to hear the industry talk about chiplets and heterogeneous integration. To create those structures, you need equipment, right? And we are, again, uniquely positioned in TSV, as an example, to do the drill and fill. That benefits the things that we do. So when I think all around this industry, frankly, and I've been in the industry a long time now, 30-plus years, I hate to admit it, but it has been that long, I've never been more excited about where we're at. And I started in the beginning of the PC or maybe -- or yes, actually, it was at the beginning. Saw mobile come, saw cloud come, and I'm stepping back and watching, okay, automotive and AI is coming, right? Semiconductors are cool again. Great place to be.

I think they've always been cool, but [indiscernible]

Listen, for so long, the kids coming out of college didn't want to come to this industry. We're hot again.

Excellent. So now that we have stipulated it a great industry, let's go through the nitty-gritty of [indiscernible]

At last. I knew you were going to get there.

Okay. So this year, you mentioned low to mid-90s. If I take away lithography, what do you think edge, in their position, are doing here as part of that broad assumption?

It's growing. If you abstract , okay, where is the growth coming from this year? Leading edge foundry and logic is actually pretty strong, right? And last year, it was pretty weak or soft, I guess, I'd call it soft. So that's a good place to be, right? I think to see the gate-all-around node invest -- and we're excited about that. Because like I said, our markets are growing there. DRAM is an exciting place to be. And I know you're going to ask me more, so I won't jump into all of the detail, though. But DRAM is critically enabling the AI training compute infrastructure, right? And we'll get into that. And you're starting to see storage tick up as well. NAND, and when I look at some of these AI servers, you have a nice step-up in storage requirements. What we're trying to understand a little bit right now at Lam is, okay, the training stuff is relatively well understood. The edge inferencing stuff, I think we're all just beginning to peel the onion back and try to understand what that looks like. And I see that as an area, and I don't have numbers for you yet today, but where storage is going to be more and more important as you begin the need to store these things on clients and whatnot. The large language model up, but that's going to be the next thing that we're all getting excited about as time unfolds.

Got it. Are you surprised that despite this excitement about AI, it's really kicked off end of '22, accelerated last year, right, this year, it has become very strong, that equipment demand hasn't really taken off with the same intensity? Are we just working off the COVID excesses, and there will be a point where we're going to get past that? Or -- because WFE hasn't really been as correlated with the excitement we have seen here.

Yes. That's a good observation, Vivek. I mean, like I said earlier, this is a real cyclical industry. And your point on COVID is important because, I think, as the world came through COVID, and especially in electronics, a lot of demand, frankly, got pulled forward, right? If you think back to when we're all messed up and not in the office, everybody was buying new PCs, new phones, TVs, stuff like that because we were all stuck at home, right? But kids were going to school, and frankly, if you remember, at one point, PC units, I think, was 350 million units. The industry, I don't know, sometimes we mistake what's going on. And that demand pull-forward, I think, some of us got a little bit confused that it was secular growth. It wasn't. And so that's the cycle part that's real cyclical. Things got pulled forward. And inventory got built with the volume assumption that was just too high, frankly. And so we're still digesting that, especially, Vivek, in memory, right? NAND is still pretty soft this year. I think next year is going to be a better year for NAND, but NAND is still pretty weak. And you and I were talking in the hall before we walked in. If you look at investment in wafer fab equipment in NAND, last year, it was down 75%. That's a shocking amount when you think about it. And it's because we overinvested for a unit view of the world that was perhaps a little bit incorrect. That happens. DRAM's better. DRAM has a product cycle in there, and it's got high-bandwidth memory in there, and we can talk about that as we go through the rest of the conversation here. But we still have a little bit of a cycle there. So when you're in a cycle and pricing isn't super good and profitability is not good, investment doesn't happen. And so that's what you're -- we're starting to see things turn the other way a little bit. I've been describing it as green shoots. Pricing is getting better. Profitability is getting better. Inevitably, and like I said, I've been in the industry a long time, I've seen lots of cycles. For every downturn, things turn back up the other way at some point. And we're still waiting a little bit for that, especially in memory, right?

You mentioned when semis go up, we call them secular. And when they come down, we call them cyclical [indiscernible].

Rightly so. It is.

Right. And you're right about PC, back, 340 million. This year, it's more like 260 million, 270 million. So we are definitely [indiscernible].

But I think they are right. The inventory has largely been worked through. And sell-in and sell-through is, from what I can tell, largely balanced. So that's a good thing. You've got to go through that digestion period.

Now Lam does have a fairly large memory exposure. So one, it is impressive that despite that 75% reduction, how the company was able to execute and generate very strong profitability last year. But if you look at the memory companies, they are enjoying this period of strong pricing, and they are saying that they are able to limit supply. And memory has usually been about incremental supply. There's always a demand driver. So doesn't that kind of work against somebody who's supplying to that industry that your customers are more inclined to keep supply tight because it helps them from a pricing perspective?

Frankly, you know that, at the end of the day, all that really matters is demand. And if my customers are being prudent, I guess, use whatever word you want to describe it, I'm happy about that. I don't like this cycle. It's difficult to manage through. I would much rather supply and demand matches every single year because I know demand is growing over time. And inevitably, when demand is growing over time to invest. And that will happen. And I would much rather have the industry invest prudently, methodically as opposed to overshooting and then undershooting. I'm happy about that, to be perfectly frank.

Got it, And then as we look at second half versus first half, which areas do you think are going to be kind of more [ loop moment ] in terms of recovery? Which area do you have better visibility in terms of sequential growth?

Yes. I think when I look at overall investment in the industry, it's a little bit second half weighted. A little bit. And if I think about that by end segment, I think everything is somewhat second half weighted, to a certain extent. We've already talked about the fact that memory is being careful in terms of investment, I would say, particularly in NAND. I don't see a real recovery in NAND until next year, potentially. It's not to say, I don't think second half is a little better than the first half. I do. But I see that, Vivek, in pretty much every segment of the business, with the exception of China. I have very clearly telegraphed that I think the spend in China is a little bit first half weighted. So you have to kind of everything else maybe is offsetting that.

Yes. On the topic of China, I know that you have been very consistent that demand is not falling off. It's not really growing. They had a very strong year last year. And there was, of course, this one customer, right, that the industry got approval quite late and then shipments were done. Excluding that customer, how is the overall China demand when it comes to trailing edge or in terms of demand or other investments?

Yes. Listen, we gave you a little bit of update on our earnings call and basically said, "Hey, this year, for the calendar year '23 versus '24, investment in China is actually growing. Now to your point, though, last year was a little bit second half weighted because of clarity on the 1 customer that you described. This year, I see it a little bit first half weighted. But for the whole year, it's growing. Now I think people, I don't know, there's a level of concern that there's going to be a pause in this. It's going to go away. Guys, that's not going to happen, right? And I say that because there is a very large tail of customers investing in a variety of different areas, at the mature node areas in spaces like analog, our image sensors, it's broad, and a couple investing in memory, primarily DRAM. And I'm sure everybody in the audience saw that the Chinese government has most recently re-upped the next big fund. I think it was $47 billion. That's going to get deployed. So listen, it's not going away. When we talk to these customers, they communicate plans, intentions, road maps that are multiyear in duration. Now is everybody going to succeed in this new set of customers? Maybe not. But next year, will there be a handful of new customers that aren't here today? I bet there will be. And so when I step back and think about what's happening, don't worry that this is going away. It's not going to.

If we look at China -- sorry, please.

[indiscernible]

Can we get a microphone for you so the people in the webcast can hear you?

Most of the semi-cap guys [indiscernible]

Okay. Yes?

Mic is right there.

Seeing this, do you expect the rest of the world [indiscernible] constant growth rate? What's the mix going forward?

So the question in the room, for those on the webcast, is what's the mix of China more steady state? It's 40-plus percent for all of us. I think it's going to be somewhat lower. If you go back before it ticked up, it was kind of in the low 30s. I don't think that's an unreasonable way to be thinking about it. And that's why I've been telegraphing. Listen, it's growing this year. I think it's a first half-weighted spend in China. I think you're going to begin to see some recovery in the rest of the world, especially in memory, that will offset it to a certain extent. I don't think it's going to sustainably be at a 40-plus percent level.

During his follow-up, you had a positive feedback on gross margin in China [indiscernible]

Yes. So the question, again, for the webcast is, what does that mean for gross margin, basically? And the puts and takes in gross margin I always describe is there's a product mix, there's a customer mix and then there is the cost component of just the efficiencies and whatnot. So right now, the reason I've been pointing to a hit of China mix -- and don't read this to be because it's a China customer, it's a lot of smaller customers. So profitability right now is really good. In fact, last quarter was an all-time high since we brought Lam and Novellus together in 2012. That likely doesn't sustain as the smaller customers in China begin to be a lower mix. I've encouraged people to think about the long-term profitability objectives. And gross margin for us is in the high 40s, call it, 48%. That's the financial model of the company. And last quarter, it was higher than that. So it will soften a little bit in the near term. And then in the longer term, you'll begin to see the efficiencies of the ramp of the factory in Malaysia that we've been talking about. That gets us, I believe, in a more consistent basis at that 48% gross margin. So that's the way you should be thinking about. Does that help? Yes, you're welcome.

On HBM, right, a critical part of all the AI accelerators.

We're all excited about it.

Exactly. And can you describe what is Lam's exposure to HBM today? Where do you see it growing? Because when we talked with a number of the end customers, they still describe it as an area of extreme supply tightness. But it's, of course, we saw AMD as they launched their next-generation product. The main improvement is really just HBM, 192 to 288 gigabytes. So what is Lam's growth in HBM? What's that exposure? Where does it [ sit ] next year?

Yes. So when you look at HBM, we're very excited about this. We view 2 things that are critically enabling to the structure. It's something called deep silicon etching. It's our tool. We call it Syndion. We pretty much own that application. We also are pretty much on the application of the metallization deposition. It's an electroplating process. We call that tool SABRE 3D. I kind of tongue-in-cheek call it the drill and fill. We do the drill and fill and put the die, then stack together in HBM 3. In my opinion, it's the most enabling part of putting the TSV together. So just to give you a few numbers on this TAM, on our most recent call, we said, when we look at our business-enabling HBM this year, we believe it triples. That if I abstract that, it's not just HBM, but advanced packaging. HBM is part of advanced packaging for us. And when I would have been here talking to you a couple of years ago, I would have described it as a business that was hundreds of millions of dollars in revenue. And then I shifted to say, and I was saying this a quarter or 2 ago, I can easily see this as a business in the next couple of years that eclipses $1 billion in revenue. We told you on the last call that it's $1 billion or so. So this is very fast growing. It's going to grow for the foreseeable future as much as I can see. High-bandwidth memory is maybe the biggest growth driver, but not the only one, right? When people talk about chiplets and heterogeneous integration and [ low loss ], all the same equipment shows up there. So we do the drill and fill everywhere. We fill other things into that as well. It's not just the drill and fill, but we don't have a strong of a position as we do in -- with the drill and the fill in other places. We do well, but anyway.

And the other you mentioned was this move at the leading edge from FinFETs to gate-all-around. Where are we in that cycle, right? Because the revenue this year, right, for yourself and some of your peers, are getting very big. So is it a case where there's a very large deployment this year than there is more utilization next year? Or do you think that we can see consistent growth over the next 2 or 3 years as the industry transitions to these new [indiscernible]?

Yes. Listen, I think everybody is pretty excited about this node, right? And I think next year, investment is going to be more than this year. We described, for the gate-all-around node, our revenue this year, we believe, will be $1 billion. And not to confuse you with the incremental addressable market around just the gate-all-around sheets, the structure. Our addressable market grows by $1 billion for every 100,000 wafer starts of capacity that gets put in place. So don't confuse those 2 things. We like throwing these big $1 billion numbers out, but there's a delineation between those 2, but $1 billion in revenue this year.

Back to NAND and how that is levered to AI. When do you think we will start to see that being more tangibly impacted by the demand for AI? For example, we are seeing more demand for SSDs, right? So when does that start to become a factor? Is that more a '25 factor? Are you starting to see some of that this year as well?

Certainly see a little bit, but we're still going through an inventory cycle, as we've been talking about. But yes, I think, today, I don't know, 75% or 80% of data stored is still in hard disk drives. As the industry transitions to QLC and, eventually, I think they call it PLC, the 5 bits per cell. The opportunity for the hard disk drive -- or excuse me, for the NAND industry to take more and more of that, I think, grows over time. And maybe, if you'll allow me, when we look at these AI servers, we get excited about what it's going to do for WFE. And let me unpack that for you a little bit. if you look at a true AI server versus an enterprise-class server and break apart with silicon, if you look at the logic square millimeters of silicon, it's 4x as much. If you look at DRAM, it's 8x. So high-bandwidth memory DRAM, 8x the content. And when we look at NAND today, it's 3 to 4x. And so that is an interesting thing to think through. What does that mean for a wafer fab equipment investment, right? More silicon. And the answer is, well, it depends on how many AI servers. Here's a data point for you to help you think about it. Because I don't know for sure what the mix is going to be. I know it's going to grow over time. But for every percent server mix, that is an AI server versus an enterprise-class server, when we think through WFE because of all the data that I just shared with you, it's an incremental $1 billion to $1.5 billion in wafer fab equipment required because of the intensity of silicon. I don't know. I see huge numbers of what's the percent of AI servers. So again, when I step back and think about the future growth trajectory, you're going to need more equipment. That, I can tell you, for sure. To what magnitude? Not entirely sure, but I know it's growing.

Is it crowding out investments in other areas? So for example, from an end market perspective, we hear that there's a lot more CapEx dollars going into AI servers, right, for all the reasons why you know how it benefits Lam. But is it crowding out investments in traditional -- so like do you see that when you talk with customers?

Listen, I think people worry about that cannibalization. Is it going that way versus this way? Frankly, I'm not the best source of an answer to that question. If I abstract, though, and think about it, just at Lam, as an example, the use of AI and the things we're looking at is incremental. I still need to run the ERP. I still need to run SAP. I still need to run the financial forecasting system, right? The geo, all of those things, the HR system, that's largely going to be on an enterprise server. It doesn't need AI compute. When I look at the things we're doing proof of concepts for around some of the AI algorithms, it's new stuff. So if I think about that at a higher level, I have to believe that there's a level of this that is absolutely incremental because these are new things. You don't need an AI-class server to run the ERP.

You mentioned, Doug, about China as a $47 billion investment plan into course of the U.S. chip stack, right? There are...

And Europe and Japan. And every government of the world has realized criticality of this industry and wants it in their border.

Has Lam already started to see those benefits? Or do you think there's still to come? Have you seen any tool orders emerge because -- any tangible tool orders because of all of these?

Yes. I think this is largely still on the comm line, to be honest, Vivek. It's maybe a little bit. There's been big announcements, right? Arizona, Texas, Ohio, New York, the announcements have happened. But the announcements don't really bring too much investment in equipment at this point. But you know it's coming. You know it's going to come over the next several years.

Right. One other thing I wanted to touch on, right, is the equipment industry has this amazing part, which is related to your services business, right, very strong...

Thanks for asking about CSBG. I love this part of the business.

Yes. So I wanted you to touch on that. And more importantly, the signal that you're seeing in different parts of the services business when it comes to spare versus the incumbent, what is that telling you about the future of tool orders to come?

Yes, great. That's a great way to kind of close our talk here. So CSBG, the Customer Support Business Group, frankly, my favorite part of the business model at Lam. Why is that? Our equipment runs for decades. It almost never goes away. It gets repurposed, repositioned, really decades. On average, over the life of our tools, we generate more revenue after we sell the tool than when we sell the tool itself. That's important to understand because that's where CSBG shows up. So what is in CSBG? Spare parts, service, equipment upgrade and the Reliant product line. And so if you think about 3 of those 4, as long as utilization isn't in a crazy place, and frankly, over the last 1.5 years, it has been, spares, service and upgrade should grow every year. Because the install base grows every year. We gave you that number at the end of the calendar year. At the end of last year, 90,000 chambers in the installed base. The Reliant product line, though, will ebb and flow. And frankly, last year, it grew very rapidly. This year, I think, well telegraphed in the industry, the investment outside of China, anyway, in the mature nodes, is a little bit soft. And so Reliant is down this year. Everything else is doing just fine. So what that means for CSBG revenue is I've suggested it looks flattish to us this year. It's not going to grow because Reliant is trending up a little bit. Everything else is in a good spot. Last quarter, actually, we talked about the fact that spares ticked up a little bit because we saw incremental improvement in utilization. That's the first indication to me, at least, in our business that we're starting to see the beginning of the green shoots in the memory cycle to a certain extent.

Right. Were they more in NAND or DRAM?

I didn't break it apart. I just said utilization has gotten somewhat better.

And historically, what does that tell you after how many quarters they're used?

All right. Yes, that's the ten-thousand-dollar question, and I'm not going to answer quite yet.

Lastly, is there a difference between the services intensity in China versus ex China?

Not really. Maybe a little bit. Maybe there's a little bit of an increased service intensity. And I say that because with a newer set of customers, usually, they're buying more from you, right? They often will want more of a turnkey solution. It's not unique to China. It's just unique to having a new set of customers.

All right. And the last 35 seconds, free cash flow, right? Again, very strong part. And I think what was impressive is just how well the free cash flow margin held up last year despite the severe pressure. So tell us about priorities. Is there any M&A possible still in the industry? Or is that pretty well consolidated now?

Yes. It's pretty well consolidated. We've done tuck-ins here and there, but nothing of any scale. And so large-scale M&A in this industry is in the rearview mirror, to be perfectly honest with you. And those that follow the industry know that. You've seen the last couple of big deals that tried to get consummated not get through the regulatory process. Real quick in the last couple of seconds. Priorities for cash is always invest in the business, first and foremost. R&D, CapEx, that's always priority #1. Second thing is, okay, even with that, we generate a lot of free cash flow. Our stated intention is to return 75% to 100% of free cash flow to the equity holders of the company through an annually growing dividend and buying the stock back. And those of you that follow know we recently announced a new buyback authorization of $10 million. So that is a clear indication of the intention of what we plan to do.

Thank you so much, Doug.

Thanks for having me. Appreciate it.

Yes. Thanks for coming everybody.