Lam Research Corporation (LRCX) Earnings Call Transcript & Summary

Good afternoon, everyone. Thank you for coming today. I'm Stacy Rasgon. I cover the U.S. semiconductor and now semiconductor capital equipment sector here at Bernstein. Today, it's my great honor to introduce our guest, the President and CEO of Lam Research, Mr. Tim Archer. Our session day is going to launch about 50 minutes. [Operator Instructions]

So how to think about Lam? So I talked about this yesterday a bit when I opened up a different session. I want to kind of reiterate some of it. Putting aside the stocks and talk of the cycle and everything else for a moment, I have to say, I remain really inspired by the semi cap industry itself. And the capabilities that these companies bring to bear to enable the fabrication of some of the most complex things that humanity has ever devised. And to that end, I do think that semi cap really seems to be coming into its own as both the underlying semi industry growth, and it does seem semi industry capital intensity seems to have to inflect higher with the contributions from the companies in this space, like Lam more important than ever before. So that and semi cap really does seem to be top of mind for many of my clients. Now there are, of course, the near-term questions on NAND spending trajectory and the impact of shortages and when things might peak, and we'll talk about some of that, I think. But I'm also increasingly getting people looking at the longer-term potential of this industry as they start to view it more on secular rather than purely cyclical terms. And to talk about these things, and hopefully, many other things, it gives me a great pleasure to welcome Tim to our session day. Tim, thank you so much for being with us today.

Great. Well, thank you, Stacy. It's great to be here. I think, as you said, we're going to talk about all the things we can do for -- to advance humanity. So maybe that forward-looking statement, you just saw, the safe harbor is very important at this point. We do want to talk a lot about our future, which we think is incredibly bright. So thanks for this opportunity.

You bet. You bet. To get started, again, I tend not to like to use this forum for short-term kind of questions. In this environment, it's hard to avoid them. I want to try to at least touch on them upfront, and then we can get on to maybe some of the more fun things. But I think initially, obviously, the industry right now is in a shortage situation and we've had shortages before. The current one is probably a bit unprecedented. Maybe could you talk just a little bit about what you're seeing in terms of your own customers -- in terms of demand for your products and your customers' demands? And any indications maybe across the 3 business segments, whether it's foundry or NAND or DRAM, some near-term views in the wake of some of the shortages that we're seeing? Anything that you see any -- I know you talked about this a bit in the past, but maybe just to refresh our memory on where you see things right now.

Yes, sure. Well, we -- clearly, we see things evolving quite quickly. And in fact, there was a -- we modified our view from our January earnings call to our April earnings call, both taking up our overall outlook for WFE to now -- we're seeing at least $75 billion or higher in WFE. And really strength across all segments and a stronger second half than first. That's an evolving view, but quite quickly evolving. And it's evolving because we're really being affected by these big 4 drivers, and it's -- first, you mentioned shortages. I think most people until they started hearing about shortages every day, didn't quite appreciate just how fundamental semiconductors have become to pretty much everything we use in our daily lives. And so that breadth, the growth and proliferation of semiconductors into everything is a big driver. And it's harder to forecast when you have so many different drivers, not just the PC refresh cycle like it was many years ago or the smartphone refresh cycle. Now it's just about everything. And then, of course, very few of us have been able to predict the pandemic and the acceleration of the adoption of technology that we've seen even in our own business, we talked about -- and what I'm sure we'll get to it later, we talk about installed base but we saw a 6x increase in remote services engagement, all of which employ the types of technologies that are so semiconductor heavy. Augmented reality goggles, high-speed data connections, massive numbers of sensors on the tool to provide database troubleshooting so maybe we're part of the problem. We're pulling these technologies in and because we couldn't travel to customer sites during the pandemic. So there's this huge acceleration that was, I think, hard to forecast, and it's creating this pretty large upside that we're seeing right now to the environment. And then those same technologies like augmented reality, virtual reality, AI, big data, they all require much more complex processing and computation chips. And I think that complexity was hard to forecast, just how fast you would need to increase the number of steps and the complexity of the equipment. And those things have just created this upside and caused us to modify our forecast in the short term. But none of those are really short-term drivers. They're all changes we're seeing in the way we work and live, and I think that gives us great confidence about the sustainability of this WFE spend. And that was probably even before we added in the fourth one, which was the involvement of governments, right?

Yes. I am going to talk about that.

And that one -- I think that one probably did come right about the time we were modifying our forecast. And now you see this much bigger push for regionalization of manufacturing. And I think that, again, will be a long-term impact given the fabs and geographic location of manufacturing concentration doesn't change overnight. It's a long-term commitment of investment, and I think that will affect us for quite some number of years as well. So...

Got it. No, that makes sense. I want to ask only one more short-term question, and then we'll get on to some of that. Specifically around like the NAND trajectory, so this was probably the biggest, I think, like perceived change in your outlook from like January until now. And I think like the prior quarter, you'd said you thought it would -- could be down in the second half, and then now you're saying up and -- I guess, 2 questions. One is, again, what is driving that? What do you see? And then number 2, I have to say, your largest competitor is giving a view that seems to be the opposite of that. And so I don't know if you have any point of view on -- and I won't ask you to speak for them, but I don't know if you have any like point of view on where some of those differences may be. And what actually drove the increase on specifically NAND?

Happy to try. I mean it's -- the good news is, and in some ways, we talk about each of these segments independently, but a lot of those drivers I talked about are -- they're pulling advanced computation and the advent and importance of AI and big data is pulling along NAND and storage and also DRAM, 5G is pulling along memory in both NAND and DRAM. And so it's -- I think, in some ways, our change or slight modification in our NAND outlook was just, again, driven by these end market drivers that we maybe didn't have a full understanding of. I think in terms of differences, as you say, it's always hard to know. One, forecasting is difficult and forecasting also depends on your particular position in the market. And Lam has a unique position within the 3D NAND market that, one, I think, gives us quite good visibility into this space. We provide almost all of the equipment for the most critical steps to manufacture the 3D NAND device and we also provide the vast majority of the tools that are required to execute layer to layer [ berth ]. And depending on your exposure to greenfield versus your exposure to conversions, you might have a slightly different view on things in any particular short period of time, like within 1 quarter, within a half year. So maybe that's part of it. But this year is a mix. It's a mix of greenfield adds and capacity, and it's a mix of conversions. And Lam has exposed significantly to both of those because of the role that etch and deposition plays in 3D NAND. And so we just see strength in that market. And I think it's just because it's fundamental to the end market use cases, as I mentioned, with 5G and big data and AI.

Got it. Okay. No, that's helpful. Let's get a little more interesting now. You talked about obviously your WFE forecast for this year, and I know you took it up in its mid-70s. Doesn't seem like that long, maybe even the last year when you were here, I mean, we're talking about a potential kind of like $50 billion in sort of normalized WFE. And now it's like $75 billion. And I think people are looking for like next peak could be $100 billion or like whatever the number is over some period of time. I guess just from your own standpoint, just in the prior to the last year, can you talk about the drivers of your route to that sort of increase at least from $50 billion to $75 billion in wherever. I mean whether it's broader trends of capital intensity, node migration, more layers in 3D NAND, just the areas where we're specifically either taking up and specifically where Lam is like more exposed to benefit.

Yes. Well, that's why I started with the -- I don't know how many times I said, the difficulty of forecasting out different drivers is because clearly, probably in our discussion last year, we wouldn't have been thinking WFE would be quite this high. But when you look at the underlying elements of it, as mentioned, it's the -- this breadth and it's the acceleration of adoption, and it's the complexity. I would say the -- it's acceleration of adoption, but really the acceleration of technology roadmaps. I mean to create these new capabilities, I would say we're seeing technology and roadmaps being pulled in for a good reason. I mean, again, it's about creating more capability in the foundry logic space, for instance, getting to that next node opens up kind of a world of next possibilities for those devices and the systems that they can enable. In memory, it's about being able to provide far more bits at that next lower cost of manufacturing, which, in turn, spurs greater demand and is sort of the engine of like new applications being created. And so specifically for Lam, what we've seen is this complexity playing out in just what you mentioned, the acceleration of -- to new layer counts in NAND, which is very etch and dep intensive, and that's obviously very beneficial for us, and we play a big role in enabling that transition. But at the same time within foundry logic, you've seen our revenue levels from a Lam perspective, hitting record revenues because the participation again of etch and deposition and new materials and atomic layer deposition, and new atomic layer etching capabilities, these things are becoming more important to create those complex devices. And when I say complexity, it's about finer patterning and the role we play in EUV. But it's also about starting to create these new 3D structures, which really, when you're creating a 3D structure, it's all about how you -- depositing films and removing them to create very complex architectures. And so we've just seen for us, an increase in intensity in addition to increases in overall just demand itself.

Got it. Got it. Maybe if we take each of those segments one at a time, like if we start with NAND and then DRAM and then foundry. So -- like I still get -- a lot of investors still think of you guys as sort of the NAND player, right? And I know historically, I think in a strong memory year, I mean, memory can be 70%, 80% of your orders and a good chunk of it, probably the majority of it, more than half of that would have been NAND. And we can talk about how that's changing by the way because I do agree with you, there are other pieces of the business now, foundry and everything else that are hitting like big numbers as well. But within NAND, one thing that I always wonder about is like how much farther does it have to go? So the transition from planar to 3D was phenomenally beneficial both to the industry and to yourself. And my own view is that capital intensity within NAND during that transition probably doubled at least, right? And a much different equipment set that played to your strengths. Is it right to think of that sector reaching sort of diminishing returns as we're getting -- you go from 32 layers to 64 like you're doubling it. You go from 176 to 192 or something, it's a much smaller increase. How do we think about the prospect for capital intensity within NAND specifically to keep increasing as layer counts are going up? And what are some of the new things that are becoming problematic that the customers need to be aware of that you guys are helping within that space? And I guess in general, how do you see your TAM or however you think about it within NAND increasing as generations are continuing to evolve?

Sure. I mean, yes, especially in the memory space, where we know that the ability to manufacture a bit at a certain cost is the key to success in kind of continuing to innovate in that area. So we don't have expectations that capital intensity rises beyond a level of affordability within the memory space. But what we focus on is the intensity of our systems and our SAM, and that's what's really important and what seems to be a big driver for Lam is your -- in terms of affordability, it's overall intensity, but important -- in terms of Lam's success, it's etch and deposition intensity. And so what we are doing is we're increasing our SAM as a percent of the total WFE by identifying new layers that we can do more productively, that we can do better technologically than the systems that are being replaced. And so for instance, you mentioned -- you asked what are the new challenges. Well, as you go from single stack to multi-stack, you introduce new challenges. You introduce challenges about how to protect the underlying tier as you now process the upper tier. This requires new types of deposition techniques and new types of deposition films that allow you to effectively process that new opportunity for Lam. As the aspect ratio of each of these tiers gets much higher, the old traditional ways of doing the gap fill of that structure no longer really meet the requirements because the aspect ratios are becoming very extreme. And so Lam now has an opportunity to introduce -- and we've introduced our ALD gap fill capability to fill that technological need. New opportunity for Lam, SAM expansion for RAM. And so we're really looking at -- we want our intensity, our share of WFE to increase even if overall WFE to produce the wafer isn't changing. And at our Investor Day last year, we lined out -- we laid out a few of those different cases, one of which, of course, was ALD, and this gap fill application within NAND is a pretty important driver for us going forward. And now this new one associated with multi-tier stacking is also a good driver as well. So that's the way we think about it, and we try to innovate new products in that way.

Got it. And to talk about DRAM. So DRAM is still on a lateral scaling path, but maybe not for that much longer, and people are starting to talk about 3D Dram as well. Do you think 3D DRAM benefits Lam over the long-term as much as 3D NAND that, that transition did? Do you guys play similarly there? Are there differences?

Well, that will be our objective. It's still a ways out, and so we are...

Few -- several years out, sure, but...

But there's no doubt. I mean, I think that when you said that investors think of Lam as a 3D NAND company, I guess if there's one thing I'd like to modify is, one, investors think of Lam as a 3D company, which means as DRAM goes to 3D, we view that as our space. We're targeted on the etch and deposition applications that will enable that transition. As logic foundry is going to a 3D-like structure with the gate-all-around. We think that's our space. And therefore, we're introducing new tools to increase our participation there. And the packaging is going 3D for 3D chip packaging. We think that's our space. And so really, we've defined our market in terms of etch and deposition intensity in 3D is going to rise over the long term. So those are the areas we want to be in. So in DRAM, we're definitely targeting that inflection. But I would also point out we plan to make progress here before it goes 3D as well. And so the next transition in DRAM is obviously to EUV and to continue to enable the 2-dimensional scaling. And there, again, memory being a very cost-sensitive application, that's where something like our dry EUV resist should play very well because that is intended to create the same quality of EUV patterning or better, maybe better, but at least the same quality of EU patterning at a lower cost. We do that through this material change and...

Let's talk about that because I was -- I used to work in a litho group at IBM once upon a time and I found this technology fascinating. I wonder if you could talk a little bit about the dry resist for -- this is something you've been working on with imec and ASML. And talk about like, what are some of the problems with like standard resist processing in EUV that the dry resist is nominally supposed to fix? What does it bring into your fold that wasn't there before? How do you think about how it increases like your opportunity in your can? And what are the problems that, frankly, that it's trying to solve?

Yes. Okay. Well, a lot of it is about materials engineering and the fact that if you want to engineer a material and deposit it precisely, we think dry -- depositing it by a dry method is, of course, one of the best ways to do that. And so it gives us very precise control over this patterning film, the resist deposition. And we're able -- effectively, the goal is to increase absorption of the photons to the point that you need less dose to accomplish the pattern. And that's how you end up making this a productivity gain for the customer. And why is ASML -- or I mean, hard to speak, but I mean, productivity as I mentioned with cost, it expands usage, I mean it accelerates adoption. And so I think that, again, we view this is something that customers will pull for when -- as it's being demonstrated. And so it's in sort of the early stages. It's in the development lines. It takes a while to qualify new techniques that really depart from traditional, say, spin on resists. But we think the value proposition is there. And what we've said is impact to Lam over the next -- we said, over 5 years, it'll be about $1.5 billion of revenue. But in the ensuing 5 years after that, as EUV layers continue to grow, I mean, it's almost certain EUV continues to be the patterning technology of choice in our opportunity to continue to grow even beyond those 5 years.

Got it. Can I ask where that $1.5 billion comes from? Does that basically take away from like lithography track like in photoresist preparation? Is that where that would come from? Or is it new TAM?

No, it would come from the -- it would replace conventional means of depositing wet resists.

Okay. Got it. Got it. Let's talk a little bit more about gate-all-around also now that we're on maybe moving into the logic space. And you talked a little bit about it in the 3D structures, but specifically, where does gate-all-around play to your strength? Is it strictly on the atomic layer deposition? Or is there more that you're offering that -- where gate-all-around plays into?

Well, it's one of the real opportunities we see is in the need for these ultra-high selectivity end processes to help create those 3D structures. And so we've introduced a suite of new selective etch products that, again, we've been successful placing into many of the R&D lines or data all around. And really, it's when we talk about selectivity, etch is always required to have some selectivity, but now we're talking about an extreme selectivity that helps to form those very complex structures. So that in and of itself is kind of new SAM opportunity for Lam, given that we haven't competed really for that business much in the past.

Got it. What was the name of that selective product, by the way?

Actually, I don't think we have formally launched it.

Oh, really? Okay. So I was wondering I couldn't come up with it.

I would hate to give our internal project name and then be in trouble. So...

Okay. I was just -- I was wondering if I missed something. That's all. That's fantastic. I want to -- another question on like maybe broader, if we step out to the broader industry. So obviously, capital intense in the industry has been coming up over the last several years, although it's still well below where it's at. If you go back late '90s, early 2000s, we had the 300-millimeter transition, which I think really helped to bring it down. We don't really have that anymore. But I'm just going through the drivers of structural capital intensity increase in my head. We have lack of black of 300 millimeters. We've got increased complexity in 3D structures. We have -- maybe it will be less efficiency as we get more localization, like, I don't know. Does Lam have a view of like how high capital intensity can go? I mean, it was -- on a broad base, it was 14%, give or take, last year. It was 20%-plus, if I go back 15 or 20 years. Do you think it's possible for the industry over time to get back to that like historical regime, especially since we don't have wafer transitions and everything else to offset anymore? Or -- I don't know if Lam has a...

Well, I think our perspective on it is I guess, driven by the fact that -- I mean, we always say we put our customers' success first. And we know they need to be operating profitable businesses in order for innovation to continue. So we look -- we say, in many ways, it's dependent on the semiconductor industry continuing to demonstrate trade, more and more value within the electronics ecosystem overall. I think we're actually seeing that. You're seeing a larger capture of [ any ] revenue of total electronics. And that creates profitability. The capital intensity today, at least by our view, is, as a percent of our customers' profitability, is actually not extreme at this point. And therefore, I believe that it's something that, again, comes back to value delivered. The way we think about it is our capture of the value should be commensurate with the value we deliver. And so therefore, I think that flows all the way through the chain. And so clearly, the last year or 2, you've seen, again, heightened awareness as to the importance and fundamental criticality of semiconductors to electronics and various industries. So hopefully, that value capture will continue to grow.

Got it. Got it. This is a sector, like when you kind of look at market share over time, like, it's hard to really parse out significant moves, right? I mean, like in any given year, strong memory year, weak memory year, things can move around. But like on a like-for-like basis, it's hard to parse out like big moves, things seem to be sticky. Maybe can you talk to us a little bit like why do your customers -- and maybe it's different in different end markets, I don't know, but why do customers choose Lam for like a specific process versus like Applied Materials or Tokyo Electron? Is it just history? Is it some sort of unique capability that Lam is bringing to the table? Like why are -- like why are customers choosing you versus other options that they might have?

Well, I mean, every tool, there's probably -- has a little bit of a unique story, but at a fundamental level, I would say that we divide -- you divide the world in our space into critical technologies, semi-critical technologies. And in the critical technology space, I mean, Lam focuses on bringing absolute best-in-class etch and deposition technology to every problem. And therefore, we try to solve the problem in a better way. And that's where we think about things like dry EV resist or our ultra-high selectivity etching or ALD. I mean, look, we did it when we brought copper plating to market back in the '90s and helped in -- add inflections. So that's -- there's a technical excellence element in the critical space that Lam will never give up on in etch and deposition. We've retained our focus. We've concentrated in etch and deposition, so that we can ensure that we're really good at where we're engaging with the customers from a technical perspective, with an intensity of engagement. But then there's the semi-critical space, and that's where we bring in architecture of the tools. And you think about Lam's systems, you hear a lot about us talk -- we talk a lot about footprint efficiency, number of pedestals or deposition chambers per system, these are all about addressing the applications where cost is keen. And that's that kind of semi critical. You have to be good enough to deliver the result, but really, it comes down to are you able to deliver the cost point. The new Sense.i tool from Lam, our new etch platform, first tool we designed ground up in the etch space in about 20 years. And there, we fundamentally said, we've got to design this so that when the need is technical, we deliver best-in-class, when the need is productivity, we deliver best-in-class. And that tool for instance, consumes on a wafer output basis, only half of the floor space of prior generation systems. You're building a new fab and you're thinking about production and efficiency, that's a huge driver. And so that -- and we did the same thing with our Strata tool on the 3D NAND stack deposition. It's the most efficient floor space consumer of any tool that can be chosen. So customers' decision factors are different process to process, but we really try to hone in on customers' key problem and then design the tool or the process specifically for that.

Got it. Got it. You do believe -- I mean, I get the impression that you do believe you can take share, right? And it's -- the reason I ask is that it's funny, like I look at you guys, I look at your competition, you're both willing to give out your models at some WFE level, like because who knows, right? They always seem to have share gains that are built into those models.

Yes.

I would say that you as well as your competition. We have done very well at executing on an absolute basis and even exceeding by pretty good margins against those models. It hasn't really been on the share side, though, at least as far as I like the data that we can get. And I know there's some hinkiness around the data even. But the market share that seems to be banking on doesn't really seem to come in, although it hasn't mattered. But I guess just given that, like how should we view the prospect for share shifts and share gains for Lam? Again, if I'm looking ahead to the -- maybe to the model horizon that you guys put forward at your Analyst Day in November, whenever it was?