Lam Research Corporation (LRCX) Earnings Call Transcript & Summary

Good morning, everyone. Thank you so much for coming. I'm Stacy Rasgon. I cover the U.S. semiconductor and semiconductor capital equipment sector here at Bernstein. It's my great honor to introduce our guest, the President and CEO of Lam Research, Mr. Tim Archer. Before I start, I want to mention, if you have questions you'd like to ask during the presentation, in your program, there's a QR code. You can scan it. That will take you to our -- what's called our pigeon hole form where you can submit those questions, and we'll leave time at the end for that Q&A. So semi cap. Look, semi cap has really been top of mind for many of my clients, especially as the strength of the last years, couple of years now sort of turns into the first kind of real cycle we've had since 2019. But even amid all of that, though, there are of course, nearer-term questions, things like memory spending trajectory and the impact of export controls and when things might trough and when they might peak -- I get all that. But I'm also increasing getting people looking at the longer term potential of this industry, especially as they start to view it on secular terms, in addition to the purely cyclical as they look for like the long-term growth of the semiconductor industry and what is going to be needed to support it with which is this. And so to answer that, and hopefully, in many other questions, it gives me great pleasure to welcome Tim to our session today. So thank you so much for joining us today. I really appreciate it.

Thank you.

I want to start with a little near-term history. Clearly, through COVID, both semis as well as semi cap have just been through a lot, and there have been shortages and supply constraints. Even the guys building the tools to build the chips have not been immune from the chip shortages themselves. And I was wondering maybe if you could just discuss I guess both the supply and the demand disruptions that we've seen in semis and semi cap as we've gone through the last couple of years through COVID, sort of landing to where we are today...

Okay. Thanks, Stacy. Just to point out, I think you all saw our safe harbor up there and take a look at that real quick, but thanks for having us. And thanks for letting me relive COVID just 1 more time before hopefully, we put that in the rearview mirror. It's -- it created a lot of, obviously, disruption on both the supply and demand side. Of course, if you look at it as a good side, I mean, the demand clearly spiked as a result of all of the electronics demand required to enable people to work from home and you saw PC growth, you saw smartphone, from a chip perspective, just growth across the more foundry, logic memory. And I think it was the urgency with which that demand came on the industry that caused a tremendous push to add supply throughout the industry. And we saw, therefore, equipment demand surge. And just as you pointed out, from the supply side, actually, companies like ours had a number of global supply constraints. I mean one was -- and the most ironic perhaps was inability to get some of the chips that were needed to manufacture the equipment that was needed to manufacture the chips. And so that sort of exacerbated the problem. But it went far beyond that, it was labor constraints, difficult bringing on enough people in the places you needed them to supply the equipment. And then on the cost and efficiency side, we talked a lot about freight and logistics and the impacts that those inflationary costs had on the financial model of the company. And while we still delivered quite strong financials, it wasn't as good as where it should have been for the volumes we were seeing in the business as those inflationary costs going to hit the company. Hopefully, that's in the rearview mirror now. We're using this slow period to actually make a lot of structural improvements to the company. We've seen us really accelerate the move of our manufacturing and supply chain towards Asia.

This is in Malaysia?

Yes, we have a big factory coming up in Malaysia. We're locating a lot of our supply chain in that same region, which allow us to respond more quickly in the future to big surges in demand, but also at a more cost-efficient close to customer's perspective. And so I think that's an important learning. We also mentioned that we're using this period to upgrade a lot of our business processes and business systems. Again, using the learnings from COVID to identify how can we be more resilient the next time we end up having to respond to sudden surges in demand.

Got it. Let's talk a little bit about that resilience. And then I want to talk about maybe some more trends in the broader market, but you bring up a good point. So Lam is typically viewed as like a memory-focused semi-cap company. There doesn't seem to be a whole lot of memory at least near term in the model. But my own numbers, I have services bigger than equipment this quarter. And memory is very small, and you're still guiding to $5. So on an annualized basis, call it, $20, doesn't feel that bad in the grand scheme of things, given -- I've seen prior memory peak cycles, remember, it could have been 70% or 80% of your revenue, and it doesn't seem to be that. Maybe talk a little bit about some of the broader dynamics or some of the areas that you've been pushing for, things you've been pushing much bigger into things like foundry and logic. The services business has certainly grown a lot. It doesn't seem like you're quite as dependent, at least for trough economics on memory anymore as you used to be.

Yes. Well, I would agree with you. We don't think $5 next quarter is a bad guide, given that -- if you look at WFE dynamics, I mean our strongest segment, as you pointed out, memory is down more than 50%. We said NAND is down in the range of 75%. So our largest markets are really seeing a...

But that was a full year statement, or that was this quarter?

Those were year-on-years.

Okay.

And so we're seeing significant reductions in spending in those areas. We also have seen significant cuts as a result of China restrictions. We said this year impact north of -- about $2 billion in terms of impact from those restrictions, hitting both the WFE as well as spares as a service as part of our business. And so we're pretty happy with the resiliency of the company. Now we've taken quick action and we saw as the downturn was coming with the need to reduce costs. And so unfortunately, we had to take some actions in the company to reduce costs. But I think at this point, we are starting to see the benefits of some efforts we started a few years ago to kind of build some additional resiliency into the business to not be quite so memory cycle dependent. And that's part of -- partly playing out there. One of those was, as you mentioned, the power of the installed base business. And if you just think the installed base from the last trough, we've grown that installed base 40%, the installed base itself.

And I guess even in a downturn, the installed base grows, you're still selling tools...

And so the installed base has grown, and that becomes year-on-year-on-year a bigger source of spares revenue, of upgrades revenue, of services revenue, and that has started to really contribute to the company in a meaningful way. And that's showing in that $5 guide, a lot of that is coming from that annuity being created from the installed base. And year-on-year, that will continue to grow. We also made a concerted effort a number of years ago to increase our product portfolio and attention to the foundry logic space. And we're starting to see that pay off, especially you need those more advanced nodes to begin to ramp, but we've been talking about share gains within the foundry logic space, and I think we're going to continue to see those as we introduce new products for AL -- atomic layer deposition. We've been talking about our EUV for dry resist, we've been talking about selective etch for gate-all-around. These are areas, again, where -- the field is kind of open, meaning these are new applications being -- that are being created by these more complex 3 dimensional devices that are existing now in foundry logic. And it's right in Lam's wheelhouse. 3D devices means etch and [ depth ], increased intensity and therefore, opportunity for us.

Got it. So I want to talk about that, but I do actually want to talk about the memory space.

Yes. Sure.

So look, it's under pressure, a lot of pressure this year. But the industry spent $40 billion, give or take, on memory WFE in '21 and '22, which were very high numbers. And I mean, we might do $20 billion this year, a lot of the forecasts have memory down at least 50%. And presumably, you can't stay here forever. So it has to grow and hopefully grows next year. But I mean -- does it ever get back to $40 billion? I'm trying to get some feeling for -- that's part of the reason we've got the oversupply now as we were spending so much then. And like does it ever -- even qualitatively, how do you think about the longer-term trajectory of memory vis-a-vis the prior peak levels that we've seen over the last couple of years and given the current dynamics.

Well, Stacy, ever is a really long time. So I think that maybe the more fundamental. I mean to take this a little bit to the longer term, I think there's pretty good consensus now around semiconductor industry that grows to the range of $1 trillion by the end of this decade. And when you think about capital intensity, you end up with a lot of WFE and therefore, a good portion of that is going to go to memory. So I think whatever time frame you want to put on this, you do ultimately end up back at levels of spending that are quite high. You also have drivers, clearly, that are -- AI being one of the big topics of the day that may end up accelerating the consumption of memory and you've heard our customers in the memory space. So they're better at talking about their end markets, but clearly, they've been signaling perhaps increased demand from some of the newer AI applications as well, both on the memory side, the DRAM side as well as the NAND side. And so I think we have to see how those play out. But from our perspective, we think the right strategy is, of course, defending the very high share position we have in memory, building that installed base, building the annuity that comes through that and then also growing our share positions and expanding our SAM and foundry logic. It just feels like that's a very well-rounded strategy for that.

No, that makes sense. And I mean to be clear, this is certainly from an industry and a utilization standpoint. This is the worst memory cycle certainly since the tech -- since financial crisis and maybe even since the tech bubble, possibly, although I guess the customers are not going bankrupt this time, so that's good. I guess I could argue that the current run rate, I mean if NAND is down 75% year -- I mean, I don't know if it gets back to -- memory gets back to $40 billion, but it probably cannot stay here forever, presumably you have to start building. I think you talked a little on the earnings call about maybe the -- qualitatively, the trajectory. And I think, Doug, you see if I would talk about probably seeing it in services first as it comes back. And then maybe the equipment takes a little longer because the utilizations are low. I guess offsetting that, though, we also have a lot of capacity in China, and we'll talk about export controls. I can't be remiss to leave those out. There are a lot of multinationals in China with a lot of memory capacity there that they can run right now under waiver, is like 1 year to, at some point, presumably -- well, I got to imagine, none of them are putting a single dollars worth of incremental investment into China now presumably, I don't know why they wouldn't. At some point, does that have to get rebuilt. I mean, is that really what if we're looking to kind of come off of this trough, is that really what leads it? Or I guess maybe...

I don't know if that's what leads it, but I do -- maybe just to highlight what we said on the earnings call around the trajectory. I do think that thinking about one of the reasons Lam is -- we're feeling good about ultimately when memory recovers and our ability to outperform in that, especially the initial cycle of that recovery is -- you'll see -- right now, customers have taken unprecedented actions really to cut utilization within fabs, and that's impacted our spares and services business. Pretty rare that customers take so much capacity offline, but that will be the first thing to come back. Customers start to reutilize the fabs they have, the equipment they already have. And therefore, spares and service business will actually improve and the predominant beneficiary of that will be Lam because of our large installed base. Second is customers will begin to upgrade the installed base. A tool that was installed in 2021 for a critical application doesn't have exactly the same capabilities you need to manufacture a leading edge memory in 2024, if that's when the recovery starts. And so therefore, those tools will have to receive technology upgrades. And so that will be sort of the second evolution of the recovery. And again, one thing we said about Lam, especially in the NAND space that may be somewhat lost is Lam gets from a revenue perspective, about the same amount of money for an upgrade per bit added in capacity as we do from a greenfield at.

Really?

And so -- and that's primarily because if you think about what the upgrade is doing in a NAND device, we're primarily wanting to create a taller stack, more layers. And what you need to do to create a taller stack is you need to add etch and deposition tools to do that. And so therefore, we capture a much higher percentage of every dollar of WFE spend from an upgrade than we do for a greenfield.

That's an interesting question. So what -- in terms of the lifetime value or lifetime revenue of the tool, so you're saying the initial purchase price, it's only maybe 50% of that lifetime value, or is it even less?

Yes, we've said that the lifetime value is quite high in those systems through both upgrades as well as spares and services. And so now back to your China question about the upgrade -- upgrading that installed base versus rebuilding elsewhere, that's ultimately going to be their decision and they're going to look at the U.S.-China situation and licensing, but I think from Lam's perspective, whether it gets upgraded or rebuilt elsewhere, both of those represent tremendous opportunities for us from a revenue perspective. Neither are happening right now because the industry just doesn't need the capacity. But whichever of those plays out, both are very positive for Lam.

Got it. Maybe that's a good segue into China, and there's a couple of aspects of this. So there's the export controls, there's the [ lagging ] edge demand, which -- I mean, even you guys are calling it out, which gives you an indication of how strong that is right now. Maybe to start with the export control. So just to remind everybody, you talked about $2 billion to $2.5 billion in -- I think it was $2 billion to $2.5 billion that would impact. More recently, you talked about maybe the -- through a rereading of the rules and nailing it down, maybe the impact was a little bit smaller than that. Maybe could you talk about that a little bit?

Sure. So the restrictions affected the NAND segment, DRAM segment and foundry logic. And what we said was that the impact of all of those would be $2 billion to $2.5 billion. There was some uncertainty around one particular part of that, though, and that was why we had this range of $2 billion to $2.5 billion. I would say that with the clarification we received and our ability now to ship some tools we thought might not have been shippable is probably more closer to the $2 billion and the lower end of that range. And...

This is in the memory -- on the memory side?

It's in what we would consider to be just trailing edge memory, yes.

Okay. Maybe on the foundry logic in China. So this is something I thought was interesting, you talked about on the call, you're sort of selling to Chinese customers that you've never sold to before. And in fact, you were actually collecting revenue upfront, I guess, precautionary action. What is driving that? And just given that these are new customers, how do we think about how sustainable that demand actually is?

Well, I think that clearly, if you look at this next 5, 6, 7 years of semiconductor revenue growth, that trailing-edge business because of growing demand for things like electric vehicles and CMOS image sensors and MEMS devices, power devices. Everything that you're seeing in this -- what's in this Specialty Technologies segment as we referred to it. It is a -- there is strong growth there and a lot of that is tied to activities within those regions. I mean China is -- has a big push for electric vehicles, for instance, electric vehicles double the semiconductor content in vehicles. And so -- my feeling is that many of these regional investments because of what happened in COVID and the impact people saw from not being able to get the chips they needed, you're seeing people building for domestic demand in a much bigger way. And so whether that's ultimately sustainable on a global basis, there's a certain global spend that's needed to fuel this trailing edge demand and the chips in all sorts of things. And I think that we'll see that build out for a number of years. Automotives and those types of devices or those -- the trailing edge. That's one of the fastest-growing segments of the semiconductor industry over the next number of years.

Are these new companies or they have never built fabs before? Or what kind of companies are these?

There are new entrants. Clearly, within this regionalization, I think you're going to see new companies coming in that some of which received government money, I mean, obviously, or -- and want to set up in the trailing edge space where they think that, that technology is accessible and they see demand.

Got it. To be clear that the trailing node demand that you're seeing, it's not just China, though, is it?

No, no, this is a -- it's a worldwide effect. Clearly, China is one area of strength for that market, but I think you're seeing similar trends in terms of the dependence on those trailing-edge chips to fuel so many other products within the global economy. And I think this will be the source. When you talk about regionalization and government subsidization, this is the market that likely will be targeted because governments have now seen the impact of not having those chips on much broader parts of their economy. You can't get very simple trailing-edge chips and you shut down an automobile factory, that's something that's very impactful for governments in Europe, in the U.S., in Japan, and so you're seeing investment in those locations driven by those needs and those resiliency concerns.

Do you think we have to be worried about a broadening of the export controls to start to hit some more of those lagging edge -- there were some new stories or rumors not that long ago about the potential. We haven't heard anything since. Is this something we should be thinking about?

I think we always have to watch any risk closely, but the reality is the line right now is drawn at foundry logic, 16-nanometer and below. The vast majority of this spending when you're talking about industrial or automotive applications, et cetera, are significantly higher than that. And so it's something we watch, but at this point, we don't have any indication that there will be an expansion of those restrictions.

What about the threat longer, again, probably longer term, but broader decoupling. So everybody always worries about the potential for local Chinese players to start to build equipment. And there are a few players, got AMEC and [indiscernible] and there's a few -- some of them I think got put on the entity list recently, so maybe we don't have to worry as much about them. But how possible or -- especially now if they have to focus 100% of their effort on trailing node and lagging edge, is that something that becomes viable? I mean, maybe it's 10 years from now, can they build a local ecosystem?

I think that many of these companies actually have been around for over 10, 20 years.

AMEC has been around -- I mean I left school in '04. They were there then, I think.

And so what I would say is that the companies over time can make progress. I mean, it requires an ecosystem, it requires partners within the industry to sort of help you. We get a tremendous amount of help from our customers, telling you what problems need to be solved. I mean, what would make your tools more valuable to them. And in many ways, I think that we worry a little bit less about them than just, again, moving our company and our technologies that much faster. And so a big focus inside our company right now is when I talked about reengineering some of our business processes, how do we go faster? How do we innovate faster, introduce new products. And at the end of the day, customers that want to be -- our customers that want to be successful, they'll buy the best equipment because that's fundamentally how they compete. And they -- I think that those local industries, they can have some success at the very low end of the market, but the real push for us is to continue to -- to grow our position at that leading edge.

Got it. For all the lagging edge stuff, is that all what you call Reliant? Or is there -- is it broader Reliant that 200-millimeter tools? Right...

Reliant includes -- can also include 300. So yes, it's mostly comprised within that Reliant business.

Okay. How big is -- I know you include Reliant within services, right, not...

Within the CSBG business, yes.

Yes, that's what it's called. Why is that, by the way. Why do you got equipment in there?

I think it's probably a little bit historical. If you think about where that business came from, primarily, the trailing edge business was serviced through mostly the sale of refurbished equipment. And so therefore, it was quite tightly coupled with the installed base because you looked at what the install -- the way customers tend to purchase in that market was, you looked at the tools that were in the installed base and you called up and you said, "I need 1 or 2 more. And by the way, if you could find me a used one, that would be best," right, because you're really just buying a little bit of extra capacity. Today, that world is quite different because now as you just pointed out, there's greenfield fabs being built specifically for these applications. And so -- to that extent, it's a little less tied to the installed base and -- but that's, I think, the historical evolution of that.

So that's an interesting segue into the whole question around capital intensity. So there's a thesis which kind of makes some sense that part of the reason capital intensity is going up is on lagging edge. Historically, there was a [ thriving ] used equipment industry. And people didn't really add a whole lot of greenfield lagging edge capacity. And now I mean they're still used tools, but a lot fewer than they were. And a lot of the factories that are out there are full because there's [indiscernible] increase always in auto and in industrial. And so a lot of these installations now are greenfield at much higher capital intensity. Texas Instruments, for example, they bought a ton of used equipment 10 or 15 years ago, and they've been running 4% of revenue CapEx and now they're going to be doing 18% to 20% for the next like 5 years or whatever. Now I guess, do you agree with that? Is that a driver of capital intensity? And do you have a view long term on where capital intensity can go? So maybe 2-part question. We did 16% to 16.5% last year. Maybe it's 15% this year, and obviously it doesn't go up every year, wobbles around, but if we're going to do $1 trillion, even at 15%, that's $150 billion. Is a $1 trillion and a 20% capital intensity, which used to sit there if you go back 15, 20 years. Is that a plausible number, if I'm looking out 10, 15 years? Do we have a $200 billion WFE?

Well, it's a great number for Lam if that were the case. But I think we're more comfortable in the -- we're more comfortable in the 15%. I think when you look at this, it's -- of course, we're always focused on our customers' ability to afford the equipment and the processes as the technology moves forward, and we're very sensitive to that. And so our job really is -- we look at it is twofold. One is to deliver the technology and to deliver the productivity that allows them to be successful with it. And that's clearly, you've seen capital intensity inching up. Partly, I think your specialty technology is that trailing edge piece certainly has played a factor. But even there, when we were offering refurbished equipment, it was because we thought that was the best most cost-effective solution for customers, and they agreed. And today, I think you're still seeing even in that segment where they now have much higher capital intensity, we're starting to now look and say, well, how do we bring some of the newer higher productivity tool sets that maybe have been delivered for 300-millimeter and port those back to 200. We're trying to bring things like looking at platforms that use like equipment intelligence, the data and artificial intelligence capabilities we have that are used at 3-nanometer and beyond. Some of those may have productivity benefits even back in trailing-edge fabs to help them lower cost do maintenance at a lower cost. And so I think that just throughout this, we focus on how to bring productivity. That will trade off capital intensity somewhat in the short term, but over the long term, I think, is great for the long-term health of the industry. And $150 billion WFE in 2030 would be -- it doesn't sound all that bad. So I think that we're -- we focus very much on the long-term sustainability of the industry as well as our position within it.

And I'd argue the industry is more sustainable. So I know people are getting worried it's a downturn. I mean, fine -- even then we might do 70, 75, maybe more, who knows. I mean it wasn't that long ago. I think you may have even been here we were talking like a $50 billion long-term normalized WFE. Now it's -- could be $100 billion, it could be $150 billion. And I think capital intensity bottomed in, I can't remember -- 2012 or something, it was probably under 10% and it's gone up a ton. Maybe if you want to talk about some of those -- what I'm trying to get at is can those continue, I guess, maybe some of the drivers of that improvement, almost 300-millimeter wafers reaching saturation in some ways NAND going from 2D to 3D. There's still tons of other like structural changes that are happening.

Well, the best -- I think there are, clearly, complexity continues to grow. And that's why I say the challenge for us is how to deal with that complexity without adding in kind of an undue level of cost.

And is that where the share gains are coming from...

Exactly. So I think that what we're what we're focused on and take, for example, the work we're doing right now in dry-resistant EUV. It's a perfect example of kind of solution for a customer, which adds productivity from the standpoint of changing the resist, makes EUV more productive means you spend less money on EUV, and we pull some of that spending over into our area. Customers' capital intensity may not change at all. It's just we've shifted that from one part of the market to another and save the customer money through a more productive process. And I think that's how we really look at in many of these steps, and those do turn into share gains for us. And I think that's -- when you can solve both the technology challenge and the cost challenge, you definitely win. The other thing that we look at is, over time, capital intensity doesn't necessarily have to rise for the whole industry, it just has to rise for etch and deposition for Lam to be happy. And if you look at what the trends are over the next 5, 6, 7 years, definitely, it's in the direction of more 3-dimensional devices. You see gate-all-around coming, which is -- we think it's the industry's first true 3D device in foundry logic, requires a lot of new processes in etch and deposition, things like ALD, selective etch. Those are all processes that are kind of open field. I mean there's no incumbents. They increase etch and dep intensity and Lam has a chance to win those positions and gain share. Maybe by the end of the decade, you have 3D DRAM programs coming. Again, increases etch and dep intensity quite dramatically. There are other elements of equipment segments that then go down in the 3D DRAM space to keep capital intensity at an affordable level but Lam can be the winner in that case. You see advanced packaging. Advanced packaging again, maybe that's an area of increased capital intensity in the packaging space that allows them to save in the chip manufacturing at the leading edge in the -- on the other equipment side foundry logic. And things like -- I know maybe you'll have an AI question, but...

We'll talk about. We'll talk about AI.

But advanced packaging is one that is a big beneficiary there, and we'll see growth.

So let's talk about some of these efforts in foundry logic. That probably will lead into AI. So you talked about gate-all-around in some of these new processes and you play in a lot of those areas, atomic layer deposition, selective etch. Have you guys ever quantified? I'm not sure in terms of the value per 100,000 wafer starts, what is the incremental opportunity from gate-all-around for you?

We just -- we're talking about the fact that it's a $1 billion incremental opportunity for us.

Overall or?

From foundry logic perspective for our equipment in gate all around -- through that transition. And it's primarily coming from new applications we said from ALD, selective etch.

Is that an annual number, by the way, or just an overall or?

That is a...

[ Just ] how it's defined.

Yes. The -- we've defined it -- it's incrementally as those processes were placed, I mean, could be in annual incremental. Yes. So I think that, again, and it results in market share gains for us because those are areas like ALD, where, again, they're replacing older technologies, other...

What gets replaced as some of these new processes come in?

It's -- as you go and build, I mean, basically, ALD is replacing older, more conventional types of CVD processes, for instance, selective etch is replacing certain types of -- of more conventional etch. But they're also all quite additive because if you think about in that device now that to build this device with nanosheet, there's more layers, more steps. And so in some cases, a direct replacement. I mean, in some cases, just new steps that are required because of the complexity of that device. And so it's really a SAM expansion as much as anything.