Thank you all, and good morning. Thank you for coming. I'm Stacy Rasgon. I cover the U.S. semiconductor and semiconductor capital equipment space, here at Bernstein, and it's my great honor today to introduce our guest, the President and CEO of Lam Research, Mr. Tim Archer. Before I start, I wanna mention, if you have questions you'd like to ask during the presentation, you should have a link to the Pigeonhole form. I think there's a QR code in the program where you can submit those questions, and we should have time for Q&A at the end. So look, I always say I have semicap in my blood. I used to—this is what I did in a prior life, and I love this space long term. The industry itself has been enjoying a pretty significant renaissance over the last several years.
I think it really has come into its own. Both industry growth and, I think, industry capital intensity have continued to inflect higher, and we've been seeing it by leaps and bounds in the numbers. I think the contributions from companies like Lam are more important than ever before, though, and people are starting to realize this. As materials-driven innovation moves to the forefront of process technology development, as physical limits get closer and closer, and to that end, semicap has really been top of mind for many of my clients. You know, there are, of course, in the nearer-term questions, like the NAND spending trajectory in China and everything else that we will certainly get into today.
But I'm also increasingly getting people looking at the longer-term potential of this industry as they start to view it as more on secular rather than purely cyclical terms, which I think is a good thing. And so to talk about that, it gives me great pleasure to welcome Tim to our session. So thank you so much for being here today.
Great. Well, thanks.
Thrilled to have you.
Thanks, Stacy. It's fantastic to be here. Appreciate you having us.
You bet. So, it's—
Oh, that's, that's for the questions. Sorry. While you do that, people can notice our safe harbor that's up on there.
Got it. So I do wanna get... I don't wanna focus on the short term, but I do wanna get it a little bit out of the way.
Sure.
Maybe if you could just review kind of what Lam saw in 2023, and maybe even how your view on the market, as well as Lam's potential, kind of changed. Because I think in the beginning of the year, like, things were looking a little worse than maybe they did at the end, and you know, you guys had gone through a bit of a cost-cutting exercise as well at that point when things were closer to the trough. How has the view of, like, what the industry and Lam did evolve as we went through the year, and kind of where are we as we go into next year?
Sure. Well, 2023, as you said, it ended up a little bit better than we thought it would at the beginning of the year. But-
But it ended up a lot better than I thought it would -
Yeah, okay, maybe. From our perspective, you have to think about the fact that our strongest market is NAND, and that was one market that really did not recover through the year. I mean, NAND WFE in 2023 was down more than 70% year-on-year, which again, from a Lam perspective, was that, that's a very severe downturn.
But, you know, we saw, did see some improvement through the year. China got a little bit stronger, obviously, continued strength in mature node spending. There was some initial buys around the initial implementation of HBM DRAM with some of the early AI look in 2023. So we did see the year progressively get a little bit better, but again, from our perspective, that NAND lack of NAND recovery is really still what we are waiting for as we move into 2024 and really as we look forward to 2025.
Got it. And I have to say, at the trough, though, I think your trough EPS guidance was $5. You beat the number. But that was a $20 annualized number for a purportedly NAND-focused semicap with almost no NAND in the model.
Well, we have worked hard to strengthen the company structurally. You know, I think one thing we're very proud of in what was a very severe downturn for Lam, from our strongest market perspective, was that we, trough to trough, had improved our operating margin by 200 basis points. And you said our trough earnings per share were quite nice.
If there's one silver lining that you can see in a year in which there's almost no NAND spending, is that the progress we've made structurally in our foundry logic positioning our DRAM positioning, becomes a lot more visible. And that's a strategy that we've been very deliberately driving for the last five years, is to... You know, of course, we love the NAND business, and we're the leader in that space, but we recognize that to be the company we want to be and truly capture the opportunity that's ahead in this industry, we want to better balance the company with stronger positions in both foundry logic and DRAM.
Right. Maybe that's a good segue. What are some of the areas that you're focusing on outside of NAND in, in foundry logic and even in, in DRAM? And I guess how is that influencing how you're viewing the market evolution into 2024 and beyond?
Well, what's great about, you know, our business, I mean, Lam is a company that is focused on etch and deposition. And etch and deposition are two of the fastest-growing segments within wafer fabrication equipment. Primarily because these are the types of equipment that are needed to build all of these complex three-dimensional structures. They're the types of equipment that are required to deposit all of these new materials and etch those materials. So they're really fundamental to several of the really critical technology inflections that you're hearing about. We look at foundry logic, we've talked about the fact that Lam is focused and is a big player in the gate-all-around inflection. You know, think about this gate-all-around transistor structure.
Looks very three-dimensional, requires lots of new process equipment in the etch and dep space, including ALD and selective etch. So we focused on those areas, introducing new products. You look at things like backside power delivery, another foundry logic inflection. I mean, Lam is very well positioned. Etch and deposition, you're now having very thick interconnect layers on the backside of the wafer. It's perfectly what we're great at: electroplating, thick film deposition with high productivity. You look at other inflections in DRAM, for instance, again, HBM building three dimensionals, three-dimensional memory structures. And so we just look at the markets we're in.... where we can focus, and we see plenty of opportunity in foundry logic and DRAM to grow our business.
Got it. Have you guys ever given any sort of TAM sizing for some of those? I know, I know one of your competitors has. I don't know if you guys have or not.
Well, we've talked about each of these inflections. We talked about four kind of billion-dollar opportunities ahead of us.
Right.
And so in many of those, in fact, even from when we put out those numbers, they're sort of accelerating a little faster. Gate-all-around, we said already this year we're gonna have $1 billion in shipments to, to gate-all-around nodes. We said our advanced packaging business is now $1 billion of shipments this year. And so we've kind of and then backside power, we said, would be about a billion-dollar opportunity for us. A nd also we've identified the last inflection, which is dry EUV resist also is a billion-dollar-plus opportunity for the company.
I want to come back to the dry resist in a minute.
So we... You know, from this perspective of Lam's size, yeah, and I think that, you know, maybe we can get to this a little bit later, but I think there's something that's very unique about Lam which is that today, if you look at Lam as one of the large equipment suppliers, one of the largest, and yet we have a rather unique feature, which is we only address about mid-30s% of WFE with our served market today. That compares to some of our peers who are well over 50%.
And so kind of when you take out litho, I would say that some companies don't have as much opportunity to find those new white spaces to grow into as Lam does. And so that means we can be very laser-focused on where we think the greatest opportunity is to apply our product and technical skills and expertise to these new areas.
That's what we've done. We have picked specifically those types of new markets where we think we can be very successful. ... based on what our engineers know how to do and what types of tools we're good at building, and that's those billion-dollar opportunities that I talked about.
Got it, got it. I guess to come back to the core business, though which is NAND. Clearly, it's, you know, we were doing $20 billion a year for quite a while during COVID, and it was like, to your point, $6 billion or whatever it was last year.
And it's probably up a little or not, maybe this year. But, I guess two questions: How are you viewing the prospect for a significant NAND recovery at some point? And, you know, on the competitive environment, as that starts to come up, you know, I get a lot of questions on Tokyo Electron and cryo-etch and I was wondering if you could address the competitive environment in some of your key markets in NAND?
Sure. Well, you know, as I said, we're excited about foundry logic and DRAM as new growth opportunities. I mean, the heart of the company right now still is t he leadership we have in NAND. You know, we look at that, and, you know, it's probably the best part of the story right now, which is our strong... Yet we're showing, you know, really nice revenue and profitability in the company. So as we look to 2024, NAND has improved a little bit but from a equipment spending perspective, very little.
What we talked about on our last earnings call was the fact that we had finally started to see some tick-up in fab utilization. But even there, that utilization is improving really only at the leading-edge equipment. And so that really is still quite underutilized and in badly, badly in need of, technology upgrades. So we think that as we kind of move towards the end of this year NAND demand is in recovering, utilization starts to pick up, and what we're really gonna see as we move into 2025 is an earnest effort by our customers to begin upgrading that installed base of systems to bring it to the leading edge, where you know, they're going to be able to make the devices at the most cost-effective technology node.
So we think that 2025 will be a strong year for NAND. B ut one that's characterized by technology upgrades to the installed base. That plays right into the strength of Lam.
Is that better than, like, greenfield capacity additions for you or?
Well, for us, it is, it's almost equivalent. A nd the reason being, the vast majority of the money spent on a technology upgrade is on the etch and deposition tools that are required to build the 3D stack and also to etch the features into that stack. And so, whereas in a greenfield, you need to buy a whole lot of ancillary equipment required to make the device. In the upgrade, you're really talking about building the stack taller, and that, that really is Lam's equipment. And so-
I guess the installed base is fairly large as well, right?
You know, it's... And so maybe to address your question about the cryo-etch and competition. You know, first of all, we've always had competition for every application. And, you know, it's a you know, we don't have any positions where we don't really have somebody trying to take those, those large positions away. But what we've done over the years is, you know, we have recognized that the installed base, and the partnership with our customer on that installed base is an incredible learning vehicle for us. And so when we talk about cryo, I mean, you know, we, we've— we feel very comfortable in this space because we have nearly 1,000 cryo chambers running high-volume production today for that application. And we have about 6,500 high aspect ratio etchers out in the field.
You know, that gives us this tremendous learning vehicle. You know, what we've done, and we've talked a little bit on previous calls and in some of our conferences about how we're using the information that's coming from that installed base as well as from our labs. We've now put that together into what we call the Semiverse Solutions.
Semiverse Solutions are our advanced simulation tools. Ah, ah. We use AI for recipe optimization you know, we, we've said an advanced etcher today has somewhere on the order of, I think the number was something like 11 trillion possible recipe combinations now. And it's just an incredibly complex o ptimization task for engineers. So now we can use Semiverse Solutions. We can use our simulation capabilities to begin to optimize recipes.
Are you, like, simulating profile evolution?
Profile evolution.
Sorry, this, this is from my grad school days, like 20 years ago, so this is...
And so when we apply that to cryo, and you're trying to build these incredibly deep- etch profiles. We've now taken that 1,000-chamber installed base of learning taken the ability to simulate profile evolution in our latest systems.
We've accelerated our cycles of learning. And that's allowed us to create some really extraordinary results. And if you wait just a couple of months, we're gonna talk about those publicly. But the technology capabilities are accelerating very fast in this space right now. A nd we feel really comfortable with our leadership.
Got it. Got it. I wanna ask about China. So it does seem to me that, you know, again, the outlook for WFE got significantly better as we went through last year. And by the way, we did... I know we were just having a discussion. We've, I think, different baselines for what WFE actually was last year, but our numbers are somewhere in, like, the low-to-mid $90s. I, you know, Lam's gonna be a little lower, but fine. But we've started the year, we thought was gonna be $70.
So I think there were a number of things that got, but I do think China was probably the thing that got the most better. And it is a controversy right now in terms of the sustainability of that China spend, and it's, it's in lagging edge logic and in DRAM and in other areas. I wonder if you could comment on your views on the sustainability of that spending environment at this point.
Well, we've said a number of times, we think the spend is sustainable. Certainly, all evidence right now points to that, in that, China's spending has continued to remain strong this year and perhaps even get a little bit stronger,
But why is that?
You know, I think that, you know, partly it's that, it's the general environment of semiconductors maybe is, is one, which is, everybody's looking at this tremendous opportunity ahead, and you know you have to build that capacity ahead. You see this also, push around the world for, regionalization or, or, a localization of capabilities for, for all sorts of reasons, whether it's, concerns about supply and, and sort of the next disruption, or it's a localization from the perspective of, just, better partnering. As we say, sometimes being close to, you know, between, with the ecosystem is, is effective for speed and learning.
But we see that occurring all over the world and in China. And, you know, so we're strong believers in the 2030 $1 trillion semiconductor industry, which means that from here to there a lot of fabs have to be built, a lot of capacity has to be put in place. You see that taking place with fab announcements in the U.S., and you certainly see that with fab announcements in China. Everybody sees that same end market opportunity in semis. And so I think at this point, there's sustainability of fab construction and capacity addition.
And then you can just argue as to in what region will that take place? A nd who will the winners be ultimately in terms of which of those companies building the fabs service that demand?
Who are these companies, though? I've... You guys had talked about your deferred revenue. I don't know if it happened this last quarter or not, but it was going up still because you were selling to customers in China that you had never sold to before and you want to kind of cash on the barrel before you, before you ship the tool.
That is true. We... You know, we're cautious when there's new players in the market that we haven't done business with for a long time. And so, you know, China is an emerging, still emerging, region for semiconductor manufacturing as they add capacity. Some of it is with new companies new fab projects in new parts of China that, you know, we're not so familiar with.
And so we've required advanced payments before we would begin building the tools just to safeguard us and our shareholders. You know, I think that's not unexpected, though. You're gonna see that in some other regions that have announced interests in growing their semiconductor ecosystem. There will be new companies that emerge.
Now, in developed countries, you know, like, that have a long history in semiconductors, they already have well-developed companies that tend to be the ones who are putting in those investments. But, in these other regions, I don't think it's a surprise to see new entrants to the market.
Where are yields and everything with these new factories? Like, do these companies know what they're doing? Do they have the expertise that they need to bring these things online?
I think it's a little early to speak to yields, and I probably couldn't do that anyway relative to our customers' information. But, I think that, you know, there is a learning curve, clearly, from the time that a new fab is built, and equipment is installed, and those ramp to to full yield. That's true regardless of where those fabs are built in the world. It's a tough process, but I think-
I would say I'm amazed any of this works at all, so yeah.
So I think that, though, that over time, again, the experience is really key in our industry. That it will take some time, but over time, you, our customers do get there.
Got it. Are you guys servicing... We'll talk about services a little later, but are you, are you guys servicing those tools?
Worldwide or specifically?
These specific ones with these new customers in China.
We would be... Where we have no restrictions on us, we we both sell and service our equipment, so-
I mean, the service is really... And I think we should spend some time in a bit talking about the service business it's become such an important part of the company. But, you know, you just talked about how difficult semiconductor manufacturing is and about that experience. The ability to service and maintain the equipment is... It's not a simple task.
I mean, it's something that again is where many times the OEM really is the best, party to to do that type of work. And, you know, not only do we do it from the standpoint of traditional break-fix, but, you know, when we talk about servicing equipment, we're really talking about bringing a whole new level of data-based solutions.
The types of analytics that allow us to very quickly understand how that equipment is operating. In these fabs now, I mean, these very, very large fabs, you may have 50 chambers of Lam etchers running the same process. And, you know, when you're trying to get yield, you need every one of those chambers to match each other very precisely, and that's where you bring in big data, machine analytics.
Big data and the ability to identify chamber-to-chamber mismatches that ultimately could cause yield loss or distribution-wide thing.
Maybe that's a good segue to the services opportunity. So I mean, maybe talk a little... I mean, you've been talking about the business already, but I mean, like, what are you doing there differently today versus, say, five years ago?
I know it's evolved, like, pretty considerably. Like, how do you address... You talked a little bit how you address some of those customer needs through services, but what's the ultimate revenue opportunity there? You know, like, how much of your installed base is on, you know, like, I guess, recurring contracts or subscription? Like, how does that business work?
Well, first, I mean, you have to understand, we have about 90,000 chambers in the field. You know, the nice thing about many years of high WFE is, I mean, that means we're shipping a lot more tools into it and so on-
What, what was that installed base, like, five years ago?
We're up about 50% in that time. And so, you know, the installed base business, therefore, you know, has grown tremendously. It's a very important part of the company. Depends on the year, but it tends to run in about the 30% of our range of our revenue. You know, but we're really transforming, you know, what we think of as the services business.
You know, a key element of the services business, obviously, is maintaining the equipment in the fab. And as I talked, we're bringing, you know, new types of capabilities AI-driven analytics to that service. You know, you think about a couple of items. I mean, one is the cost of building a new fab. $15 billion-$20 billion for a leading-edge fab.
Time to money is how fast you can install those tools, get them qualified and running. And again, back to chamber matching, we've shown that you can use the data that comes off of those tools to dramatically shorten the time to get a fab, at least the Lam equipment in that fab up and qualified for production. And so, you know, we look at helping the customer save money in that way. Same thing with tool availability.
You know, you've paid a lot of money for this equipment. The need to keep those tools up and operating, ready for production, at a higher rate now than it was even possible in the past, is, that much more important. And so, you know, we've looked at that, you know, the use of that data. The second is, you know, when you talked about this regionalization of fabs, and one of the things that's gonna happen with regionalization of fabs is, there are fabs being built in areas where, you mentioned, do they have the expertise? They even ask: Do they even have the labor required to operate some of these fabs? So we're looking at a lot more automation.
One of the things that Lam has been focused on recently is looking at some of the more labor-intensive and also labor-sensitive, meaning if you don't do the procedure correctly, the tool might not come back up into production. And we're automating those. W e've introduced the use of collaborative robots, cobots, to do some of the more labor-intensive tasks on our tool.
On the tool maintenance?
On the tool maintenance. And so these cobots, engineer will wheel this up to the tool, and basically then, for the next six-seven hours, this cobot can do-
Like it will replace the pumps and-
- cleaning, cleaning the changer, cleaning the chamber, replacing some of the parts, and then basically the tool's buttoned back up, and the qualification, the requalification success rate when the cobot does the work, is significantly higher than when even our experienced engineers do that work.
You charge for that, I assume.
And so when you talk about... We're working on the model for that, but this would be something where, of course, we're not looking to get into selling the cobots but we're looking to sell that service, that results-based, outcome which is, that tool is now up at a higher percentage than it was without the use of the cobot.
The savings for the customers, obviously, is the need to hire and train and retain people to do what is, not the most fun work. I mean, cleaning an etch chamber, you know, every couple-
Do a lot of customers do that work themselves right now?
It's a combination, but often it is-
I've cleaned etch chambers, by the way. Like, you're right, it's not, it's not fun.
Sometimes it's done by us, sometimes it's something they contract out to others, or sometimes they do themselves. I think that... Yeah, so we're really looking at that 90,000-chamber installed base as this tremendous opportunity for us to innovate, you know, new services that are really win-win for us and the customer. I mean, they... And in many places, that need to find the workforce is not an easy task.
Uh-huh. So it sounds like you're suggesting to me that you do still see opportunity for not just installed base chamber growth, but like, I mean, in dollars per tool, I don't know how you have the right way to think about-
Oh, no, in fact-
- services, revenues per tool.
You know, if you think back to our last I nvestor D ay we talked about the fact that not only do we see expansion of the chamber count, but our goal is to capture greater revenue from every one of those chambers out in the field through new offerings and new capabilities.
Yeah, got it. And I know you've said historically, you generally see services growing even in a down year for WFE, correct?
That, that's right. So the services, you know, one-
Because of this.
Yeah, because of this, but also we've said that services business, maybe I should just explain a little bit. Also, includes the services, and it also includes our spares business. So even in a down year, our installed base is getting quite a bit larger. And so what we've actually said is that we expect the revenue from our installed base business to grow faster than the growth of the installed base itself.
Which is an objective that implies just what you said, which is we'll capture actually a greater revenue from each of those chambers year on year on year, through the introduction of new products and services. And so that's, we've been traditionally doing that, except for last year.
That was the export controls.
Export controls took out some of the installed base. Nobody ever really gets rid of tools that are in the installed base, but the export controls prevented us from being able to service- certain customers that we'd previously been doing business with. But also, the second was, we had never seen in the past, utilization cuts to the degree that we saw last year, especially in the NAND space.
And so the spares business, that usually is a year-on-year grower- actually took a step back. So the export controls and utilization cuts, a ctually combined for sort of the perfect storm in the services business last year. But that's kind of behind us now, and I think that as you look forward it's again, we're back to, we would expect year-on-year growth in that business.
Got it. Remind us what the overall impact of the export controls were on the overall business?
We'd estimated them between $2 billion-$2.5 billion of impact. And so, pretty meaningful for Lam. And so again, you combine in 2023, export control impact and NAND utilization and NAND downturn. It's why we think that, actually, the company performed quite well in the face of those two setbacks.
Do you think there's any risk of, like, further export controls? I mean, moving to, you know, more trailing node stuff across the different markets or...?
Yeah, you know, I mean, obviously, we can't forecast you know, changes in U.S. export policy. You know, it's something that we're in constant conversations about what it would mean both for, for U.S., equipment companies as well as, just our general competitiveness. We'll, we'll have to see, but it's something that is very difficult to forecast.
Okay, got it. I want to go back to, you know, you talked about $1 trillion of semiconductor and my own view, by the way, we'll get there. I don't know if it's 2030 or 2034 who knows? But we'll get there. What do you think about capital intensity as we move toward that? So, like, you know, capital intensity over the years in the industry has moved around. It was running 20% +, you know, in the pre-2000, and then the transition from 200 mm to 300 mm took it down pretty significantly.
I think we bottomed at 8%-9%, probably in 2010, and we had the 3D NAND transition, which you guys benefited from greatly. My own view is that capital intensity, NAND probably doubled, maybe even more, as we did that transition, and now it's logic's turn and everything else. Like, at $1 trillion, like, where does capital intensity go? Like, I think you or some of your competitors talk about, well, we can do 15%, we're doing 150%. Is it plausible we could have a trillion-dollar industry and a 20% capital intensity of a $200 billion WFE at some point? I mean, is that like, is that within the realm of possibility?
Well, it'd be a good outcome for us so I don't want to say it's not plausible. But, you know, I think that, you know, one, from where we are today, which is sort of mid-teens to a little bit higher than that, in capital intensity, I would say that, you just have to ask yourself, I mean, are devices and semiconductor manufacturing getting tougher or easier, more complex or less complex?
You know, I don't actually see anything on the horizon that materially drives capital intensity lower. Y ou know, the economics of the industry, the economics of our customers, you know, part of the way we compete is by bending that cap, that complexity cost curve.
And that's what makes a solution, you know, really appealing to our customers, as we bring some new technology. Take ALD, for example. A lot of people can do ALD. It's atomic layer deposition-
Oh, sorry
- by the way. Atomic layer deposition. You're depositing films, atomic layer by atomic layer. You can do this in very complex three-dimensional structures that's needed for gate-all-around and some of the new, new applications. It's been known for a long time. Many people can do it. The trick in winning in ALD and why Lam is a winner in ALD: productivity. Meaning you deposit ALD in a way that customers can afford it. And so I think that you know, looking forward, we're very sensitive to the fact that, you know, capital intensity unchecked without a focus on productivity, yeah, probably is 20%+.
I mean, just the scales with complexity. But I think that the winners in this industry will be those who recognize that and introduce new capabilities that keep that productivity better. And I think that when you, you asked about, for instance, cryo-etch. One of the attractive things about cryo-etch is the etch rates are higher.
Not only can you etch deeper features, as Lam is demonstrating, but you can also do it faster. Therefore, it's both a technology step forward for customers, but also brings with it some productivity enhancement. So as they go, as customers scale to 1,000 layers of NAND the cost of etch didn't scale proportionately to that. That's how, you know, Lam can win, and our customers can win, and the industry can continue to grow.
Got it.
So, but I think, you know, again, we feel very comfortable to say at $1 trillion, $150 billion WFE is a fair number and likely at the low end of where that WFE would be.
Got it. What do all of these innovations imply for how you price? Like, we've had this conversation before, but I always... I look at what you guys deliver I always wonder, like, why do the gross margins have to start with a four? Like, I've asked you this question before, but like , how do I think about the evolution of that going forward?
Yeah, well, we're always trying to drive that higher. Obviously, it's a fine balance between, you know, again, customer affordability c ompetitiveness in the market, and your own profitability. But I think that structurally you know, as you pointed out, the industry, and the equipment industry, and Lam, we've all gotten dramatically healthier.
And you look at the gross margins of Lam in the last trough was very, very healthy. And, and what we have coming, I mean, there's an element of pricing that certainly will help that. I think we get paid for the technology innovation we bring. But beyond that, there's a little bit of self-help that I think is here in the very near term for Lam, which is our large investment in Malaysia.
And so, you know, if we look at sort of the pre-COVID, pre-2020 timeframe to today, we've dramatically restructured where we manufacture, where our supply chain is located. And I think that we've built a global footprint for our manufacturing supply chain that makes us significantly more resilient, allows us to scale much faster to the demand that we see coming in the future, and also improve our gross margin, looking forward.
And, I think that, you know, as we move through these next cycles of industry upturn companies and our output hitting new highs, you know, the real power of that new manufacturing and supply chain infrastructure will really start to come to bear in our profitability.
But why is Malaysia so much cheaper, margin accretive? Is it just bigger, is it labor costs? Is it just you're closer to where you're shipping tools or...?
It's a few things. In fact, it's all of those. You nailed it, Stacy. It's, you know, I mean, it's obviously we were able to build a facility that's of significantly greater scale than we had had before. So you get benefits from that scale. We've built a... You know, when you really start from greenfield you can also build in a lot of the latest technologies.
We just opened, not too long ago, our newest, largest warehouse, fully automated. You know, it's like, if you want to see robots and cobots operating, this is a really impressive facility. We also are very close to our supply chain. That supply chain is obviously operating in a low-cost region, so that helps them pass on their savings to us.
And then transportation-wise, still, you know, 70%-80% of our business is in the Asia region. So now transportation, freight, logistics, all of these sorts of things, just... You're just chipping away at the costs. And as I said, one, it was for cost reduction. Two, it's for resilience. I mean, you know, as much as we'd like to think that the COVID disruption was the last one we'll ever have, you know, we've been mindful to construct the infrastructure of the company in a way that, ideally, will be much more resilient to any future shock that would come.
So, you know, we saw during COVID, freight and logistics costs skyrocket. Even though we were at very high output, we should have seen tremendous leverage in our model. We didn't, because you had chip inflation because of shortages. You had freight and logistics inflation because of all kinds of supply disruption and shipping and transportation disruptions. And so some of those things were really in the back of our mind as we built this new facility and decided where to locate it. I think that'll bode well in this next upturn.
Got it. Got it. I'd be remiss if I didn't ask you about AI. I gotta ask everybody. What does it mean for Lam? You guys have put out some numbers. I think the last one I heard was, I think, was it a 1 percentage point increase in AI server penetration is $1 billion in WFE, was the number, I think.
Yep. $1 billion-$1.5 billion.
$1.5 billion. So I guess a couple things. Like, like, number one, where does that number come from? And number... Like, like, in general, like, how does AI, like, drive that? Like, what, what is it about AI that's actually driving that, that, that number? How does it help Lam?
Well, I think that specific number is actually pretty easy to calculate because it's based simply on the semiconductor content of an AI server versus a traditional server. You really start at just the specs of the GPU. You know, you really look at the number of advanced foundry logic chips and the GPUs, going up, dramatically compared to a standard server. Amount of DRAM going up amount of NAND going up. And so it's a pretty simple calculation to say, if you use a lot more servers that are highly intensive in their use of semiconductors, you'll drive more WFE.
So that's where that came from. But when we dig a little bit deeper, and we say: What does it mean for Lam? What does AI mean? Again, it comes back to the technology requirements of AI. I mean, it is. You know, you, when you're doing AI, you're looking for compute power, you're looking for transistors that can operate with you know, better performance, less power. You know, those are driving things like gate-all-around. They're driving backside power distribution and delivery. They're driving advanced packaging. These are all tremendously important inflections for Lam. And so, you know, AI, as an accelerator of those technology trends is beneficial for Lam. On the DRAM side, obviously, the move to HBM for AI, you know, has two big impacts for Lam.
One is the die size. You know, you've heard our customers put out numbers like, it takes three times the number of wafers to build one bit of HBM memory versus conventional DDR5 memory. I mean, that's, you know, as an equipment maker, three times the number of wafers. We're wafer-based. I mean, if you need to build three times more wafers, you need three times more equipment, effectively. And so, HBM is a big driver for us. And technology-wise—you know, we are the leader in TSV applications for—
Through-silicon via.
With little holes through the- Sorry, we should have passed out the-
Probably should have passed out a glossary.
- glossary. But for TSV, the holes that connect basically allow you to stack the DRAM die, LAM has 100% share across all the major players for the etching and the deposition of those the etching and the electroplating of the films.
You know, this is also a big driver for us. I would just say, when we look at AI, it starts with demand and the growth we're seeing there, but really translates into acceleration inflections. All of which increase Lam -served market, and all of which we think we're very well positioned to win in.
Got it. Got it. I said earlier, I wanted to ask you about dry photoresist.
Yes.
Please talk about what that is exactly. How does it benefit... What's different versus, like, sort of traditional methodologies for EUV lithography, and what are you doing in that space? I find it fascinating, personally.
So again, both in the foundry logic space and the DRAM space, I think many people are aware that the lithography is changing to EUV to be able to print smaller, smaller features. And up to this point, that's worked pretty well. But as you continue to try to print smaller and smaller features, you need to change the chemistry involved in the resist itself. And also, we believe you need to change the methodology by which that, those, the resist films are deposited. So today, they're, they're conventionally deposited using wet, wet films that are spun on the wafer. And Lam has developed a means by which we can deposit those same types of materials dry using equipment that looks very much like our etch and deposition equipment.
And what we're finding by doing that processing, both the resist deposition as well as the development post-exposure dry, is you're finding much better control, the ability to create higher fidelity patterns with less defectivity. And, you know, and the real benefit is the class of materials that we're depositing now also requires less EUV dose to expose the film. Which means that the productivity of the EUV tool goes up. And so as the-
How much less dose?
You know, we estimate somewhere between 20%-30%- less dose. And so-
Is that just because the resist is thinner, or?
It's because the resist, the material we're using has more EUV photon absorbers. Therefore, basically requires less EUV- dose to expose that film. And so if you're talking 20%-30% less dose, directly translates to-
Throughput.
- throughput for the EUV tool. Obviously, everybody knows that EUV tools now are incredibly expensive machines. So again, it's a win-win for, y ou know, Lam gets to introduce and sell into a new class of a new category. It's really, this is 100% served market expansion for Lam. We don't compete in the resist deposition space today. Traditionally, that's done on a track system.
This would replace track, basically?
It would, it would replace track. It's a great new market opportunity for us and for our customers. It helps them save capital expense in an area that really has become a very large portion of their capital budget.
Okay, got it. That's... Where are you in terms of adoption of this right now?
You know, we're engaged with every user of EUV has our, has our equipment. You know, what I would say is that, we had hoped to be, probably had hoped to be a little bit further along in terms of actual production implementation.
But, you know, this is a major disruptive change. Tracks have been used for decades and decades. This has been the means by which this is done. And so there is a little bit of a hurdle to get over in terms of adoption inside the fabs, but I think we're this is on the horizon. And especially as we see the demands and the cost of High NA on the horizon, the value of this, this technology becomes even greater as we move towards that.
Okay. Yeah, that, that's interesting. I, I thought the biggest reason to use it was pattern collapse, but it sounds like it's much more than that.
Well, pattern collapse is one of the biggest advantages of the dry develop portion of the process. So, you know, after you've exposed, and now you're going to basically, you'll have the very fine pattern feature. At that point, using any wet process in the wafer can result in pattern collapse.
What happens when you develop this using a wet process, the surface tension can pull your features over if they're too tall relative to how wide they are?
That is, that's a big benefit from that dry develop portion.
I see. But I hadn't realized there was a benefit on the exposure side as well. That's actually really interesting. Sorry. I wanted to ask about the 200 mm side of things. Because I know you include 200 mm in, also in the services.
Yes, we do.
It's Reliant is the name in-
Yes.
I understand you but—clearly, the industry has been selling actually a lot more greenfield 200 mm than we. I guess, why is that, number one? Are there innovations that you're still bringing on 200 mm, or is that just like a static kind of thing? And I guess, how big is that market for you, or how big has it been over the last several years as a, maybe as a percentage of revenue, or percentage of your services business?
Yeah, I don't think, I don't think we've broken it out. What I would say is that, 200 mm continues to be an important part of the market especially for certain specialty applications. You know, if we look back, you know, probably just five, 10 years ago, we would have thought of that market as—w e often called that our refurbished tool market, meaning that, you know, you found old 200 mm fabs were being shut down. We'd often buy those tools back, we'd refurbish them, and we'd resell them. It was a good part of the business, but it wasn't one that we focused on a lot.
Now you're really seeing a lot of innovation occurring in the materials. You know, if you think about the types of things that are being done in 200 mm, it might be things like, silicon carbide or GaN. You know, some of these new materials that really don't exist at 300 mm. In those cases, we're actually innovating new products, new technologies. One of the ones that we've talked about r ecently is, we've taken atomic layer etching. So atomic layer etching is one of our most advanced capabilities. It's used for patterning extremely fine pitch patterns with can be with EUV, or it can be for features like gate-all-around o r, you know, the atomic layer etching is, yeah, it's a 2 nm and below -type of technology. We've applied that now to 200 mm wafers to do atomic l ayer e tching of GaN.
The reason being, you know, when you start working on some of these specialty materials, you find them to be very sensitive to plasma etch damage. An atomic layer etch process becomes a very gentle etch. Capability allows us to very precisely layer by layer, etch those materials. And so, you know, it's, it's just an example of where, you know, we're really bringing, like, learning from the very leading edge.
You're bringing it back to a technology today that most people, if you define technology as the critical dimension width or the critical dimension, wouldn't consider it to be advanced at all. But the materials themselves make it a very complex solution that's required. And so, we're still investing in 200 mm and it's a good business for us.
Got it. So we've got about five minutes left. Should we go to the lightning round? We've got a few questions here.
Sure.
Again, if you have any questions you'd like to submit, you can please submit them to the Pigeonhole form, and we can ask them here. What's the sort of local Chinese semi cap players, in your opinion?
Well, I think obviously in the places where we can no longer sell. They have, they have gained share. But I think, in the longer run, I mean, the advantage of, companies like Lam is, again, back to that installed-base learning. The know-how, to help our customers get up and running and yield in a fab, still, I think makes our equipment and the equipment from other global players extremely attractive.
But, you know, there's no doubt in certain segments of the market and in certain restricted customers, they're gaining, gaining share and gaining size c ompared to where they were just years ago.
Okay. Do you have any particular ones in China that you guys look at more or less? Like, you have AMEC that does more etch, right? But-
Yeah, we have. Obviously, there are some who are in our space. And again, they've continued to improve their capabilities as they've gained more experience in some of the restricted fabs. But we look at the market broadly for, for all comers. We, we're always looking at competitive threats, where they might come from.
Got it. What do you think about the CHIPS Act in terms of driving WFE spending higher?
Well, it is driving WFE spending higher. I mean, I think that, again, the regionalization, as I talked about, it is whether it's in the U.S. CHIPS Act, it's the European Chips Act, it's the funding in Japan, the announcements in India. All of these, I think, over the next, you know, five to seven years are gonna be driving-
Does it really drive it structurally higher, though, or is it just, like, pulling it forward or just disaggregating it? I mean, presumably, it's demand that should determine how much gets built, not subsidies.
It's a little bit of both. I would say it's a little bit of both. You know, one, I do believe it's probably pulling it forward a bit. But two, in the long run, there is a benefit to consolidation in my, in my personal view.
Which means, large, large locations where you have, you know, a large number of fabs. As you regionalize there will be some loss of efficiency. So I think that, you know, there can be some overall uptick in the total WFE that's required in a more decentralized semiconductor ecosystem.
Got it. How are you thinking about R&D investment levels going forward in the near to medium term? And maybe even ask like, what are the areas that you guys are excited about, that you're investing in, that are new? Maybe, maybe I would add that as well.
Okay. I mean, I would point out, we, we had said at the beginning of this year that 2024 was an R&D investment year for Lam. And what we meant by that was that you know, while our operating margin, like I said, has actually improved significantly and we're quite happy with it now, it's lower than where it should be for this, this revenue level, because this year we need to invest for some of those inflections I talked about. Things we're excited about. We're obviously excited about advanced packaging. We're excited about our dry EUV resist.
We're excited about backside power and what that means for us. We're excited about gate-all-around. Really excited about HBM. I mean, these are all just things that we know that right now is the moment we have to capture those, and we're committed to introducing the new tools required to win in those spaces.
But I think as we see revenue, we see the NAND market recover next year. I think for us, we think our R&D levels normalize back to similar to what they've been in the past. And so, you know, R&D investment for us, it's you know, we live and die by our technology so we always make sure, first and foremost, we invest what's needed for the future growth of the company. But, 2024 is a little bit higher. We're also very committed to make sure that we spend the money, you know, spend all of our OpEx well. I mean, right now, we're spending about 70% of our OpEx goes to R&D. So you can see the priority we place on it in the company.
Got it. Got it. Maybe one more. I think this is a China-focused question, but, question on: How does yield improvement impact Lam demand?
Sorry, say that again?
Sorry, I think it's a China -focused. As yields improve over time, how does that impact demand for you?
Yeah, I mean, I think as yields improve in general, you know, theoretically, demand should go down, but often, as yield improves, I mean, you find ways to find new applications for chips. And so, you know, from our perspective, it's why this cost and demand issue is they're so interrelated.
Moore's Law was driven by the fact that if you could make things cheaper they get used more. And so that's, you know, I think that we're always looking to make yields go up, make our customers more successful. That ultimately drives more business for us.
Got it. So we've got about 20 seconds left. Maybe a little more, but I'll, I'll give you your soapbox at this point. Why should investors buy your stock?
I get 12 seconds for my soapbox?
12 seconds. You can go a little over.
Well, you know, it's everything I just said about the tremendous opportunity. We're a fantastic industry, one that's set to grow. Lam's in a very unique position to grow, not only in our served market, but also our market share across all those technology inflections. And I think maybe, just if that's not enough for you, we just announced our authorization to buy up to $10 billion of our stock back as part of our cash return to shareholders, and so we're buyers, and we think all of you should be as well.
Got it. I think that's a good way to close it out. Thank you so much.
Thanks.