Lam Research Corporation (LRCX) Earnings Call Transcript & Summary

Great. Welcome back, everyone. I'm Joe Moore at Morgan Stanley semiconductor research. And very happy to have with us today the management team of Lam Research: CEO Tim Archer and CFO Doug Bettinger. I think before we get started, Doug is going to read the safe harbor, and then we'll jump right in.

Yes. Give me 30 seconds with this. I always like to keep the attorneys happy. Today's discussion may include forward-looking statements that are subject to risks and uncertainties. Actual results may differ materially. Additional information concerning these risk factors that could cause actual results to differ in those forward-looking statements can be found in the risk factors disclosed in our public filings with the SEC, on Form 10-K and 10-Q. And with that, Joe, we're ready to go.

Great. So maybe we could start. You just did your Investor Day a couple of weeks ago. So to freshen in everybody's mind, can you remind the audience what message you want us to take away from that meeting?

Sure. Yes. Thanks, Joe. It's great to be here. It was just a couple of weeks ago, we had our Investor Day, first one we've done in 5 years. And so we had a whole bunch of messages we wanted to deliver. But really at a very high level, the message is focused around Lam's expectation and our confidence in our ability to outperform WFE as we enter this period of vertical scaling. Basically, if we look at what's on the road maps for every device type, whether it's NAND or foundry/logic or DRAM, we see increased focus on increasing performance, increasing scaling by going in the third dimension. And what that means for us was the deposition and etch intensity will grow. We've seen this coming. We were obviously the leader in the transition from 2D to 3D NAND. We saw it coming years ago that it would eventually be here in DRAM and foundry/logic. And so we've been preparing a number of new tools. And so we introduced 2 significant new tools last week as well we can talk about, which will allow us to gain an increasingly higher share through those inflections. And then the third key message, since we hadn't updated in 5 years, but we've talked about it on every earnings call was our CSBG business, which has just become an ever increasingly important part of our growth and revenue story and especially our profitability and stability story because it has proven to be a lot less -- a lot more resilient to the kind of WFE cycles that we see in our industry. Every year, our installed base of system grows. And every year, we're working hard to increase our ability to capture more revenue from that growing installed base. And so really at those levels, WFE moving in our favor, the right products at the right time and an installed base that helps us through cycles.

Great. Very strong message given your history with vertical NAND, as you mentioned. Can you talk about -- maybe this is for Doug, talk about some of the market assumptions behind your 2028 financial model, both the WFE and share?

No, sure. Yes, listen, when you're forecasting something that's 3, 4 years into the future, lots of assumptions going to it, Joe, as you might imagine. Maybe I'll unpack a little bit of it, and then you can ask me clarifying questions, if you'd like to. I think overall, first, let's start with our outlook for WFE. We think WFE grows at a CAGR of, I don't know, mid-high single digits, a good working assumption, I think. And then maybe more importantly, when we look at the markets we're in, to the point Tim was making, we see things inflecting in the third dimension with gate-all-around, the NAND stacks continuing to grow, DRAM going from 6F to 4F to eventually 3D. When we look at our addressable market, we see that going from the low 30% of WFE to the mid-high 30%. So that's important. It's just -- we're in a good neighborhood, that's how I like to describe it. Thinking through that then, where is that growth coming from? It's coming from every end market, right? You've got -- we see NAND from current node to future node growing 1.8x; DRAM 1.7x; and foundry and logic actually where we've done really well, 2x. We're obviously excited about that. Like I said, we sit in a good neighborhood. And so when we look at that, Tim talked about the new products that we announced, right, Akara, right, first new bottoms-up conductor etch platform in 20-odd years. We feel great about the strength of the product portfolio such that in that growing share of WFE, we believe we gain a 50% share of that incremental growth in WFE because of the strength in the product portfolio, Joe. Maybe I'll stop there at the top line. You might want to ask later about profitability and stuff. I've got other stuff I can talk about there.

Yes, I would love to go into that as well. But maybe sticking on the topic of WFE, if you could talk about this year, you guys are looking at mid-single digits. Which end markets are driving that? Kind of how do you bottom up, get to a mid-single-digit?

Yes. When we look through it, you're getting a little bit of an uptick in NAND, primarily conversion-related spending. It's not a huge amount, but coming off 2 years of almost no investment, Joe, that's good for us, right? We get a great share of spend in a conversion environment. That's what we see happening in NAND. DRAM, when you unpack it, outside of China, DRAM is growing nicely, right? You get the move to DDR5, walking down the process nodes, high-bandwidth memory, showing up where we do extremely well in the through-silicon via application. So we feel good about what's going on in DRAM. And then foundry and logic, leading-edge foundry and logic, continues to be strong with the move to gate-all-around and the incremental addressable market for us in that, that's what's driving a little bit of the growth from '24 to '25.

Great. So maybe we could double-click on some of the markets, starting with NAND, and it's so impressive how well you guys have done the last couple of years as NAND has really been tough, right, down 70% or so from peak. You talked about the need at the update for 400 -- $40 billion, sorry, over the next several years to upgrade. Can you talk about how much of that is your SAM and what your target market share is within that SAM?

Sure. What we laid out was our view, I mean, obviously, the last couple of years for NAND have been incredibly tough. I mean the economics of the industry have been pretty challenging coming off of the peak in 2022, where we saw almost $20 billion of NAND spend down to something in the kind of mid-single-digit spend at its low. But what we really laid out was the fact that 2/3 of the industry's bits now find themselves being manufactured at layers, less than 200 layers on the device. And that $40 billion in upgrade spending we see over the next several years is such that customers can begin to migrate that installed base to higher layer counts, higher-performing NAND bits, implementing things like QLC, implementing many of the different stacking technologies in that NAND device to make themselves more competitive and more suited to where the demand is in the marketplace. And so for us, what it means is that we will be selling upgrades to our existing installed base, that large installed base that's manufacturing today at the 128 and 1xx node. But also, we'll be able to provide new systems that help enable that layer count growth. We talked about a few of these at the Investor Day, things like PECVD carbon gapfill. It's a critical technology for enabling stacking so that you can stack tiers of layers on top of each other to build ever taller devices. There's a tool we sell that helps manage the stress on the wafer by depositing a film on the backside of the wafer that compensates for the front side stress. And then, of course, we also introduced a new tool that deposits moly, molybdenum metallization, which helps with the resistance of the metal lines which becomes increasingly critical as you go to higher layer counts. So in the upgrade market, it's a combination of both those installed base plus new tools. And we estimate that our SAM is over 2/3 of the WFE that gets spent on upgrades, and we capture a very large percentage of that SAM. So the upgrade environment is actually very favorable for Lam.

Yes. I mean it's impressive because your peers didn't really show the same uptick in NAND that you did. So exactly what you're saying is that spending on upgrading your capacity to maybe enterprise-capable NAND and things like that rather than adding supply. And then you've talked about the ability for you to get back to NAND kind of peak revenues without getting to peak WFE. I think it's some of the effects you just talked about, but can you give more color on that?

That's right. I think it's exactly the effect of that, if we're capturing over 2/3 of the WFE that's our SAM, and we're capturing a large portion of that in that upgrade market, clearly, we don't need the overall WFE spending to reach that new peak before Lam's capture rate allows us to reach new NAND revenue peaks. And so we haven't given an exact time frame for that, but that $40 billion upgrade market that's going to take place over the next several years definitely presents an opportunity for us to achieve that new peak revenue.

Okay. Great. And then maybe just to touch on a question we get a lot. Tokyo Electron has been pretty aggressive about cryo etch and their opportunity to gain some business in NAND through the use of that tool. Can you just talk to that generally?

Yes. So we have talked a lot about this in the last year. We spent a lot of time working on high aspect ratio etch many, many, many years. Obviously, we've been the leader in that space since the advent of 3D NAND. And today, we now have what we call the Cryo 3.0 process. It's a process that delivers what we believe are the industry's best results for creating very high aspect ratio, very vertical structures, again, with good repeatability and good productivity. And so we think we're extremely competitive there. We also have another very favorable offering for the customer, which is that technology can be upgraded to our existing installed base. And so again, we view our customers in this space, while the economics may start to improve in NAND, everybody is always looking for the lowest cost, easiest means of upgrading to that next technology node. And so the upgradability of our tools to now implement Cryo 3.0, I think, gives us an edge for any future investments made by our customers. And so we feel very good about that market.

Okay. Great. And then just another development in NAND, the use of hybrid bonding, something that SanDisk and Kioxia talked about in their recent IPO-type events, something that I think there's been some discussion of Samsung licensing, some of that capability. Can you just talk about the importance of that for Lam?

Sure. So I think if you think about hybrid bonding and NAND, it's really part of a much bigger trend, which is people have figured out that packaging type technologies, advanced packaging technologies are very suited to helping with scaling. And so we've seen that in other devices and now it's coming to NAND, where you're going to put the array underneath the cell. You're going to bond it. And in that case, it creates opportunities for Lam because that bonding process now introduces the need for new tools that help ensure that, that bond is done in a high-quality way. And just one example, we sell a tool which deposits a film around the bevel of the wafer. In that way, during that bonding process, we get a very high-performance bond. And so again, it will be important. We show that occurring at some customers at the 200-plus layers and others at nodes beyond that. And so we do think it is the future trend of the NAND road map.

Okay. Great. So maybe on the other markets, starting with DRAM, you talked about 3D DRAM. What do you see as the time frame of implementation? And what are the opportunities for Lam?

Yes. Big disruptions like 3D DRAM will take some time. While we haven't put an exact time frame on it, it's probably the, in our view, probably the early 2030s. But importantly, on the way there, you're going to get some important learning and some important opportunity for Lam and that is through the transition to 4F squared, which starts to introduce the idea of vertical scaling into DRAM in the first sense. And basically, the gates become more vertical, and you end up needing high aspect ratio etch. One of the new tools we just introduced, the Akara conductor edge tool, what its real specialty is this very high aspect ratio conductor etch, very suitable for the challenges that you're going to see in 4F squared. And those same capabilities will then translate nicely into 3D DRAM. So we do see this evolution taking a number of years. But over those next 7, 8 years, they'll be learning along the way, and DRAM will begin to scale much more vertically. And as Doug mentioned, in that case, our SAM expands from where we are today through 3D DRAM by 1.7x.

Okay. Great. And then maybe similarly for logic, if you could talk to gate-all-around, and the opportunities for Lam around that investment.

Yes. Foundry/logic, I mean, it's been a real bright spot. I mean we always start every discussion with NAND, but probably we would now start -- if we really want to talk about where we're making progress, it's foundry/logic. Two fronts: gate-all-around, we talked about the fact that just in this past year, we shipped over $1 billion worth of equipment to the gate-all-around nodes; and the second story in foundry/logic is advanced packaging, where we also said $1 billion of shipments in advanced packaging between foundry/logic and HBM, but foundry/logic really being one of the critical drivers there of especially the advanced technologies. So we see our move into foundry/logic really being quite successful as deposition and etch intensity continues to grow. Gate-all-around also creates new market opportunities: atomic layer deposition, selective etch. And these are areas where, again, Lam, looking to expand our SAM, began the development of these tools a number of years ago. And now we're starting to see those processes begin to ramp, and that's contributing to that $1 billion in shipments in gate-all-around, and we just expect it to grow. And if you look beyond gate-all-around to CFET, the devices become even more vertical and therefore, debt intensity increases even further.

Okay. Great. I wonder if you could talk to the cyclical environment in logic spending because it seems like to me, when I look at the spend, there's a lot of spending in gate-all-around. There's a lot of spending with which you guys see through Reliant and the more legacy nodes. And there's actually not that much spending in kind of 3-, 5-, 7-nanometer, which are the workhorse nodes for kind of everything that we spend all they talking about. So, so much of the investment is going to advanced technology, advanced packaging and legacy. Is there an opportunity to see a rebound in kind of the more mainstream technologies as well?

I guess Joe, I think about it in 2 categories. First, the leading edge stuff, where we talked about the SAM expansion. Investment there is very strong driven by AI and things like that, the stuff everybody in the room is excited about. So I think it's pretty well understood what's going on there. That was strong last year, we believe, continues to be strong this year. Then when you broaden it out and look at some of the more mature node spending, I think about it in 2 buckets, right? You've got a decent amount of spending occurring in the China region, which I think everybody understands, that shows up in our CSBG business, in the Reliant product line. That was quite strong last year. It continues to be reasonably strong this year. And then if you look outside of China, we're coming through a classic semi cycle in analog and microcontrollers and things like that. I think that's pretty well understood if you follow semis. Investment there was soft last year, not insignificant but softer. And I think it continues to be softish this year. But as we move through that analog cycle, that will come back around as well. And we do extremely well in some of that mature node investment, Joe, as you know.

Great. So I wonder if you could talk about China a little bit. Just starting with your kind of views on 2025, you seem to be conservatively approaching that market. Obviously, some incremental export restrictions that you have. Can you just talk to your expectations?

Yes, our view when you look at China this year is, as a percent of total revenue, it's going to be down this year. I think you're hearing that from all of the equipment companies. We're no exception. Part of the down percent is a little bit of a softening in the spending in China, but also a strengthening outside of China. So that will move that around a little bit. There were some incremental new regulations that showed up in December and we talked about the impact of that. That was, I don't know, roughly $700 million of revenue we had forecasted, Joe, that would have shown up for us in the second half. And so the fact that, that is no longer accessible contributes to that downtick, I guess, in the China percentage.

If you look at those China rules, is there a judgment call about what you could ship and what you can't? I remember 2 years ago, everybody had different treatments. And then that's right, you could ship stuff you didn't think you could 6 months later. Is it pretty clear cut what you're allowed to ship and not allowed to ship? Because I know there's also some customers that aren't on the entity list that you're also not shipping to out of various other restrictions?

Yes. Listen, the way I look at this, and Tim, feel free to add, there's end use and end-user restriction. End use is technology nodes. We do a deep technical assessment of the capability of our tools and make sure we understand that really well. So we know where we can and can't ship. And then there's end-user restrictions, which are certain companies ended up on the entity list that might not have been before, you just can't ship to them anymore. And so it goes into those 2 buckets.

Yes. I think, Joe, the only thing I'd add is, I mean, to your point, we do understand what we can and cannot do. And if there's any question, obviously, we simply don't. Yes, I think we feel very comfortable about understanding those restrictions.

I mean I asked the question because you guys have a slightly larger headwind than your 2 closest peers from a dollar basis and a percentage basis. Any sense for why that would be?

Joe, it's always hard for me to answer a question like that because I have no idea what somebody else saw or thinks or contemplated. When we give you that $700 million, it was a forecast from those customers that just no longer we can ship to. So we can talk about what's going on here at Lam. Hard for me to guess what might have been going on at another company.

And then what do you think about these export controls from a competitive dynamic standpoint? You have companies like YMTC, which have been on the entity list for years, have gotten to approaching 10% share of NAND by the end of this year. You have companies like AMEC and NAURA, which is not really direct competitors to Lam, but they're very excited about these rules, Chinese capital equipment companies that get a shot in the arm from all of this, and a very resourceful and resilient economy that wants semiconductor self-sufficiency. Does this all create headwinds for you 5, 10 years down the road that you're sort of creating this level of competition?

Yes. I guess maybe I'll start and Doug can add. But I think clearly, local equipment companies in China have gained traction where we have not been able to compete. Obviously, where we've had to exit, they've stepped in and been able to take some of that business. Where we are able to compete, we still do extremely well. It's -- there's a value to the maturity and the reliability and the productivity of our tools and the support that we provide as a global leader in these spaces. But I think that as I step back, we think the future for Lam Research -- and if we look at where deposition and etch intensity is growing and where we're investing in new products and a lot of the growth we projected at the Investor Day and we've talked about today, it's all coming in leading-edge customers, the far leading edge, whether it be foundry/logic and DRAM and NAND. And so we're really focusing our attention there and making sure that with those customers, those global leaders we're running as fast as we can. We're helping them run as fast as they can. I think that's the way you stay ahead of any competitor regardless of where they are in the world. And so while it's always disappointing not to be able to compete everywhere, we really don't see it as a real impediment to the long-term growth of the company as we've laid out last week or 2 weeks ago.

And it's hard to ask about the future when it comes to things like export controls because none of us knows. But I guess you guys do seem to have something of a voice in that conversation. I know you've had some strong hires in the government affairs area a few years ago. And do you feel like -- because obviously, semiconductor companies are going to have their desire about what can and can't be built in China. Do you feel like you have a voice in that conversation as well?

Yes. I think we have a voice. I mean, obviously, we always lobby and advocate for a level playing field for us. We would like to be able to compete globally. But we also understand that there are greater considerations at play at times. And as I said, we always adhere to whatever export controls are in place wherever we do business.

Great. So maybe CSBG, I actually have 2 questions here, both for Tim, but I know Doug loves to talk about it. Maybe give an overview why you're so excited about that.

It's my favorite part of the business.

I know, and you always get mad at me if I don't ask you. And then Tim, I want to follow up on some of the technology there.

No problem.

Yes. Listen, CSBG is a great part of the business model, right? On average, our tools, after they're sold, will generate more revenue over the life of the tool than when we sell them themselves. And we spend all of these meetings talking about WFE, that this is an important part of the business. It spares service upgrades and then the Reliant product line. And our tools literally run for decades. And Tim can talk about some of the first tools he worked on that we believe are still running in the fabs today. It is a wonderful part of the business model, very cash generative, very profitable. It's just an awesome part of what we do.

Yes. And it has stayed that way for a long time, for sure. Maybe you could talk a little bit about the Reliant portion of it. We get the question a lot, how big a portion of that is Reliant. And are you seeing a return to an actual refurbished tools business, which is where we began with that? Or is it still...

Yes. No, Reliant for the most part, we're selling older models, but new equipment. We're always looking opportunistically for something in the aftermarket that we can sell and refurbish. That used to be largely how we competed in Reliant, but not so much anymore, Joe. The biggest individual component in CSBG is spares, always has been. Reliant is #2. It was pretty strong last year, softening a little bit this year primarily because of the China stuff that we were talking about. Although when we look at CSBG this year, we think it ends up being flattish year-over-year because of the strength in the upgrade business.

Great. And then you also talked at the Analyst Day about the cobot aspect of that business. I wonder if you can give us an overview of what's happening there.

Yes. I think that there's a couple of things going on. One, with respect to semiconductors kind of re-globalizing, meaning that many regions that haven't done semiconductor manufacturing for a long time or maybe countries that are doing it for the first time, they're running into challenges with starting up these fabs and then finding the skilled personnel to maintain them. And so that's true whether it's an advanced location like United States or it's one of the newer countries entering the market. And so one of the ideas we have is to use equipment intelligence, which essentially is our ability to use data coming off the tool to make smart decisions about when to perform maintenance, how to match chambers to each other and then marry that up with cobots, so basically to do automated maintenance at a level of precision that's beyond what humans can do. And in our first implementation, we've implemented this in several memory fabs now. And what we're seeing is that not only is it eliminating the types of work that our engineers don't like to do, it's also doing that work much, much better, and we're seeing the tools come back at a higher percentage right first time after preventive maintenance. We're seeing less unscheduled downtime. And this has real benefits for the operating cost of a fab. And so the use of -- I think this idea of the use of robotics for maintenance of systems is just going to grow. We're expanding the number of applications. It was first done on our high aspect ratio dielectric etch systems because that was pretty intensive and frequent maintenance that had to be done, but now we're expanding it to more systems, more applications. And I think it's going to become a big driver of what we call advanced services growth into the future.

That is pretty cool. Is that a different revenue model for you? Or I know a lot of what you do in the services is sort of ensure results on the wafer thing like that, is that...

Essentially, it is what we would probably refer to as a results-based contract. We're basically taking over the maintenance using these cobots and using this equipment intelligence and the customer is getting a result, a performance result out of their tool in exchange. And so I think it's a win-win for us and the customer. And as I said, in some of these places, it resolves this issue with not be able to hire and find enough skilled talent in that particular region.

Great. Okay. And if we could follow up on the financials. If you could talk about idiosyncratic growth drivers to gross margin and operating margin?

Yes, let me take that one. Yes, we just gave an updated financial model for '28, and then we have this cute little model of what a $1 trillion industry might look like for Lam Research. For the first time, Joe, we put a gross margin number out there that started with a 5 handle. We were pretty excited to be able to do that. A lot of that, frankly, comes from an expansion of our Asia factory network, right? As business grows, we will expand increasingly in the Asia regions. Malaysia, actually today, is our largest factory. And so as business grows, we can more efficiently support that growth in business. And it's not just about labor content. It's also about localization of the supply chain. It's also about freight lanes that are shorter distances on inbound freight and things like that, that enable -- we're just moving stuff around a shorter distance. So you've actually already seen some level of benefit from that, Joe. And I actually feel really good, really proud of the company in terms of when the memory market turned on in early '23, we told you we were going to go do this and restructure things and we did. And it's showing up now as we're beginning to see some resumption of business growth. You've already seen a 100 basis point improvement in gross margin. So as we look at where we are and where we're going to go from a gross margin standpoint, it's largely about that, right, expanding top line and supporting that for manufacturing and doing a lot of the operational work in the Asia region, where the customers are, frankly, right? It's just being closer to customers. And so there's that. And then there's the digital transformation stuff that we began to talk about that.

Yes, I'd love to hear more about that.

Yes. So let me unpack that for you a little bit. It's interesting. Business has grown meaningfully since the last time we did something like this at the company, right? I don't know, we're probably 5x as big today and maybe even more than the last time we did something like this 20-plus years ago. And the last time we did something like this, we were a single product company. We hadn't yet brought Lam and Novellus together, a single factory location. We are now a global company with a global footprint, both manufacturing, customer. And so as we began to look at business expansion, we realized, "You know what, we need to get on and modernize things here. We need better systems to support not just where we are today, but where we see this growth into the future." And so this is going to be a multiyear program. We call it digital transformation. You'll hear us describe it just as DT, easier to use an acronym, where we're modernizing business process. We are modernizing the ERP. We're going to upgrade the ERP across the totality of the company. We are making ourselves AI-ready or AI capable in a lot of ways, right? You got to cleanse data and have things set up for the hooks that these compute algorithms need. And so we're doing all of that stuff. So we're going to go through a couple of years where this is a drag on profitability, Joe. I don't know, 3 or 4 years from now though, it will be accretive to operating profit. And from the trough to where it begins to generate future returns, it's an improvement of 150 basis points to operating margin. And so that's also a part of how we get from where we are today in the low-30% operating margin to that mid-30%. In addition to the gross margin stuff I talked about, DT is going to help get us there as well.

Joe, I guess one thing I'd add, I mean, there's a common thread running through a lot of what Doug just said, whether it was -- whether we're looking at supply chain and manufacturing, whether it's DT, a lot of this is about speed and efficiency and being able to service our customers better. One element we haven't talked about, but it has been a key part of our strategy the last several years is putting a lot of our development capability very close to our customers. This is an industry where our leading customers work with us on our most advanced tools. And the closer you can be to make sure that, that interaction occurs with that speed and efficiency, the faster we can turn out new tools to solve their problems, those tools tend to be much better fit for purpose, meaning they're more competitive. And so you've seen us do that. So whether it's the supply chain, whether it's our development centers we put in places like Korea and Taiwan, whether it's DT, whether it's the semi verse and our virtual simulation and digital twin activity, all of these things are designed to create that next iteration of process or equipment solution faster for the customers. And I think that, that's going to be highly valuable for our competitiveness as we look forward in the next few years.

Great. Let me see -- open it up to the audience, see if they have any questions.

Yes. One up here in the third row, I guess. Go ahead and shout it out. We'll repeat it.

Can you maybe comment on how much share you're going to get from 2 nanometers?

So the question was how much share are we going to get in 2 nanometers. I don't know, do you want to answer that? Listen, 2-nanometer is all about gate-all-around. And so there's an expansion of creating that nano sheet, right, selective etch, ALD. We feel really good about the strength of our positioning. SAM is expanding and our product portfolio has enabled some level of share gain. We haven't broken down what magnitude is share gain versus SAM expansion, but it's a little bit of both.

Yes. I think the -- when you think about any one of these particular technology nodes also pulls in, whether it's 2 nanometers or just beyond, it's gate-all-around, it's the backside power distribution, you end up with a lot more advanced packaging. All these are areas that are highly suited and highly targeted to deposition and etch technologies, and that's -- so it's both SAM expansion plus share gain. And so I wouldn't tell you quite so much to the specific numerical node as much as those technology inflections like gate-all-around, ultimately CFET, backside power distribution, advanced packaging, those are the types of drivers for us that are highly...

And how much does it matter if you have 1, 2 or 3 foundries supplying those nodes? Like does it change the opportunity?

Well, it's only an opportunity in timing. Eventually, when you look at it, all customers, I think, ultimately end up on the same technology road map. These things are being done -- take backside power. It may exist at different points on different customers' road map. But everybody wants better performance at lower power. I mean that's what the backside power distribution delivers. And so eventually, everybody gets there. It's the same as in NAND. There may be different cut-in points for, say, moly. But everybody wants lower-resistance metal, better performance, lower power. So everybody gets there within maybe one or 2 nodes of each other just by their own strategic choices.

Other questions from the audience?

We've got one over here.

On the higher aspect ratio etch, are you guys having suppliers for power delivery for the higher aspect ratio etch tools, so like the plasma power delivery systems?

Power delivery systems in our system. Well, we work with a lot of leading companies for different innovative power delivery systems to our tool. Relative to high aspect ratio conductor etch, we recently talked about a new innovation on the Akara system, the most recently introduced tool, which is what we call direct drive, which allows us to -- it's basically the industry's first RF matchless system for plasma etching. And what that allows us to do is switch the -- and control the plasma about 100x faster than previous approaches. Today, almost all applications for high aspect ratio include very, very fine pulsing sequences and that finer control with an RF matchless system allows us to do that. So the power delivery and power source for our plasma is a technical differentiator for the company.

We have less than 1 minute. If there's any more questions? One more over here.

Yes. One last one here.

The $40 billion NAND upgrade, [ how do you ] size up the low end of the scenario versus like -- if it's being spent in 3 years, obviously, on a per annum basis, the WFE number will be higher. If it's being spent on 5-year -- over the next 5 years, it will be a smaller WFE number. So I'm just wondering how to think about the risk guardrail on that. And obviously, if the NAND end market does better, that's helpful. But I'm just wondering how you guys think about -- in the context of Samsung spending the other guys are reluctant to spend at this point.

Yes. So the question was just about the timing for that $40 billion. We didn't put a time frame on it. We said next several years, and that's frankly because our customers and the end market will determine exactly the rate and pace at which those upgrades occur. But what we are certain of is that customers will spend that amount by our estimation to move from the 100 -- the 1xx layers up to current state-of-the-art. So whether it's 3 years, 4 years, 5 years, it's less important for us. We're positioned with the right products. We've got the installed base. Our SAM is over 2/3 of all that WFE spending, and we're going to capture a large portion of it. So it's a great opportunity for Lam.

Awesome. Well, with that, we'll wrap it up there. Tim, Doug, thank you so much.

Great. Thank you.

Thanks for having us, Joe.