Right. Morning. Thank you all for coming today. 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 Lamb Research, Mr. Tim Archer. Before I start, I want to mention, if you have questions you'd like to ask during the presentation, you should have a link or a QR code, I think, to the pigeonhole form where you can submit those. I will get them up here, and we'll have time for Q&A at the end. So semi-cap, So after enjoying, you know, it's had a bit of a renaissance already over the last several years, but the capital equipment industry has really shifted into overdrive recently as AI has gone mainstream, with the demand profiles that we're seeing calling for at the end of the day just simply more, more chips, more wafers, more tools. And in that world, the contributions from companies like LAM are becoming more important than ever, and to tell us all about it, it gives me great pleasure to welcome Tim to our session today. So thank you so much. I have to say, I've been super excited to sit down. This is actually my first session of the conference, and I'm glad it's starting with you guys. I mean, let's just jump to it. I mean, let's talk about AI. I mean, clearly, as I said, it's created a renaissance in WFE. You know, it wasn't that long ago that $100 billion in WFE was just sort of in this aspiration that was out there. I mean, we've blown through that now, right? I mean, we'll probably do. I can't remember where your number is, 145 or something?
140 with a bias to the upside.
Yeah, bias to the upside. And as strong as that is, I mean, it's a constrained number. Like, we were short on clean room space. I mean, maybe just at a high level to start us off, maybe just talk about what AI has done for wafer and equipment demand. I mean, maybe buy-in markets like leading-edge logic, DRAM, flash, packaging, even lagging edge. What has it done for us?
Yeah, it's, I mean, first of all, I agree with you on this point of this is a renaissance and it's driven by AI demand that has really touched every segment of the equipment market, the devices, in-device markets.
I mean, in semis, by the way, you could have owned anything. It's gotten so big, it's dragging everything along with it, it feels like.
I hope some have been better than others. And I think that we'll talk a little bit more about the markets we're in. But, you know, as you said, AI, I think what's been most amazing about it is that it has driven performance across every device type. I mean, there used to be, there would always be a discussion of what's good enough. But, I mean, in AI right now, more is more. And so you want to have more leading-edge compute power at the Foundry Logic side. You need more storage. You need more memory. higher bandwidth, lower latency, and that's just created changes very rapidly to the architectures of the devices, how they're packaged, the materials that are used to manufacture those devices and so forth. An etch and deposition focused equipment company, I mean it's a very
exciting time. And you know like you know this time is different is always sort of a dangerous kind of statement, but I could make the argument that maybe it could be something like How would you respond to that? Is this just a function of just demand is overwhelming? Is that what's different? Is there something different about the types of things that that demand is driving as it relates to you guys? How would you respond to that?
Yeah, I think, you know, as I was talking about, I think what is different is this almost insatiable desire and need for more performance.
I mean, we've always wanted more performance in the space, haven't we? Yeah, but I think that in this case, it is really, it makes the end product that much more capable in a way that I think is a little bit unique.
It's always been there, but it is also quite unique that it's performance across all types of devices and including packaging. And I think that the other thing that's a little bit unique and maybe different is that we have visibility to these waves of demand. And so we've started in training, and that drove certain performance requirements around certain types of devices and really accelerated this push towards more advanced foundry logic to get the compute, and HBM to get that capability. But now, as we move into inference, and then we move into agentic AI, and ultimately into physical AI, we see different parts of the market and different requirements. We clearly now see much stronger demand for NAND, for instance, than people had anticipated. Again, as you start to make this transition from training to inference into a Gentic. We've seen the same thing with CPUs and therefore different parts of the market, different process requirements for some of those devices. And so I think that's perhaps what's different is the visibility into this roadmap and march of performance requirements that really allows us to marry up our roadmaps for new equipment, new materials, and really engage customers to be ready as those waves come in.
How much visibility are you actually getting now? And what does that look like versus what would have been typical a few years ago?
You know, it's interesting. We look at visibility in two different ways. I mean, what's most important for us is visibility of the technology roadmap Because the reality is, we are already engaged and qualifying for device architectures and materials that won't ramp into manufacturing until probably the early 2030s. And so having that visibility, that customer intimacy where you're really exchanging what's possible around architectures and materials, that visibility has always been strong. And I would say today, in many ways, is even stronger because of the criticality of our tools to their ability to deliver products to their customers. Capacity, visibility, and those needs, you know, I think, you know, one, we could never ask for enough and we'd never get as much as we want.
But I would say... Will you say that, do you mean your own capacity?
No, no, no, customer demand. I mean, meaning that, you know, at this point, if you look at our WFE forecast, you know, every call, it's sort of been, we've used this term, we're in an accelerating demand environment. What that fundamentally means is every time we come to, you know, that next checkpoint, there's more than we thought there was. And so I would say that the visibility is not as good as we would desire, but it's much stronger. Therefore, it's exactly the good kind of lack of visibility. And it's quite far into the future at this point.
Got it, and maybe to go back to that point, I mean, clean rooms right now, what do you see, is that the primary constraint on things right now?
Yeah, we've been clearly vocal about the fact that clean rooms are constraining the ability to ship more. Obviously, there's a lot of activity in that space to try to bring new clean rooms online, but frankly, it just takes time. It's a big process to build one of these very advanced semiconductor fabrication facilities. And so I think that we've, that's why I'd say the visibility at this point, once a fab starts construction, you kind of know where your demand is coming from maybe two years out. And so I'd say we have quite nice visibility from that.
How many fab projects are you guys tracking right now? Have you ever given that number?
I don't think we've given that number, but it's a lot. Hundreds, is it? I don't know if there would be, in terms of new greenfields, I don't think it's quite that number, but clearly every fab is also going through continual technology upgrades and refreshes to make sure that they have the capacity needed to meet the performance demands that really is where the market is today.
Let's maybe take a look at the different end markets. Let's start with NAND. I know you guys used to be thought of as the NAND player, and you really still are, but not anymore. It used to be almost fully dependent on it, and it's clearly not the case. and I still remember, it seems quaint now, but it was a few years ago during the NAND trough, and it was before the stock split, but you guided something like a $5 EPS for the following quarter for a nominally NAND-focused semi-cap company with no NAND in the guide, right? It was fine, right? And so the diversification story has done really well. At the same time, though, you know, we are, I don't know that we've seen a ton, especially given all the demand that there clearly is for NAND and for storage, we haven't seen a ton of greenfield builds yet. It's been primarily upgrades. My view has been the NAND guys decided they'd like to earn money for a change, which is a little new, but I guess maybe talk a little bit about the differences between the NAND upgrades and expectations potentially for Greenfield, because I feel like they've got to start adding Greenfield at some point. I don't know how far the upgrades can carry us.
Yeah, actually, I'd love to answer that question, and I will in just a minute. But first, I have to sort of at least clarify a little bit of this. LAM is the memory company versus what we've become.
Yeah, because you really think it's really transformed, I think.
And we still wear that memory leadership proudly, but the reality is we've become so much more. We went back five years ago. LAM, our business is about 60% memory. Last year, it was 60% Foundry Logic. And that was a deliberate strategy of ours to not give up this tremendous position we built as the enabler of a lot of these memory roadmaps, but it was a strategy to diversify the company in a way that we could participate in a much bigger way in the device architecture inflections and the advanced packaging advances that were coming for Foundry Logic. And I mentioned the visibility that's important to us is we see where these roadmaps are going five, six, seven, eight years in advance, And so really, about seven, eight years ago, we said, there's a huge wave coming for Foundry Logic. Advanced packaging is going to become a pretty big thing. And so we pivoted the company in terms of where we put a lot of our R&D spending, the new products we were developing. And now, it takes several years to get those products completed. It takes several years to get them qualified in customers. And now you really see that rolling through. And I think you've seen LAM gaining, both expanding our SAM through this new portfolio, but also gaining share with those new products. And, you know, I think every earnings call we've talked about, you know, low-K ALD. We can get into all of this a little bit next.
I always have to be careful in this forum because I can start to go down rabbit holes.
So we'll come back to your NAND question. But, you know, the company has reshaped, but we have not given up that NAND leadership. And I think that, you know, now you're starting to see that roll through, not through Greenfield yet, but last year at our Investor Day, February 2025, we had talked about this upgrade cycle. And, you know, in the NAND business for years, the upgrades had been sort of the natural way of making use of that installed base and just scaling the number of layers.
So this is simplistically using the install base to put out wafers with more bits on them?
With more bits on them, yes. And so what we had said last year was that it would take several years, but the customers would upgrade the tools they had in their installed fleet from kind of the 100-plus layer counts to 200-plus layer count devices and therefore get more bits, and also those bits might be higher performance, more capable bits. We've seen a dramatic pull-in in that upgrade cycle. We said several years, and now we, on our most recent call, said that entire $40 billion upgrade spending would likely be complete by the end of 2027.
Now, that's a... Have you said how far along we are on the $40 billion, by the way?
We haven't really, but we're, again, back to this accelerating demand environment. I mean, most likely, you know, it's from this point forward more back-end weighted. But it's an indication of two things. One, the role of NAND within the AI hierarchy is becoming more prominent. There's stronger demand, whether that's for enterprise SSDs or use of NAND in KVCache or any of these new data storage projects that you've seen from companies. But it also is the indication of your cleanroom shortage, which is there just weren't enough clean rooms available to satisfy all the demand of HBM and this rising demand of NAND. And I think that NAND has found itself first going through an upgrade cycle, but then eventually, as you've seen, as there's been shortages, effectively needs to start building Greenfield. And you've just started to see some of those announcements, but it takes a while to build those fabs, so it's still pretty far out there on the horizon. and I think that it's probably more of a 28 and beyond event, which I think from our perspective is fantastic because, again, we can focus right now on a lot of these upgrades, get the installed base up to that technology, and then as a next wave of growth for the company, really see these greenfields come in and the additional shipments that come from that demand.
Maybe two follow-ons to that. One is, do you care upgrades versus Greenfield? Like, what does your exposure look like between the two? Is one better than the other? And then number two, just to clarify, the upgrades is really an installed base kind of business. You guys control the vast majority of the installed base that's out there in NAND, correct?
Yes, we do. We are clearly the leader in NAND. And we're, more importantly, the clear leader in the critical technologies that are required to enable the upgrade. Meaning, if you think about the way NAND flash is manufactured, with that big memory stack and then the etching and deposition done inside of that to create the devices, LAM has a significantly high market share of all of those steps. And so the customer wants to build a taller device. Typically, the majority of the spending is actually on LAM equipment to build taller stacks and create those devices. So upgrades are great for us. We have a very, very high percentage of the SAM, of the WFE that's being spent on the upgrade. But actually Greenfield's great too because it grows our installed base and it makes that next upgrade cycle even that much bigger. And so I think that what I also want people to understand is when we put out the number that said $40 billion to upgrade from the 100X level to the 200X level. That's not a one and done. Because then a couple years down the road, 200X needs to go to 300 or 400. And then that needs to go even bigger. And so you see our customers have been, you know, there have been stories out and talk out about building, you know, 6, 7, 800, 900 layer NAND devices just recently. And so, you know, that's a roadmap that fundamentally we look to to help enable in the years to come and so both both are in the upgrade we capture a higher percentage of the customer spend but in the greenfield we build an even bigger install base for an even brighter future and so it's a both are both are good got it let's talk about some of the
other segments and you mentioned advanced packaging yes that's clearly been on a lot of radar screens as Moore's laws slowing or stopping you innovation is not stopping, and what it means is we have to do other things, and packaging is clearly one of those other things. So what does LAM actually do in advanced packaging? Where are you actually, like, where's your primary contribution? And how big is this business for you now?
Well, you know, maybe I'll reach all the way back into, I think, the 1990s when the Novellas portion of our company was known as the king of copper. Today, all these years later, we're still the king of copper. and so if you think about advanced packaging a lot of the copper plating everything that's done to to create the the metal interconnect is is an area where we're incredibly strong etching of course things you whatever the device if it's HBM or it's advanced packaging TSV etching silicon etch is a big
through silicon vias by the way through silicon holes that go all the way
through the wafer, and so that etch. But also, you know, advanced packaging has become quite a complicated architecture, and so there's a number of dielectric gap fills, inter-die gap fill, we call it as well, to go in between chips, just a number of steps. And so, you know, if you look at the advanced packaging and you think about where the money gets spent, a very large percentage of the dollars are directed towards etch and deposition applications, Essentially, it's a building and interconnect type of architecture, and so it's really great for our portfolio.
How big is that for you now, and how's it growing?
Yeah, we said that this year would grow over 50% from our prior year.
Did you give a dollar number?
You know, we've sized it in the range of around $2 billion. I mean, it's kind of like maybe that's probably about as close as we're going to get.
But clearly growing and certainly, I think, gaining in importance as the complexities of these chips continue.
Yeah, no, it is a fast-growing area. Again, it enables, again, as you said, for the end customer to create something that's very difficult to create on the chip itself. And so, to your point of when shrinks aren't the best way to get to an answer, the industry's finding ways to create an alternative path to now get out and do that.
Got it, got it. And so, you were talking about memory before, but you were known as the memory player, but it was still mostly manned. Like, what do you guys do in DRAM? Clearly, there's some of the packaging piece. I mean, for HBM, you're going to need TSVs and other things. If I'm looking at the wafer volume inside one of these big GPU racks, by the way, it seems to be dominated to me by HBM, by DRAM. It's like, where is LAM's exposure in the DRAM space?
Yeah, within DRAM, as DRAM dimensions continue to shrink, the criticality of our etch platforms has become even greater. In recent earnings calls, last few quarters, we've talked about wins in conductor etch. Things that are tied very closely to the formation of those very small features that are now on these DRAM devices. We also have had good success with our Dry Resist product.
I want to talk about that a little bit.
Yep, we can talk about that. But Dry Resist is, again, DRAM is utilizing more layers of EUV. And so, again, LAM's strengths are around patterning and the feature formation. And so that's become a lot more critical as our customers continue to try to push DRAM to its performance limits. On the deposition side, again, you mentioned in HBM, through Silicon Via, copper plating, those are markets and applications where we're incredibly strong. And so our share of DRAM has been growing as we've continued to broaden our portfolio.
Got it. And then I guess finally, I mean, to round it out to Foundry Logic, I guess, both Leading Edge as well as Trailing Node, what are you doing differently there now that maybe, if you look back five years or ten years, these used to be relatively small parts of your business, they're quite sizable now. There's been clear share gains, adding customers, it's like, what have you been doing there and what are you seeing there?
Well, Stacey, maybe if I can add a little humor here, I mean, the 3D NAND transition was huge for Lamb. I wish they had called, like, gate all around 3D logic, and maybe Lamb's participation would have been a little bit... It's not quite the same. It's not quite the same, but at that transistor level, if you kind of think about what a gate all around looks like, it really is starting to get into that realm of 3D processing. You now have these, you know, nanosheets, and you have to... And these layers that you've got to, like, deposit and then fill and then selectively etch the material out. And so many of the things that you would do in a NAND device, clearly the processes are different, but the concept is the same. And so we've been... You guys need better marketing. We need better marketing. As we've gone into... As FoundryLogic has transitioned into more three-dimensional device architectures, the demand for LAM products like our Selective Edge have grown pretty dramatically. Atomic layer depositions have grown. Of course, you're also talking about the shrink to 2 nanometer and below, which means conductor etch and the ability to etch the pattern that's been printed by this very precise EUV process. You've got to be able to translate that into the underlying device. So I would say that the demands of Foundry Logic today are much better suited to the portfolio of products that we've been working on. And as I said, that was part of that shift. A lot of these products have been introduced just in the last several years, like low-case spacers using ALD. These are products that we started to develop as we shifted from this, not just a memory focus, but really wanting to own the logic transitions as well. and I think we've done quite nicely there.
Can you give us any color on some of the things you're looking at now that would be more in that 2030 plus kind of time frame? Or is that not something you can talk about?
Probably on a product perspective, I'd rather not do that.
But general concepts?
Well, I think general concepts. Anywhere where you see a push for, one, new performance enhancing materials. Obviously, if you think about what we have going on right now, We're leading the push towards molybdenum, like within NAND and ultimately within Foundry Logic. And, you know, so when you kind of see that, I'll leave it at that. It's like we've built an incredible base of deposition and etch technologies. And I think that now it's about how to continue to expand the applications of those into forming these new devices. But there's a lot of new architectures coming, too, out there in that 2030s roadmap.
So as we're going to four sheets and CFETs.
You know, CFETs, and then, you know, across every device, there's new things coming. And so what I'm happy about is we today, and we say this, and, you know, you can kind of see it if you look at the full lineup of products. I mean, we have the broadest, most competitive product portfolio in the company's history. And so we feel like we are well-positioned to, like, address these new inflections that are coming.
on it. Let's talk about some of the more internal aspects of the company now. Let's start with maybe margins. It's funny. In this session, I always ask you why do your gross margins have to start with a 4? I don't have to ask it anymore now because they don't start with a 4 anymore. Now they start with a 5. Maybe talk a little bit about what that process looked like to get there. Some of it's going to be mixed. I don't know how much of it is pricing. Some of it is new products and new value add. But how did we get there, and where can things go? And I know you gave, you had the analyst day, but a little over a year ago, and you gave margin targets, I think it was for 28 that you have now hit. I think you've talked about we may be getting new targets later on in the year. I don't know if you're going to have another analyst day or you're just going to give them to us, but. Yeah, probably not a new analyst day at this point. I'm assuming you're not ready to give them to us today. I won't. I will not give them today. But I mean, in general, it's just, what is the, and there's been other things, right? But, I mean, I think you guys have been out in front of adding supply. You certainly built up Malaysia. And I remember Malaysia was a margin headwind for years while that was happening. And now that's paying off benefits. And it looks like you've actually added to that now as well. Maybe talk just a little bit about some of the internal things that the company's done around the execution margin, supply. Anything else that comes to mind? Because I think it's another reason, beyond just, like, the market lifting all boats, that LAM has clearly outperformed over the last, like, year. I can't remember where the stock is up year over year, 200%, 250, like whatever it is.
I rarely look at it.
Sure.
No, it is, yeah, no, a lot of it has been self-help. I mean, we, but it was with this eye that we were going to become a much bigger company. We've had that confidence. And so many of the items that you talked about, first of all, I do like living in the 50s neighborhood better than the 40s neighborhood. So that's good. But some of it did come from our large manufacturing expansion. Obviously, as we went through COVID, we saw that and the big boom there and the difficulty to supply to the demand. I think we all sort of internalized we don't want to let this happen again. And so we've made significant investments both in our manufacturing capabilities and our supply chain over the last several years. And I think that's starting to really pay off. As you said, for a few years it was a bit of a headwind as we were filling that up. We've now had no problem filling it. And, in fact, later this year we'll be opening our second facility in Malaysia that will be approximately equal size to the first one.
Will it be equipped, or is it just like floor space?
No, it will be in use in the second half of this year. So the good news for us is it does not take quite as long to build an equipment manufacturing facility as to build a full-on clean room, and so we are able to respond more quickly.
Your CapEx is what, 4% of revenue?
Our CapEx runs 4% to 5% of revenue all in labs, manufacturing, everything. So we're pretty CapEx-lite from that perspective. So it's a combination of becoming significantly more efficient in our own operations, plus a lot of the applications I just talked about where we've moved into these spaces where they're really in technology-enabling capabilities tied to the customer's latest devices, You know, Gate All Around or HBM, High Layer Count NAND. And we've built products where we've, you know, we're delivering significant value. And so I think that it's a combination of all those things has driven the margins up. I would also point out that, you know, from a mix perspective, in general, some of the that would have been known in the past is higher mix portions of higher margin portions of our business have become a little bit lower part of the mix. And we'll probably talk about China a little bit. But, you know, so what it actually says is the core part of our business, non-China, you can imagine the margins from our own work have done even slightly better.
Got it. And I guess are tariffs a thing now? I don't know if you had, like, cost increases or anything. I assume you're offsetting that.
All contemplated in the numbers we have reported and put out. And so I think you can see that even if there are, it's part of the margin improvement story.
Yeah, so maybe, I mean, that is a good segue into China. And you're right. I know for a while you were getting a boost when China was a bigger piece of the business. And China's fallen off a bit, which I don't think investors are terribly sad about, frankly, as long as the rest of it's growing. And yet the margins still look really good. But what are you seeing in China? Like, I mean, clearly we've had general decline in, like, lagging edge logic. I don't think it's all China, by the way, but, like, a lot of it is China. I get a lot of questions clearly about Chinese competition and local players, which my general view, by the way, has been that they are real companies. Like, they're not, you know, they're very capable of what they do. They can't do everything. And they will probably take more share than they would ordinarily deserve to take because of some of the regulatory issues. They have no choice. But in general, like, what are you seeing there? It doesn't seem like, it's been an investment controversy and a worry, but it doesn't seem like it's really slowed anything down at all for you guys or, frankly, for the industry as a whole.
Yeah, I think that two things. One, maybe just to address the regulatory issue, we comply, obviously, with all regulatory requirements. And so we don't do any business in places where we're not allowed to.
So you're not shipping like YMTC or anything?
So therefore, local equipment suppliers have filled in for where we can't compete. But places where we can compete in China, which tends to be, as you say, the lagging edge nodes, there's still tremendous value that's delivered from LAM tools, even though those tools were likely engineered and developed by LAM more than 10 years ago. And so I think that that speaks to, one, the quality and capability of tools LAM builds, plus the local support capabilities that our team provides. But the business itself, as you say, we see China in general as kind of flattish to maybe roughly up a little bit this year. So compared to the last few years of growth, I mean, it's clearly moderated.
Yeah, I mean, it was, what, 40% of your revenue or something at the peak?
Yes, and now come down quite substantially. And so I think that that's, as you said, look, that's the kind of balance we want to see in the business. And so it's an important region for us, and we satisfy demand from the customers we can serve with tools that we can sell. And I think that that's – but it's no longer like quite the headline story. I think every earnings call we used to get questions about China's sustainability, and I think now it appears to be in this relatively sustainable run rate right now. Our story now has transitioned much more towards our growing SAM, growing deposition edge intensity at leading edge driven by AI, and I think I like telling that story a lot more at this point. Yeah, it's certainly more fun to tell.
Do you have any thoughts on some of the Huawei announcements that we've seen over the last couple of days? Logic folding, and I don't know if you've followed it.
I've seen it, but no.
I haven't internalized enough to comment at this point. Maybe just one more question as long as we're in China and lagging edge. China's a big piece of your lagging edge business, but it's not all of it, I guess. How much is in China? And maybe just a few comments on that business. This is the Reliant business, right? Mostly the 200 millimeter business. Although I guess all of your lagging edge probably is in all 200 millimeter, I'm assuming.
That's right. There's quite a lot of mature node technology that's at 300. In fact, quite a lot. And within our company, actually, some of it to a mature node, some of it will come out of Reliant. Some of it will actually still come out of our primary business units. So it's, therefore, a little bit harder for you to track down. But, you know, we do see, obviously, the reality is complex electronic systems today, they require, you know, leading-edge chips, and they require a whole lot of...
We're starting to see growth in the lagging edge now, too. We're in an industrial cyclical recovery and clearly the AI like power semis have all been on a tear. I mean, you're not seeing any, we're not seeing any signs of real recovery in trailing node yet.
That's right. We are, we're seeing some growth in that area. But I think more importantly, what we are also focused on is, again, this idea of engaging on where those markets are going. So it's a little bit less about, we serve capacity that comes and demand that comes for the things we already have. but we see an emerging opportunity in what we consider to be the specialty technology space. You know, again, there's a lot of new applications where there are new materials required, new deposition techniques, and while none of them are quite the market size that you see from AI, they're very good businesses for us to pursue, given that quite often we either have the equipment available and it's really about application development, or it's equipment that can easily be derived from the fundamental technologies that are within our company. And so we have a group that focuses on those applications. These would be things like photonics, maybe some materials for RF applications, special materials for power devices. And so those are all, in many cases, they're materials, they're deposition or etching applications. And so very well suited to what LAM can do. And I would say that we intend to pursue those as well as we, again, look to broaden our portfolio across these markets.
Got it. Now, those Reliant tools are in the services business. This is a good segue to services. I still got to ask, any plans to move it out of there? I think AMAT moved their 200 millimeter back into equipment. It seems like it makes more sense.
Yeah, we haven't.
I would add.
We understand the comment and don't have any plans at this point.
But let's talk about services. So again, this is an installed-based business. Even in a down year for equipment, it tends to grow because the installed base grows even in a down year. I think your services revenue per tool has been going, I can't remember the CAGR, I have the number somewhere, but it's been growing at a pretty good clip. I think services you've talked about is sort of like a sustained double digit. I think the only time it didn't happen is when you had some of the China issues due to the sanctions a few years ago. But just talk in general about the services business. Where's the value out there? What are the different pieces in there? Again, I went over a little bit of the growth, but how do you see the growth structure going forward and what does that add for you guys?
Sure, well it's a really important part of our business. In fact, the services, or CSBG as we refer to it, is about a third of our business at this point. Which, because a lot of that is based on the install base and some of the larger components in there are things like spares and upgrades and, you know, really installed-
But it's all that NAND upgrade revenue is actually in services. It's not in-
Well, I'm going to make your life a lot. Not all of it's in there. But, you know, it's very installed-based focused. And so our installed base right now is about 100,000 chambers and growing. And again-
In operation.
They're all running.
What's the lifetime typically of these tools, by the way?
A couple of decades, or more than a couple of decades. In fact, a few years ago, I had the team go look to find. My first job was actually installing tools at the company. I was like a process engineer inside the customer fabs. And I said, go find that very first tool I installed. And they actually tracked it down. That was about 25 years after I'd done that job, still running.
We had, by the way, my lab at MIT 20 years ago,
we had an old lamb rainbow TCP. OK, well, there you go. I didn't install that one. But the reality is, like, the installed base, you know, it's the gift that keeps on giving, right? So it has multiple revenue streams. Of course, we service them, maintain those tools in many cases, spare parts, a lot of proprietary parts on our tools. So, you know, as the installed base gets bigger, especially when you're in these high WFE years like now, installed base is rapidly expanding. that's a future revenue stream for the company that continues to grow but the more exciting thing about services is what's happening you know as a result of all of two things one the rapid expansion of fabs everywhere in the world uh often in places where resources trained resources are not readily available and you combine that with the fact that the ai is demanding the most the highest performance, most precise device manufacturing we've ever seen, which means that the tools condition and the maintenance on those tools is kind of reaching the engineering or technician limit in terms of precision and repeatability. And so, you know, for several years now, Lamb has been working on what we call equipment intelligence, but think about it as building AI models around the data that's coming off of your tools. And we've also been working on cobots. Cobots, yeah. And so we call it a new product we have called Dextro. And so combining...
Does it help like physically servicing the tools?
Physically servicing the tools.
Taking off flanges or I don't know.
And reinstalling and cleaning the chambers and all the things that an engineer actually would hate to do anyway, the cobot can now do. And what we've seen, these are now running in production, maintaining tools in several of our customers' fabs. And what you see is you see a dramatic improvement in first-time right after maintenance. You see an improvement in maintenance-sensitive measures like edge uniformity around the wafer. In an etch tool, you install a plasma tuning ring. And we asked the engineer, the focus ring, and you asked the engineer to basically align that thing to a 50 micron tolerance. It's just difficult. Robot has no problem. because the COBOT can have a vision system. It can basically take measurements, install this.
Is the robot mobile, or is it parked next to a single tool? No, it would be mobile.
Yeah, from a cost perspective, remember, you don't need to do maintenance that often on our tools, and so one COBOT can service, depending on which type of tool, a large number of modules. And so that service opportunity for us, it adds significant value. It eliminates this resource constraint that has become a challenge for ramping new fabs and frankly just from a technical perspective the precision repeatability of Cobot driven maintenance is Dramatically better and shows up in on way for performance. So how does that work?
Do you sell the robot or is it part of you guys do subscriptions or part of a service agreement? Okay subscription
We don't you often use that term but service agreement. Yeah
Have you ever talked about how much of your like services revenue? until they were actually under agreements versus one-off stuff?
We haven't, really. I don't believe we've given that number. But it's growing. I mean, a cobot becomes a very sticky type of service simply because now you're relying on that piece of equipment to effectively maintain your other equipment. We combine that then with what we call equipment intelligence. And this is another big focus of ours, which is, you know, we've always collected a lot of data from the tools, but now as these processes are becoming so much more complex, we're finding that, and also AI engines are becoming so much more capable of handling disparate types of data input that we've started to focus a lot more on sensor technology. You know, what new types of sensors, what, you know, it used to be when you measure powers, pressures, flows, all kinds of these things. but now you can start to do things like video analysis you can and you start feeding those into the engine and effectively look at the wafer while attaching and yeah look at the wafer look at the plasma condition these sorts of types of data streams and what it's really helping us do is is start to really match tool performance across very large fleets of land equipment running in different fabs that might be located one in Asia one in the US and and you want to run the same process for the same customer. And our customer would want to do that. And the ability to match those fleets of tools really is becoming dependent on this equipment intelligence capability to do that. And it has a meaningful impact on tool-to-tool matching and performance. And so again, that is also a service that we would provide to customers.
Presumably, you were able to charge for this. Yes, presumably, yes, we are. I mentioned I wanted to give you a chance to talk about dry resist. What is dry resist? Talk to us about it. So, I just think it's phenomenal.
Yeah, no, no, dry resist, I mean, it's the way LAM is disrupting some of these technologies that, like, didn't appear to be disruptable.
Yeah, I mean, there was this big controversy back in the days that EUV was going to be bad for you. Doesn't seem to be the case. Turned out not to be the case. Seems to certainly push back on that.
Yeah, well, a little bit. I mean, one, it turned out that EUV wasn't bad for us because the smaller features you printed, the more critical our tools became. And so that argument of multiple patterning versus single print just made our tools actually much more critical for the capability to etch those single print features. And now you have multiple patterning anyway. So in any case, that part of our business has been fine. But we looked at dry resist and this idea of, like, how do you continue to shrink and push kind of the limits of lithography? And one of the limitations we saw was in the resist itself. And not only in the material, but really in the deposition technique of the resist. And so for, I don't know, five decades or more, I mean, basically you've spun the resist on the wafer. And the resist is delivered as a wet chemical. And, you know, in most cases, wet processes over the years have gone to dry for controllability and repeatability, defect control, all sorts of reasons. And so I guess about six, seven years ago, LEM introduced dry resist, probably a little ahead of its time. But what dry resist does is it is a material that replaces the wet resist for the most critical lithography applications on certain devices. And so it took a while for us to gain customer confidence to disrupt something that they were so comfortable using. But we introduced a suite of tools we call Ether. These Ether solutions consist of an underlayer that helps with photon absorption from EUV. the resist itself, and then a dry develop process. So it's a suite of tools.
So it turns the developing almost into an etch process rather than like a, they currently use a liquid, it's almost like developing a photograph, right?
Correct, it's like a deposition and etch type processing environments. And so therefore, much more controllable, and looks like many of your other processes. And so it is now in ramping in production at two memory makers.
So it's process of record then at? Two memory makers.
So memory first. these solutions you know interestingly um i think it's just the pace at which the pace at which some of these uh you've got to be very very far ahead in the in the foundry logic world to to get that cut in and and so i think that as as we continue to push forward to um the below two nanometer regime i mean as etching can as patterning continues to become more critical uh we'll find the we'll find the adoption there as well but memory is one that moves a little bit can move a little bit faster, and the volume of wafers is quite high, and EUV adoption in memory is also accelerating. And so I think that it's on its way now.
You guys gave some revenue targets a while back. I don't think we've ever upgraded.
Yeah, we haven't really changed them at this point. We said over a five-year period, about $1.5 billion of revenue back in Wade, and so we're now into that five-year period.
So we've got about five minutes left. Should we go to the lightning round? Let's see what we've got here. Okay. Where are you seeing or do you see the biggest bottlenecks within the AI value chain right now? I would actually rephrase that. Is semi-cap itself a bottleneck? Is it going to get worse?
You know, we've worked really hard. We talked about our operational investments, manufacturing supply chain to not become the constraint. So, you know, I think that we're not, but I think you see where the constraints are. As we said, clean rooms have been one we've talked about quite a lot. That problem can get resolved. You know, constraint that we're addressing, we just talked about the equipment intelligence and cobots, is how long it takes to then ramp and start up and get to mature yield. And that will be a little bit of a time constraint. I mean, it's solvable, but we're trying to address that as well. But I think right now it feels like there's a bit of shortages everywhere, but most of them are being addressed.
Where do you think WFE growth will be next year if clean room availability was not an issue?
Oh, I'm not going to give you an unconstrained number, but we've said 2027, even with the constraints in place,
is going to be a year of compelling WFE growth. It's safe to say there would be more if there was no constraint.
I think it's safe to say there would be more if there were more clean rooms.
I won't hold you to a number, though. Let's see here. Okay. As packaging, co-host, HBM stacking, et cetera, becomes a gating step for AI compute, wafer fab and packaging, the boundary between wafers and wafer fabs and packaging is blurring. Is LAM's roadmap converging with the packaging equipment players like Bessie?
Well, what I would say is, I mean, we have a very clear roadmap for advanced packaging. We play a critical role in that market. I would say that the biggest investment we're making is, you know, we're believers in this transition to larger format advanced packaging. I mean, from an efficiencies perspective, from an enabling...
When you say larger format, do you mean like panels?
Panels. Panels. Panels of various sizes at this point. There's not just one. And so as a critical supplier, one of the largest for wafer-based advanced packaging, it's a natural evolution for us. We acquired a company several years ago that has large panel packaging positions, and we've significantly invested in that company now based out of Austria. We just opened a couple days ago our center of excellence for advanced packaging, for panel packaging in Austria. And effectively, we're wanting to lead through this transition. You know, as AI devices themselves become larger and larger for the packages, that's a larger format is becoming almost a necessity.
Any thoughts on when that goes mainstream, panel level?
We're making shipments now. I think it's within the foreseeable future. It's coming.
Okay. All right. We've got about two minutes left. I will ask you the same question I always ask, and we've got a whole room full of investors here. Some may be new to the story. We talked about it through the session, but you can sum up. Why should investors buy your stock? Well, great. Normally, you give me like 30 seconds. I'll give you two minutes.
I've got two minutes.
It always feels a lot.
No, no, really. We've touched on a number of the key points, but if I were to sort of sum it up, we are, like the rest of the industry, excited about these multiple waves of AI demand. And so from a demand perspective, the move from training to inference, to agentic, to physical, fits extremely well with kind of Lamb's breadth of position across advanced foundry logic and then into, of course, HBM, but then really the excitement that coming is kind of further use cases for NAND. And so I think that sets us up for multi-year opportunities for growth in WFE and specifically growth within our segment. I think the LAM specific story then drills down to the increasing role that etch and deposition play in enabling the technology devices and new architectures that are coming. The world is going 3D. I joked about 3D logic, but, I mean, the reality is, as you said, CFET, 4F squared, ultimately 3D DRAM, you know, these are all 3D implementations, you know, leveraging that vertical scaling dimension because two-dimensional scaling, while still occurring, is becoming so much more difficult. Etch and deposition is synonymous with 3D. And so we see etch and deposition intensity as its share of total WFE spending continuing to rise from now as far as we can see these roadmaps. And so it says LAM is in exactly the right markets, and that's where we put all of our focus, etch and deposition leadership. And then finally, I think the reality is we're gaining share. We, you know, if you look at the last two years' performance, I mean, we've gained share of WFE. Partly it's that our markets are expanding faster than WFE, but partly it's that, you know, we have, I think, at a faster pace than most, refreshed our product lines. We just introduced a new Acara conductor etch tool with this direct drive RF capability. I mean, so we're sitting here with a great product portfolio in the right markets and really excited about the future growth opportunities.
right place right time great stuff there you go perfect with that I think we'll close it out thank you so much thank you