There you go. Perfect. Good morning, everyone. Thanks all for your time. Really delighted to have with us, Doug Bettinger, Chief Financial Officer of Lam Research, for our next keynote. Joe McCormack and I kind of cover the tech space here at Evercore. We're gonna keep this session to about 44, 45 minutes. Halfway through, we're gonna stop for some Q&A, so if you have any questions, please feel free to kind of jump in at that point. Our hope is this will be a fairly interactive session. You know what? I guess, Doug, first off, thank you very much for your time. Really appreciate it.
Yep.
Hopefully everyone read the safe harbor statement that was up there.
Put it, put it up real quick. I gotta be on the record with it, or my attorneys get mad at me. Real quick, I don't plan to say anything new that you haven't heard me say before, so I don't think you need to look at the safe harbor, but please do anyway. To the extent that I make any forward-looking statements, the safe harbor is relevant for anything that I decide to go rogue and say today.
Always thought it would be fun if you actually put something material in the safe harbor to see how many people actually read it and find the nugget of information. That would be kind of fascinating.
Yeah, it's actually kind of funny. We actually look at this every quarter and modify the wording. It's kind of funny, but anyway, that's not what we're here to talk about.
All right. Perfect. Well, thanks a lot for your time. You know, I guess maybe just to kick things off, and we have a bunch of questions to go through. I'm sure the folks here have questions to go through. You know, I feel, Doug, long term, everyone is really excited about semiconductors and the growth that's gonna happen in the space, and you presumably need a lot of tools, semiconductor equipment tools, to drive this growth.
Semiconductors don't build themselves. You gotta have the equipment to support whatever the output of the industry is. So yeah, that's-
Yeah
... absolutely true statement.
So, I guess maybe touch on two parts. What does that mean for the longer-term growth framework for Lam Research as you go forward? And, you know, maybe also touch on what do you see is happening near term to WFE trends. I think you talked about this number being around $70 billion for 2023, but within that, you had some, let's say, very dire assumptions on the memory side. Just talk about how is that stacking up, longer-term stuff-
Yep
... and then near-term stuff as well.
Yeah. Boy, I can ramble on for 10 minutes on that, so I'll just say a few things. First, listen, it's a down year. If you've followed semis, and I'm looking around the room, I see a lot of familiar faces, you know this industry. This is a growth cyclical industry with an emphasis on growth and an emphasis on the cyclical, but both are relevant to understand if you're an investor in the space. We're in the middle of a pretty significant kind of downtick, especially in memory this year. So when we came into the year, my recollection, we started the year thinking the industry was gonna spend low mid-70s. We're now looking at the industry and believe it's in the mid-70s this year. That's down 20+% from last year, so that's your cycle comment, actually.
What's going on? I'll talk about the near term, and then-
Mm-hmm
... I'll come back for the optimism on the longer term. Yeah, I mean, there's a lot of inventory and memory out there, memory and storage. And when we parse the level of investment this year, NAND is really quite dire, frankly, right? I have never seen-
Mm
... a downturn, this significant in memory in my entire career. I've been in the industry for a while. Memory, in total, down north of 40%, kind of sequentially year on year from an investment standpoint. NAND is down 75%. It is substantial, and when you look at what my customers are doing, utilization is dialed back more than I've ever seen. That's what you need to do when we're in the down cycle, I guess, but I've never seen it dialed this far back. It is what it is. At the end of the day, we, the industry will invest whatever it will invest. So, NAND is pretty dark.
DRAM a little bit better, but when you put memory in total down, like I said, north of 40%, Foundry and Logic is doing relatively good-
Yep
... good amount better. And frankly, and I know you're gonna ask me some about Foundry and Logic, so I'll park that for a minute, but, you know, everybody thinks of Lam as the memory company, right? The memory equipment company. But frankly, if you look at the composition of our business right now, that's not true. 70%+ of my business is Foundry and Logic right now, and frankly, business is holding in pretty well. So that's one thing to kind of get in your mind, that we are not just a memory company. We are actually quite strong in Foundry and Logic. So that's the near-term stuff-
Mm-hmm
... and I'm sure you'll, you'll ask me more about it as, as we go forward here. But that's not why people get excited about what's going on in the industry. I know you had NVIDIA in here just before, before us. Everybody is super excited about AI and what's going on there, and, and I am too, and we can talk about what I think that means for equipment investment. But it's gonna be significant, it's gonna be substantial. It's not hard to go find your favorite trillion-dollar semiconductor industry report. There's a half a dozen of them out there. I'm a believer in that. I am a firm believer in that, and frankly, that's the reason to be excited about what's going on in the industry.
Mm.
Industrial, automotive is gonna drive long-term secular growth. AI is too. And so, you know, if you wanna kind of look forward and be excited about something, that's what's to be excited about. This is a growth industry. It's a growth cyclical industry, and if the industry is gonna print $1 trillion revenue, you need a good deal more equipment than is out there right now, and you need investment across the-
Mm
... total spectrum of leading-edge foundry and logic, trailing-edge foundry and logic, and memory. All this stuff goes together in a system configuration at the end of the day. So, our optimism when we look at that trajectory is completely intact, independent of the fact that, yeah, things are a little, are a little bit soft this year. Also, stepping back to tell a little bit or foreshadow a little bit of the Lam story, when I look at a very kind of macro level, you have these, changes in architectures that are inflecting in the third dimension, right?
Mm-hmm.
NAND did it years ago... the industry is looking forward, still pretty far away, but DRAM is eventually going to adopt a 3D structure. You're beginning to see it in Foundry and Logic with Gate-All-Around. When things go 3D, Lam does well. We are an etch and deposition company. We put film down on the wafer, and we remove it with our-
Mm-hmm
... etch tools. That is the definition of 3D. Everybody's super excited about Advanced Packaging today. Well, guess what? To make Advanced Packaging happen, you need to deposit conductive material, and you need to create space to deposit that. That's etch and deposition. So when you think about the evolution of the industry heading in the third dimension, we, in my humble opinion, are the best-positioned company in the industry for that, broadly speaking.
Perfect. You know, I always feel like investors, markets have misconception about stocks or conceptions about stocks sometimes, and, you know, I think for Lam, it's always been, it's a memory-centric company for the most part. And you touched on this a little bit, but you really don't have a lot of memory business today.
Nobody does.
You know, you're still doing. You're on track to do, like, $27-$28 of earnings this year, right?
Yeah.
It gives me two parts on this, right? A, just talk a little bit more about the diversity, the recurring business you build that's enabling you to do that, versus that misconception of Lam is memory. And B, how do you think about recovery across both NAND and DRAM over time?
Yeah, there's a couple questions packed in there or buried down in there. Let me start a little bit. You're right. If you go back a year plus ago, you would have seen two-thirds of our business was memory-oriented.
Right.
If you fast-forward to this year, it's completely flipped. Two-thirds of our business, actually a little bit more than that right now, is Foundry and Logic related. We are not just a memory company at the end of the day. Listen, I love my memory customers. We love where that's going. At the end of the day, you don't have a logic device without memory and storage in a system configuration, so you need all of it together. And we enable a lot of the memory architectures, but frankly, we are not just a memory company. Fully understand that. We have printed in the last several quarters, record Foundry, record Logic. We've talked about with a large logic customer, doubling our share node over node as they go forward.
There is some really good stuff going on with our business and our tool capability in Foundry and Logic. So understand that, we are not just a memory company. And when we look forward-
Mm
... we see that continuing. Like I said, one of the things, we're excited about right now is the evolution of foundry going to Gate-All-Around. The first true 3D structure, in my opinion, in Foundry and Logic, by the way, both are obviously going to adopt a very similar architecture. To create that structure, it's very etch and deposition intensive, right? To create the sheets, you need to lay material down, and to shape those sheets, you need a whole different approach to how you etch things. It's Selective etch and surface preparation. And so when we look at this, this is a very large opportunity as we look into the future for our business. We've described an opportunity of close to an incremental $1 billion for every 100,000 starts that gets put in place.
Mm.
Not that we're going to win all that, but we will aspire to win at least as much as we hold across the rest of the industry. It's a big opportunity. We've been talking, oh, I think, for three years, or actually a little bit longer, about a new Dry Resist approach to putting material down as we look forward to the adoption of High-NA EUV in Foundry and Logic. Brand new business for us, something we have not done in the past, and we've described that as an incremental $1.5 billion opportunity cumulatively over a five-year timeframe, growing towards the latter part of that five-year horizon. So I'm super excited about what we have going on in Foundry and Logic, and when you think about the future, the future looks good.
Got it. You know, listen, you've talked a little bit on the 2024 WFE expectations on memory and everything else, but, you know, maybe just flush that out a little bit, how you think about the different parts, Foundry and Logic more specifically, but also, you know, you touched on memory already. But importantly, what does recovery look like? And I think that's a question everyone has, is what does NAND recovery look like?
Yeah. I'm staying away from, like, what shape is the recovery going to be? When is it coming? I don't know for sure, frankly. I don't. It feels to me right now that we're kind of bouncing along a bottom, but I do not see things right now getting better. And what you need to pay attention to is how much inventory is out there, what pricing looks like, and all of that kind of stuff. It's classic memory cycle stuff, and like I said, looking in the room, I see a lot of familiar faces. You guys all know what I'm talking about. The metrics are out there. You can kind of go modulate this on your own and figure out kind of when it's going to begin to happen. It's not happening right now, though.
I want to be very clear about it. But like I said, utilization is dialed so far back-
Mm
... I have a hard time envisioning it getting worse than it is right now. But I don't see it getting better right now either.
Yeah.
I'm not going to give you a shape of recovery in the next year.
In your seat, what are sort of the leading indicators you look at to say, "This is when it happens"? Like, historically, when you look, reflect on it, what do those indicators look like, and what are they?
Yeah. You know, I have the benefit of having somebody at my company talk to every one of my customers every single day. So that's what I have.
That helps.
That helps. You guys can't do that. You can come to conferences and listen to them talk and help parse it that way. Maybe what would be helpful, though, is, you know, we described how this is going to show up in our business, so maybe that'd be helpful to talk to-
Yeah
... everybody in the room a little bit about, in Lam's business, how are you going to know things are starting to get somewhat better? Well, first, the first thing that will happen is, you know, inventory will get more in balance. Hopefully, that's not too far down the road. Pricing will stabilize, profitability will improve. The first thing that our customers will do on the memory side is they'll begin to bring utilization back up, right?
Mm.
Because that's the logical thing to do. For our business, where that will show up is in the customer support business group. You've heard us talk about CSBG. CSBG is a composition of four different businesses, two of which will first show benefits when utilization comes back. The first area is spare parts. Spare parts right now is dealing with a headwind, in that utilization is dialed pretty far back. So we'll see it show up. Our customers will begin to need to consume more spare parts, 'cause they'll start running the tools more significantly. Similarly, with service, service will need to come back when the tools actually come back online. So those two things, if you listen to our earnings calls, you'll hear Tim, our CEO, begin to talk about that.
I think, Ram, you'll probably write it in his script, not mine, but that- that's what you'll hear us talking about, is, okay, you're starting to see somewhat of an uptick in spares and service. The next thing that'll happen in our business is customers will start to upgrade the tools that are in the installed base again.
Mm-hmm.
When utilization gets dialed back, generally speaking, the oldest tools are what get dialed back, and so that will be the stuff that then will need to get upgraded. That's also in the CSBG business. So the upgrade business, by the way, it's a great part of the business. It's an awesome value proposition for our customers because they get new tool capability at a fraction of the cost-
Mm
... of buying a new tool. So you'll start to hear us talk about recovery in the upgrade portion of the business. And then eventually, new wafer capacity will need to get put in place. So that'll be the thing, and by the way, that's kind of the new equipment stuff that everybody thinks about when they follow Semicap. But anyway, I'm rambling on here a little bit, but that's how it's gonna show up in our business.
Seems like talking to customers every day is a much easier way to figure this out than the second option, but-
Usually speaking, it is. Although sometimes I wonder, they don't always know when things are gonna get better either.
Fair enough. You know, maybe shifting gears a little bit, on the AI opportunity, I think on the last earnings call, you know, you folks had talked about, hey, each one percent penetration of AI servers-
Yep
... is gonna have some impact to WFE start.
Yep
... over time, right? Maybe just talk about that and how you think about Lam very specifically, how your opportunity when it comes to AI deployments.
Yeah, I mean, everybody's getting really excited about AI, so we felt it was important to describe, "Hey, what is this gonna mean on the equipment side?" And so we did our best to sit down, spend some time talking to customers, but also analyze the market, and anybody can go do this. The data point that we gave you is, for every 1% of server volume that shifts to an AI configuration-
Mm
... versus an enterprise configuration, what is that gonna mean for the investment in WFE? So first, if you look at true AI configurations, what do you see? First, it's most advanced microprocessors with the biggest AI, so there's capacity at a certain level there. There's nominally 8 GPUs in that configuration versus 1 or 2 in an enterprise-class server. There's 8 times the DRAM content put together in high-bandwidth memory configuration using a TSV structure, which is awesome for us, by the way. I'll come back to that in a minute. And then you've got nominally 3 or 4 times the NAND content. So if you look at what I just described and analyze that 1% pivoting from enterprise to AI, that's a $1 billion-dollar incremental WFE required to support that configuration.
Now, I've had some people pushing back to say, "Yeah, but the, that's just a reallocation of hyperscale budget from enterprise to AI." Maybe that's somewhat true in the near term. It's not true in the long term. AI is gonna be incremental in the medium and long term, for sure, and I can talk about why I confidently say that. It's just based on what I see happening in my own business and how we're looking at what AI is gonna do for our own business. But that's kind of how you get to the math that we put out there.
Yeah. No, I mean, to your point, I think IBM, for example, has talked about 30% savings is what you can get if you deploy AI properly in your cost structure. And so if you can start saving billions of dollars, I imagine these deployments will be incremental and not just cannibalistic over time, right?
Listen, that's how I think about it, and maybe why I say that is right now, when I look at... Well, let me paint a picture of what we're doing at Lam. All of us are trying to figure out, what does this mean to your business?
Mm.
I know I'm just like every other large company in the world, which is: I'm not sure. We're looking at all these proof of concept, different areas where we've got pain points in terms of information sits over there, work is being done over there, and, and how can you more effectively put these things together? I see hugely compelling opportunities in our own business, and I know I'm not unique in that regard, in that every large company has these pain points. Everybody's business is a little bit different. Everybody's got different things that challenge them from a productivity standpoint, but this is gonna be a real thing.
Right.
There's gonna be real business value from this, and I don't exactly know where, but I know it's there. I know when you see stuff like this, it ends up being incremental investment over here to save over there.
Right.
There, there's a big ROI there, I believe.
Yeah. No, it's a little bit like, I think when iPhones came out, no one knew you'd use apps on it, but eventually, as that happened, you saw massive spending happen around it, so-
Absolutely
... I think AI and the equipment needed for that will go down that same path. I'll pause there, maybe give it to Joe to ask the smarter questions now.
Doug, maybe just go back to CSPG for a little bit.
Yeah.
You had covered most of it, but maybe just to touch on the Reliant business as well there.
Sure.
you know, lagging edge tools-
Yep
... been a bright spot this year. Maybe do you want to just remind us on, like, the diversity, both from a number of customers as well as, like, a geographic location standpoint for that business, to think about, you know, sustainability?
... Yeah, maybe if you'll allow me, 'cause I'm sure there's some new people listening to me talk for the first time. Let me describe what is CSPG first, and then I'll come to your direct question on Reliant. For us, CSPG, again, the Customer Support Business Group, is the money we make from the installed base in the customer's fab, for the most part. It tends to fall into four buckets. Not tends to, it is in four buckets: spare parts, service, equipment upgrades, and then what you specifically asked about, Joe, is the Reliant product line, which is more of our specialty node, or some people think of it as legacy node, equipment sales. That all sits in one business unit at Lam because it is how we monetize equipment that we've designed in the past, right?
It's not necessarily requiring a lot of new R&D, although there is some R&D in there. It's more kind of the, the dollars made from prior period investments. And so when you look at it, today, actually, it's actually quite amazing. Before the down cycle, CSPG was roughly a third of the company's revenue.
Mm-hmm.
Last quarter, it was 47%. That's as much a statement around how much the investment in new equipment is down, but it also is a statement around the relatively resilient, CSBG business. In particular, to your point, that Reliant product line or the specialty node product line is actually doing pretty well this year. It's perhaps a bright spot in a more of a down market, and that's because if you think about what's going on there, this is a broad footprint geographically. It's supporting customer bases like industrial, automotive, IoT, analog, power. It's very broad. There is certainly a component of it in China, which is something everybody's been asking a lot about, and you guys probably will ask me more about it as we keep talking here.
But that is doing pretty well. You know, if I would've been here two years ago talking to you, I would've said the largest component in CSPG is spare parts. I now, when I describe CSPG, describe it as there's two big components in CSPG, spare parts still, but Reliant has actually grown quite a lot over the years and is doing real well this year.
Makes sense, and maybe just touching on CSPG longer term. You kind of pointed to this typically grows every year. Do you wanna touch on, as we push through 2024 and beyond, how to think about a more sustainable growth profile?
Yep
... for this business going forward?
Listen, a couple of things. Yeah, I used to say: Hey, this is a business that grows every single year. Well, it's not growing this year, but it's not down too much, and it's not growing this year because utilization is dialed so far back. I never envisioned utilization being pulled so far back. Obviously, that impacts spare parts, service, and equipment upgrades. All that stuff is facing headwinds relative to utilization right now. And so that's why it's important also to listen to us. When we finish the calendar year, the December earnings call, we always give the market... Here's the number of chambers in the field. If you compare this downturn to the previous one, the number of chambers in the field for Lam has grown by 40%.
That's why, generally speaking, and I describe this as a business that grows every year, I think people sometimes don't realize our equipment runs for decades. It's not something that you sell, and then it goes away. It actually gets repurposed, it gets upgraded, it gets moved to sometimes a different customer, although right now, that's not happening any longer. But that's the Reliant product line, things that go into some of the legacy nodes. So that's an important thing to understand. Our equipment runs for decades. We've got stuff that's been in the field 30 years, and so if you think about that, what does that mean? Well, the revenue generated for an average Lam tool in the field, over the life of the tool, generates more revenue than when the equipment is first sold. We've spent-
Wow
... most of the conversation, we're talking about, Hey, where's WFE? It's up, down, or sideways. Yeah, that's important, but understand, a key part of how we make money at this company is from the annuity that comes off the fact that we've got 84,000 chambers in the field, up 40% from the last down cycle. That's why usually I describe this as a business that grows every year. Because all of those chambers consume spares, they need to be serviced. There's an opportunity for them to be upgraded, sometimes multiple times over the life of the tool, and then every once in a while, we actually will buy used equipment back, refurbish it, and resell it. There isn't any used equipment available anymore, but that used to be part of the Reliant product line, business model.
We're now selling new, old equipment, if that makes any sense to anybody. It's a great part of the business, by the way. Like I said earlier, it doesn't require a huge amount of R&D. The R&D is basically a sunk cost when the tool's developed, right? And then the opportunity to, like I said, upgrade it, spares and so forth, happens over time. Enormously free cash flow generative in CSBG, like I said, because it doesn't require a whole lot of R&D. So it's an awesome part of the business model that I believe is truly underappreciated by people. And maybe I don't talk enough about it, and I need to try to do a better job describing this part of the business. It's a great part of the business.
Yeah, that's great.
Mm-hmm.
Maybe one last one before we open up to-
... questions is, you know, you've touched on a few of the, you know, kind of key opportunities moving forward, whether dry resist or AI, anything else that you want to kind of just call out there as it relates to what investors might underappreciate from your competitive positioning? NAND moving forward, as you know, layers continue to increase, you know, just anything that you feel like is really misplaced as it relates to how investors look at you.
I think the most important thing to understand is every new node is somewhat more expensive than the one before it. Like I just tried to describe earlier, when I kind of abstract and look at the evolution of architectures of our customer, generally speaking, they're inflecting in the third dimension to a greater or lesser extent, depending on the architecture. When things inflect in the third dimension, we do well, 'cause to create structures that are three-dimensional, it's materials intensive, and you need to etch the material that you put down on the wafer. Maybe if you'll allow me, one of the things we're really excited about right now is Advanced Packaging. So I'm gonna ramble on about that a little bit here, 'cause I wanna make sure I get it out there.
One of the ways things inflect in the third dimension is Advanced Packaging. You'll hear my customers describe chiplets, Heterogeneous Integration, Advanced Packaging, what have you. This is an opportunity, when you think about it, is you're putting multiple die in some kind of structure or package, or sometimes just silicon on silicon, to create an interconnect to help manage cost, performance, and then just putting different device types together. When you do things like that, you need to create a conductive pathway for the electrons to flow. Generally speaking, when you look at how that is deposited, it's a copper electroplating process. Frankly, we own nearly all of, maybe all of-
Mm
... it's a tool we call Sabre 3D. To create the space for the conductive material to get deposited into, it generally is a silicon etch, a deep silicon etch. It's a tool we call Syndion, which we have nearly 100% market share in that. So again, when you think 3D, think Lam Research. It's all we do. It's what we do. And maybe if you let me digress for just a minute. Everybody is super excited about AI right now. One of the key enablers of AI to feed that parallel compute architecture, you need a lot of data flowing through. To make that happen, low latency DRAM. Generally, when you look at how it is put together in these AI structure architectures, you've got high-bandwidth memory in there using DDR5.
It's 8 die, eventually going to 12, put together using something called Through-Silicon Via process architecture. To make that happen, pretty much 100% market share for us in silicon etch and in that electroplating process. So yeah, memory's, you know, not spending a lot of money, but one of the areas actually where customers want tools quicker than we can get them to them right now, and we're expediting things, is in TSV for high-bandwidth memory-
Mm
... because of AI. We make that happen, full stop.
We'll pause this, see if there's any questions in the group. Let's get the mic over there.
We have a mic in the room? Yep.
To Ankur right there.
Yeah, Ankur. How are you, Ankur?
I'm good, Doug. How are you?
I'm good.
Just a quick question on the math around the 1% AI penetration and the effect on Wafer Fab Equipment?
Mm-hmm.
How many AI servers replace 1 enterprise server or 10 enterprise servers? Like-
It depends on the application. Like I said, in the near term, maybe this is a push from one to another. In the long term, I think this is all incremental, Ankur, and I say that because when I look at where we're looking at, these AI servers, these are new activities. I still have to upgrade my enterprise infrastructure to go to the next instantiation of SAP S/4. So to me, it's not a this replaces that. This is going to be incremental.
Mm. Okay.
Maybe not in the near term, but in the medium and long term, it's going to be incremental.
All right. Well, if anyone has questions, feel free to raise your hand, we'll get the mic to you. If not, we'll keep going. You know what, I guess, we just shift away from revenues and what's happening to memory. Gross margins last quarter were extremely impressive.
Pretty good, yeah.
They were good. It was helped—I think mix was good. You also had better cost improvements than expected. Maybe just talk about like what enabled that, and then importantly, how do you think about the sustainability of this as you go forward?
Yeah, let me describe a little bit about, I don't know, strategically or operationally, what we're doing at Lam this year. Like I said, business is down. And so what we're, what we're doing, as a leadership team, as a management team, is taking the opportunity to make the company better so that when business gets better, we're gonna be better. And what that means is we're going through a reasonably sized restructuring activity, taking cost out, eventually setting the company up for when growth resumes, to be able to do it more cost effectively. We described that, I don't know, back in the January call. What I would describe to you right now is we're ahead of schedule on that-
All right
... the project, the program. And so that is a component of, I think we came into the year, and I set an expectation that we would improve gross margin by 100 basis points. Frankly, we've well exceeded that already. I guided to 46.5% gross margin. Starting the year, we're at 44%. A key component of that is this program is ahead of schedule. And by the way, a portion of what we're doing in this restructuring is related to kind of figuring out how to make AI work at the company, right? We're undertaking a digital transformation project that's delivering some of this. Although a lot of it is just taking advantage of the manufacturing footprint that we have in place.
We've been talking for a few years about, you know, this large factory we've built in Malaysia. When business does resume, a lot of the incremental growth will be supported from that factory in Malaysia.
Mm-hmm.
It's beneficial because it's got good labor rate, but frankly, more importantly, it's closer to customers, right? Most of this industry, the fabs are in Asia, obviously. And so being closer to the customer enables you to take advantage of a supply chain that has a better cross footprint. But, equally importantly, and maybe more importantly actually, is, you know, to get our tools from where we build them to where the fabs are, we're flying them around in aircraft. If the distance you need to move stuff around is shorter, and obviously, if the factory is in Asia and the customers are in Asia, you're saving on freight, that's gonna be a key component of it, too. So a big part of what you're seeing from us relative to gross margin performance is that program-
Got it.
that we're working really hard on. And then, yes, I will acknowledge there's a favorable mix as we sit here going into the September quarter that is gonna show up. So both of those are important right now when you look at where we're at on gross margin.
Got it. And, you know, you touched a little bit on China, or mentioned China earlier, at least. How much benefit are you seeing from higher-than-normal domestic China revenues, let's just say, that could potentially fall off in 2024? Maybe just level set that dynamic a bit.
I guess the first thing I would describe to you is, you know, back in October of last year, a new set of regulations came out from the U.S. government. That was a big negative hit to our business. We described, back then, an impact of $2 billion-$2.5 billion in revenue that we would no longer be able to support. So that's important to understand. This isn't like in a multi-year context. This is a headwind, not necessarily an opportunity.
Although what we talked about more recently is we did get clarification that there was some business we weren't sure we were gonna be able to support, that we're now able to support, and that was a component of when we described where's WFE this year, we moved it up from low- to mid-70s to mid-70s because there was some incremental business.
Mm-hmm
... that came back in, related to that. That's part of the, like I said, the uptick in our view of WFE.
Got it. And, you know, I think longer term, you've sort of talked about gross margins eventually getting to, like, 47.5%-48% kind of target.
Yeah.
I think it was back at the analyst day you folks had. What do you think is the path to get there? Is that a... Is it, is it revenue dependent? Is it more on the cost takeout stuff continuing on? Like, what are the levers that get you there?
If I go back, we put that model out, and you got the numbers exactly right. I think it was March of 2020, just before COVID, and then a lot changed, obviously. We had some inflationary headwinds that came in that don't necessarily go away, so that was a bit of a negative component to gross margin that we had in that model. What we're gonna do to offset that, though, is what I've already described to you, which is we're working on operationally making the company better right now. And so when you think about those two things, you had a little bit of inflation come in. You now are doing more operationally.
Those are gonna net out, such that at the end of the day, we're still committed to get to that 47%-48% gross margin that was in the financial model. We're just gonna have to pull slightly different levers to get there.
Got it. And then, you know, I think this year, 2023, supposed to be a really good, strong free cash year as well for the company. You know, but, but I think within that, you also talked about, hey, inventory maybe may not come down as quickly as expected or as you would like-
Mm-hmm
... to the pre-pandemic levels. Just, I mean, just talk about what, what's happening with working capital that's delaying that, and when do you see that kind of normalizing?
Yeah, let me go through it a little bit. I mean, it may be somewhat counterintuitive. When business goes down, oftentimes cash flow improves. Why does that happen? Generally, it's because working capital comes down, accounts receivable comes down, inventory comes down, and cash shows up, I guess, a simple way to describe it. And that is very much what you're seeing from us this year. Very strong cash flow last year, or excuse me, last quarter. Free cash flow last quarter was north of $1 billion, so that's great. Actually, on a run rate basis, it's higher than it was last year, because last year, business was growing, this year it's coming down.
Having said that, though, there's a little bit of a lag I see us being able to deliver this year relative to inventory. Why is that? If you think about the rate at which business declined, right, if you go back to the December quarter last year, we printed $5.3 billion in revenue. Last quarter, it was down to $3.2 billion. And so when you have a business reduction that occurs that quickly, you're ending up buying less material than you previously told your suppliers you were gonna buy, right? We canceled a lot of purchase orders. And so as you go through a process like that, inevitably, what you end up needing to do is make sure your suppliers are hanging in okay.
Yep
... and you end up taking sometimes more inventory than you need in the near term, knowing at the end of the day, this is gonna be good inventory... concerned at all about risk of obsolescence or anything like that. It's more the pace at which business decline meant that we were canceling a lot of stuff that we previously told our suppliers we needed. And so as, like I said, as part of that, you end up taking some inventory that you don't need in the near term, that you'll use later. And so when you look at previous cycles, you know, we've been able to manage inventory down more quickly. The pace at which this one showed up meant that there's gonna be a little bit of a lag to that, but it will still happen, right?
You know, we're still good operators, we still know how to manage the balance sheet, we're still focused on managing the balance sheet, but at the end of the day, you also need to be focused on the health of your supply chain, and that's part of what's going on there. Inventory's not coming down as much as I wish it was, but for good business reasons.
It almost seems like demand came off a lot quicker, but your purchase commitments are out there.
That's-
You kind of still have to honor these, and-
That, that's exactly right.
And, you know, you wanna make sure suppliers are around the next time you need them, so...
At the end of the day, this is a long-term relationship, right?
Right.
We think of them as remote factories in-
Right
... in a lot of ways, and so the resiliency and the ongoing health of our suppliers is important to us.
I want to draw some analogies to your prior company on that statement, but I won't do that.
Yeah, don't do that.
Well, Doug, we have a few minutes left, maybe to swing back to some technology-oriented questions for a bit. You know, you've talked about advanced packaging.
Mm-hmm.
Do you wanna just talk a little bit more about, like, the current status of demand? It seems like some of your peers are looking for upticks in DRAM-related revenues into the September quarter. So just maybe when you look at high-bandwidth memory and the generative AI-related demand, you know, how quickly do you expect, you know, that, you know, couple hundred million dollars to move to the, the billion dollars over time that you've talked to, as it relates to advanced packaging?
Yeah, I guess what I would describe is, you're right, the way I get asked: "Hey, how big is this advanced packaging stuff you're talking about?" And I've described it, I don't know, in the recent past about, at Lam today, it's $hundreds of millions. I more recently said, you know, in the not-too-distant future, I have not a hard time envisioning it being a billion-dollar business because of the evolution of things that are happening in the industry, not just high-bandwidth memory, although that's a key bright spot right now in a fairly dark memory spending environment. Like I said, we're trying to pull things to be able to ship them sooner to the high-bandwidth memory component. So that's a part, but you hear everybody talk about chiplet architecture and all, everybody's doing this stuff, right?
CoWoS, all of this stuff requires everything I've already described with the Syndion and SABRE tool.
Makes sense. Then, maybe just to touch on Gate-All-Around a little bit further as well. Do you wanna talk about, you know, whether it be metal or dielectric, how you guys are positioned to kind of, you know, win out there and, you know, how your new products, you know, in Selective etch and otherwise are, are really gonna, you know, kind of benefit as you think about that, you know, close to $1 billion per 100K?
Yeah, when you look at the composition of that incremental opportunity, it falls broadly into two categories. There are ALD applications in there, and there's selective etch applications in there. Probably a few more selective etch than ALD, but both are critically important. We're well positioned. You know, I feel really good about it. We wouldn't be putting numbers out there if we didn't feel good about it. And it's just around the corner.
Okay, maybe one last one. You've talked about, you know, optimizing the cost model, you know, in the current downturn to, you know, look more attractive out here, but obviously, probably still spending to, you know, kind of support growth into a trillion-dollar-
For sure
... semis industry. So anything else you want to call out, whether it's the Sense.i platform or, or otherwise, as it relates to your R&D spend, that you guys are focused on to, you know, really benefit into the back half of the decade?
Yeah, I guess that's one thing that's important to understand. In this business, you don't cut R&D independent of wherever the business cycle is. You really don't, because these are multi-year investments, and our customers today are making decisions that are three nodes from now. So independent of whatever the business cycle is doing, you have to stay committed to your long-term R&D programs, and we very much are. We've got a new etch platform, we call it Sense.i, that's out today, but requiring incremental investment for sure, and we're still investing in the older platform as well, 'cause customers are still buying new equipment there. There's a whole bunch of new ALD investment that's occurring. The dry resist is all incremental business for us and all incremental R&D. So those are examples of things that are driving the R&D profile.
Maybe just to remind, on dry resist, you said $1.5 billion over the next five years-
That's right. Yep.
... starting this year, and maybe a little bit back-half loaded as it relates to-
More weighted to the back half of that time, that timeframe. In, in year 6, 7, and beyond-
Yeah
... it'll get even bigger. And like I said, it, it's hard in this business to find something brand new to go, to go after, to go invest in, to go generate incremental revenue, but this is an example of that. We don't play in this part of the equipment space today. The customer pull for this is very strong. Every one of our customers that uses EUV has our hardware in the lab, evaluating the capability, and what's important to understand is that lab space is extremely valuable. If the customer doesn't see a benefit to what you're bringing to them, they won't put your equipment in the lab to look at it and evaluate it. We've already announced three tool selections.
Two DRAM customers have selected this, and more recently, we talked about a large foundry logic customer also has selected this, to ramp into production. So, customer pull is quite strong. The value proposition, I think, is quite compelling. The momentum behind where this is going is continuing to build. So I'm, I'm very excited about it. From our point of view, this is a question of when, not if, this ramps. And we're, we're making really solid progress.
Mm-hmm.
Yeah, it's great to hear. I mean-
Perfect. There's a question up here?
Yep.
Yeah, um,
Can we get the mic so we, real quick, so they can hear it on the webcast?
So just two questions on NAND.
Yep.
One is like, you know, we talk about like how the AI has a positive impacts on logic, on DRAM. Like, just if you think about the AI's impact on NAND over the mid- to long-term, just wondering what your thought on that? And number two is, like, obviously, NAND is in a very bad cycle right now.
It is.
If you listen to the memory guys, like, they're not only, they're also kind of like talking down, like, the midterm outlook for the bit growth-
Mm
... on NAND as well. Is that, like, a typical mid of down cycle behavior that, like, the near term is so bad, is it very hard to not feel a little bit worse about the long term as well? Or is, like, there is some truth behind, like, you know, the long-term outlook from then as well? Thanks.
Yeah, listen, I never question my customer's perspective point of view, and I'm not going to now, right? They, they're closer to the marketplace than I am. I would share an observation I have with you, which is, in this industry, we suffer from a recency bias. When things are good, we just draw a line up and to the right. When things are bad, we draw a line down and to the right. I don't know if that's the case here. I wonder if it might be a little bit. But yeah, I think my customers have down-ticked a little bit in terms of what they think long-term bit demand looks like. I don't know what's changed, except that we're in a down cycle. So... But I'm not saying they don't have this exactly right.
They're closer to the market than I am. And then to the beginning of your question, today, best I can give you is when we look at these AI server configurations, it's got about three times demand content as compared to an enterprise server. Whether that is what it sustainably looks at as this gets to be a bigger and bigger component of the market, I'm not entirely sure, but that's what I see today.
Perfect. I think we're almost up on our time, but maybe to wrap this up, free cash flow generation. I mean, based on just how well you folks are doing over here, love to understand your ability to stay within this targeted 75%-100% free cash flow returns in 2023, and then any changes really broadly to capital allocation priorities as you go forward.
No change. Listen, like I said, it's gonna be a very strong free cash flow year for us this year. We are planning to continue to return 75%-100% of free cash flow. In fact, a week ago, we raised the dividend by 16%.
Mm-hmm.
We plan to grow the dividend on an annual basis. I know people like that. So no change.
Perfect. I think we're up on our time. Maybe I'll pause here, turn it back to you if there's any closing comments, anything you and I did not touch on that you wanna make sure investors are aware about.
No, listen. I think you guys covered a lot of what I'm spending all my meetings talking about. We're in a downtick. Kind of is what it is. Customers will spend whatever the customers will spend. At the end of the day, I can't, we can't control that at Lam. But what we can control, we're working on making this company better for when business gets better, right?
Mm-hmm.
You're seeing that show up in the P&L. I'm pleased with that. In the long term, I am as optimistic about this industry and our unique position in it as I ever have been, independent of wherever we're at in the cycle. Don't lose sight of the fact that this is a growth cyclical industry. Yeah, there's a cycle, I, I will never say there's not, but there's awesome growth in front of us.
Perfect. All right. Thank you very much for your time, Doug.
Thank you.