Let's get started. Good afternoon. Welcome to this afternoon session. I'm Vivek Arya. I cover semiconductor, semi cap equipment at BofA. I'm really delighted to have the team from Lam Research join us, Doug Bettinger, the Chief Financial Officer. What we'll do is go through some of my Q&A, but please feel free to raise your hand if you'd like to bring up a question. With that, warm welcome, Doug, and I know you have-
Let me-
Some exciting announcement.
Let me do an exciting, just real quick, safe harbor to keep all of my attorneys happy. Let me read this real quick, and please take a look at what's up on the screen. "Today's discussion may include forward-looking statements that are subject to risks and uncertainties, and actual results may differ materially. Additional information concerning factors that could cause actual adult results to differ materially from those forward-looking statements can be found in the risk factors disclosed in our public filings with the SEC on Form 10-Q." With that, we can actually talk about the business.
Excellent. Thank you. Maybe let's just start with the state of the union, Doug. I mean, versus what you thought at the start of the year in terms of the planning assumptions.
Mm.
of demand in your various end markets, you know, if you do kind of a mid-year check, how are they panning out versus your original assumptions?
Yep. No, that's a great place to start. Not all that different, frankly, Vivek, when I think about it. We started the year. It did change a little bit, so I'll unpack that slightly. It is a down year, right? That statement is especially germane in the memory side of our business, and I'll talk a little bit about that. We began the year with the view that wafer fab equipment spending would be mid-seventies, and we took that down just a little tiny bit after the last quarter's earnings. Now we're saying low mid-seventies. We began the year thinking things were a little bit first half-weighted. As we sit here today, we now think it's a little bit second half-weighted. Subtle movements around in there.
It's probably no surprise to people that follow the semiconductor industry to know that memory got a little bit softer. Probably got a little softer, too, relative to 90-plus days ago or something, right? That was part of the downtick in total. Offsetting that to a certain extent, and part of the reason why it's a little bit second half-weighted, is we got some clarification on the rule, the regulations relative to what we could ship into China. That provided a little bit of an upside that we believe we're gonna ship in the second half. I describe that for Lam as a few hundred million dollars that will show up in the second half. When you unpack the whole thing, we see WFE down more than 20%, memory down more than 50%.
Within that, NAND's down well more than overall memory, and DRAM a little bit less so. That's kind of how we see things.
Got it. We, you know, go through the different end markets, but I have to ask this obligatory question about AI.
I love this question.
Which I know you're very excited-.
Yeah.
to answer.
I am the best company in the industry to ask an AI question. It's kind of tongue in cheek, but, yeah, please.
Yeah. What's your sense, right? There's a lot of excitement, right, in parts of the data centre.
Mm-hmm.
How do we connect the dots from that excitement to what it means for Lam?
Listen, when I think about this, there are two incremental growth drivers as I think forward over the next several years, one of which is what's gonna occur in the cloud and driven in a big way by AI. It's new stuff, new use cases, new compute architectures that are gonna show up. Frankly, today, even though there's lots of excitement about it's still a fairly modest amount of real spending at the end of the day, right? If you think about total servers that are true AI servers, it's probably only 5%-ish, something like that, but it's gonna meaningfully grow over time. When we look at that and, like, study the motherboards in there and the architectures that we see, you tend to have very big logic die. It's a statement around microprocessors, accelerators, like GPUs, it's obvious, right?
FPGAs, other custom TPUs and things like that. Maybe equally importantly for Lam, is the memory content that shows up in these servers. When we look at the DRAM footprint, it's nominally 8x as much as an enterprise server, and in NAND it's 3x to 5x . Today, there's tons of excitement about it, and I'm super excited about it too, but the future is really gonna be where it shows up and impacts our business. Frankly, right now, it's not really doing anything with equipment orders. Not yet. Right? You'll see it at leading-edge foundry, and you'll see it in memory.
Got it. Is that because on the memory side, data centre still has inventory?
That's part of it.
That's really.
That's part of it. When you think forward, there's a new set of CPUs ramping, Sapphire Rapids and Genoa. It's gonna pair itself in a bigger way with DDR5. That's gonna begin to help the industry move through the cycle when that begins to be a bigger contributor to the total bit demand.
Got it. Just one last thing there, Doug. Longer term, do you think that, you know, data centre and AI is incremental to industry growth, or is it merely going to offset, right, the maturation of the consumer part of the industry?
I think it's gonna be incremental. I mean, if you go read these studies that are out there about the trillion-dollar semi industry, it's almost obligatory, and I sometimes like to stay away from that number. Anyway, be that as it may, if you actually go read the stuff, the people who have done the work, there's two things that contribute to incremental growth. One is this, in AI and very high-end compute architectures, and the other is the content growth in automotive. Both, I believe, are gonna be incremental as we go forward.
Got it. One other kind of near term is, you know, we heard that TSMC might, you know, change their CapEx towards the lower end of their outlook. I was under the assumption that, you know, when you gave your outlook or others, that you have kind of contemplated some of that softness in the leading-edge also, right? Was that news to you, or was it kind of already sort of contemplated in how you were thinking about the year?
Yeah, generally speaking, you know, when our customers make public announcements, usually they've communicated with us beforehand, usually well in advance. We have to plan, we need to be positioned for whatever might be changing. Almost always, when our customers are making public statements, we've known about it ahead of time, and to the best of our ability, contemplated that in how we describe what's going on in the industry. This is no different. I'd also tell you, we have a pretty robust process. When we get on an earnings call, we've gone through a cycle of talking to all of our customers. What are your plans? Has anything changed? That's one, it's a bottoms-up process.
My corporate analytics and IR team goes through and does a tops-down analysis of the market, and do these two things kind of hold together to the best of our ability to analyse it. When we communicate something to you, I want to be as precise and accurate as we can be, so we go at this in a pretty rigorous way.
Got it. Now, if you take a step back, Doug, last year, you know, the discussion was the industry would be at $100 billion WFE, unconstrained number, even higher. Here we are, right? I think what you suggested is, what, low to mid $70s billion.
Low, mid-70s, what we see, yeah.
How much of that is the cycle, and how much of that could be the fact that maybe 100 was a very, you know, abnormal I know it's all hindsight, but maybe that number was too high to begin with.
Listen, with hindsight, it's very clear to me what occurred. When it was, when we were in the middle of it wasn't apparent to me. When I think about where we sit today, often the cycle in semis is more about supply than it is about demand. When I think about the last year and a half, it's been as much about demand and change in demand outlook as it is supply related. And by that, I mean, if you take yourself back in time to the beginning of last year, I think all of us believed there was going to be unit growth in PCs and unit growth in smartphones, and the industry configured itself according to that point of view. Well, that didn't turn out to be what happened, right?
I think some aspect of it, we perhaps mistook secular growth a little bit for a demand pull forward that occurred because of COVID. Anyway, with hindsight, I can look at it and say, "Yeah, that's probably what happened." So there was that aspect to it. It was confounded, actually, and made it a little bit worse, and this is maybe as much supply. You know, we went through this crazy period of everything being completely constrained, and the entire industry rallied itself around the stuff you could get. Maybe let's get a little bit more, right, just in case. Waiting for that mythical golden screw. We all did it. And when you put those two things together, it created a cycle that we're sitting in today.
I can look back on it, and intellectually, what I just described, I've thought deeply about, and that's my conclusion. When it was happening, it wasn't apparent to me, anyway, that what, that's what was going on.
Got it. As we start to think about, like, the next two, three years, what is the right formulation, you know, if you are in our seat, to predict what WFE can be, right? Not like a specific number, but when you have to make your predictions about WFE, like, do you start with semiconductor industry growth and say, look, if in the past, semis grew X, right, WFE grew 20%, 30% faster than that, you think that kind of formulation is.. Let's say if I say semiconductor industry growth is 10% next year-
Yep.
Does that inform you what WFE might be next year?
It helps. I mean, the first thing I would say is, listen, I'm a firm believer in the growth aspect of this industry. This industry is going to be meaningfully bigger 5, 10 years from now than it is today. That's an important thing to believe, and if you don't believe that, you're in the wrong room, invested in the wrong sector. I think we all believe that, or you wouldn't be here listening to me talk about this. That's one thing, right? The future growth is important, but this is a growth cyclical industry. Every year does not grow. As much as I wish it did, it just doesn't work that way, and it never has, and as things have consolidated, there was a kernel of hope I had at one point that it would've got somewhat better, and frankly, it hasn't.
It is what it is, a growth cyclical industry. The growth piece is important because you have to be prepared for that future growth. For Lam, as I think about, okay, we're in a down cycle right now, what are we doing? We're making sure we've got the best product roadmap we can possibly have, right? Even though we've, we're trimming spending a little bit, we're keeping R&D full steam ahead because you must have the best products in the industry to win, and that is what we're intending to do and what we are doing. Our product lineup's never been stronger. You must be convicted in that, and we are, right? Don't sacrifice the long term to manage the short-term P&L. Can't do it.
You also must construct a company and have an operating model that has a level of variability to it. We've done that extremely well over the 43-year history of the company, and we keep refining that. We're fine-tuning that. It's not fun to go through the downturn where you got to kind of trim the workforce and whatnot, but that's what we just did. Frankly, we are setting the company up. Here's something else I've been saying: At the end of the day, in the near term, I can't control what the customers are going to spend. Kind of is what it is. What we can control is make sure you got the best products coming out. We're doing that. New Sense.i platform, new drivers here. We got a whole bunch of different things that we're extremely focused on.
Trying to make the operating model better, right? We are working extremely hard right now on, maybe restructuring is not quite the right word, but I'll use it anyway, to take advantage of the manufacturing footprint we've grown in Asia over the last three or four years to pivot to it in a bigger way when growth does come back, so that we're a more efficient, more profitable company. That's very much what the leadership team at Lam Research is focused on this year, is control the stuff you can control and be better when business gets better, is how I've been describing it. I believe we're going to be extraordinarily well-positioned when that growth does come back, whenever it is.
...A few more things, talk about the back half, and then we'll talk about the longer term. The incremental approvals for China, is that a one-off business, or do you think that is kind of then becomes a baseline, and it will do then whatever overall China WFE does going forward?
The first thing I want to make sure people understand is the rules didn't change. There are no new rules. The rules that came out on October seventh are the rules. There was some level of uncertainty we had with a specific customer at a specific process node. We had a view that it was not crossing that technology line that was out there. We needed kind of some level of concurrence.
Mm-hmm.
The industry actually working together, kind of, we came together and went and had that conversation with the government and got that clarification. The rules didn't change. I think maybe we didn't describe this well. There was some people that thought, "Hey, did something change?" No, not really. We just got clarification. As I think through that's part of that. It's a little bit second half weighted year, because we didn't ship any of that in the first half. It'll show up in the second half.
Got it. Is that then the new baseline?
I think so.
All else being equal, then it is part of the new baseline, right? For whatever China WFE does.
Yes, it is.
-going forward.
It is.
It's not like one-off that was from last year, that was something.
Right.
Um-
No, but having said that, let me remind you, this was a big impact on our business, right? We described a $2 billion revenue that we believed would have shown up this year, that we it's not there any longer because the rules restricted our ability to ship. We had to respond to that, and we have. It's still a big impact, but a little bit less than we previously thought.
Right. You think these China restrictions have reduced how we think about WFE going forward? Because, you know, their intensity used to be higher. Even if, let's say, overall industry sales are the same, is it possible that we should be preparing for somewhat incrementally lower intensity, given that, you know, the customer with the higher intensity may not be there because of these restrictions?
Yeah, I don't necessarily think so. You know, at the end of the day, in the longer term, really what matters is demand for semiconductors.
Right.
Geographically, where it's supplied doesn't matter all that much. The demand for semiconductors is based on form factors. We're all buying digitization of everything going on in society. It's the consumption of that at the end of the day, matters. Geographically, if something moves from one place to another to another, we're just shipping to a different location.
Got it. One last one, kind of more semantics than anything. If Lam's view is kind of low to mid-seventies, and your competitors are saying mid-seventies, is that just an end market difference, right? Like, what is that just noise, and you're reading much into that?
I don't read too much into it. frankly, there's. None of us get this perfect. we generally know our markets pretty well. In fact, perfectly is too strong a statement, but pretty well. we estimate the stuff that we don't participate in, like litho and CMP and ion implant and whatnot.
Right.
We don't always get that as correct as we get our own market. I think there's a little bit of noise from that.
Right.
Frankly, I think everybody would say the same thing. This year, because there was some level of, like, deferred shipment that spilled into this year, it's maybe even a little bit more confusing what the hard number's going to be.
Right.
The important thing to understand, though, at the end of the day, the market is the market. It'll be what it'll be.
Got it.
It's not gonna be different from one company versus another. It's the same market.
Got it. Then on the deferred revenue, you know, kind of unchanged at that $2 billion. What was your original plan when the year started? Like, where do you. Is this a new normal for deferred revenue now, and why is that? Why has the normal-
Yeah.
Level changed for deferred?
It's probably a little bit higher than it used to be, but it no longer has anything to do with all of the, like, deferred shipments that we had going on. When we came through the supply chain, we had a phenomenon where we couldn't get everything we needed, and we were shipping somewhat incomplete systems to the customer so that we could begin installation and qualification, and then we would ship the final components to the customer fab and assemble it and then turn it over to the customer. That created a level of deferred revenue, 'cause if the tool is incomplete, you can't recognise the revenue.
Right.
We always have a little bit of that, but it went well beyond what was normal in terms of how we generally like to be running the business. It's what the customers wanted, 'cause we could start the installation process sooner.
Right.
As we exited the March quarter, that phenomenon largely is back to normal, meaning there's not some kind of abnormal volume of these incomplete tool shipments out there anymore. We're caught up. We're back running the business in a normal fashion. However, what occurred in, also in the March quarter is deferred revenue stayed flat. What offset that reduction in these incomplete tool shipments was cash in advance payments, which oftentimes, when we have a new set of customers who haven't done business before, they pay us ahead before we deliver and ship the tool. That offset it. We have several new, relatively newer customers-
Right.
-that are in that bucket. The way I've described that is, yep, that is where it is. I believe the majority of that revenue is by the end of the calendar year. It's maybe still a little bit elevated. It'll come down to a more normal level as we exit the calendar year.
Got it. Next, on memory. Would you agree that this is kind of the biggest X factor for when WFE really starts to recover? If yes, then, you know, as we sit here today, what's, what are the, you know, typical, you know, landmarks you're looking for to suggest when that investment comes back? Sure, it's, you know, 70% off of highs, we said, you know, when it was 50% off of highs, it's, you know, down a lot.
It's still.
70%.
It's still...
So.
listen, here's Chintan, you guys can pay attention to a lot of this stuff, too. First is inventory, right? You parse everybody's earnings call, look at their balance sheets. Is inventory coming down in the memory space? That's the first thing that has to happen.
Right.
There's a lot of inventory out there, more maybe than I can remember seeing, for the reasons we've already spoken about. As that begins coming down, that's essentially a reduction in supply, right? Econ 101, supply comes down, pricing gets better. That's another metric you should be paying attention to, is where's memory pricing? It's still going down. Maybe the rate of decline is getting a little bit better. Those two things are important. That then shows up in my customer's P&L and their cash generation. When that begins to happen, the way we're gonna, I believe, the way it will show up in our business is utilisations. The first thing that will happen is utilisations will come back to a more normalised level. We will see that in our spares and service business.
In the customer support business group, a component of it, in spares and service, will begin to get better. That's when, if you listen to our description of what's happening, that's when you'll know it's starting to show up at Lam. First, utilisation will come back up to a more normalised level. The next thing that'll happen is, you know, there's a large installed base out there in memory that's not being upgraded right now. The node conversion, the pace of node conversions has slowed to a certain extent. That equipment we shipped in 2021 and 2022 is not being upgraded. When we get past the increase in utilisation, the upgrades will begin to happen. That's also in our CSPG business.
That's the next thing you'll begin to hear us talking about whenever this does begin to get better, is, "Okay, the upgrades are happening again." Then eventually, you get to the point that new wafer capacity begins to get put in place again. I am very hesitant to communicate any time frame for any of this, because I, frankly, I don't really know. My phone's ringing, sorry. That's how it, I know it'll show up, because that's how it always shows up.
Have you seen any green shoots, any changes at all in behaviour in those?
Nothing I'm seeing right now.
How much of an early warning system is that, right? In that, let's say if you're not seeing those green shoots today, that doesn't mean, like, effectively, you know what the situation is for the next six months? Or do you think that, you know, memory, you know, when things change, they change very quickly, so things could change three months from now, right? Or you think that that's too optimistic?
Yeah, I'm being super hesitant on a time frame. I'm just not gonna engage the conversation because frankly, I don't know for sure. Here's what I do know, though. I've been in the industry 30 years. It's been a while. I'm kind of long in the tooth here in some ways. For every downturn, the upturn always comes. The pace and the shape, I don't know.
Right.
I do know it will happen, because it always does.
Got it. One other behaviour, right, we have seen that, some of the memory customers have, written inventory off, right?
Mm-hmm.
Before they increase utilisation, is it possible they first sell that?
Very possible.
Because it's
Yep, very possible.
zero cost, right? Their shareholders have already paid for it.
Very possible.
They first sell that, and then they start increasing utilisation.
Mm-hmm.
You start getting the orders.
I think that's what will happen.
That's what?
Yep.
Got it. Okay. You see any differences in behaviour, NAND versus DRAM, or you think that's pretty much the same?
It's not all that different. Frankly, three of my customers do both.
Right. It won't matter.
Not all that much. They're not in exactly the same timing, cadence, pricing-
Right
... inventory is a little bit different between the two, but it's not terribly different.
I see. On lagging edge, like, as you must have heard, that, you know, there is a concern that there is pull-in, there are over-investments, right? If you could address that both in terms of China versus non-China, what part of lagging edge investments and growth that we are seeing this year is sustainable, and where is it kind of unsustainable?
Yep. Again, I kind of think about this very simplistically. The first thing you have to think about when you're trying to ask-
Right
answer a question like that is: What is the secular demand growth driver? What's going on? Is there secular demand that is requiring the investment? There is in this case, right? It's, if you think about what's going on in industrial, automotive, security, analog, IoT-
Right
Is there secular growth in all that? There is. Is it a segment of the business that's probably been somewhat underinvested in over the last several years? Yeah, it probably is. That's important to get your head wrapped around. If you believe that, then it'll make more sense when you start decomposing. Okay, who's investing, and why are they investing? When I look at the mature node spending, it's pretty resilient this year. That shows up in our Reliant business, so I've described that. It's a global set of customers investing. It's not any one or another region because of the secular demand. That's important to understand. Now, there is an aspect to this that is, yes, there's a set of customers in China investing.
I would tell you that that set of customers is relatively new to some of this stuff. If you think about the level of investment they're likely making relative to the amount of output they're generating, it's very likely less efficient.
Right
... than somebody that's been doing this for 40 years. Is that over-investment? I don't know if that's over-investment. It's just the state of where a newer customer finds themselves, and they will get better and more efficient over time. That's the way I think about this, is first, is there secular demand? Yes, I believe there is.
Right.
Is it an area that needs the investment? Yes, there is. Are there unique demand drivers in here? Yeah, there are, and that's the most important thing. You decompose who's investing and at what stage are they, You got to then go decompose it sort of customer by customer.
Got it. Is there a way to quantify what is Lam's lagging edge exposure? How much has it grown this year?
It's the Reliant product line has done very, very well and continues to do very well. I've now been describing it as there's like a number one and number two size of business in CSPG. Spares and Reliant are kind of one and two.
Got it.
I haven't put hard numbers on it.
Got it. The CSPG business, right, very attractive part of the portfolio.
I love CSPG.
Very resilient. Why not start giving a backlog for that business? If it is resilient and if it needs to be valued as a subscription and software-led business, then why not provide a backlog?
I don't provide backlog for anything. It's not a business that backlog is actually all that relevant for or all that important for. I mean, yeah, the service stuff clearly has long-term contracts, but there's a lot of it that is. Just think about spares consumption.
Okay.
A lot of that is transactional, based on where the fabs are and where the utilisation is, and it's a very large part of the business. Backlog isn't really all that relevant to consumption spares. Same with upgrades.
Right.
Really not all that relevant to backlogs. Why would I provide backlog when it's not relevant for a significant portion of the business? I don't, that's why. But that doesn't underestimate the recurring nature of this second section of the business, right? I've been reminding people today, and sometimes I forget to do this, our tools actually generate revenue for decades.
Right.
On average, over an average tool sold at Lam, over the life of the tool, it generates more revenue than we sell the tool itself. We've spent the whole 30 minutes talking about WFE and where is that at, and so forth. This is an important component of what the business model here is, and how we make money, and how we generate free cash flow. It's a very recurring portion of the business. Before this year, I used to kind of run around shaking my head and saying, "It's business that grows every single year." This year is not going to grow because utilisation is down so much. I didn't anticipate that when I used to say it grows every year. It is only down a little bit.
Right.
-with utilisation down as much as I've ever seen it. It's an important statement on the quality of this business.
Got it. Look, even in a kind of a trough scenario, Lam is still, you know, putting up almost $5, right?
Just got it to $5.
Yeah.
Precisely.
Nice. Significantly above the last... If I were to look at margins, right, they're about the same as they were kind of on the trough, right? About 44% to 45%.
44% gross margin.
44, right. Your revenues are significantly higher than the last trough, why shouldn't trough margins be higher, given that your revenues are significantly higher than the last trough?
This is not a fixed cost business, and you know that.
Right.
I'm just reminding you, right? When the revenue goes up and down, so goes cost of goods sold here, and that's part of the operating model and why we show resiliency in the down part of the cycle. Actually, I'm quite proud of where the profitability of the company is performing. Frankly, one thing I've been pointing out today, and we talked about on our earnings call not all that long ago, is, frankly, if you go to the last memory down cycle in 2019, our installed base is 40% higher than it was then. That's part of the reason why earnings and cash generation, frankly, are as resilient as they have been. Gross margin will ebb and flow with volume. It just, it's the way this business runs.
The earnings resiliency has a lot to do with CSPG, which people often, I believe, and when they look at our business, under-appreciate the quality of that business. It's a great business.
Right. Got it. Just one last conceptual one, Doug, which is, you know, if in, if in the next several years, so WFE grows, but EUV and litho takes up a lot more of those WFE dollars, doesn't that reduce the, you know, SAM for dev and etch markets?
Why would you think that? I mean, you don't buy a litho tool in isolation. You don't just put the print on the wafer and then not go create the features. You got to deposit the films and etch the films to create the transistor in the structure. If you're buying EUV, you're buying etch and deposition as well. EUV and etch define the patterning module.
Right.
If EUV is doing well, we're going to do well also, right? You need it all together. Frankly, there's some new stuff we're doing, we've been talking about this for a while, right, like dry resist.
Right.
-through partnership with ASML. We're going to do our best to actually make the photon absorptions with the resist we're putting down more efficient and better.
Right.
There are also incremental opportunities for us as EUV ramps. You shouldn't think of it as it's a zero-sum game here. If you're buying a litho tool, you got to create the features, too. You need the process tools.
Got it. The last one, Doug, is just, you know, there's been some excitement about maybe CHIPS Act funding adding to fab investments, right, in U.S., and obviously, every country has their own version of that.
Seems like every region of the world is trying to do something.
That's right.
Yeah.
Doing their own.
Exactly.
Have you seen any of those benefits come through, or they're all overwhelmed by the cycle? Or are those management teams, right, because they need to partner with, you know, folks like yourself to make sure that those fabs come up in time? Have you seen any of those benefits, or all that goodness is to come, or it's not that big to begin with?
Not too much. Again, I'm often a little cautious when I describe this. Don't lose sight of the fact that the demand for semiconductors is what matters at the end of the day. If there's some government assistance, it'll move the fabs around, and if you have more fabs than fewer, it's probably going to create some level of incremental investment in WFE, but it's not just pure upside.
Right.
Demand for semiconductors is what matters. Frankly, that's why I'm so excited about this industry, because demand for semiconductors in the longer term is going to be extremely strong, I believe.
That's the most important.
That's the most important thing. Don't lose sight of that.
Right. Excellent. Thank you so much, Doug. Really appreciate your time.
Thanks for having us.
Thank you. Thanks.