Applied Materials, Inc. (AMAT) Earnings Call Transcript & Summary

Great. So let's get started. I'm Joe Quatrochi, the semicap and [ TMT's ] analyst here at Wells Fargo. Excited to have Applied Materials, Brice Hill, CFO. Join us this morning. Thanks for joining us.

Thanks for hosting, Joe. Great venue. Happy to be here, and thanks for inviting us.

Awesome. So maybe to start, right, there's a little bit of news this week. China export restrictions, anything you guys put out a short kind of release earlier this week, just kind of addressing that stuff, but can you maybe [indiscernible] on the takeaways and just kind of how to think about the impact there?

Well, certainly, yes, certainly not unexpected. There were rumors that new rules are being considered. So the timing was unknown. But I think from our perspective, the first thing on our press release, we said that we're not changing our range for the quarter. And the nuance there for investors is the midpoint is -- will be a little bit different because there will be some impact from those rules. There's 200 pages plus Joe, to go through. And I think from our perspective, there were pros and cons, there were some tools that we thought there would be controls on that were "demoted" that there won't be and vice versa. And so there's a lot to go through. But for Applied Materials, most of our business in China is already in the more mature nodes, what we call the ICAPS nodes, IoT, communications, auto, power, sensors, those types of products. So I think besides companies that might be put on the entity list, most of those nodes, we were already impacted by previous rules. So anyway, back to the not changing. So what we decided, we didn't need to change the range in the quarter.

Okay. And you're kind of -- I guess we didn't wait until February when you guys report as we think about like just the impact in calendar '25?

Yes. I think I think you know our practice. We haven't guided calendar '25, so there's a lot of puts and takes. And as I mentioned, the rules, there's some positive, some things that we think move in a positive direction and some that don't. And so we'll have to evaluate that. It will take some time.

Okay. Okay. Fair enough. Maybe just kind of as we move on that and just kind of -- as we look into next year, you guys just recently reported earnings, but just to kind of frame the discussion, I guess, like as you look into 2025 and you think of WFE, like is there anything you can kind of help us just kind of shape, at least like how it looks into next year and just kind of to form the discussion?

We definitely try to -- as you know, we don't guide the year, but we definitely try to give as much information as we can about the pace of the market and what's happening. All of last year, leading edge was at a low level, but it's been accelerating through the course of our '24 and into '25. And that's certainly where we see the strongest market dynamics. If you think about high-bandwidth memory, DRAM, advanced packaging in general and then leading logic. When you look at leading logic where is the best utilization across the ecosystem. The best utilization is at the very advanced node where you're making GPUs, CPUs, accelerator chips, et cetera. So we think that's where there's the most energy in the ecosystem. Last year, because leading was lower, ICAPS was very strong, DRAM was very strong. And obviously, there's a growth year for Applied 5 years in a row of growth for Applied. As we look into '25, if there's any maturity in the spending of ICAPS, then the question is, well, leading -- will that demand on leading be strong enough to grow the business. We don't guide that, but that's certainly what we've been seeing in Q4 and Q1 is still growth quarters for the company. Our Q4 growth year-over-year, our Q1 is a growth year-over-year in our guide. So we think it's that dynamic that's the strong pull in the market right now.

Okay. I guess as you maybe ask it a different way, like when you look at '25 and you think about, right, the pull for leading-edge technologies strong and accelerating next year, what's the biggest -- you put China aside for a second, but like or at least the export restrictions, what's the biggest unknown or like in that you're just kind of like we just -- we kind of need to see how things play out like as you try to think about '25?

Well, I think the biggest unknown is that ICAPS investment in China. So sorry, I didn't put China aside, but...

Well, export restrictions, strict demand, right, like...

I think for us, we see the acceleration in leading edge. The reason it was lower last year was because companies are on the cusp of moving gate all around process technologies from pilot to high-volume manufacturing. And so everybody's expecting this investment cycle. We will build out a large node on the Gate-All-Around node. It certainly got great performance characteristics. So if it meets its schedule and ramp requirements, it should be a good landing spot for lots of designs and be a large node. And so where is, the more uncertainty in the market relative to that, it would be -- we've had significant ramp in China over the past 3 years in terms of ICAPS capacity. Now when we look at that, we have a number of ways of triangulating to say, is it too high? We think the country has a goal of being self-sufficient from a capacity perspective. We don't think it's reached that level yet. The utilizations in China are at a good level. And so it will just be a question of whether that continues at the same pace or not.

Yes. I guess like on the domestic China piece, right, like you've seen there's a lot of new customers there, right? And as we move into next year and maybe there's some digestion a little bit just because they've spent a lot, right, trying to kind of ramp up their technology. Are you seeing like where there's consolidation of that ecosystem yet? Or is that still maybe something that is in front of us?

No evidence of consolidation, at least from my perspective. We see a growing customer list, and a lot of these customers to your point, are smaller. It's just a long tail of customers, some are large, but there's a long tail of smaller customers. And so the positive from that perspective is -- we've got over 30 factory projects that we see. We've got even a longer list of customers that are adding capacity in facilities that already exist. It does make sense to me that over time, I have a mental model where I say, okay, maybe there's 8 companies that are starting an image sensor business. It does make sense over time that they won't all be successful and there'll be some consolidation. I don't think we're seeing that at this point.

Okay. Okay. That's helpful. I mean you talked about maybe let's say we tend to [indiscernible] a little bit longer. You talked about kind of the normalized rate of 30-ish percent of revenue. I mean that's still, I guess, the right way to think about it. I mean it can ebb and flow. And then how do you break down the services versus -- systems versus the display business, right, that maybe is a little bit different than your peers, that I think historically has been a little bit more driven by China than other regions?

It is. So first of all, since there had been a few quarters where we had a very high percentage of our business in China. That was largely driven by the DRAM shipments that we could do to once the DRAM node was confirmed that we can ship within the rules to that. That lasted for about 3 quarters, I think our prior Q4, Q1 and Q2. And then we've sort of normalized, and that was as expected. And we call 30% normal because we sort of looked at our historical average. Longer historical average, and we think 30% is a good number. We also have a perspective that China consumes about 25% to 30% of semiconductors globally. And so it makes sense to us from that perspective if they're working to put that capacity in place. And I think with respect to the mix of the business, I don't think we really would call out significant changes in intensity between our services business and our equipment business. So I think 30% is a decent number to think of. It is true that our display business is more intensive in China. So the number of the other two is probably just a little bit lower, but I wouldn't I call that out.

Okay. Maybe one of the things that with the rules that came out Monday, right, and some of the things that have been happening in China is like they've also been trying to develop their own equipment, just ecosystem. And it feels like maybe the rules on Monday that came out could hamper that pretty significantly. Do you feel like that, that's an opportunity for you guys to maybe like kind of gain back some share that you've kind of like ceded away to some of those players? Or how should we think about that?

It's definitely a positive from that perspective that the rules may also have some impact there. And so I think what we're describing is if the local China vendors don't have access to some of the supply chain because of the trade rules that maybe they're impacted. Joe, one thing you'd have to think through in that thesis is, assuming those vendors have expected it, they've probably also taken steps to stock up on inventory and work on their supply chain. So it's unclear what the impact will be, but then backing up just to an applied share thinking perspective. The local China vendors have grown share over time, while at the same time, Applied has been growing share. And so, we understand that companies in China will try some of their applications at lower-cost hardware vendors. We understand that's part of the model. For us, we have two -- our strategy to address that, first of all, is to just grow the TAM, grow the addressable market of our business faster than anyone can duplicate a hardware offering that we have somewhere else in the road map. And that's what you see happening, the leading-edge nodes, leading-edge logic, DRAM, the applications grow so quickly and because Applied has integrated equipment and unit equipment that can address those demands. Those applications are growing more quickly than any share we would lose on something that can be duplicated from a hardware perspective. Then the other piece of the strategy sort of head on is when Applied sells in China or anywhere else in the world, you're bringing a top to bottom, and you're not just selling a hardware capability, you're bringing an experienced supply chain, high-quality supply chain. You're bringing a services business that can service the equipment and make recommendations how to keep it high yield and how to get it to high yield in the first place. You're bringing the installation capabilities that we have for the company and a road map of how to improve those tools over time. And so there's a lot that comes with that. And so we understand that customers might try an application with the local vendors for lots of reasons, including being incentivized, but we have ways to address that competitively.

Okay. That's helpful. Maybe we'll move on to China for a bit. We'll stick with ICAPS though. I think outside of China, like analog semi demand relatively still kind of bounce on the bottom. A lot of the customers that have spent a lot on tools over the last couple of years to add capacity. Now they kind of don't need to, maybe outside of one that's kind of building some strategic capacity. So like how do we think about, I guess, them returning to normalized utilization rates? And what's the right way to think about the long-term kind of growth algorithm for that part of the business? And then also I guess, like where are you seeing the areas that can still grow from just kind of secular trends?

Yes. I think, first of all, I would -- there's I would back up and say, all right, the process technologies we're talking about for ICAPS, our definition is greater than 7-nanometer or more mature than 7-nanometer. So anything that's EUV or newer. EUV starts to have anything that's EUV or newer we'll put in leading logic and then the ICAPS process technologies will be the ones that are larger than that or older than that. If you look at those, they're larger as an ecosystem, the actual factory footprint and wafer starts of capacity that are in place. They're larger than all of the other markets that we're talking about put together plus a factor. NAND plus DRAM, plus leading logic, ICAPS is much bigger than all that. All of the third-party forecasts call for those devices every single device, micro-controllers, analog devices, image sensors, power devices, et cetera, all of the third party forecasts say that those will grow at mid- to- high single digits over time as especially the EV renewable energy, all these end markets continue to grow. And we agree with those forecasts. So if you look at that large footprint, our view is, as you step back, that footprint will grow at that speed over time. Now the question is, today, where are you on that curve? You're right. The end markets have been slower, the utilization is probably somewhere in the 70s and the rest of the world. And so they can take -- there's a little bit less pressure to add capacity right now. So I think it's just a question of when does that normalize and then they get back on the path of adding 5% to 7% capacity every year. And as you know, when I think about the trends, we like to think -- we look at GDP, GDP is typically 2% to 3% from a global perspective. Every bunch -- every once in a while, it doesn't meet that, like 2009, you have a year off from that or something like that, but it's a pretty steady drumbeat. If you look at semiconductors they're usually a few points higher than that 2, 3, 4 points higher than that from a growth perspective, but it's more volatile, right? And then if you look at equipment, it's a couple of more points on top of that, but it's also more volatile, it amplifies that line. So it's difficult to forecast 1 year, but we're confident in the secular trends that the industry is going to grow to $1 trillion, all those device types will grow to $1 trillion and that large capacity profile of those more mature nodes, the ICAPS nodes will continue to grow.

Okay. Maybe, I guess, like you mentioned it, right, everyone's kind of said, look, $1 trillion, but pick your time frame, 2028, 2030, whatever, right? But I think like what -- and that hasn't changed for years -- a few years, right? But I think what has changed is the mix of that, right? GPU is taking a much bigger share of that. So I guess how do you like kind of handicap that in the growth expectations?

Yes, it's a great question because I think when people ask the question about the GPUs, they're thinking about price right? Because if you get to $1 trillion by selling 1 GPU for $1 trillion, that doesn't help the equipment business, very much. But a lot of people ask us this in the sense of intensity, what's the equipment intensity. If you take the equipment spending for a year and you divide it by the semiconductors, we have an intensity rate that we think about, it's been elevated, and it's been increasing over the past number of years for various reasons. One, because intensity increases. The second is because of the build-out in China that we just discussed. So we've sort of expected it to come down a little bit from an intensity perspective. And one of the unknowns in that equation is, what is the price movement of semiconductors relative to volume. I guess what we would say is the range actually has widened. People say $1 trillion, now we're starting to see estimates that are as high as $1.3 trillion, some are lower. And certainly, the mix toward leading edge is improving. So I'll just say there's -- it's a range, but if we back up, it's good news. We don't want to turn a good news story into bad news. The good news is leading edge logic, high-performance compute systems, the search for power efficiency and the ability to put more and more compute in data centers and serve these models. I mean there is a lot of energy there today. And sure, it might change the intensity and it might change the ratio of semi revenues versus equipment. But we're not actually -- it doesn't cause us to not sleep at night. For us, it just means boy, we need rapid innovation on power efficiency on leading-edge process technologies. We need to support our customers, building out these factories. And we need to have plans if you think of the net-zero stuff and ESG, it's very important that we drive the energy efficiency, and we have to work with our customer base on how are we all going to improve the grid and be able to serve the AI demands successfully.

Okay. That's helpful. Maybe moving to leading edge more broadly. I mean you guys have been pretty vocal about your opportunity to gain share as the industry moves to Gate-All-Around, 2-nanometer. Can you talk about just kind of double-click on those opportunities to gain share, where what steps are becoming more important dates all around and your position there?

Yes. So Gate-All-Around is a new transistor. So we were on FinFET for a number of years in leading logic. We're moving to a new transistor. The reason that's done is because that transistor is more reliable and more power efficient than the FinFET transistor. The difficulty, of course, is it's a more complex architecture built at even smaller geometries which offers Applied, since we bring these connected tool solutions. Someone described to me, Joe, if you look at a Gate-All-Around transistor, there's like 3-nano sheets, if you've seen a diagram of this. And in between those nano sheets, there's just literally few atoms thick or angstroms thick of operation space. And in between, I think we're putting 3 or 4 or 5 different types of materials in between those sheets. And so the tools required and the environment required to combine the techniques, to put those films, put those layers, put them with precision at high yield. It's very complex, and that's sort of -- that is the bread and butter of Applied Materials is, knowing the road map for many years forward with customers and then working collaboratively with them to develop the tool techniques that will build those architectures at that small level. So Gate-All-Around, it actually increases a broad variety of tool types needed to do that. I think the steps for the process, we've added -- or the industry has added over 200 steps in terms of the number of passes to get a Gate-All-Around type of processor versus a FinFET type of processor. So that's a lot of different passes and different tool types. For us, what that's done is grown the addressable market by about 15%. So we said for the transistor itself for 100,000 wafer starts of capacity, it was a $6 billion addressable market for FinFET, it's now $7 billion for Gate-All-Around. And it's actually the same math if you add the wiring for backside power delivery, if that's being employed, it also grows from $6 billion to $7 billion of addressable market, and we feel we're well positioned. We do feel we'll gain share in that transition. And part of what drives that is those integrated solutions, because of the complexity of what we just described, having tools that can work in a vacuum environment that are connected together where you never let the device see air and oxidation that comes with that. There's advantages Applied brings to that equation to deliver those solutions.

And on the integrated solutions, I mean, can you remind us what percentage of revenue that is today? And like where do you realistically see that going?

It's approximately 30% of our business. That means tools that are integrated and co-optimized and we do have -- it is accretive from a margin perspective for the company. So it's better for the customers and it's also better for Applied, to build those environments. And we've seen it grow over time. So we haven't -- I don't think we have a number that we've shared in terms of the pace, but we expect it to gradually grow as a portion of our business over time.

Okay. Okay. Maybe just kind of back on 2-nanometer. I mean, outside of the architectural changes, I think there's some excitement in terms of potentially kind of the size of the node relative to prior nodes here, any kind of help you can provide about -- thinking about that?

Sure. I can just give you my intuition. I mean, what I've heard from designers is it's a big node. I mean if you think of nodes 10, 7, 5, 10 was a small node, 7 was a large node. And I think 3 is a fairly large node. I guess my expectation is that 2-nanometer will be a good landing spot. Power, shrink, the reliability of the transistor type itself. If they stay on schedule with demand and yield, then we think it will be a large landing spot.

I mean has that changed at all over the last several months, just given that I think there's been some movement in some of the customers just planning and things like that?

If you -- from our perspective, the size of the node, and I think thank you for starting that way, because it's a great way to think about it. The size of the node is going to be driven by the markets. PCs, smartphones, data center, how many -- how much product and how many wafer starts are needed. Now whether it's the mix of foundries that serve that market, we think from a macro perspective, it doesn't affect our planning very much. So let's just say that we expected this to be a 400,000 wafer start per month node. And I don't have that expectation. I'm just picking that as a number. then we would expect that to be put in place regardless of what the mix is between foundries.

Okay. Okay. Maybe I guess maybe one last question there. I mean, tool reuse in terms of 2-nanometer. I think maybe 3-nanometer demand has been a little bit surprisingly stronger this year than expected. So can you help us -- any expectations in terms of reuse?

I don't think it matters early. So when you're starting the ramp of a new process technology, you can't use -- you generally can't use the existing tools that you have to start that ramp, because as you're describing, if it's 3 in that example, 3 is being used so you can't use those tools. So you -- generally the first part of your build-out of a new node is all [ Greenfield ], all new equipment. And then at some point, you're taking demand from that old node as designs move over, and then you can reuse some of that equipment. And we just think that's a positive. That's always been part of the leading-edge math. It's there's a planned amount of reuse, and I don't really see any difference in the dynamic there.

Okay. Let me shift gears a little bit to DRAM. You guys have been gaining a lot of share there. What's been the key driver of those share gains over the last several years?

I think there's a couple of things. One, some of the integrated tool offerings factor into that space. At a high level, we think the architectures in DRAM are starting to employ many of the techniques we've used in leading logic. And so for us, bringing that to bear in the DRAM space, has offered the company the opportunity to gain applications and gain share. And as we look at the DRAM road map, it becomes more and more vertical architectures going forward. So the company has talked about 4F squared architecture that we're working on the roadmap, that will offer Applied an opportunity to provide even more value from materials engineering perspective and then much further out on the road map is the 3D DRAM. So just the complexity of those architectures, the more 3D nature of those architectures, offers similar to the logic, it offers many more passes and many more steps of materials engineering type applications.

Okay. HBM obviously has been a big area of focus for investors and [bigger and] nicer area of growth. I mean, how do you think about like the setup into '25 after a really, really strong '24?

Yes. High bandwidth memory, another indication of this leading logic pull, and the pull of high-performance compute because those high-bandwidth memory stacks are being put in the highest performance systems right now. So first, sizing that, we think about 10% of DRAM wafer starts today are allocated towards high-bandwidth memory. And we think that's growing at a 30% rate. So a fast -- it's helped increase the utilization on DRAM in general. And last year, the company, we had started with $100 million of revenue but last year, the company delivered $700 million of revenue in the HBM area. So obviously, it grew a lot during the course of the year.

[indiscernible] only, right?

That's right. It's -- it's -- when we quote our advanced packaging number of $1.7 billion, it includes that $700 million of revenue. So when we think of '25, they're going to continue to add capacity. We don't think it will be at that same rate, because there's an initial build-out and then you just start adding capacity. So there's sort of that first jump off in '24, we'll see it just continue to grow at '25.

Okay. That's helpful. Maybe on the flip side, right, like a lot of conversion of conventional DRAM capacity to HBM to basically, it's also helped improve utilization rates. I mean, in the discussions with your customers, like how are they thinking about conventional DRAM capacity looking into next year, I guess is that an area that are starting to maybe do some planning in terms of adding capacity?

Oh, I think so. I think we've tried to call out that the DRAM market has just continued to be strong even after those quarters that we shipped that China demand, it continues to be strong, and I think it is from the pull that you're describing. And the DRAM market continues to add capacity footprint overall. It's a little bit different from NAND. NAND it's approximately the same wafer start capacity footprint mostly advances by upgrades. DRAM also advances by upgrades, but is also continuing to add to its wafer start footprint. And that's what we see in the forecast and we called out in our Q4 and our Q1 even after those China shipments, it's still strong from a DRAM market perspective.

Right. Anything to worry about in terms of like fab readiness, footprint, whether like clean room space?

Well, it's a good question. So I'm not very close to those -- to the NAND -- or sorry, the DRAM footprint. So no, I don't -- I'm not aware of a road block.

Yes. Okay, okay. That's helpful. Maybe shift gears a little bit. I thought this past quarter, on the gross margin side, you guys had some really positive things to share. You're talking about 48% kind of gross margin baseline. I don't think many people had expected that to hear you guys talk about that at least maybe not this quickly. Can you talk about just kind of, how you got there? What are the puts and takes of it? And where can we go from here?

Yes. I think it's been gradually improving. So at least from my perspective, if you go back 2 plus years, the company was operating at close 40% or just over 40%, and so the supply chain issues and COVID coming off the back of COVID certainly caused a depression in that gross margin. And so what you've seen over the last couple of years is we nailed the low-hanging fruit relatively quickly. Like when I first came to the company 2-plus years ago, a lot of things were being air freighted and expedited and you have all these kinds of extraordinary costs to deal with the environment. That was relatively low-hanging fruit. Since then, we've been doing the harder work of continuing to reengineer components, shaking out inventory. If you look at our inventory on our balance sheet, it's at a much better days of inventory position than it has been over the prior years. I feel really good about that. Along with that comes excess and obsolete material, a great job of managing those costs down. All of those are benefiting. And then on the price side, so a lot of cost drivers. On the price side, it was also a factor, although less of a factor. And one thing I highlighted is that I expect our pricing -- our ability to improve pricing to mature as we go forward. We've been working on that process, and that will become a more meaningful benefit in our gross margin trajectory going forward. And so -- and lastly, just to reiterate, we had said we think our -- our guide for Q1 is 48.4% gross margin, but I didn't want people to think we were at 48.4%. It just happens to be a rich quarter from a mix perspective. So it's better to think of it as approximately 48%, and then we're going to continue to work on improving that.

Okay. I mean when I look out 3 to 5 years, I mean, do I dare start to say gross margin starts with the 5?

Well, we're not going to give that guide. But I'll just say that we're committed to continue what we've been doing. Yes, I have to try.

Maybe shift gears a little bit on the AGS side, you guys have seen nice reacceleration as utilizations have increased. Can you talk about this, the confidence of low double-digit growth kind of looking into next year still is the right way to think about it?

Yes. Well, again, we didn't guide the year. But when you look at our Q4 and Q1 guide, we're certainly growing at low double digits. And we think from a macro perspective and a through-cycle perspective, that's the right way to think about that services business. The installed base continues to grow every single year. Our attach rate, if you will, to that installed base is growing also, because there's more service intensity to the more complex tools and because a lot of the customers, maybe because of global incentives are building in newer places, and they're going to be more interested in working with Applied and getting our labor into their factories to help them ramp their factories and keep them at high yield. So we think there's a few tailwinds, and we're providing more service capability, Joe, in terms of -- if you think about AI, we've got this fleet of tools. We've got the ability to see how they operate and help the customers tune them so that they're operating at the highest levels. That is sort of increasing our service offerings and capabilities as we work across the ecosystem. So we think that's a good long-term prospect for that business, demonstrating it in Q4 and Q1. And more and more of that business has a subscription recurring revenue type flavor. So we talk about how 85% of the revenues approximately are recurring business, meaning either spares or service contracts. About 2/3 of that is under contract. And those contracts are at a high renewal rate and actually our being renewed for, on average, longer term as it helps us with our planning and it helps our customers with their planning, if those are longer-term service agreements. And then from a finance perspective, that gives you a little bit more of a recurring revenue flow, a little bit more confidence in that. And so we've also sort of attached our sizing of our dividend expectations to that flow.

Yes. That's kind of perfect way like 30 seconds left. But like capital return, thinking about that dynamic and that's exactly what's going is services and the dividend, right, growing that dividend. And then I think maybe there's a little bit of CapEx kind of requirements over the next several quarters, but just kind of briefly talk about capital return.

Yes. So no change in the overall plan, return 80% to 100%. If there's little M&A in the environment, it's close to 100% being distributed to shareholders. We have been on a path to raise the dividend. We are thinking of that related to that recurring business that we just talked about. And then you mentioned CapEx, we are expanding our lab footprint so that we can do this collaboration work with our customers. That's a project primarily in Sunnyvale, where you see that. And that will elevate our capital spending for '25 and '26.

Okay. Perfect.

All right.

Leave it there.

Thanks very much.

Thank you.