Of Citi Global TMT Conference. My name is Atif Malik. I cover U.S. Semiconductors, Semiconductor Equipment, and Networking E quipment Stocks at Citi. It's my pleasure to welcome Brice Hill, Senior Vice President and CFO from Applied Materials. We also have Liz Morali and Marty Parker from IR team here. Brice is going to open with a few remarks, and then we'll dive into the fireside chat. I'll go first, ask my questions, and then give you an opportunity to ask questions. If you have a question, please raise your hand and the mic will come to you. Brice?
Great, Atif, thank you, and thanks for having us here, and nice to see everybody. You know, we just had our earnings call for our Q3 not too many weeks ago, and one of the things we pointed to was just a lot of energy around the ecosystem with respect to AI and AI data center. We've talked about data center for a while, because data center is probably the fastest end market in terms of wafer starts that's been growing, and what we've seen over the last three quarters is very strong DRAM, specifically high-bandwidth memory, that goes with some of these high-performance systems. We see an acceleration in the advanced packaging, especially equipment for high-bandwidth memory again for those systems. We see our leading-edge business beginning to pick up.
That's been accelerating as we go towards the back half of the year. Applied talks a lot about gate-all-around, the next transistor technology, and how we're on the cusp of moving that to high-volume manufacturing across the ecosystem. So just a lot of energy around those leading-edge systems and the growth. And in fact, when we listen to the ecosystem and what's happening in the ecosystem, a lot of CapEx investment going into data center and specifically AI data center. In fact, all the major companies were raising their forecasts pretty significantly. So a lot of energy on the leading-edge. At the same time, we've talked for about a year on how strong, you know, what Applied calls ICAPS, our IoT, communications, auto, analog, energy , sensors, all those types of devices that go in the more mature technologies across the world.
That's been growing very strongly for the last couple of years. And while we had a little bit lower leading-edge investment and a lower NAND investment, ICAPS, those mature technologies, have been very strong. That's been led by China in a lot of that growth, but it's also a global function. And so we also see a lot of innovation, and we're investing in innovation, even in those older technologies. So when you think about the industry, we kind of tie it all together between the AI and the data center-driven cloud ecosystem, plus that IoT ecosystem, which is sensors and data-collecting devices. Those things are connected in our mind, and there's a lot of energy around those things.
For us, again, just, you know, Applied has continued to have good results and growth during the course of the year because we have exposure to those markets, some of the fastest-growing markets. That's the way we orient the company, is by investing in the inflections that give us access to the newest technologies in those fastest-growing markets. I know we'll get into that, but you know, that was a flavor for us in really the last couple of quarters.
Great. Brice, two words that we hear a lot from Gary, who's a product maverick, and also Prabu, are technology inflections. Applied has been talking about materials engineering playing an increasingly important role in many of these technology inflections. You guys always keep us busy with your Master Classes on gate-all-around, and backside power, and HBM, advanced packaging, and whatnot. Can you share your views on the role of materials engineering in these inflections? And also, if you can provide some insights, as a CFO, how do you guys stay ahead of the curve, where these inflections are going?
I think that's a great question, and part of it really leans into you know the fun of working in the semiconductor industry. You're always working on the future. In fact, we like to say, "Make the future possible." We're envisioning what's five years out and what's 10 years out. And if you think about a semiconductor, of course, we love semiconductors. If you think about a semiconductor, you know, a chip the size of your thumbnail with 10-20 billion transistors, with that many functioning devices, if you look at. If you take one of our Master Classes and you look at the architecture for a gate-all-around transistor, one of the things you'll see in there is three nanosheets. Those nanosheets are built, they're only a few atoms tall each.
They're separated by different materials, and those layers themselves are only a few atoms thick. All of these materials have to be built perfectly. They have to be functional. They have to operate in different temperatures. And if you think about the layer upon layer that's used to build these devices, you have to have the ability to deposit materials, shape materials, treat them, put films down, measure them with a degree of complexity, with, like, the eBeam inspection tools. So it's really incredible as we get smaller and smaller and smaller. You're talking about engineering and construction really at the atomic level, and that's the fun of being in this industry. And when you ask, you know, "What's the role of materials engineering?" The semiconductor advancement used to be. I won't say easy, but it used to be routine. Companies would look for-...
A shrink that was enabled by a new litho capability, and that has slowed down a little bit. Companies are looking more and more on the leading-edge of how do we change the device types? When you change the devices and you make them more power efficient and more capable individually, you need to be smaller and smaller, and that means new materials and new capabilities. That's the fun of being in this industry. Financially, that also means staying dedicated to R&D. Applied Materials and, you know, like companies, do fundamental R&D on materials research, and then we also do the development of the equipment that can go into high-volume manufacturing to make these transistors and make the wafers.
And that's also part of, you know, what excites me about the industry is you do the fundamental R&D, you do the materials technology, then you do the development of equipment that can build at that level, operate at that level. Then you sell it to customers, and then you service that and help the customers keep that equipment operating at top yields, ramping quickly. And that's also, you know, a part of the business that's fun, where we can collect data and provide insights as to how to do that across the ecosystem. So the role is really inventing the future.
All right. Let's start with one of the technology inflections, advanced packaging, which many are calling the new front end. You know, back end has always been more kind of commoditized, not so kind of leading-edge, but you know, tools on the front end are increasingly being used on the back end now.
Mm.
Because of the requirements on packaging for AI chips. So on advanced packaging sales, including HBM, you're expecting it to grow to $1.7 billion plus this year, for $1.1 billion last year, and then expecting to double in the next several years. What are the main drivers, and how much of this growth is coming from HBM?
Yes, so this year, almost all the growth comes from HBM. It's $600 million from $100 million on the HBM side, so a lot of excitement around the high-bandwidth memory, the stack, you know, stacks of high-bandwidth DRAM that are in some of these high-performance computing devices. The way I would think about it, Atif, is, it's really the search again for performance. And so these chips that companies want to stack, the reason they want to stack them or put them close together is, if they could build them on one piece of silicon with a, you know, silicon interconnect, they would, but you can't put it all on the same silicon. So instead, the closest thing they can do is stack them or put them right next together on another substrate.
You know, hybrid bonding would be a good example of this, where today, when you stack packages or you stack silicon, there's micro bumps that are connecting those silicon in a stack. At some point, we run out of gas on the micro bumps, and you need something that's even smaller and will be even thinner, so you can stack more of those chips, and hybrid bonding is a technology that will do that. It's really the search for: how do we continue to get more density in terms of everything that's in that silicon? And the advanced packaging allows us, allows the industry to figure out, how do you put these chips together, and how do you make them operate as if they were on the same piece of silicon?
You do that by connecting them through, you know, a substrate that is silicon, a substrate that might be glass. There's all sorts of different investigations right now as to what the best interconnect technologies will be, but that's allowing companies to basically design a much larger system than if they were only able to do what can fit on one chip in the reticle. Does that make sense?
Yeah. And Brice, a lot of our investors want to kind of simplify things, and, you know, they ask us, like, who’s the most exposed to DRAM, who’s the most exposed to NAND? And I think they’re just trying to play the different end markets. And mathematically, your share in the DRAM market, you know, is stronger than any of the peers in the U.S., and some of that has been kind of secular shift to high-k materials in traditional DDR4, DDR5 capacitor-type architectures in the past. But when we look at HBM, can you point to maybe a couple of products or areas where you have very strong leadership and where you’re kind of dominating?
I think for us, there's the way I understand that is, if you take a DRAM process, it may have seven hundred steps. That's an example that we've used before. And if you want to add in the HBM stacking that happens at the end of that process, then you add approximately nineteen steps. And Applied provides the majority of those additional steps or materials for the majority of those additional steps. So we're super excited about HBM. We're super excited about DRAM. One of the things we've been talking about is the DRAM technology itself will continue to, you know, advance. There's inflections in DRAM with new process capabilities that, from our perspective, will be more and more 3D and, you know, use more and more give Applied more opportunities to win steps in those new processes.
I'm talking about 4F squared is one that we've talked about, 3D DRAM is another. These are out on the roadmap, but they'll allow Applied to advance its available market as those technologies continue to evolve. And I think that's just a good example of what's the strategy for Applied. The strategy for Applied is to put our R&D in collaboration with our key customers, so we're working together, which to me raises the confidence and the success of those R&D dollars on the technologies that are five to 10 years away. And then we basically co-develop the solutions that will go into high-volume manufacturing as those new devices or new architectures are being developed.
That's the strategy, and so if we, if we're successful, and we pick the right inflections, and those are in the fastest-growing areas of the market, then that helps Applied grow faster than the semiconductor industry. And if you just step back one step, we think the semiconductor industry has been growing at about 3x GDP over the long term. So if GDP grows at 2.5%, we would say semis should grow somewhere in the order of 7.5%, and then Applied should grow faster than that because we'll be investing in those inflections and hopefully have an outsized share opportunity. And because our services business grows faster than the core business, that should allow Applied to, you know, have a faster growth rate than what is already secularly positive with semis.
Great. Just staying on that collaboration theme, you guys have been working with BESI, and hybrid bonding is something that's to us, it's as big of a deal as EUV lithography was for the industry. So, but it's a little bit unclear in terms of the timing of it. Can you talk about the ramp in hybrid bonding? And you guys, you know, at SEMICON West had kind of a picture of a configuration that you're looking at, and if you just pull the curtain a little bit and help us understand where is the ramp for hybrid bonding.
We still think it's over the next two to three years. I think that's the way we describe it, Atif. It's going to depend really on the need, at least from my perspective, and that goes back to if you're a system architect, and micro bumps or, you know, some lesser technology is still going to be able to satisfy the system performance, then you're not pushed towards hybrid bonding. But some hybrid bonding is in production today, and we expect it to, you know, begin to accelerate over that horizon.
Great. Then shifting gears to gate-all-around and backside power, you guys have talked about Applied's SAM growing from 6 billion to 7 billion per 100 wafer starts, and you capture 50% of that SAM. Just help us understand your capacity kind of assumptions for gate-all-around ramp. Are you engaged with, you know, all the four major fabs? Is it more of a linear ramp? Anything on that?
Yeah, it's, it's so interesting. We are engaged with all the major players across the ecosystem, for gate-all-around technology. You know, we said that, we shipped $2.5 billion, or are shipping $2.5 billion of, equipment destined for gate-all-around application in 2024. We said we have the opportunity to double that in 2025. That would suggest, you know, that's, just by math, that would suggest you're getting towards, 100,000 wafer starts of capacity. But, the way I would think about it is, you know, the ecosystem is in pilot right now. We're moving towards HVM, equipment orders and installation, and that's why you see the orders starting to accelerate. And we articulated, you know, in our earnings that the leading-edge buys, were accelerating through the course of the year.
And so that's the way I would think about it. We don't talk about any specific customers, but it's moving from pilot and, you know, pre-product testing into HVM plans, and you need to be ahead of HVM, high volume manufacturing. You need to be ahead of high-volume manufacturing with the equipment buys and installation. So we're sort of in that period right now, where companies are moving in that direction. The other thing that, you know, to think about is, it's you can't just take the products off the existing technology. So if you have products running on the last FinFET node, you don't just move them in one day to the new process technology. We'll use three nanometer as an example. Just move them from three on two and turn on two. That's not the way it works.
You have to put a certain amount of 2-nanometer capacity in place that's ready to accept the new volumes before you can unload three nanometers. So there is some sort of overlapped investment across the technologies as you move forward, if that makes sense. So a certain amount of that gate-all-around capacity will be put in place for initial ramps. And you know, I think you have to ask the companies that are doing that, how much they're putting in place.
Yeah. Brice, this question is coming up a lot in my conversations with investors. But kind of big picture, when you look at the foundry landscape, if you know, if a certain foundry is lagging and one foundry is kind of leapfrogging the other one, from an equipment perspective, is this a bit of a zero-sum game for you guys? It doesn't matter who adds capacity, you still spend or... Because one of the MPU makers, they cut their CapEx before your earnings, and we did not see an impact on your results or the outlook. So help us understand, and you've been with Xilinx and in other roles in the past, like, help us understand from an equipment perspective, is this a kind of a zero-sum net neutral game?
We definitely, the way we check ourselves on our own demand is at the macro level. So we think about, you know, we talked earlier, what's driving wafer starts? Smartphones are driving wafer starts, PCs are driving wafer starts, data centers are driving wafer starts, and of course, all the mature technologies that we talked about. So we look at those end markets and calculate the simplest way is just by intensity. What's the amount of equipment you need relative to those end markets of silicon? We check ourselves that way, and I think, Atif, what you're alluding to is the way we think about it, is we're not sure which foundry vendors are gonna make that supply.
And we're certainly hopeful that all of them are successful because more competition is, you know, better for everybody and drives the beat rate of the industry. But, you know, I guess we would be, you know, we're not surprised if the levels of equipment that we sell are constant or relatively constant, no matter which foundries are providing that capacity. So I think what you're suggesting makes sense to us. And then going back to our earnings, yes, the way it works for us, and when we give our outlook, we have just recently, you know, been in touch with all of our customers.
So I think I said on the call, somebody asked me a question and said: "Are you, you know, are you changing your outlook because of any changes in the ecosystem?" We're constantly getting updates, so our foundry customers, you know, update us routinely, and they're always changing their capacity envelope. You know, they have their own customers. The volumes are changing, expectations are changing. So this is a, you know, sort of a routine situation, and for us, our outlook for gate -all -around and our outlook for leading-edge didn't change very much and hasn't changed very much.
Great. Let's talk about ICAPS. In July quarter earnings, you used the word robust many times to describe ICAPS demand. And is the strength or stability being driven by China, or do you see strength outside China? And can you remind us roughly how much is ICAPS within semi sales, and by region?
Sure. So ICAPS, you know, I started off earlier, for Applied Materials, when we talk about ICAPS, it's the sale of larger geometry equipment. So for us, the dividing line is 10 nanometer. If it's 10 nanometer or older, if you will, then that's going to fall in the ICAPS category. And this is like image sensors, microcontrollers, power chips, analog chips. Really the, you know, horses of semiconductors across the ecosystem that isn't leading-edge. ICAPS is about, has been about 50% of our logic business. It's more than 50% at the moment because it's continued to grow and leading-edge was slower. But our long-term model is that it'll be about 50-50 between ICAPS and leading-edge long term.
And we just think that this is a very powerful, very global market. We would go back to, you know, all of our third-party research says this is going to grow mid to high single digits. This has not only been a China phenomenon. We have multiple geographies and regions that have been growing, and certainly over the last three years, they've all been growing. And so, you know, we think that utilizations have been improving. We think that the outlook through cycle outlook is very positive, and so this will be a very important business for Applied going forward. The last thing I just want to say about that is, we do focus, even though it's, you know, quote, unquote, "lagging edge," we do have a significant amount of R&D focused on innovations in the ICAPS space.
If you think about electric vehicles, you think about renewable power, there's you know a similar search for what are more power-efficient devices, power-efficient materials that can be you know provided in this space, and companies are you know exploring. I said robust because I think it'll be a record year for us in ICAPS for Applied Materials' fiscal year. You know, it was great to be exposed and serving that market over the last couple of years. Significant growth.
Great. Another topic that comes up a lot with investors is China, and I feel like you guys have managed expectations very well this year. It started at, you know, high kind of forties of sales, and then you've been saying low thirties going forward. You know, do you expect this low thirties to be kind of the normal rate of spend for China, or what are the puts and takes there?
Yes. Thank you, so we do expect 30% is kind of normal for us from China, and we get there two ways. We looked at our history, and for us, this is 30% inclusive of all of our businesses, so inclusive of services and inclusive of display. But 30% has been, you know, probably the average over many years for Applied. And then we think about it from another perspective. When we look at consumption, end market consumption of semiconductor devices in China, we think that's also around 30%. So, you know, if our business is largely the ICAPS business in China, that mature technology business, because we can't ship to leading-edge, we're not really shipping to memory.
If you think of it from that perspective, then, you know, 30% of the ecosystem sort of makes sense, from a volume. So that's how we get to that number. And to your point, the reason we were elevated a, above that for several quarters, there was a DRAM technology that we were allowed to ship to, and we were able, over the course of three quarters, to make those shipments to DRAM customers in China. And then it, it's not zero, but it's closer to zero as we, finish this year. Whether there'll be another DRAM technology that we can ship to, we don't know yet, but, for us, primarily, it'd be an ICAPS market, and we think 30% is the right number.
Great. Let me pause here and see if there are any questions in the audience. Hi.
Hi, thank you very much for taking the time. We're just wondering, you guys very helpfully, kind of, on calls, break out, like, you know, content wins that you're gonna have, like, say, for gate-all-around and things like that. Would it be possible, I know there's a lot of moving parts, but what do you currently expect in terms of kind of like the net impact? Because as, you know, as production switches to, like, 2 nanometers, some other nodes get transitioned out. Is that, do you think, like... How do you think about, is that a meaningful impact? Thank you.
Yes, thanks for the question. So when we think about gate-all-around, I think the or any next technology, I think what you're asking is, you know, is it purely incremental, or does it replace some of the FinFET business that we were selling? And certainly, it's a combination of the two, so it's both. It grows our SAM. What we're saying is, as you ship, you know, tens of thousands of wafer starts of capacity of gate-all-around, we'll have a larger number of tools that are required from Applied to produce those wafer starts, so the share grows within that shipment. And sort of to your point, that means that customers aren't buying the prior FinFET technology. They're investing there instead. So it's not purely incremental.
It does replace some of the existing business, but what is incremental is sort of the added share, and then if there's larger volumes in the market in general, then that also grows the TAM for the business. 2-nanometer and gate-all-around technology, we're certainly very hopeful that that will be a strong node for the industry. There's performance improvements. I think it's on the order of 20%-30% in gate-all-around. That's, there's also density improvements related to the process. So we're hoping that's going to be a strong landing node for the industry, and it'll be widely deployed, and that's, you know, TBD as we look at the product plans for 2-nanometer. But, you know, finally, it's just a good example of these new technologies.
We'll deploy new device types, in this case, gate-all-around, and then eventually backside power, and those are the things that we're working on to enable.
Front.
Hi, Brice. I think the NAND business is about as close to zero as it's ever been. And I think it peaked back in 2018 with a lot of the capacity that was built out for 3D NAND. Can you talk about any of the potential of inflections there, or what sort of catalyst could cause that market to grow again?
Yeah, it's a good question. First of all, our long-term view on NAND would suggest that there will be an improvement going forward. Can't say exactly when, but our long-term view of the market from a wafer perspective or an equipment perspective would be that two-thirds is foundry logic and one-third is memory, and the memory would be roughly split between DRAM and NAND. I think what's happening in the short term is that there's been so much productivity in the bit production on a wafer for NAND that it's keeping up with the bit demand.
So in other words, you know, if you have, let's say you have 2 million wafer starts of NAND technology, that same 2 million wafer starts will provide enough bits to serve 20% growth in the bits because their process node has evolved and is giving more bit capacity as they've moved up the technology. And so you really haven't needed a lot more wafer starts or a lot more capacity. It's mostly upgrades, and I think that's slowed the investment on the NAND side a little. But long term, we expect at least, you know, I'm not a system architect, but I don't see any lower need for storage or any lower need for that type of memory. So long term, we expect that'll be an important market.
Okay. Brice, let's talk about AGS and services. What are the different drivers for you to get to low double-digit growth for the AGS business?
Thanks for bringing that up. You know, from a CFO perspective, definitely, really focused on the service business. I like to talk about how the profits from our service business, they're very stable, we'll talk about that in a second, and they're more than enough to pay our growing dividend. We raised our dividend 23% last year, 25% this year, and, you know, our outlook for our services business is low double digits. We signaled our intent to, you know, two years ago, approximately double our dividend, and it was based a lot, largely on the strength of that services business and the sustainability of that services business, which has a lot of recurring revenue. But that low double-digit outlook, you know, where does that come from if semis are only growing at 7.5%?
The first thing is, every time we ship a tool, it grows the installed base of tools that we can service. So, you know, think of that. We have fifty-some-odd thousand tools in our installed base. Every day and every year, we're adding to that installed base. Companies around the world are being incentivized to go in different places with their fabs, and they're more interested, in a lot of cases, in using our services because it's difficult to get the labor, and it's difficult, certainly, to get the trained labor. So our attach rate to the equipment that we're selling is higher than it was in the past. And the last thing is, the equipment is ever more complex, requires more spares, requires more information, and requires more service to operate at that top level, and that also grows.
So when you add those three factors together, you get low double-digit growth rate for us. That business, you know, we have a bit of 200-millimeter equipment that's in that business, but 85% of that business we think of as recurring revenue. So those tools are operating every day. They need to be serviced, they need new parts. That's recurring revenue. About two-thirds of that is under contract, long-term contract. We disclose, you know, 2.8 years, I think, for the contract length, and we have a very high renewal rate, over 90% for those contracts. So, you know, for people that like the attraction of a subscription business, that like the attraction of recurring revenue, it's an element of our business that provides those characteristics. And again, when you look at the profits for that business, very sustainable.
Even in the last couple of years, you see, you know, steady growth in that business. So we use that sort of to inform our ability to pay dividends over the long term.
Great. And then, just touching on the display market, which has been dormant for quite a few years. There was news yesterday that Apple could be using OLED material on all their iPhones in the future. I'm sure that's positive for you guys, but we're just looking at inflections in the display market, foldable phones, you know, there's talk about those maybe in 2026. Like, and your lead times are very long, they are 12 months. Are you seeing anything in the funnel that could be exciting, maybe on a steady kind of recovery?
I think, Atif, you sort of hit it. Definitely the, you know, what we think of as the OLED wave. OLED being deployed in PCs, notebooks, phones, that's really the next move for us. Possibly eventually broader in TVs, but the very next step will be those devices: PCs, notebooks, tablets, and already in phones. And so if it happens in notebooks and tablets in volume, then there needs to be a significant amount of investment in the manufacturing equipment in OLED, and we provide several technologies in that space. We think we're well positioned to serve the market from those investments perspective.
You know, in the past couple of years, when LCD has been smaller and the demand for equipment has been smaller in the display business, we resized the business from a spending perspective. So we feel, you know, like we're well positioned, even economically, if the business expands and the top-line perspective, that our profits will expand in that business. Last thing I'll say is, you know, it's a positive return for Applied, it's a positive cash flow business, so really for us, it's the option on that OLED growth.
A question on gross margins, if you can remind us what your long-term target model is, and or, should we even looking at it and just focus more on operating margins? Because there's so many moving parts in your gross margin, the China mix and the services, and so just kind of remind us how you're looking at those two numbers.
I'm glad you brought up the moving mix, because, of course, when we talk about gross margin, we talk about it typically at the company level, and I've said we're trying to get to 48% next year, or higher, which we're sticking with that goal. The moving pieces, to your point, if display or services grow faster than the equipment business, then that's a headwind on our overall gross margin. So you really, you know, if you're trying to tease out what kind of progress we're making in the semi business, you really have to keep that in mind. We may be making more progress than is visible on the bottom line. But anyway, we're at 47.4% today. We're actively investing in cost reductions that, you know, take a period of time to implement with our customers.
We're actively working on our pricing for value. That's been the case for last year, which is why we've improved our margins over last year. We'll continue to do that. We plan on reaching 48% at some point next year, and that's an interim milestone for us. We think margins will go higher as we continue to work on inflections and continue to provide more and more, you know, value in the collaborative ecosystem. And it's definitely a focus for the company. We are focused on that. And then as far as, you know, should you focus on gross margins or operating margins, you know, the company has excellent operating margins. I think the R&D is very efficient. And what I would say to that is back to the collaboration.
What's interesting to me, you know, coming to the business is the R&D isn't sort of solutions looking for some way to implement or some market to implement. This is development that you're doing with your customers that's informed across the industry. So we know what we're shooting for, we know we're engaged with, you know, the right players. They're evaluating the technology with us going forward, so I think to me, the probability of success is much higher than maybe just, you know, open-ended research that you might find in other industries.
Great. And just quickly on capital allocation, how do you think about dividend versus buyback?
We do both dividend and buyback. We just talked about how we're raising the dividend, and we're confident that we're going to raise it going forward. Tied that to our services business and the profits of our services business. Not directly, so we won't calculate the dividend based on an algorithm from our services business, but just that gives us the confidence in continuing to raise the dividend. You know, we talk about distributing excess profits 80% or more to our shareholders. M&A is sort of not much of an opportunity in the large scale for the industry at this point. So ultimately, all of the excess profits go back to the shareholders, either through the dividend or through buybacks.
If you look at our history, three years or five years, approximately 80% has been distributed that way, so no change from that perspective. We don't have a quarterly cash target, so, you know, the buybacks depend on the specific quarter, but, over the long term, we'll stick with that model.
Great. We're almost out of time and wrap it up here. Thank you, Brice, for coming to the Citi Conference.
Thank you very much. Nice to see everybody. Good luck this week!