Applied Materials (AMAT) Earnings Call Transcript & Summary

Okay. We're going to get started. I'm Tim Arcuri. I'm the semi and semi equipment analyst here at UBS. I'm very pleased to have Applied Materials with us. We have Brice Hill, who is the CFO. Thank you, Brice.

Tim, what a great venue. Thanks for hosting us, and great to be here and appreciate everybody's interest in Applied Materials.

So Brice, let's just start and talk about just WFE in general. You're pointing to the first half of next year being sort of flat to modestly up in the second half, I think you think will be a little stronger, so a little more of a back half loaded year next year. What are the drivers for the second half of the year to kind of accelerate versus the first half? Is it mostly due to DRAM and the fab readiness issue?

Awesome. Well, you probably won't be surprised if I bring up AI as a significant tailwind for the industry. And for Applied Materials, people who know Applied Materials, we love materials engineering. And I think there is not a surprise that the world is just hungry for more and more energy-efficient compute. And that's what we work on as a company. And I think there's been a concept for a long time of pervasive compute. If you can make compute less and less expensive, higher performance, lower power, then you're going to find all kinds of new opportunities and solutions. AI is a great example of this. Tim, you mentioned DRAM, probably the largest grower we see in the near future is leading logic. Leading logic is -- leading-edge logic is 100% utilized on some nodes right now. There's a big pull from the accelerators, the GPUs, the CPUs that run these AI data center components. And then, of course, DRAM goes with that. So you've got a pull on leading-edge logic and then you've also got a strong pull on DRAM. And so those are the 2 components. When we look in the second half, we've been working with our customers for the last quarter on getting more and more visibility into the road map for the next couple of years. And the reason is we started hearing discussions about more capacity being put in place, and you have to prepare your supply chain if you're going to do that. It's one thing for Applied Materials to have capacity, but we have over 2,000 direct suppliers, all of those suppliers on the same page. And so we started making sure we have good visibility from our customers on what their plans are over the next couple of years. And yes, that's what we see as strong growth for leading-edge. Most of this is oriented towards the second half calendar, as we mentioned. DRAM will also go along with that. And then HBM, the high-bandwidth memory that also pulls on certain packaging capabilities that we provide a good portion of the equipment for. So that will be the strongest pull in the market. Now I'll just -- I'll give you the offset for a second. The offset that we have is the mature node technologies. Applied Materials calls those the ICAPS, IoT, communications, auto, power sensors, the components that are built on the more mature process technologies. We still expect some digestion in that market, and I know we'll talk about that more. But the strength and the growth that we expect will be in the leading-edge in DRAM.

So you would say next year, it sounds like you think it's more of a leading-edge foundry year and maybe '27 is more of a DRAM here?

To me, I think they go together. I think leading-edge will be stronger in '26. That's just what we see talking to our customers. But as you know, it's a forecast when you're talking about a full year. We usually guide 1 quarter and the practice is because there's a lot of variability in demand. These things are -- the demand schedule, people ask us, well, why is it second half calendar year? Why not more capacity today? Of course, it depends on your customer schedules. The customers have factory build teams. They have installed teams. They have products to ramp. So each customer has their own schedule as to when they can take equipment, install it and start executing and running that equipment. And so it's just customer dependent, and it happens to be more oriented towards the second half of the calendar year. Yes.

I mean I think this is my 27th year covering Applied. And I have never seen anything like what we're seeing over the next few years, whether it's '26 or '27, depends on when the fabs are ready, but we have memory companies talking about CapEx being up '26, '27 and even '28 now. So can you just talk about sort of what you see from your customers? Is this really all being driven by AI? And I know you're not going to answer this. But I mean, how -- like when you think about how high WFE can go when you plan out your business and how much cash you're going to generate, how do you link WFE to sort of what's going on in AI? Do you think that at $1 trillion in semiconductor revenue, I think you probably think it's 15% WFE intensity. So that would be $150 billion in WFE. How do you kind of think about that?

Okay. Lots in that question. I think first, it would be wrong to think that it's only AI. Our perspective, even if you back the clock up 2.5 years before a lot of the AI sensation, we would say that semiconductors are secular growth. And it goes back to what I was just talking about, the belief in the company and focusing its research material engineering for lower power solutions, higher performance solutions. We believe that, that's on its way to pervasive compute, and you'll find more and more uses for compute. And so the company believes that it's secular demand for semiconductors, secular demand growth, secular demand growth for different compute platforms. And if you back up and just look at leading-edge logic for a second, smartphones and PCs are the biggest share of that market from a device perspective. Data center has been growing the fastest. This year, it's overtaking PCs. We think it will overtake smartphones in 2029. So it is -- what you're describing with AI, it is a big tailwind for the industry. But if you took that away, there would still be -- we would still think there'd be solid growth from a compute perspective. And if you look at the ICAPS devices, our forecast for the ICAPS device is looking forward is mid- to high single-digit growth for that market also. So let's put that there and say, no, we don't think it's only AI. We don't think that if AI demand suddenly dropped that there would be a significant change in the semiconductor equipment forecast. We don't think so. Now you ask, change the topic a little bit and say, okay, what do you think about gigawatts and how much demand is being driven by AI accelerators and AI data center. And the way we think about that is approximately 15% of the leading-edge logic capacity right now is oriented towards the data center, AI data center products, so GPUs and accelerators. And similarly, about 15% of DRAM wafer starts is oriented towards supporting HBM and AI components. And that 15% of those 2 large components of the industry is growing at, I'll estimate a mid-30% CAGR right now. So if you can imagine that capacity in place, you're talking many factories and growing at a 30% CAGR, that is the tailwind that the industry is looking at. And yes, they're adding to their capacity plans because of that.

And do you have a metric? I know since Jensen has been talking in gigawatt terms and equating revenue to gigawatts, everyone's asked how to connect their revenue to gigawatts. And I know Lam has tried to put up a number and their number basically equates to something like $4 billion per gigawatt. I think it's closer to $3 billion. But is there some way that you tried to tie your -- or the WFE market to the number of gigawatts that are being installed? And if you're going to do that, how would you actually do that?

I think it's very difficult. But we replicate -- we can replicate the calculations people are doing to get those numbers. What you have to think about is when you're estimating the number of wafers required to build the components in a gigawatt of capacity, once people calculate the number of wafers needed to build those components, at least what I've seen, they typically multiply that by a capacity cost that's per month, okay? So you might multiply it by x billion for 100,000 wafer starts per month of capacity for wafers. So really, what you've just calculated is that you produce a gigawatt every month. And so you kind of have to figure out, is that what you want or you want to say that's 12 gigawatts a year. So it's a difficult metric to use that way. That's why we thought it's easier for people to think about how much capacity is supporting that today, likely not on the very leading-edge, but close to the leading-edge. What's it growing at? And then our customers that are thinking 2 years down the road, 3 years down the road, they're taking that amount of capacity, and they're growing it at 35% or higher CAGR and trying to put that in place for the products of the future. So hopefully, that describes the metric is just hard to use because I don't think you can take those numbers and just simply multiply them by the number of gigawatts and get the capacity.

Yes, it's hard. Okay. Got it. And then let's just talk about your share. So this year, you did lose, I think, in the range of 200 basis points worth of WFE share. I think a lot of that was because of a bad mix. So can you just talk about that? You sort of tried to address some of this on the call. And how much of a factor is China in that lost share?

Yes. So when we look at share for 2025, I was just looking at my wife thinking, there's a few days left, Tim. We'll see how it shakes off. But in China, U.S. companies are more restricted from a trade perspective than non-U.S. companies. So we have trade restrictions in China. There's some customers that we can't serve. There were additional customers we couldn't serve in 2025. We had previously described that we were restricted from approximately 10%, a little bit more than 10% of the market. That was a 2024 number. In 2025, that's more than doubled. In China, over the last couple of years, it has been about 40% of WFE. And so the restrictions we have probably cost us about 1 percentage point in share just by itself, the incremental restrictions this year. So yes, we lose approximately 1 point of share. And then also this past year happened to be a strong year for NAND. NAND grew significantly in the year. We have lower share in NAND, and it was a good shipment year for lithography equipment and we don't sell lithography equipment. And so those mix components, we didn't participate as directly. We don't think that that's indicative of what we expect this year or even in the near future. Back to the initial comments, we expect leading-edge to grow significantly. Applied Materials is #1 from a process equipment on leading-edge. We expect DRAM to grow significantly. Applied Materials is #1 in process equipment for DRAM. We expect HBM to grow rate with that. We have the #1 position on HBM equipment, about 14 out of the 19 additional tools that are needed to do the processing and stacking. And so our perspective is the demand drivers are in the right place for Applied Materials. We're doing the work to increase our share in each of those areas, leading-edge logic and DRAM and HBM. And we think that market is kind of moving towards us partly because of the tailwind of AI. And so hopefully, that answers your question. Yes, probably between 1 and 2 points of share loss this year, China and the mix, and then we expect our position to be really strong looking forward at the market.

So obviously, you believe you're going to be a share gainer next year. So however, much WFE grows next year, you're going to outgrow that number? Sounds like that's what you think.

Well, yes, and partly the other reason I didn't mention that I should mention, we don't expect any incremental trade restrictions in China. So in '25, we had incremental trade restrictions. We don't expect those at this point. So we should be able to compete and sell to our large number of ICAPS customers in China. We also think that, that market, if you look 2 years ago, that market was kind of focused on 60-nanometer, 45-nanometer, those kind of technologies. We think for the next few years, it's going to be more oriented towards 28-nanometer. A lot of the new customers that we're getting even over the last year are 28-nanometer customers. Applied has a very strong position in 28-nanometer uses some of the same types of copper wiring and capabilities that you see on the leading-edge. And so we think that, that will help us compete in China.

Let's actually talk about China for a moment. So there was a piece of affiliates rule came in and cost you revenue and you said $600 million next year. And now that has been revoked and there's a year reprieve on that rule. So assuming now that the government's open back up, I would assume that you're getting licenses to ship to those customers again, but you probably gave away those slots at least in the current quarter. So there -- those become additive as we look out throughout next year. But when you think about that $600 million number, the way that these customers typically operate is if you give them a year reprieve, they're going to try to get as much as they can in that year before they know that they're going to get banned again. So why would the add back not be more than $600 million now?

It could be, Tim. Of course, we don't know that, and none of us know that at this point, but there's another factor that would go along with the question you have. And that is the $600 million was a point in time. Those are the orders that we can see. And to your point, if you stay open for business, usually you get more orders. And so it's possible that, that book of business grows. You're right about the linearity. It will be through the year as we'll have to put those back into production and secure the right specs from the customers and go through that process with the supply chain. So it'll sort of be spread throughout the year. But yes, since we can serve those customers, we'll -- it's very possible we'll get more orders. The other thing that I wanted to mention was we had also said that we had an impact of $110 million in our just reported Q4. And what happened with that was we were able to put those slots to other customers. So we essentially swapped revenues. So we still made our revenue. We beat our guide for Q4. And what we did was we built -- use those slots, those build slots for other customers, ship that business and that $110 million that reappears in Q1, and we're able to support that. We were partially built on all those tools. So that part ended up making no difference for us.

One thing just on the topic of China, one thing I heard on the call that was new was Gary was talking about building purpose-building tools for China that are lower cost. Can you talk about that? That's the first time that I've heard Applied talk about this, and I thought that this was something that the industry should always do to try to compete with these domestic Chinese tool companies. But can you talk about how you're going about that?

Yes. The way I think about it is they're not purpose-built for China, but they are purpose-built for either new technologies, especially in the power space or mature logic nodes, possibly memory, too, but the way -- the examples that I've seen are mostly in the mature logic space. And so I think the basic idea is you don't need the same level of precision. You don't need the same level of features on the platform as you do at the very smallest geometries. So we have been reengineering -- since I've been at Applied, which is almost 4 years, you've been working on those platforms. So they are more focused platforms, more cost reduced. And yes, I think we'll have announcements later in the year about these products. But it fits that exact example. You don't necessarily have to take all of your features back to 28-nanometer that you're currently developing today. And that relieves also price pressure on your leading-edge equipment. So we think that's a good strategy. And the only nuance there was I don't think it's China only. We will have -- it will benefit our competitiveness in China, but it will benefit our competitiveness everywhere.

And you're going to be able to go in and offer performance that's commensurate or better than the domestic Chinese companies at lower price. Is that the...

Well, the -- in China, you face subsidies, you face all sorts of dynamics in China. So I won't make a comment about the price, but we certainly think TCO, the total cost of our equipment, especially with services will be pro competitive. It will be very competitive in China.

And I think just to follow on the same topic. So you and everyone else are talking about WFE and China being down next year. Now of course, that's what you thought a year ago and it turned out to be up this year. Can you just talk about why do you see it is down? I know you talked about some digestion in ICAPS in some of the mature node. But we've seen a lot of that this year. Your China business is down 20%. Some of your peers is actually up. So it seems to me like maybe you're undershooting what China is going to actually be next year.

Well, the first thing I'll admit is we could be wrong. We thought China would be lower -- we said China would be lower each year. And we expect that again, and the main reason is the whole world has to reckon with the amount of capacity that's been put in place in China, especially on the ICAPS nodes. And I think you see that around the global companies trying to figure out how much are they going to be able to sell in China versus how much is made inside country. And since that build was so large in '23, '24, '25 and the utilization, Tim, is, I would say, not high yet at this point and not high across ICAPS. So there is some reckoning with that amount of capital that's been put in place. And that's the main reason. And of course, we talk to customers and we have forecast from customers. And to your point, they have been lower than what has happened in each year. They typically come up in China. But right now, because of how much capacity because of the low utilization, we're still expecting a lower year for ICAPS and a lower year for China, and that's what's built into our expectations.

Got it. Can we talk about AGS, your services business? What's the right growth rate for that business going forward? And can you talk about this AIx platform? And have you seen benefits from that yet?

Yes. So for our services business, AGS. One thing we've just done, we announced during our earnings call is we've done a reorganization, and we've taken the 200-millimeter equipment that was part of AGS, and we moved that into our semi equipment business. So now when you see us reporting in the future, all of the equipment will be in the semi equipment business and all the services, it will just be services, so just recurring revenues in the service business. And I say that because it should be easy for investors now to see that our expectation is it will grow low double digits, so no change to that. It's difficult to see that last year because the 200-millimeter equipment have a lower year. Now that will be crystal clear, We expect it to grow low double digits. And the reason it grows faster than WFE, every single quarter as we ship tools, we add to the installed base. So that's like your opportunity set growing that you can sell services to. And each quarter, our capture rate has been rising. A lot of it is because customers are building new fabs and new cities, new areas. They need the help from qualified labor to help ramp and install the tools and service the tools. And lastly, to your point, we're inventing and adding new products in the service business all the time. We call those AIx, those are actionable insights. And some of that is AI-generated insights for customers to optimize the yields on the tools. And that added product count for the equipment that we have is growing the revenue per tool. And we think this is a win-win for customers when they use our services, they operate at higher yields and higher performance. And of course, we get to supply the product for that. So it's a win-win with the customers.

Is there a way to characterize your revenue per tool in the installed base and how much that's changed over time?

It's not something that we've shared, but it does increase. I think it's increased every year for the last few years.

And is that primarily because you're learning how to price for the performance? Is that because customers want -- they want to sign service contracts, and you can price better in that way?

I think it's -- this is a services question, right? I think it's really the added value of the services. So like I mentioned, we have new products every year. Some of these products are data products, some of these products are better service products, and I think that grows the value. So I say it's more of that than the pricing component for the services business. We do have a pricing process for the services business. A lot of that is to recognize inflation from the supply chain and different FX changes, et cetera. So really, I think it's added capability in the services.

Got it. And that kind of dovetails into gross margin, which I wanted to talk about. That's been a real bright spot. You've done a great, great job there. I mean, at similar China portion of revenue in July '22, gross margin was 46.2%, and you're guiding it to 48.5% with a similar contribution from China. So if you try to break down what that 230 basis point delta is, what is that? Because you did talk about raising prices? And does that explain most of the increase or are there other things?

There's other things that mix certainly plays in there. But in the last year, our gross margin increase. I think we averaged 48.8% for the year in 2025. And we finished at 48.1%. We guided 48.4%. So we're right in that ZIP code right now. In the last year, most of the growth, 120 basis points has been because of our improved pricing. On the cost side, we did improve costs, but we also faced some tariff headwinds. We also faced other additional costs we had to address like adding strategic inventory and those sorts of things. And so costs ended up being a little bit of a wash. The pricing component, and I've mentioned this before on some of the earnings calls, we've had a concerted effort the last 2 years to revamp our pricing and revamp our ability to study the value of both new tools and new capabilities that we're providing the market and then the existing and then set a price target for the sales team and go through that process. It's, I would say, more disciplined than it was before, and it's helping us ensure that we address the value. And of course, we all know that our customers, a lot of the large customers are also benefiting. AI is a good example from higher prices. And so when we're able to enable some of that capability with our engineering solutions, then we can participate in some of that value. And I think you've asked us that question before, Tim. As we focus our portfolio on more and more enabling, Applied Materials is focused on 3 generations ahead, 4 generations ahead, working very closely with customers. If you're successful in moving your portfolio to more and more enabling solutions, then it should come with higher gross margins. And so that sets the right landscape for us to tackle that.

And you talked before, Brice, about leading-edge foundry/logic. So can you give us a sense of how much you think the etch and depth intensity in your SAM expands as you go from, say, N3 to N2 to A14?

The way we've described this is the gate-all-around process technology, the SAM that we articulated for that. For just the new transistor, the gate-all-around, our SAM grows from $6 billion to $7 billion. And then if a customer also deploys the backside power distribution, then it's exactly the same number set. It's a $6 billion SAM growing to a $7 billion SAM. That's the way we described the gate-all-around transition. So -- and within that, we also have, I think, some additional share gains that help us in the original $6 billion that we have. So gate-all-around, huge inflection. We think it's a very powerful innovation for the industry. If you look at the power performance of a gate-all-around transistor with backside power delivery. If you look at the shrink capabilities and you look at the device performance, all significant improvements over the prior transistor. And so we think it will be a powerful node. And I've said, we expect it to ramp to 300,000 wafer starts of capacity per month. And we're somewhere in that journey probably still in the first half of that journey.

Great. Can we talk just for a moment about PVD? It's about 1/3 of your revenue based on Gartner numbers, even more than that of your earnings because it does have high margins. Can you just talk about the health of that franchise? I get some questions about the franchise and about what's going on in China. You hear some of the Chinese companies saying, "Oh, we're going to fully displace them for PVD applications". Can you just talk about the health of that franchise? Because it is probably your most important franchise.

Sure. We think that -- well, first, I'll just say, I just gave you the forecast again for our gate-all-around nodes. And a lot of that is built. PVD is the metal deposition that Applied uses to help build the copper interconnect wiring that connect all the transistors in a device we've talked about 100 -- over 100 miles of copper wiring for a large device. So you can picture transistors and basically up to 18 layers of copper interconnects on top of the transistors. Those are built by our tools. Those are built, PVD as one tool in that technique, and we're not changing our forecast. There's no change. We think that's a multigenerational solution. It's critical for low resistivity solutions. And yes, I think no change in our expectation for that franchise.

Great. It sounds like the next few years are going to be great for you. So thank you again, Brice, for the time.

All right. Good to see everybody. Thank you.