Welcome, everyone, to Cantor's Day Two Technology Conference. My name is C.J. Muse, semiconductor equipment analyst, and it is my pleasure to host Applied Materials, and we have Brice Hill, Chief Financial Officer. Welcome.
Good morning, C.J. Thanks for inviting Applied Materials. Great to be here.
Awesome. I thought for all semi-equipment companies, I'd start with your favorite question, the industry's favorite question. Where are we in the cycle? I think across the board, most equipment companies espoused a view of WFE up mid-single digits, with everyone taking share, which likely means maybe the growth rate could be a little bit better than that. To take a step back, I guess, how are you seeing the landscape? What are the puts and takes that you're focused on? If you have an early view or thoughts on 2026, we'd love to hear that as well.
All right. Thanks, C.J. We definitely view the market as secular growth. If you look at the last 10 years from a semiconductor equipment perspective, I think there's growth in eight out of those 10 years. Either in the two that were not growth were very small, so very small declines. Our perspective is that this is a secular growth business. If you zoom out far enough, you should expect it to grow every single year. If you look at our planning, what Applied Materials does from an investment, from a CapEx, from an R&D perspective, certainly that should come through in our business model that we're planning for secular growth in this business. Over those 10 years—oh, sorry, I lost the mic.
You have the mic? Is it OK? No?
All right. Thank you. How's that? Yeah? Good?
Great.
OK. Secular growth, we're planning for secular growth, we're investing in secular growth. If you look over those years, C.J., semiconductors grows at a multiple of GDP. That's our expectation. The view underneath that is that semiconductors become more and more core and more and more integral in the economy. Part of the underlying view of that is the world needs productivity. There's, in a lot of places, fewer workers. We need the power and productivity of semiconductors to continue driving growth for the industry. Applied looks and plans for that secular growth. Where are we in the cycle? I think it's an interesting question. Mixed markets, as you know, leading edge is certainly the highest energy area of the cycle right now.
When you look at the cloud service providers making their CapEx forecast, I think the CapEx forecast I've seen is over $325 billion for the cloud service providers adding high-performance computing for AI and other uses as we go forward. That's an example that's driving so many underlying technologies in the industry, and we certainly feel that at Applied. You think of leading edge, you think of leading edge DRAM and memory capabilities, you think of a High Bandwidth Memory that's being stacked in those systems to drive performance. You think of the packaging requirements. One of the systems design companies we recently spoke to said there was 8,000 mm squared of leading edge logic on a system that they're planning. Not just one chip at the reticle limit, 800 mm squared, but 10 times that from a silicon perspective.
Of course, all of that has to be interconnected on some sort of substrate. Those are driving all types of new technologies in the industry. Back to the question of where are we in the cycle, from that perspective, it seems early in the cycle. It seems like there's going to be a build-out of capability, especially around those technologies for the next many years. We can feel that in the industry.
Perfect. Maybe we talked about the haves. How about the have-nots? How are you thinking about those markets and any signs of kind of a recovery at some point, whether it is second half 2025 or 2026?
Yes. The market that's been slower than the other markets has been the NAND market. We do see some improvement in the NAND market. It's kind of an interesting story, at least from my perspective. There's no decline in storage memory needs. In fact, that's increasing at the same rates as probably it always has been. What's been happening in a good way is as the technologies evolve, you have bit density that's increasing faster than demand. They upgrade the layer count, and you can create more bits on the same silicon size, 30%-40% increases in some of these nodes. That's outrunning supply. From a wafer start perspective, the NAND business is actually able to more than serve full demand from the existing footprint of wafers, which makes it mostly an upgrade business from an equipment perspective. The market hasn't actually been weak. It's just that the investment required to deliver that output has been lower. I think underneath, if you look at the systems and you look at the requirements for storage memory, it's still running at a very significant rate.
Makes sense. Maybe specific to Applied Materials, I think Gary Dickerson was pretty early with PPAC. Most companies now in the industry kind of refer to that as a key area of focus. Using maybe that as the starting point and thinking through the technology inflections that are most exciting specific to Applied Materials, could you kind of walk through what gets you excited into 2025, 2026? First, maybe starting off leading edge foundry, I think you've talked about a 15% increase in TAM related to Gate-All-Around and backside power. Could you elaborate on that, how you see that progressing, what kind of visibility that gives you to growth this year, next year?
Sure. Let's start with you brought up the acronym PPAC, which the company has talked about regularly. PPAC, power, performance, area, and cost. When the semiconductor companies are designing a new chip, they're looking to put more transistors on that chip. Hopefully, the transistors are what delivers performance. They're hoping to do that at lower power and ideally same cost, so more performance at the same cost. The new technologies, Gate-All-Around as a starting point, that's a new transistor. It operates at a lower power level and more efficiently than the FinFET transistor. The dimensions of that architecture are more complex than the FinFET architecture, which is why companies like Applied Materials are TAM grows in making that new transistor.
Similarly, the backside power delivery moves the power delivery from being congested on top of the wafer and on top of the transistor like it used to be and puts that underneath the wafer. That opens up more room for more transistors. It creates another density advantage that the designers will love to have. If you look at the combined benefit of a new transistor that has probably 20% or more and more performance, and you take the power wires and you move them out of the way and bring them underneath the wafer, that also gets you a performance improvement in terms of power and density. Those are two big generational improvements that have been difficult to get without the typical Moore's Law, litho-driven density functions in the past.
It is very exciting from a silicon design perspective, huge performance gains. What Applied Materials does for those transistors, there's a significant TAM increase, a $1 billion TAM increase for the Gate-All-Around portion, and a $1 billion TAM increase for the backside power portion. Applied Materials expects to win more than 50% of RSAM for that incremental capacity.
Perfect. Maybe moving over to kind of the overall foundry logic landscape. I think a common theme from yesterday is that with the challenges at Intel, coupled with the fact that TSMC has sold out on 2 nm into 2026, 2027, that something has to change. I know you will not speak about specific customers. I guess, do you agree that at some point this year, a decision will need to be made to add incremental capacity somewhere to support the pent-up demand from foundry logic? Whether it is Intel internally or externally at TSMC, that is a tailwind that we should have some sort of clarity on in the next three to six months?
I think we would agree with that from many different perspectives. One, we started with the market and the macro and all of the energy around leading edge compute. As you said, the leading edge nodes are 100% utilized, and there's a lot of demand there. Continuing, if you look at our estimate for what we're selling for the leading edge nodes, it implies about 100,000 wafer starts of capacity between 2024 and 2025. We understand that the 2 nm node has lots of designs on it. The design rate tape outs is very high. That means it should be a good landing spot from a node perspective. You would expect a lot more capacity being put in place. I think we would agree with that expectation. It's 100% utilized. There's a lot of demand on the leading edge. We expect this particular node, because of the benefits that I just described, Gate-All-Around , backside power delivery, we expect there to be a significant amount of investment there.
Perfect. Maybe tangentially, you have a very strong business in advanced packaging, $1.75 billion in fiscal 2024. How are you seeing both CoWoS and HBM trends here in 2025? Is there some digestion that's going on from a capacity perspective specific to HBM? Is there sort of a necessity of Samsung getting qualified at HBM4 to kind of open up the aperture of spending? How are you thinking about kind of the growth trends for that business?
Yeah, I think we would say no digestion to HBM, only because that's probably not a good way to think about that market. About 10% of DRAM wafers right now are being allocated to HBM, high bandwidth memory production. That's growing. Our data tells us that's growing at about a 30% CAGR. Every day, capacity is being added for HBM memory. The issue when people ask about digestion or think about digestion, I think it's that at the end of 2024, we had a very high rate of sales for capacity for HBM because it was initial launch volume for our customers. They continue to buy. They continue to add capacity. That should be a 30% growth function. It's just not quite at the same rate as it was at the very end of last year. It's going to be a great market. It's going to be a huge build-out. For Applied Materials, there's about 19 extra steps to the DRAM process that are associated with bonding those memory chips. Applied Materials has a preferred 50%+ share of those steps. It's a good market for us.
Perfect. Maybe following on with CoWoS, obviously, an HPC-led market should be very strong as well. I would assume there's less kind of digestion there relative to HBM. Is that fair?
Yeah. I don't have a sense for the capacity utilization, but I would say demand function is extreme at this point. Every system designer that we're aware of is looking at these 2.5D and 3D solutions. 2.5D means it includes a memory stack in the complex of chips. 3D means it actually stacks some of the logic chips on top of each other in addition to the memory stacks. You have basically all of the complex system designers looking at ways to interconnect these chips more and more efficiently. Going forward, that will mean new interconnect capabilities, new substrates. There will be a lot of experimentation in this space. It is a focus area for Applied Materials. We recently announced an advanced packaging platform that we have in Singapore and had an event with a lot of our partners that do development with us. This is a large focus for the company in how to advance the packaging capabilities.
Maybe just to double-click on that, could you elaborate on kind of what's exciting there? Is that a focus in terms of hybrid bonding and kind of the partnership you have there? Is it moving to glass substrates? What is the kind of critical areas of focus for Applied?
All of the above. Every area that you can think about, whether it's different substrate types, silicon, glass, and others, whether it's new interconnect capabilities, also just the hybrid bonding. I think in our earnings call, Gary mentioned, our CEO mentioned that we're in advanced qualification for our hybrid bonding capability, which is an integrated tool, six chambers from us, and then one with a partner in that particular tool. We expect these bonding capabilities and stacking capabilities, and then the different techniques to build a complex of chips. We expect that to be one of the frontiers of innovation for the next 10 years.
Now, we're already doing stacking for NAND, but within the design winner or the partnership, or not partnership, but the work that you're doing, is that on the DRAM side?
I think my understanding is it would be more logic to logic, but I suppose there's not a reason why you can't stack the DRAM, but I think it's logic to logic.
Gotcha.
Yeah.
Maybe moving to ICAPS in China, obviously, has been a great market for Applied over the last handful of years. When I talk to investors, that's the one kind of area of concern. Could you level set kind of where are we in terms of your outlook for that business in 2025? What were the implications to the incremental kind of embargoes and restrictions, and where do we go from here?
Sure. First of all, great market for Applied Materials. If you look back to 2024 and 2023, we had a low NAND market. We also had a low leading logic market. To your earlier question about is there underinvestment in leading logic, we certainly went through a couple of years where leading logic was lower. Applied Materials is operating at record levels. We just had record earnings in terms of our non-GAAP earnings. We have record revenue. What was happening, what's happening is the ICAPS market, which is IoT, communications, auto, power, sensors, the more mature technologies. That market had just exploded over the last few years, a lot of that driven in China. Significant investments in that market. In fact, it's the largest market that we serve at this point. When you zero into China, there's been a lot of investment there.
We think it's been a national strategy to create self-sufficiency in a lot of those nodes. We don't view that as being in a cliff situation or ending. We think we're tracking a large number of factory projects there. We're adding customers as time goes on. There doesn't seem to be any change in the plans to continue to build out capacity. The only thing that's changed is it's not quite at the same rate it was in 2024. 2024 may have been stimulated by concerns of upcoming rules changes, of those sorts of things. There was a heavy investment in 2024, but we think the investment just continues. We expect the chip market itself to grow at mid to high single digits over time. Utilization is a little bit lower in the ICAPS space. It's probably in the high 70s at this point. The rate of investment is a little bit less, but we expect this, going back to secular trends, we expect this market to grow over time and be a very healthy market.
Maybe drilling into the restrictions, I think you talked about $200 million impact to April. Is that now out of the model and completely de-risked from a go forward perspective?
Sure. I think definitely from a modeling perspective, it's de-risked. We talked about a $400 million impact to Applied Materials. About half of that would be in our Q2, and it's roughly split between our services business and our equipment business. If you think about the services business, there are just some customers that we can't serve anymore. You take that business away in Q2, and after that, we expect to resume growing at the low double-digit rate that we've talked about in the past. If you had a line that was your forecast line in the past, you just step it down to what we guided in our Q2, $1.55 billion, and then it should grow at the same rate from there, would be our perspective. For the equipment, it's a little bit more sporadic, but I don't think it'll be a reconciling item going forward.
Perfect. Maybe just going back to the utilization comment of high 70s, do you have a view on kind of multi-net versus domestic China? Is there much variance?
Within ICAPS?
Yeah.
I'm just trying to think. If you go outside of China, the rate of utilization in China has been higher, actually, than the rest of the world. That's something that we've shared before. That's probably true in a number of the markets. From that perspective, China has been healthier, at least from a factory perspective, than the rest of the world.
Perfect. I guess maybe just to finalize kind of the thinking of drivers in 2025, you talked briefly about AGS. Just to confirm, we should be growing low double-digit year over year on kind of a new reset number. I guess as part of that, are there kind of design wins that are coming off warranty that could enable a step up or no?
Low double-digit growth is what we've shared in the past. Yes, that's still consistent with our view. What happens in the services business is the footprint of installed capacity grows every day. Basically, every day we ship a tool, the installed base grows. Our attach rate to those tools right now, it's about a third of the tools that we actually put under contract. The footprint grows all the time. Our service intensity on those installed tools grows as the generations become more complex. We're adding new services. When you think about AI and you think about the capabilities that data provides us to give to our customers to help tune their tools, that adds to the value that we're providing or the dollars per tool. Those three things multiply together to give you a faster growth rate than the core business.
The installed base is growing. You add service capability, and it is more intense. Those things add to a faster growth rate, which should be low double digits. You may have seen that our board announced a 15% dividend increase just last week after a 23% and a 25% increase the prior years. We are working to get our dividend at approximately the level of profitability of the services business, both to highlight the growth there in that business and also attach it to those recurring revenues and recurring profits. Our services business, about 85% of that is recurring revenue. About two-thirds of that is under contract, so subscription-type agreement with high renewal rates. Those terms are fairly long. I think 2.9 years is our average contract length at this point. If you think from an investor perspective, we have attached our dividend stream to the recurring revenues and recurring profits of that service business, which makes sense, I think, to a lot of investors.
Perfect. Now, maybe moving to display. I guess how are you thinking about the business today? I know there's excitement around your OLED vertical deposition tool. I guess, is that something that could deliver incremental growth sometime in the near future? From a kind of right-sizing perspective, are you at the right place for that business today, or are there considerations to try to figure out ways of delivering higher profitability?
We're definitely excited about this business. It's been obviously low levels of investment for the past few years with LCD technology, sort of mature and lower demand function over the last few years. What's happening going forward is OLED should become a higher and higher penetration for laptop screens, for tablet screens, as well as larger form factors going forward. We recently announced a new technology. It's a patterning technology for OLED that we think is a breakthrough, better brightness, better lifetime, lower power. If that demand function and if that penetration of OLED across those larger screen sizes plays out, which is what we expect, we'll have a much healthier business and good growth rates. We did announce that technology recently, so we're nearing the point where we expect that demand function to improve. No announcement today, but we think that's right in front of us.
I can't recall. Is that a Gen 8.5, Gen 10?
It's 8.5.
8.5. All right. Maybe to finalize kind of thoughts around 2025, I think, well, I'll give you my view, and then you can tell me what you think. That if I were to think about where there might be upside, it would be domestic China and Samsung, given the pullback in their spending and requirement of getting kind of qualified HBM4 as really like the two incremental drivers. I think the third would be if TSMC and/or SK hynix could get clean room space. But dinner last night with another equipment company suggesting just there isn't that availability. They'd like it, but it's a challenge. Would you generally agree with that kind of vision for 2025?
I think so. In the short term, what's changed for us over the past couple of years is our China forecast generally comes up as the year goes along. It is just because you have a larger number of customers in China, our visibility. Those are less mature customers from a demand forecast perspective. Their lead time and their signals to their equipment providers are not as robust as the larger, more mature companies. I think in the short term, C.J., you're right. Without reformatting capacity footprints and factory spaces and things like that, it would just be that the demand function in China could improve, which we've actually seen so far this year. We saw that the last couple of years. I think that's probably number one. Number two would be, as you suggested, leading logic. There is just a lot of demand there. That's where there's 100% utilization. I think companies will work to see how they can increase capacity there.
Perfect. I guess longer term, focusing in on kind of where you're investing from a technology inflection perspective, we hit briefly on a few areas. Anything that we missed? I know as an example, we were out at SPIE, and we met with you, and we discussed eBeam or DRAM for 4F². You recently added power electronics to your list of technology inflections. Is there a kind of, and I don't care about the rank order, but is there, in your view, areas that we should be talking about today? If so, let's go.
I think probably the largest one would be the DRAM for 4F². That'll be another inflection for us that will add to our TAM. I think there's a $500 million uplift. No, sorry, that's the wrong number. I think it goes from $600 million to $650 million. It's a $50 million TAM uplift per 100,000 wafer starts for the 4F². That's an added function for us. When you get to, you're talking about eBeam. eBeam, I don't know if we would call that an inflection. It's just that when the technologies, when the dimensions are becoming so tight, defect investigation capabilities are much more important. eBeam is a much better technology for looking at specific defects. It's just that it operates at a slower rate than the optical, but it's becoming more deployed as the technologies become more advanced. I don't think of that as an inflection in the same way, but the demand function's increasing for that type of inspection capability.
Okay. Taking a step back, Gate-All-Around , backside power, advanced packaging, those are the real trends.
Those are the big ones with DRAM also. Yes.
Okay. Perfect. Maybe moving to gross margins. You just reported a robust 48.9% and China contribution slowing, guiding to still robust 48.4%. I guess as you look ahead, what can get you kind of to that hopeful 50% kind of number? What are the key drivers to that?
There are two things primarily. The first is pricing. We've talked about pricing. We've been asked before, where are we in our pricing journey? I've said that we're probably in the third inning. When we went through the COVID impact, we went through supply chain impacts, we saw inflation. We had to change the pricing dynamic for the company. In the past, you didn't have inflation. You didn't have some of those supply chain functions. I think the company had been used to offsetting any inflation with cost reductions and being able to roughly keep pricing flat. It's no longer true. Now you have those tensions. You really have to think through your value proposition for the equipment and what the right pricing function is relative to the cost structure of the day. That's just step one, CJ.
Step two is you have to have the capability in the company to train your salespeople and to have those conversations with your customers. When do you schedule them? You think about an industry that does that regularly, perhaps the auto industry. You have a new model every year and a new price every year. We did not have that sort of construct in place with our customers. You have to build the capability to do the analysis, to set a price target, and then have the conversations with your customers. It is a whole new infrastructure that has to be put in place, which we are implementing at this point. I think we are still at early days.
Cost reductions. We have added a significant amount of effort in terms of engineering on working with our supply chain and also working with our tool designs on cost reduction opportunities. Those things combined have helped us improve to the levels that you just described. Plus, it was benefited by mix in Q1 and Q2. We expect to continue to make improvements using those two levers, cost and price, over time. When I've characterized that, I said it's slow improvement.
Perfect. Maybe on capital allocation, since your arrival, I don't recall kind of a set strategy. I think it's pretty clear that you're returning all excess free cash flow. Previously, some of your peers really drive to kind of a net cash breakeven. That's kind of where you are now. I guess just taking a step back, is there a philosophy that investors should kind of think about in terms of how you and the board think about returning free cash flow?
When you ask about a target, we don't have a cash target. We look at cash as flexibility. It provides strategic flexibility. We will move that cash number depending on what the opportunities are within a quarter. I guess the biggest takeaway would be in an environment where you don't have much M&A, and we certainly haven't been able to do any significant M&A, you're going to return all of your uninvested cash to your shareholders. We say 80-100%, but without M&A, it would basically be 100%. That is our strategy. You see that. We use a grid each quarter to guide our share repurchases, and that has worked fairly well.
Just to follow up on that, in terms of the grid, does that mean you have a 10B5 in place so you can be more aggressive at a certain price?
We usually do. Yeah. What I would say also is if you look, when you think about cash and you think about cash management, again, if you look at the investments for the company, Gary likes to talk about our EPIC project, which is about a mile west of our Santa Clara campus, where we're building a large lab infrastructure. That is a collaboration platform with our customers that we're putting in place called EPIC. When you look at the company, we're increasing our R&D. We're growing our footprint. We have record profits and record revenues. You can see from the investments that we're making the confidence we have in the business and back to the secular growth of semiconductors and the underlying demand function we see in the market today, especially driven by AI and cloud service providers. You can see in our behavior what the forecast is that's implied.
Perfect. Appreciate the time. Thank you very much.
Thanks, C.J. Appreciate it. Thank you.