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

Good afternoon, everyone. Welcome to the final day of Citi 2021 Virtual Global Technology Conference. My name is Atif Malik. I cover U.S. semiconductors and capital equipment stocks. It's my pleasure to welcome Dan Durn, CFO from Applied Materials. The format of our discussion is going to be fireside chat. I'm going to kick it off with my questions first. If you have questions for Dan or the AMAT team, you can text them in the e-mail box in front of you, and I'll go over those questions towards the end. Welcome, Dan.

Hi, Atif. Great to be here with you today.

Dan, Applied Materials has a unique value proposition in the semiconductor equipment supply chain. I view the company as the [ Texan ] of equipment makers with a broad product portfolio. Your end market mix is the most balanced with around 20% market share across foundry/logic, NAND, DRAM, wafer fab equipment markets. You have services, displays, ICAPS, advanced packaging, which adds further diversification. And in some way, this makes you agnostic to the mix of WFE and exposes you to multiple mega trends in the space. As you look into the next 3 years, where do you see the biggest opportunity to change this end market mix or your expectations that the mix is going to remain fairly steady and your market shares are going to remain fairly steady?

Yes. So what I really like about the way the company is positioned against the opportunities we see in the market, as you know, we've got a very broad portfolio. And I think that serves us really well in this environment. Typically, when the road map pivots from 2D shrink-driven world to something that has multiple vectors around the power performance road map. It's new types of architectures, new types of materials, going vertical on the chip, new types of packaging, new ways to shrink. The company is really well positioned against that opportunity. So I like the broad portfolio. I like the momentum embedded in that portfolio. You've got epithermal up over 70% this year. CMP process control, up over 60%. You've got implant up over 50%. ICAPS, packaging performing well, strong growth. And so that growth is broad-based. Then when I look at customer-by-customer, we're really well positioned with each customer on what they're driving today, and node-over-node, we see our opportunity increasing with those customers. And then I break it down by end market, and we're very balanced between foundry/logic and memory. And when you take a year like 2020, the overall market was up mid- to high teens. And against that opportunity, our Semi Systems segment was up over 26%. So strong relative performance against that opportunity. But the interesting thing about 2020, NAND was the strong grower, almost 2x the overall market. DRAM grew about in line with the overall market and foundry/logic under-grew the overall market. And our broad customer footprint and our broad product portfolio and our broad market exposure led to outperformance. And then you look at 2021 and we talk about the market being up sort of mid-30s this year, over $80 billion, it's up mid-30s, plus/minus. But the rank order of growth is exactly the opposite. Foundry/logic is the fastest-growing market. DRAM is a little more than the overall market. NAND is going to grow, but it's going to undergrow the overall market. And against that opportunity, which is the mirror opposite of 2020, we're showing strong outperformance again. We've got a market that's likely to be up mid-30% this year, and our Semi Systems segment in the first 3 quarters of the year is going to be up over 50%. So I really like the broad portfolio, the momentum top to bottom from a product standpoint, the momentum from a road map standpoint, and it's leading to outperformance in the current environment irrespective of how the spend profiles. And I think that makes us fairly unique in the industry from a portfolio strength and end market exposure perspective.

Yes, I would agree. The understanding on that broad portfolio subject -- there's an acronym that Gary uses. You guys do throw a lot of acronyms at us, but this PPACt acronym, which stands for power, performance, area, cost and time to market, essentially, this is the modified Moore's Law now. 30% of your company products are currently integrated. So -- and how can you use the breadth of your products and tools to add more value to the fab managers and job they're doing? And can you monetize some of this know-how and license it to your customers?

Yes. So the great thing about how we're positioned in the market, the strength of the company is always going to start with the absolute best technology around the things that we're trying to do. And then when you think about the road map and how it's progressing, again, we're pivoting as an industry away from a 2D shrink-driven world because we're running up against the laws of physics. But innovation is not going to stop. It's just going to happen differently going forward than we've seen in the past. And you're going to get multiple vectors to the power performance road map. And increasingly, combining those technologies under vacuum and combining those best technologies, co-optimizing between adjacent steps in the flow are going to become increasingly important to solving our customers' highest-value problems. And so we're really well positioned with the breadth of that portfolio but also a strategy that's looking to co-optimize adjacent steps and integrate steps under vacuum. One example that Gary points to is we've got 7 technologies combined under vacuum on a single pool that's focused on reducing wiring resistance in a chip by 50%. That's a whole node improvement. That problem is addressed by 1 tool. And that 1 tool is going to produce multibillions of dollars of revenue for this company over the next 4 or 5 years. And so as the road map pivots away from 2D shrinks going to new types of materials, combining technologies under vacuum, that's just 1 example of how we're benefiting as a company from those unique technologies. And again, that 1 tool which can provide a full node improvement in the power performance road map can be worth billions of dollars for us in the next 4 or 5 years. The momentum you see in our DRAM market, strong momentum in performance in 2020, you see it again this year, that's driven in large part by co-optimizing adjacent steps to solve problems for customers. So it's a real benefit to us as the road map pivots, the breadth of the portfolio and this concept around co-optimization and integration of technologies, and we see that playing out in terms of momentum of the company solving those highest value problems for our customers.

Great. Dan, Applied Materials was early in 2018 and talked about the fourth wave of computing and waves of new end markets like Big Data. A team expects that wafer fab equipment spending will be above $80 billion this year and all business segments, including silicon systems, to grow next year. You have talked about $150 billion WFE view by 2030, assuming flat or 15% WFE intensity. A common question we get from investors is if there's a ceiling to this mid-teens WFE assumption and why can't it be 20% if some of your customers like TSMC are talking about 15% plus component growth rate for the next 3 years?

Yes. So I don't think there's any reason why it won't be biased to the upside. I actually think it's conservative to hold that expectation flat. And what we've seen over time, 2013, you saw the bottom from a capital intensity standpoint, and it's risen consistently over time. And so a couple of things. The wafer size transition to 300 millimeter, the highly automated factories, the large-scale facilities clustered geographically, the consolidation of the customer base over that period of time and used equipment flooding the market periodically, that all kept capital intensity down. Now you've got all of those efficiency gains built into the infrastructure of the industry. There's no way for size transition. The road map is migrating away from 2D shrinks when Moore's Law was alive and well to more vectors of innovation and more complexity to drive those road maps. And then when you think about distribution of the global capacity footprint, when you think about the distribution of that geographically as a result of what governments want to do to secure supply of semiconductors into their home geography, you're going to see a regional distribution of capacity versus today, where you've got very concentrated ecosystems around the globe to supply global demand. You're going to see that get distributed. That's going to cause capital intensity to rise as you build a larger number of smaller-scale facilities to bring global supply online. So I think there's a number of drivers over time that cause that bias in capital intensity to go up off of these levels. Then you ask yourself the question, is it affordable? And what I would point to is capital intensity bottomed in 2013, and it has been on a slow, steady rise since then. And our customers are investing a lot of money in capital equipment to bring that global supply online. But if you look at those investments, WFE to their EBITDA, capital intensity has been coming up, but WFE to EBITDA has been coming down over that same time period. In memory, it's down 40%. In foundry/logic, it's down by 1/3. And so our customers are finding more effective ways to monetize those investments than they ever have. So it is affordable by the industry. And I do think that you're going to see shifts in global profit pools. Profit pools of the global economy are going to shift. More of it's going to come to the electronics ecosystem, and I think all market participants in the electronics ecosystem are going to benefit as that happens.

Excellent. Dan, the next question is on a very topical topic of government spending. Equipment makers like Applied are clear beneficiaries of U.S. CHIPS Act and government-sponsored foundry spending initiatives in Europe and Japan. You were at GLOBALFOUNDRIES in your past life, and you have a good sense of the foundry model. Some investors fear that equipment makers have pushed the semiconductor spend consolidation as a positive for the space as it drives lower CapEx volatility and a favorable shift, and emergence of new foundry and IDM 2.0 models could reverse all that. How do foundries sustain suspending and don't cause an oversupply down the road?

Yes. So I think it's important to start with the foundry model and how it operates, and then I'll come to government influence as capacity gets distributed around the globe and the implication that, that has on industry economics. The foundry model, the way it operates, is our customers will have looked through visibility on their customers and the demand that they have for capacity. And our customers will make investments when they have that relationship with their customers, and there's a demand statement that sits behind it. And so I think that's why you've seen less volatility in the foundry/logic market over time than, say, memory. It's given the way that industry operates where you have actual end market demand sitting behind capacity investments that are made, which makes them more stable and sustainable over time. As we think about the model of global supply today, the most efficient way the industry brings capacity online is very large-scale facilities clustered geographically to take advantage of scale benefits of an ecosystem in a home geography. That's the model today, and it's very efficient. What's going to happen as governments begin to play a role in distributing capacity around the globe, governments have taken a point of view semiconductors are increasingly important to economic growth and national security going forward, and they want to secure supply of semiconductors to feed their economies. We know that the model of bringing global supply online today is very capital efficient. And so when you distribute that supply with smaller -- larger number of smaller scale facilities, that creates an inefficiency for our customers. Governments are going to step in to offset that economic inefficiency, make our customers economically neutral from today's model to how the industry will evolve going forward. That's going to lead to an upper bias in capital intensity. And as it does, then I think the capital equipment industry will benefit as that trend plays out. That dynamic is not going to fundamentally alter the long-term supply-demand balance in the industry. These factories are very expensive, even by government standards. And so global supply will be brought online to meet global demand. That's how the foundry/logic business has evolved over time, and we think it will continue to be the way that it works going forward. It's just how that supply gets brought online and where it sits geographically that's going to fundamentally change. But the long-term supply-demand balance won't be altered based on the dynamics of the industry, lead to an upper bias in capital intensity, and I would expect to benefit as that happens.

Very interesting. Dan, when are you expecting equipment contribution from government-sponsored fabs?

We think that we start to see market impact from government participation in the 2023 time frame. If it happens sooner than that, then I think that's an upward bias to the expectations on where we've set in the market. But we see it starting in 2023 and lasting for multiple years once it starts to impact our markets.

Got it. And some of the major foundries have talked about $100 billion CapEx plans over the next 3 years. And it's been news that they're asking for pricing discounts from equipment and material makers to lower CapEx burden and gross margin headwinds. I understand these customers already get some kind of volume discounts because they have been big for you historically. Is there a risk to equipment gross margins from rising wafer prices or it just gets passed down to your suppliers?

Yes. So there's always going to be commercial tension in the system. Any industry that operates the way we do that focuses on being efficient, there will always be that commercial tension. But I would take a step back from that dynamic at where we are at this point in time in the industry. The broad context around this industry has always been driving value for the end markets and the end customers, bringing in incredible capability online that multiple industries have benefited from. So there's this concept of value versus cost. In an environment where -- I think it's been definitively answered that a competitive technology road map brought to market on a time scale that allows a value proposition to the end market is an enormously valuable statement. Our customers understand the difference between value and cost, and I think there's a bias in this market to bring a very robust process technology road map online and the time element of the PPACt equation. That time element is super important. So I think these -- I mean these conversations always start with enabling technology, the value of your technology, bringing those technologies together in unique ways that solve high-value problems and do it as quickly as possible so that our customers can get their technology to market in a way that creates an enormously valuable proposition for their customers. I would say that the industry is much more focused on that dynamic today than just thinking about things in a more commodity, like, cost way. It's really a value-driven dynamic and at the core of that is innovation and technology. And I think we're pretty -- I think we're very well positioned as those trends continue to play out and this emphasis on time and solving tough problems.

Got it. Dan, the next topic is on services business. I think you guys have done a really good job in helping investors understand the recurring nature of your services business. I was quite surprised to hear that 87% of AGS sales are recurring, services, parts and software. You've talked about a 10% to 15% component growth rate for the future. You have 40,000 systems in installed base. And you mentioned a goal to achieve 20% increase in sales per system through 2024. The question is what actions are you taking to drive increased services sales per system?

Yes. So as you point out, Atif, it's a great business for us. And what we report as services, each company is going to report different things. What we report as services is a very high-quality look-through on true end market services for our customers. So we feel good about that level of transparency and disclosure. As we think about that business and what drives growth, you always start with the size of the installed base. We've got the industry's largest installed base over 40,000 systems, 160,000 chambers, like you point out. And then every year, that installed base is going to grow. So that is a great foundation upon which a high-quality services business is built. The next thing I would point to is when we ship a system, it's typically in production for decades, 3, 4 decades. And so it creates a very long lifetime of that tool, which creates a nice opportunity to monetize it. If I look at what we ship today from a complexity standpoint versus, say, what we did a handful of years ago or a decade ago, service entitlement over the lifetime of what gets shipped today is much, much greater than, say, what we shipped a decade ago. That level of complexity, what we're shipping into the market today to solve these problems for customers, is going to create a very nice long-term annuity on that tool over its lifetime. So rising complexity of the technology serves us well with respect to this business. Then the question is how do we engage with our customers commercially? And so if it was just a transactional relationship where the customer needs something, they call us up, we ship a part. That's one model. We put in place an engine, a strategy around penetrating our customer base with what we call long-term service agreements. These are subscription agreements around the tools we have in the field. And we've grown that proportion of our revenue from 40% to 50%. Now it sits at 60% of our services revenue that we drive with our customers through subscription agreements. We talked about over the next 3 years between now and 2024, driving that to 70% of our revenue. And why that's important to us? We talk about revenue per tool, and we talk about increasing that 20%. And we're already halfway there between now and 2024. We're already halfway there to meeting that objective. When we put a tool under a subscription agreement, the potential value of that, we get probably 3x the revenue on that tool than we would if it was just transactional. Our customers spend the same amount of money, but instead of spending it with other parties or within their own organizations, they spend that money with us as part of that subscription agreement. So our revenue goes up. As a result of that, customers still spend the same amount of money, but they get better outcomes, better performance out of the tools, more uptime, better yields, more consistency. So there's a value proposition for customers. So the combination of industry-leading installed base, rising complexity of technology, then a strategy around turning those relationships and those commercial engagements into subscription agreements creates a growth engine around our services business that we think serves us well, serves the customers well, and it's going to serve investors well as that continues to grow over time.

Super. Let's move and talk about the end markets. First of all, personally, I think the team has done a really good job in explaining the challenges around different end markets, memory, logic, advanced packaging in the master classes of investors. Who have not watched those videos yet, I will highly recommend all the 3 videos. On the DRAM market, Applied has talked about DRAM being second half weighted this year. Any changes in DRAM spending plans given customers' customer inventories look a bit elevated and pricing is peaking?

Yes. No, we don't see any changes in -- from a capital equipment standpoint, we don't see any changes to our customers' plans in terms of deployment of equipment. We see the DRAM market continues to be robust this year. We think that's in line with our customers' expectations of where those markets are going to go over time. And I know there's a debate and narrative around what's defining the current market and the debate between is there a true softness in end market demand or is demand for DRAM influenced by a supply chain that's still recovering from pandemic and you can't get all of the components that sit around the DRAM, creating a bit of softness, temporary softness in the near-term environment. I think that debate is alive and well. But if I look through that debate and think about the actions of our customers, we don't see changes in their deployment plans around capacity for the back half of the year or really as we look into next year.

Great. And on the NAND profile, you've talked about flat half over half. Any concerns that the industry is overspending? One of your peers talked about maybe push out into next year. And have you seen any movement in NAND?

Yes. So there's no change to our perspective on what we've seen. For several quarters now, we said flat, possibly down half over half. We just need more time. Hard to know what our peers and competitors bake into their forecast. And we know there's been some changes in their view of the market, but it's hard to know what's driving that change in perspective over the last several quarters. As I look through to the expectations that we've had around that market, the conversations we've had with customers, we don't see any change in where we set the expectations. NAND is going to grow this year. We see it being flat to down half over half. We'll see more time elapse and see what happens in the back half of the year, but that's been our expectation for a while. We've really seen no change in the expectations set by our customers.

Okay. And then domestic China has been a strong force in first half. What are your expectations for second half? Are you seeing any impact from credit issues from certain memory makers or anything on the U.S. restriction side?

Yes. So I think our expectations around this market are going to be what we've seen now for several years. There's a commitment to build an ecosystem in an industry. We see that commitment being executed in a slow, steady, disciplined way. We see investments in the ecosystem. We see investments from a technology road map standpoint. And where they've got viable technology road maps, modest capacity additions that sit behind that, we see greater than 55% of the investments into the foundry/logic market. We think the ICAPS nodes, it's everything larger geometry sizes than the 3 most leading-edge nodes in foundry/logic, so most leading-edge nodes 3, 5, 7. The ICAPS nodes would then be defined as 10 nanometers and above. We see the majority of the investment in the ICAPS nodes within foundry/logic. That's a slow, steady building of the ecosystem. I'd also say that we see that market this year growing a little bit above the industry average. We'll see where it ultimately ends up. But the behavior that we see, slow, steady building of the ecosystem, it's been that way for several years now, and we would expect that to continue as we go forward. In terms of the credit comment that you made as part of the question, we operate under letters of credit there. No impact to any of the things that we've seen in the market get announced. No change from a customer behavior standpoint. And from a risk credit exposure standpoint, given how we operate commercially there, we're just not exposed to any of those dynamics.

Good. Dan, some investors were positively surprised to hear ICAPS around $4 billion in sales or 50% of foundry/logic sales this year. You shared that number last week in your master class. I think historically, this number has been more like 1/3 kind of the lagging or the trailing edge and 2/3 leading edge. What is the sustainability of this piece?

Yes. So what we see in that market today we're actually encouraged by because, I think, it's a diversification of the end market drivers of foundry/logic. And it's not a 1-year phenomenon. If I were to go back in time, in the current environment, we see balance between leading edge and the ICAPS nodes. If I were to talk about 2020, we see balance in those markets again. As I look forward to 2022, we would expect balance in those markets. So the current environment is defined by diversification and balance. At our Analyst Day a few months ago, we talked about 2024 and our expectation between those 2 markets. And again, we see them roughly profiling in line with each other, perpetuating this balance we see in the current environment. And if I were to go back in time and observe what's been going on in foundry/logic, if I were to go back a decade, 90% of the spend would be on the leading edge, then that diversified to 80-20, 70-30, 60-40 over time and now it's balanced. So there's been an evolution of that market. And if you think about the drivers of that market, used to be a PC-centric world back in the day. You'd put a lot of investments around the leading edge as new operating systems would come out, and then you'd migrate down the node profile. And I think what we've observed is that with consolidation of customers and migrating down the node profile, it was a pretty robust used equipment market that would serve the long tail of market participants. And as you've gone from a PC-centric world over time and evolving more towards edge, cloud, customer base is already consolidating. We're seeing not only those nodes persist in time, but we're seeing those broad profile of nodes actually come up from a capacity statement over time. And it's just led to a diversification of that market that gets us to where we are today, which is not a 1-year phenomenon. It's been an evolution over time, and we would expect that dynamic to persist going forward as the world moves more towards an edge, cloud view of the world. And I'd also like to connect to the concept of machine-generated data in 2018 being greater than human data. That was the first year machines generated more data. And embedded in that dynamic is intelligence being pushed to the edge. That's what creates machine-generated data. And if you think about the trends by 2025, 99% of the data being generated machine to machine, there's going to be a demand for intelligence at the edge, as that data thesis plays out, that's going to serve this market well. I think some of our competitors talk about that market growing faster than the leading edge. Our view is that those markets are going to profile roughly in line with each other. As we go forward, more intelligence gets pushed to the edge and more data gets generated machine to machine.

Great. On the leading-edge logic, 3D devices is going to be a mega trend for the space as we move from FinFET to gate-all-around. With 3D NAND, I believe the scaling was more in one direction, only Z, which helped your peer a little bit more on the etch and deposition side. But when I look at the logic transition, which is more truly 3D applied is much better positioned in PVD, CVD, epi and CMP. You have talked about $1 billion TAM for gate-all-around. When do you start to see this [ bearing ] in?

Yes. So I would -- I'll defer to our customers to talk about when they're going to cut this transistor technology into the road maps, but you should start to see it towards the back end of the 2024 or mid-range model that we've put out for investors at our Analyst Day. Towards the end of that time period, you should start to see this new transistor technology begin to influence the outcome. And we're really excited about the inflection. We think we're really well positioned. We think that, that technology inflection is going to rely on technologies. We call them our leadership technologies where we've had traditionally high sources of share and are very enabling to transistor performance. We think this inflection is going to be really in our sweet spot. We're engaged with a number of customers on this, and we think we're really well positioned. You rightfully point out, it's an incremental $1 billion of opportunity for every 100,000 wafer starts a month that our customers put in place. And as I think about FinFET technology and transistor performance, when you go to gate-all-around, think about it as a 20%, 25% improvement in power and like a 10%, 15% improvement in performance for just that technology inflection in transistor performance. And it's all about how you engineer and control the channel. When you're doing it within FinFET, it's lithography and etch-driven dynamic to construct that channel on the vertical dimension. In gate-all-around, it's a horizontal construct. And so it's going to emphasize technologies like epi and selective removal. Again, these are leadership products for us. We're engaged with the customers and the road maps, and we think we're really well set up to benefit as those technology inflections get driven into the market. And again, you'll start to see it in the -- towards the end of our mid-range model 2024, but we'll let our customers be a little bit more specific about the timing of when they start to cut that in.

Makes sense. Moving on to the financial model, really nice execution on gross margins this year. You are running ahead of the 48.5% base case in your fiscal '24 model at a lower revenue run rate. What are some of the variables that can impact the gross margins if you stay in the COVID environment longer?

Yes. So in the current environment, we think we are performing well. I don't think wherever we're at, we'll be satisfied. We'll always look to push that higher. In the current environment, I'd say, we've got maybe 50 basis points of headwind. So we'd probably be right on top of that long-term model. And when we put that out there, we said that we would hope to do better. And the company is performing well against that opportunity. So over time, as those COVID headwinds begin to fall away and the world becomes more efficient at operating again globally, logistics channels free up, some of the adders we see from a labor standpoint having a higher level of head count in our factories so we can deal with some of the nonlinearities we're seeing around the globe, we would expect to become more efficient as a company. And then I always think gross margin discussions start with technology. Are you delivering the absolute best technology to the market that the customers find valuable and enabling? And so I think we're really well set up given the breadth of the portfolio, the momentum we see node over node and the market inflection away from 2D-driven shrinks to something that's more pervasive and has more vectors to the power performance road map, new materials, new types of architecture, going vertical on the chip, new types of packaging, we think we're really well set up from a technology standpoint to be a key enabler of each and every one of those drivers of the PPACt road map. And so I think we're well set up from a technology standpoint. And then from an execution standpoint, continuing to focus on the things that we've been focusing on the last several years that are serving us well in terms of operating discipline and productivity gains. How we drive that discipline into the company? How we think about designing new products, leveraging best known methods, BKMs, to increase the hit rate of successful products? Where we insert new products from a gross margin standpoint? It will typically be the lowest point of that product's life cycle when it gets first introduced to the market. How do we bias that higher on insertion and then improve from that higher starting point? And then trying to understand more about the value of the technology that we're bringing to market in terms of key enablement, solving high-value problems, what it ultimately means for our customers as part of their road map, what value it's adding and use that perspective to have a conversation around what's the right value for the innovation that we're bringing to market. So there's a number of things that come together over a period of time that lead to the type of outcomes that we're seeing. We would just hope to do more and more of that over time.

Great. And let me squeeze in one question I've received from a client. What are the key drivers for market share gains by end market in calendar '21 and calendar '22?

Yes. So from -- so we've got a lot of momentum in each of the end markets. So I would say that we will do well in the NAND market. We're going to do even better in foundry/logic and DRAM. And where I go in DRAM is part of the road map. We talked about the best technology. And then we talk about optimizing and co-optimizing across adjacent process steps and then integration of those technologies under vacuum to lead to outcomes for our customers. So from a DRAM standpoint, when you think about co-optimization of steps and then you think about our process control technology and you think about our e-beam metrology that can get to sub-1-nanometer resolution and then bringing it together with our AI(x) platform and process recipe optimizer to speed our customers' road map and time to market of the next node, these capabilities are coming together real time that's driving outperformance against that market. Then when we think about foundry/logic and we think about bringing capabilities together under vacuum, Gary talks about lowering wiring resistance. We're bringing 7 different technologies together under vacuum. And again, that one tool alone is going to provide a one node improvement in power performance just by itself, and that's going to lead to billions of dollars of revenue over time. So I think we're well positioned against these trends. I think we're going to do well in NAND. We'll do even better than that in DRAM and foundry/logic as those inflections play out.

Dan, thank you for your insights and coming to the Citi Conference.

Always good spending time with you, Atif. Thank you.

All right.