Applied Materials, Inc. (NASDAQ:AMAT) Wells Fargo TMT Summit December 2, 2020 2:40 PM ET
Dan Durn – Chief Financial Officer
Conference Call Participants
Joe Quatrochi – Wells Fargo
Great. So I’m Joe Quatrochi, the Semi Cap analyst here at Wells Fargo. I’m happy to welcome Applied Materials’ CFO, Dan Durn, as well as Mike Sullivan from IR. I have a list of questions to go through but if anyone on the webcast would like to ask the question, feel free to email me. So maybe Dan, thanks, and Mike, for joining us.
Maybe first, walk us through kind of your view in terms of how do you think about the Applied Materials story and maybe what investors are misunderstanding or under appreciating the most. And as you look out over the next three to five years relative to that and relative to today, what do you see as the biggest opportunities for the company?
Sure. Thanks, Joe. And I really appreciate the opportunity to join you and others in this session. So thank you for the invitation. We appreciate it. So a couple of things I’d point to, as I think about maybe what’s underappreciated about the story. I think the end market we operate in, I think there was a period of time, where it was a no growth cyclical market, growth of our industry was decoupled from the overall semi industry.
I don’t think that’s the case anymore. The industry hit an inflection point, I think about in 2012 where it fully absorbed the efficiency gains of the transition to 300 millimeter, wafer size transition, customer base consolidated. All of those efficiency gains are now behind us in our industry for almost a decade, not quite but almost a decade is back on a growth trajectory along with the semi industry.
I don’t think that’s fully appreciated yet, but we’re going to continue to post more growth going forward. And I think pretty soon it’ll be understood that this is an upward sloping trend line. We still have a cyclical overlay, but it will be higher highs, higher lows over time. And for the first 35 – 35 years, 3.5 decades, our industry grew in line with the overall semi industry. Then we went no growth cyclical. Now I think we’re back on a trend line consistent with overall semi’s and we get pretty excited about that as an industry.
Second thing, I would say is, I think foundry/logic is a historical position of strength for us as a company. But I think we’re narrowly viewed through that lens of being strong foundry/logic and underserved in memory. And what I would observe in a year like 2020, where memory was the predominant growth driver, foundry/logic 55% of the spend but it undergrew the overall industry. DRAM grew in line maybe a little bit more than the overall industry. And NAND outgrew the industry two to one. If the industry is 15% grower this year, NAND outgrew the industry two to one.
Against that backdrop of 15% growth for the overall industry where foundry/logic undergrew the industry, our systems business is going to be up over 25% at the midpoint of our guide for fiscal Q1 or calendar Q4, however you want to describe it. So I think and you look at our share position in markets like DRAM significantly outperforming the market. It’s a great pocket of strength for us. And I just think that’s an underappreciated part of the portfolio.
And as I look forward into 2021, foundry/logic strong again, DRAM the predominant driver of growth next year in memory and NAND being flattish from a spend standpoint, that’s set up around us as a company and the strength we see across our portfolio. I don’t think is fully appreciated today in the market.
The third thing, I’d say is, we get a lot of questions about display and while it’s at a cyclical low, as we sit today, we think from a planning standpoint, we’re planning for return to a more attractive point in investment cycle in 2022. But we’re seeing green shoots and we think the bias is towards the upside as those green shoots take hold. I think it’s prudent to set expectations that that materializes in 2022, but if it happens faster, we’ll certainly benefit from that. And so display is going to be very quickly a good adder to growth, revenue, earnings cash flow. And I just think it’s an underappreciated part of the portfolio as we sit here today, given the fact that it’s poised for a bias to the upside.
From an opportunity standpoint, I guess, what I point to is classic Moore’s laws hitting a wall, it’s run its course, you no longer get simultaneous benefit of power, performance, area-cost, as you shrink in a two dimensional plane. And what that’s meaning is the industry is going to a much broader portfolio of innovation to drive the power performance roadmap. Gary will call it the new playbook. You’d look at what customers are saying, I think Intel has six elements. TSMC has five elements. They all mirror the same thematic drivers of growth, new architectures, new materials, new structures, new forms of packaging and different ways to shrink in a two dimensional plane.
So the innovation is broadening and given the breadth of our portfolio creating, shaping, modifying, analyzing, we’ve observed this trend now early, we’ve been talking about it for probably three or four years. Gary has for three or four years, we were on it. We’ve got an innovation – pipeline of innovation that’s targeted at these new inflections. And we really like how we’re sitted or seated next to this opportunity as it grows, really like how this company is positioned, given the inflection we see. We are on it early and I think we’re well positioned to win.
That’s really helpful. And I think over the past couple of quarters, it seems like you guys have started to provide more metrics around some of the sub categories within your semi systems business. I think you talked about metal deposition of 32% last year, last fiscal year record etch revenue of 30%. Can you talk about your confidence in that competitive position? And then how do we think about your positioning for some of the application wins or penetration at subsequent technology nodes relative to maybe what’s an actual high volume manufacturing today?
Yes, so I think we’re really well positioned across multiple fronts. We talk about metal deposition, the business was up over 40%, I guess, 42% in the fiscal year, you see our etch business up 30%. You can see the combination of PVD, CVD, epi up eight points a share gain. You can see our inspection business up over 40% and was 46% in this recent fiscal year. You can see momentum on the DRAM front. You see continued momentum on the foundry/logic front. And we think we’re well positioned to NAND and we’ve got a number one position in packaging.
And so we think the portfolio is performing really well. As we disaggregate the market and whether we’re talking about foundry/logic or memory, node-over-node, 5-nanometer over 7-nanometer, our opportunity increases significantly 3-nanometers over 5-nanometers, we see it increasing again fairly substantially. As you look at our revenue opportunity per wafer start of installed capacity in the industry node-over-node, we see momentum, we see our revenue opportunity increasing node-over-node because of the product portfolio, it all starts with best-in-class technology, then the opportunity with the breadth of the portfolio to stitch those together.
In integrated systems where our customers can manage the interfaces between the different layers of the chip in a way that drives better performance of the semiconductor devices, it gives us a very unique position in the industry as these roadmaps progress down the node profile, technologies becoming harder and we’re positioned to be more influential in how the – in terms of how the industry solves those difficult challenges, given the breadth of the portfolio and the ability to help our customers manage the interface. And we call it interface engineering, super important principle going forward.
And so we see our opportunity strong momentum today, and the opportunity increasing, we like how we’re positioned.
That’s perfect. And so, maybe going back to some of the demand side. One of the pushbacks that we get from investors a lot is the sustainability of foundry/logic spending, just given how high of a level of spending we’ve reached, currently over the past 18 months or so. So maybe talk about what you’re seeing there as we look into next year? And maybe how do we think about just capacity utilization at some of your foundry customers and the demand they’re seeing? And I guess, how should we think about maybe the number of customers spending the broaden aspect of that spending?
So 2020 is playing out very consistently with how we thought four, five quarters ago. We said that foundry/logic is going to be strong throughout the year. We said, there is going to be a broadening of the customer spend, multiple customers, multiple nodes. And so the current environment is playing out as we expect. And as I think about what’s driving the current environment and what gets us excited as we look forward. Go back to fundamental demand for semiconductors. There is more demand for compute power today than there is ever been.
And what’s driving that, there is a handoff in terms of end market growth, that’s from a consumer oriented device called the handset to something that’s far more substantive material, secular theme around the data economy. It’s the largest companies making substantive investments in their own capabilities, so that they can serve their customers better, manage their supply chain better, drive more efficiency into their operation. I think it’s existential for these companies. They either invest or they don’t. And if they don’t, it becomes an existential issue for them.
And the total economic pool that’s being tapped into is much, much larger, probably orders of magnitude larger than a consumer oriented device. I call it non-consumer discretionary spend. And so, at the core of that is the demand for compute power that’s greater today than it’s ever been. And this is something that’s going to play out over several decades as the data economy takes hold and develops more momentum.
And if I look at the drivers today versus where we were say four quarters ago pre-pandemic, they’re stronger and clearer today than they were back then. We think these trends are accelerating as a result of the current environment, not slowing down. And so, it doesn’t surprise us that there’s robust demand on the leading edge. It doesn’t surprise us that there is tightness in the system around 7 nanometers and 5 nanometers and there is a strong pull to get 3 nanometers online as soon as possible.
Our customers’ roadmaps are accelerating, not slowing down. There is a stronger poll to get that technology online. What we’ve also observed as part of use case specific architectures. You’re seeing a broadening of investments across node profiles, trailing node geometries, leading edge of geometries, you’re seeing intelligence push to the edge, you’re seeing sensors, data generation is accelerating, you’re seeing a substantive broadening of those demand drivers, which dampens volatility over time. So we don’t think this is a one year phenomenon. We don’t think this is an 18 month phenomenon.
In fact, I’ve been having the same conversation for the entire time I’ve been sitting in this seat. Once the foundry/logic spend is unsustainable, we see that strength continuing in the next year. And what we see is a multi-decade secular trend playing out in our industry, semiconductors are going to go structurally larger. We are back on a growth trajectory with the overall industry, and that has structural growth aspects for our industry.
If we were sitting here five years from now and talking about high 50s WFE, I would be disappointed. I think you’re pushing into the 60s. And then I think you’re going to push into the 70s on your way to 100 billion plus types of WFE in support of these trends we’re talking about. This is what gets us excited going forward.
That’s helpful. And I think looking into 2021, one of the things you highlighted last quarter is some of the specialty markets as maybe – they’ve been kind of impacted from COVID, maybe being some sort of kind of maybe an upside driver in 2021. How big of a revenue contribution is that? And I guess, what type of projects are these? Are these more a R&D type projects that have been delayed or are they production type products? Just kind of help us maybe understand a little bit better what does exactly are?
Sure. So as we take a look at trailing versus leading edge technologies. I think definitions are important. This is one area where everybody will have their own definition of what they mean. And so for us, when we talk leading edge technologies, today it’s the three leading nodes that customers are investing in. So we see investments at 7, 5 and 3 nanometers today. Everything 10 nanometer and above, we consider trailing node geometries, when you talk about specialty nodes for us that’s 28 nanometers and above, and so definitions are important.
When we look at leading versus trailing node geometries, you go back a decade, 80%, 90% of the spend was on the leading edge, very little of it was on trailing node geometries. And then over the last decade, as trailing node became more prominent from an investment standpoint, broadening that demand and dampening the volatility, you saw it go from 80-20 to 70-30, 60-40. That was a couple of years where it was fully split 50-50 in the market. For the last two years, we see it coming back to more 70% leading edge, 30% trailing node, but given the aggregate size of the industry and how it’s grown, the trailing node investments at 30% are still very substantive. And we think the growth profile of the trailing node geometries probably is greater over time than overall WFE.
So we really like our historical position across these node profiles, and we think we’re well set up to win as this industry continues to further diversify demand drivers, dampen volatility and drive structural growth off these levels. I think this is an industry that’s in the process of maturing in a very material way and the long-term implications of that for an industry that historically has seen a lot of volatility, I think bodes really, really well to investible themes around structural growth in a super important industry, that’s at the foundation of all the technology trends that are shaping the way people live their lives. We think we’re incredibly well positioned to do well against that backdrop.
That’s helpful. And with the demand that you see on the trailing edge side, is that – how do we think about the mix of when you sell into that demand? Is that mostly new equipment or is there also some refurbished in that? And then just maybe how do you think about the split of that? And I assume there’s also a kind of a price differential in that equipment as well?
Yes. So the way we see it, if you were to go back in time up until maybe a couple years ago, we think there was enough used equipment out on the market where we could go take the cores off the market in the used equipment market, refurbished them, put them back into the market. And it was a very nice profitable business for us, but there was enough availability of that used equipment to support the growth and development of the market. And something happened a couple of years ago where you were seeing a lot more robust activity in these older trailing node geometries. And we’re back – actually a couple of years ago, we started building new systems, brand new systems to support the demand from our customers.
From a profitability standpoint, gross margin on this segment of the market is probably a step down versus what we would see on the leading edge, but because there’s less investment from an R&D standpoint, from an operating margin and a cash flow standpoint we’re fairly indifferent. It’s a very nice business for us to drive. And so as that matures, I really liked the aspect of dampening overall industry volatility and still providing a nice driver of operating profit and cash flow for us based on our strong position historically in this industry.
That’s helpful. Maybe kind of switching gears a little bit to the memory side, in DRAM you talked how you feel really good about your competitive positioning. Maybe talk about that, and how do we think about your position going into kind of, let’s call it a higher spending cycle for DRAM equipment relative to maybe the last one. And there are certain types of; I guess applications or areas where you’ve gained share that that we should be thinking about?
Yes. So a great question. And I think it’s one of the underappreciated parts of the story of applied lately. Let’s start in 2012, in DRAM we had less than 15% share; today it’s 20-plus percent share. So the team has done a great job since Gary’s been here, really building out our footprint and innovation platform, serving that segment of the market. And as I look at the strong out-performance this year in that market and the set up in the next year, I think we’re going to continue that momentum as we look into next year.
What’s driving the growth? Since 2016 we’ve gotten – we’ve gained 30 points of conductor etch share in the last four years alone. Then you look at what’s happening from a multi patterning standpoint. This is a multi-year growth story for the industry, our position co-optimizing CVD patterning films with our Sym3 etcher really important part of the story. We’re delivering great performance enhancements for customers and get them to accelerate their roadmaps as part of this co-optimization of steps between CVD and ETCH. When you look at the uplift that that provides from a CMP standpoint, it’s a nice adder from a CMP perspective.
And as high-k metal gate gets driven into the periphery of the logic transistors to create higher I/O speeds on and off the chip. Traditional sources of strength, it’s great for our metals business. Our PVD business is up over 40% this year, and it’s just been a real strong driver for us. So I think how we’re situated, the innovation we’re bringing to market and what’s happening in that segment of the marketplace really sets us up well to continue strong performance going forward. And again, given the setup around next year and the growth of the DRAM market that I think is a really nice setup as we look into next year, given that momentum that we’re seeing.
That’s super helpful. And, maybe on the NAND side, Dan thought you made an interesting comment a couple quarters ago, where you talked about the inefficiency of spending for bit production back in 2017, 2018 timeframe. From your vantage point, I mean, how do you think about us getting back to a point where the industries may be absorbed that capacity and we’re starting to kind of move to a situation where we need to add net new wafer starts to dry really drive big growth?
So I think we’re in the early innings of absorbing that that level of spend, and as I took a – take a look at bit production today, it’s far more efficient than it was two, three years ago. Level of investment and productive bits out. You saw concurrent investments in plainer and 3D technologies at the time; much like you saw in 2000 with concurrent investment in 200 millimeter and 300 millimeter geometries, concurrent investments in technology tends to spike capital intensity. And we saw that in the 2000 timeframe. I think we saw that in the memory market in 2017 and 2018 when bit production for the level of spend, I think it was fairly inefficient for the industry. And I think what you’ve seen since then is the roadmaps have become a lot more efficient in driving that supply growth. And you haven’t seen the industry need to add wafer starts to dial in bit supply commensurate with that demand.
And I would say that they’re probably going to be that way for several more years. I think you’re going to be in a technology node transition mode. You lose wafer starts out of the factory as you migrate to the next node. You add a little bit of Greenfield back. So implicit in this is there’ll always be a little bit of Greenfield ads, but the overall manufacturing footprint of the industry wafer starts per month. It feels pretty consistent at this level for several more years, as the customers just continue to push the technology roadmaps and the node migrations.
So, on that, right, I would think, and maybe you’re interested in your thoughts is the plans of your customers to transition to next generation nodes or higher layer accounts. I assume that visibility is maybe a little bit better than some of the Greenfield wafer expansion plan. So that give you, I guess additional confidence or visibility in the demand profile on the NAND side?
Yes. So I think from our perspective, what really shapes our perspective is, is where is the demand in the end markets. For NAND mid-30% bit growth plus or minus depending on which customer you talk too. DRAM it’s mid-teens plus or minus, but mid-30s, we don’t see that changing. And so now the question is, is how is the industry going to respond to that? And so, as long as that demand statement holds together, and I don’t see any reason why it wouldn’t as a result of these secular trends we were talking about earlier that’s going to create a strong demand statement that our customers will meet with supply. And the interesting thing is, is today you see it happening through node migrations, but when you build new capacity, what is the longest lead time item? It’s the shell, it’s the clean room space, it’s the utility footprint.
And the industry has built a clean room footprint to support a structurally larger semiconductor industry. So they can add that capacity very quickly as the demand materializes. And you see it across all the device types. That is a strong lead indicator. Again, getting back to all of the data points that continue to point along the line of structural growth of the semi-industry. Today $500 billion on its way to $750 billion on its way to $1 trillion industry. The infrastructure is being put in place to support that level of structural growth.
So we see no migrations today, but eventually you’re going to see the way for start ads and you see it consistently and foundry/logic. You’re going to see it over time in DRAM, and you’re going to see it over time in NAND. And again, it’s another thing that just gives us a lot of strong confidence that we’ve got a good read on where we think the industry’s going, and we see this as a strong lead indicator confirming that view.
Perfect. Maybe switching gears a little bit, I think, let’s talk about something that I think is still kind of underappreciated for semi-cap in general, and that’s the services business. You guys have continued to report some pretty impressive revenue growth on that side of the business. Talk about one of the drivers of the outperformance and maybe some of it, some easier comps in last year as we kind of start to increase our capacity utilization in some of the fabs that were taken down on the memory side?
Yes. So we think we’ve got a great service business. And as I look at the growth of the business, the team is executing really, really well on the strategy. It’s a business that over the last, I think four years has outgrown the overall WFE industry by about 50%. So the team has been executing against the opportunity. What are the drivers of growth? It all starts with install base, largest – industries largest install base, very large footprint in a good year and a bad year that installed base footprint grows. And right now we do about 2000 systems a year give or take and a good year we’ll do more than that. In a down year in WFE, we may do slightly less than that, but the install base grows about 5% per year. Then you think about the increasing complexity of what we sell today versus what we sold the decade ago that has a higher service entitlement on it.
So there’s just a natural upward bias on the service business, based on the complexity of the systems that are shipped today, versus what we’ve done historically. The last thing I’d point to is, again this gets back to the strategy of the business. We’re going towards more long-term service agreements that have a subscription like revenue model versus what we used to do, which was more transactional. Even though services is a steadier part of the portfolio, what we’re driving today drives more consistency in that revenue stream. If you were to go back in time, a few years, 40% of the spares and service revenue was from long-term service agreements. In 2018, we crossed above 50% today. It stands at 60% is in the form of long-term service agreements.
And then the other dynamic you see is, is if we were to go back in time the vast majority of what we would sign with customers had about one-year duration. In 2020, a full third of what we signed from a long-term service agreement had a tenure of three years or more. And so really strong proof point that we’re adding a lot of value to the customer base, willing to sign up for multi-year journeys. So I liked the strategy. I like the execution against the opportunity. And today, as we look at our install-based business, which is the AGS reporting segment, plus 300 millimeter upgrade, it’s a full third of our revenue. And so I really liked the setup, the execution and what that means for the profile of the company over time, as we continue to grow that segment.
That’s helpful. And maybe that for some that aren’t as familiar with the services business, I mean, we’re talking a lot more than just delivering spare parts, replacement parts. Talk about some of the value that, that the services bring and why a customer would want to sign a long-term contract?
Yes. So the interesting thing is, is when we sell a new system what we do is, it’s on a warranty. And what we’ll do is we’ll put a data server – high-speed data server right along in that tool. What we’ll do is for the first year it’s on warranty we’ll be extracting data off the tool and we’ll understand based on a global footprint and fleet of systems, what’s operating really, really well.
And what’s lagging from a performance standpoint, and you get an opportunity to drive performance throughput yield, performance characteristics of the tool based on a global footprint and experience. What happens when it comes off warranty customer has a choice, sign a long-term service agreement, keep the data server in place and drive higher performance or not sign a long-term service agreement, and then the data server comes out.
So there’s a strong incentive to drive much higher levels of performance. The other thing I’d say is as we’ve got customers who operate in a home geography and then have fabs built in other geographies. And what we’ve observed is a service entitlement and outside the home geography is higher than what we see inside the home geography. And the interesting thing is, is the tools that are under that higher level of service in many respects, operate at higher levels of performance than what they’re able to drive in their home geography.
And so it changes the nature of the dialogue with customers when they see that difference in performance. So this is a real high value added part of what we offer to our customers. It’s definitely not just transactional. The phone rings, we pick up the phone, we take an order and we ship apart. There’s much, much more going on in this business.
That’s perfect. Unfortunately I think I could go on for another hour, but we’re out of time.
So guys, I appreciate you attending the conference and look forward to speaking soon.
Great seeing you, Joe, take care everyone. Thanks for joining us. Enjoy the holidays.