|A few excerpts from AMAT's CC, with a focus mostly on comments about memory. Full transcript here:|
I have added a couple of my own comments to the transcript below, enclosing them in brackets. There are more questions in the Q&A, but I only put the ones that have to do with memory here.
Gary Dickerson opening remarks
Let me start with the big picture. This is an incredibly exciting period for the electronics industry, with a broader set of drivers and a wider spectrum of companies making very large investments to advance semiconductor and display technology. In the past, PCs were the dominant factor in semiconductor demand. PCs drove the technology road map and enterprise refresh cycles drove industry upturns and downturns. More recently, global adoption of smartphones, combined with social media, have created a much more pervasive consumer demand, expanding the market and making it substantially less cyclical. Today, many new demand drivers are emerging that layer on top of traditional computing and mobility, technologies like the Internet of Things, big data and artificial intelligence are transforming industries, including transportation, health care, entertainment and manufacturing. The way these industries create value is increasingly dependent on capturing, transmitting, understanding and storing data. In turn, this means they are more and more reliant on advanced semiconductors. The impact for Applied is twofold. First, our markets are growing and becoming more stable. Second, this is a period of incredible innovation in logic, memory and display and we're in a great position to provide the critical building blocks needed to move the industry forward.
Having provided this context, I'll now describe the major inflections and investments taking place within our markets. Sales of memory chips are at record levels, fueled by increasing content in smartphones to support better cameras and VR applications as well as the need for more and higher performance storage in data centers. Market fundamentals remain strong and as a result, we're seeing robust investment from our memory customers. We believe we will see double-digit growth in DRAM spending this year and we expect NAND investment to be even stronger. Our view of 2018 and beyond is also positive for a number of reasons. The explosion of data storage requirements created by IoT, big data, AI and streaming video has only just begun. Data generation from new categories, such as Industry 4.0 and autonomous vehicles, can potentially dwarf existing applications within a few years. As 3D NAND bit density increases and cost per bit falls, new segments of the storage market are opening up for solid-state drives. And at the same time, the bit density increase from generation to generation is slower for 3D than it was for planar NAND. To compensate, the industry needs more wafer starts and greenfield capacity.
Robert Halliday opening remarks
Before going into the details of the quarter, I'll touch on 3 things that are sustainably better for Applied, one, our markets; two, our improving position in those markets which is enabling stronger gross margin performance; and three, our profitability and free cash flow. First, our markets are more attractive. We believe wafer fab equipment spending will be higher and less volatile for the foreseeable future. Display equipment spending also looks higher and more attractive, for Applied Materials in particular, because we're greatly increasing our served addressable market. Many of you have added valuable insights as to why WFE has become larger and less volatile. I believe it boils down to three things, one, the semiconductor market is growing and becoming more diverse; two, equipment intensity is increasing; and three, capital investment is measured and rational.
This better industry environment is sustainable. As Gary mentioned, the semiconductor industry we enable has also become larger, more diverse and more critical to big innovations in the global economy. Today, we're seeing the emergence of new silicon-rich devices needed to enable the Internet of Things, cloud data centers and artificial intelligence. Data is becoming more valuable and harnessing the value requires silicon. For Applied, this translates to higher demand for logic and memory capacity at both the leading edge and trailing geometries. I believe the industry has seen a new wave of sustainable growth with economic value creation that justifies the cost of investment.
Another way to measure Applied's stronger position in the markets is to consider the changes in our share by semiconductor device type. In 2012, our share in foundry was over 20%, but our share in logic, DRAM and NAND was under 15%. Today, we believe our share is at least 22% in all 4. In fact, we're the #1 equipment company by revenue in NAND, DRAM, logic and foundry. So regardless of the spending mix, applied is well positioned.
[Halliday goes over here the extremely impressive details of AMAT's quarter and guidance which anyone interested in the company should certainly read but I will omit except for the last summary paragraph:
"To summarize, one, our markets are sustainably better than at any other time in the history of the company; two, our competitive position and execution is sustainably better, giving us a strong pipeline of new and disruptive products and gross margin expansion; three, this growth, combined with disciplined investment and spending, is generating higher free cash flow and we're committed to returning the excess to our shareholders."
The key theme for both men is that they believe that (a) their growth is sustainable, and (b) this time is different from all other times in the history of semiconductors, mainly due to the centrality of semiconductors in the modern post-industrial economy. IMHO, it has become the equivalent to energy in the industrial economy.]
[And the first question is about that key term: sustainability]
I guess question around sustainability. You're guiding WFE $40-plus billion here for '17. And curious, what gives you the confidence to be so upbeat looking into '18 and beyond? And I guess as part of that, would love to hear your thoughts in terms of the contributions from areas like AI and public cloud investments versus, I guess, more traditional areas like smartphone and PCs in the past.
Sure. Let me see if I can try to give you context. I think context rather than point answer is useful, give you context over time and context over breadth. So if you look at context, we thought back at Analyst Day last September that 2016 WFE would be about $33.7 billion. And in fact, it ended up a little over $35 billion. We thought at Analyst Day last year, this year would be $34.5 billion. And in fact, it started with a 4 number this year and it looks very healthy. So it's trending up. So then you say, well, where is it trending up? It's kind of up across the board. Earlier in the year, we thought it was kind of 5% up across the board. Now we think the whole thing is up 15%.
We're seeing in DRAM, NAND, foundry. Tactically, we're seeing high utilization in all the fabs. We see no diminution of ordering patterns. We don't see double bookings or anything people might fear. So we see -- when you see strength across virtually every customer and virtually every device type, that goes to some root cause questions. So in my opinion, we're facing what we've been talking about for a while which is, there is more root cause demand drivers for devices. Now longer term, those root drive cause factors are going to be things like cars and AR/VR. What we're seeing now is tactically more content in the phones. We see a lot more content in the phones around NAND, DRAM and we also see strength within the processor side on both the cutting edge and the lagging edge. Lagging edge, again, this year, will be over 40% of device types. But the cutting edge is pretty strong, too. So we're seeing broadscale demand.
My belief is that in the intermediate term, there's more content going in phones and more going into the big data centers which are both driving this which will help us. So then the question is, why is there more going in? We believe that there's, what I call broadly speaking, more applications for processor technology, more apps on top of it that are driving systemic demand. And I think, frankly, we're in the early innings of it. So if you ask me, okay, is the second half good? Yes, we're okay. If you ask me, is '18 good? Yes, I think it probably starts with a 4 again. It's pretty good. If you ask me long term which is what I spend a lot of time on, I see more root cause drivers for growth in semiconductor and silicon growth than I've seen in years and I think it's sustainable.
The question I had was WFE running at 40 billion, 40 plus billion this year and next year, where do you think 3D NAND WFE is within that? And if the prices for NAND roll over, the reality of economics kick in and 3D NAND makers scale back the CapEx. Or do you think demand is strong enough to continue investing in capacity for the next 2 to 4 years?
Sure. So the question was, is 3D NAND going to keep expanding? Is that -- yes. So let me take a shot at that. So 3D NAND spending is up this year. It was up again last year. If you look at it, there's about 1.6 million wafer starts in the world of 3D NAND -- NAND in total. By the end of this year, they will convert at 750,000 wafer starts per month. So there's still a fair amount that's going to convert. I think that the vast majority that's going to convert over time because 3D performs better than 2D. I think what also helps memory in general, particularly NAND, is that more and more customers' customers are starting swap out 3D NAND over time for hard disk drives. So overall, demand for memory is going up. The share of NAND, far versus how this drive will trend up and 3D will be the NAND of choice versus 2D. So they're early in the build-out, 750,000. They got to get to 1.6 million. And we think that the installed capacity is going to go up over time because of this demand. Then if you look at greenfield versus refresh, we're kind of agnostic, frankly. So if you look at the cost of doing a greenfield 3D NAND factory, we do well. In fact, we're gaining share at virtually every product, I believe. If you look at transition of one 3D to a higher level of 3D, the revenue opportunity for us is about the same.
One thing I'd add to that is that I think everybody is seeing a large increase in the amount of data and also the value of the data. If you look at deep learning, the amount of data that you process is going to go up. Right now, a lot of data's thrown away. So you've got an increase in the amount of data and then the amount of data that's processed is also going to increase. So both of those factors make me personally pretty bullish about the memory business long term. And certainly, that's also what we're hearing from our customers.
I wanted to follow up on some of the questions on the upward revision in WFE, particularly with regards to memory. We look at sort of memory cash flows and make some assumptions that there's just an algorithmic percentage of that peak cash flow that gets spent which is probably more negative than what you guys are describing that there are sort of people who are looking at the current situation and deciding that they need more supply. I mean, I guess it's more semantics, I guess. But how do you characterize when you see the upward revision that we've seen in just a 6-month window? How much of that is just because prices are better, companies have cash flow and they use it to improve their competitive position versus something that's more sort of structural and dynamic? How do you interpret it as more structural aspect versus just better pricing?
We look at a few things. Yes, we dug into this ourselves, too. We looked at what's driving this data. We looked at some cosmic data. So we looked at the stuff like how much data is captured out there. And we capture -- it was like 8 zettabytes of data was captured in 2016, I think, close to 25 zettabytes of data in 2020 and then it goes to like 45 zettabytes of data. So more and more data has been captured. It's gone up a lot, even in your homes, your cars, your industry, your companies. Then we look at the transmission, processing, source of that data we talked about. So when we start to look at root causes, we've seen a lot of this stuff, right? So then you go look at what's going on more close to planet Earth. And so we looked at this and we said, well, what's going on with NAND and DRAM? We split them up. DRAM's up but it's the still not through the roof. It's pretty good but still kind of restrained because a fair amount of this budget's being spent on NAND. So DRAM price is good. NAND price is good.
Supply and demand's okay. So the DRAM, I would say, has probably gotten a little more upside than downside, frankly, in '18 because this year it hasn't been through the roof. It's up from what we thought earlier in the year. And then if you look at NAND, they're selling all the 3D NAND they can build. So it's still pretty good and I do think it goes to these longer term trends they have at NAND. So then if you look at the -- the final thing is just, well, how come you guys are low early in the year? And I think it goes to 2 root causes which, frankly, inflict all of us. One is I think we're a little bit a prisoner of past thinking. So okay, it can only go so high. But it goes to what Gary said earlier. There's different fundamental demand drivers at a simple level. It was, hey, use PCs. Then we agreed, okay, in 2010, mobile's coming in. And then -- but now you had all this big data stuff in content which is going up. That's what's really up last year. So it's content in the phones, content in the data centers, right? And so when you start to look, what were we inflicted with? We were saying, oh, we're going to roll over. But maybe there's real demand drivers here, number one. And number two, sometimes we listen too closely to our customers on a tactical basis.
They have to give us good demand forecast for kind of 6 months out so we could build the tools. Beyond that 6 months, maybe they're a little conservative to what they say to us for various reasons. So I think there are tactical reasons why we're a little too conservative. And I think that will -- data that shows it's not overbought now and there are longer term reasons that make you think this is very sustainable.
You mentioned earlier on the call regarding bit density and 3D NAND from generation to generation being slower than the planar NAND conversion. So just wondering, could you give us an idea on how much more fab capacity might be needed annually to drive, say, 40% bit growth and what the incremental opportunity can be for AMAT?
Sure. Some of this goes to what I said earlier. So we see and I think we're a little bit on the low side sometimes, that NAND bit growth is going to be probably in the high 30s or something like that. Now last year was much higher. If you look at content in phones, it was up like -- the data we got from an outside service provider was 57%. So there are indications that we're on the low side on some of these things. Plus we're not sure we've captured all the movement to hard disk drives. So I'd say I'm more on the over camp in that number than the under. So then if you go look at it, as I said earlier, the 750,000 wafer starts by the end of this year, we think that they're going to have to continue to spend in dollars WFE content similar to this year for a number of years. Now if they spend on greenfield or they spend on conversions, we're kind of agnostic from a revenue point of view. But we think the total spending is similar to this year for a number of years.
Just want to circle back on China. You guys mentioned expect a pretty significant pickup in spending in China in 2018. Maybe if I -- we focus on the memory side. We've seen like some of these customer announce pretty aggressive road map for the 3D NAND technology. Just wondering, is this bigger spending in '18 predicated on those customer hitting those targets or rolling out those devices and if they have issues going on with the devices that delay the spending that you expect in '18?
We have a range of forecast for China. So as I said earlier, it's one of our strongest regions in both semiconductor and display. We have very, very deep relationships with many of these customers. Right now, what we believe for 2018 is that the business will be up meaningfully from 2017. But frankly, that's also at the most likely and low risk end of the forecast. There's a higher range there that would be going up at a much faster pace. We're looking at early indicators for all of those new projects, but I think right now, we're pretty confident that the increase in China in 2018 will happen. Maybe it's $1 billion, $1.5 billion in terms of total wafer fab equipment. So that we have pretty high confidence in terms of the engagements or the information we're getting from all of those customers. There is upside potential, but that's our current view.
Robert Halliday (concluding comment; back to sustainability)
Sure. Thanks, Mike. One of Gary's favorite expressions around here is innovation's about connecting the dots. Let me see if I connect a few dots that I -- we believe and I hope you heard today. One, we believe the wafer fab equipment market is sustainably higher and less volatile. Also, display is higher and for Applied, it's sustainably a better market because there are more technology inflections and because we're growing our served market. Second, Applied's position in the market is sustainably stronger. And we're executing better and better. And third, we're generating more free cash flow and returning the excess to our shareholders.
Speaking of free cash flow, one of your own interesting note comparing Applied Materials to some of the top names in the industrial sector. Relative to the average of the companies, Applied has higher revenue and profit growth along with higher ROIC and free cash flow margins. Compared to these companies, I believe we're being discounted for being more cyclical even as we become demonstrably less cyclical. I believe that over time, we'll be viewed and valued in a new way. So thank you for your time today. And we look forward to seeing many of you in person over the next few weeks.