Western Digital Corporation (WDC) Earnings Call Transcript & Summary

Good morning, again, and welcome back, everyone. This is Tom O'Malley, Mid-Cap Semi Analyst at Barclays. Pleasure to have Western Digital here. We've got Peter Andrew, Vice President of Investor Relations; and also Dr. Siva Sivaram, President of Technology and Strategy. I think we'll hand it over to Peter to read some disclosures first, but gentlemen, very happy to have you both here.

Thank you very much, Tom. Yes, I've got to read the safe harbor disclosures here very quickly. We'll be making forward-looking statements and I ask that you refer to our SEC filings for the risks associated with these statements. We will also be making reference to non-GAAP financials and a reconciliation between our GAAP and our non-GAAP financial results can be found on our website. So with that, Tom, let me turn it back to you.

Perfect. Great. So, given the -- given the level of knowledge you have in the technology background, Doctor, can we maybe start at a very high level. Can you -- can you just share your thoughts on what makes Western Digital unique from a technology perspective? You can start on whatever side of the house you want, but what key areas do you -- are you differentiated and that make you very unique from a competitive standpoint in the market today?

Tom, thank you for having me here. Let me start from the drive side, from the hard drive side. As you know, Western Digital has a very rich history of true technology breakthroughs over a very long time, whether it's on the head or in the media, you've seen what we have done with respect to this big technological improvements with energy-enhanced PMR, or whether it is on the mechanical side with triple-stage actuator or our leadership in Areal Density and more recently, what we've done with OptiNAND, which is a use of Flash-based technologies in the control plane of HD, which naturally leads me to the Flash side of it where we have been the leaders in multi-level cell technologies for a very long time. We have always been the pioneers taking 1 bit, 2 bit, 3 bit, 4 bits per cell. And ability to laterally shrink and get technologies into this right level of leverage with respect to the Z-direction. Vertical integration where we have our own controller, firmware and memory, all working together seamlessly. And the segmentation into the product where we can take it from retail to mobile, to client, to enterprise in the same Flash coming in going and doing. So, we have done this very, very well over a long time. And what now is unique is storage. The need for data is exploding and we are the only true full-service, entire full spectrum storage company there is. Given the more than 1-zettabyte, 1.5-zettabyte of hard drive bits being shipped and 0.5-zettabyte of Flash being shipped, we are growing in both at a substantive manner, fully leveraging our market knowledge from one with respect to the other -- technology, one with respect to the other. And of course, the channels and market between the one and the other. So, the 2 verticals of hard drive and flash, both servicing that big growing data center market. We are unique in it. We are a technology company, always founded and growing on technology, but these 2 big growing vertical markets provide the tailwind to us.

Very helpful. And I think to a point you made, if you look at the environment in the market right now, you've seen all of these supply issues and you've seen inability to get product. And you talked about your vertical integration, right? And your ability to source controllers in-house, et cetera. And how do you think that your ability to keep more of your ecosystem internally has helped you through this? And has it given you a competitive advantage just having all of this product very close to home?

Yes. So, it is an interesting aspect of the way we do business. The wafers are coming out of the fab, they are the same wafers when we grow from generation technology to the other. What we have the freedom in vertical integration is how we allocate the bits. How we optimize the allocation of the bits for the maximum revenue and profitability. So, when there is, let's say, a controller shortage on 1F from a foundry in Taiwan or elsewhere, we are immediately able to, hey, I will re-trade these bits to the higher value, higher density segments, so that we can optimize for earnings. It's always a problem when you can have semiconductor shortage. But our ability is in this flexibility of the supply chain to make it to the -- in the end to obtain the right financial results for us.

Very helpful. So, why don't we start on the Flash side? So, I think just given the past year, you've heard a lot of different conversations about movement in the industry, potential M&A. But in the end, I think what's really important is that a lot of people that look at your company don't really understand the Flash joint venture and the reason why it's so special in working with your partner for so long. So, why don't you take a moment? Can you just describe the dynamic and why it's beneficial from you from a competitive perspective? And then also from a technology perspective?

Let me take some -- allow me to expand on this for a little bit longer. The Flash JVs between Toshiba and SanDisk Guard started in late 1999, 1998 time frame. So, this is a JV that has survived 22 years, a phenomenal length of time for any technology partnership in any industry. Between the 2 of us, the scale of manufacturing, even though each of us is 15%, 16% or 20% on the other side. Together, we have the largest install base of Flash manufacturing in the world. So, each of us gets the leverage of other. Even though we have only 15%, 16%, we run as though we are a 36%, 37% market share fab. So, this, we probably -- the joint venture fabs in Japan are probably the third largest fab complex in the world. Meaning, we produce close to, I want to say, 500,000 wafers a month. I mean, that's the kind of install capacity we have between the 2 of us. So, manufacturing scale is a very important number. The second is R&D scale. Because we pool our resources together to develop process technology, we probably for the 15%, 16% of the bits that I have, I have the R&D scale also of a 35%, 36% player. We are consistently been the technology leader with our own unique bend of how we want to get there. So for instance, you can see us optimizing for bit growth, capital intensity and cost reduction consistently. Year-over-year, we are the market leaders in cost reduction. And you've seen this. If you run your own models, you'll see our cost reductions have been stellar. This comes from the fact that the 2 of us are able to look at the market in completely different perspectives. We look at it in our retail and client dominated bits and where we are going with enterprise. They look at it with their own customer base in mind. Together, we are able to look at the entire spectrum of the marketplace to optimize the technology for the future. And that's a huge advantage. We continue to be growing the JV. The JV is already signed and in place for the next, whatever, 2029 or 2030. All that time we are already in place. We've started a new site in Japan in Kitakami where those fabs are signed in place till 2034. So, this is a very long-term joint investment where we pool our resources to provide the best leverage for each of us with respect to getting the most of out of technology and manufacturing.

That's helpful. So, you talk about cost reduction and then you also talk about technology development. Those 2 go hand in hand. So, could you take a moment to just talk about the Western Digital Flash roadmap and where that sits versus competition today? Can you talk about the several nodes that have been announced thus far and then the time line for those nodes over the next couple of years? Just give us a reminder from what you said already and kind of where that technology roadmap is set in?

Let me put some perspective on this. We were in 2D NAND, the leaders in technology for long time. We also had -- the JV had the first 3D NAND technology announced. Even though we had the technology, we always said that till the true customer-oriented cost reduction happens with respect to 2D NAND, we won't introduce a new technology, even though we had the technology. So, when competitors are talking about, oh, I have a 24-Layer, I have a 32-Layer, I have 48-Layer, we said at 64-Layer is where the true cost and benefit crossover happens. We will make sure only at that time will we introduce it in very, very high volume. And we did that and we were the cost leaders when we started introducing 64-Layer. So from that point, we've also been very prudent in the way we move technology back. The most important thing we want to watch is the size and scale of that single operation in Yokkaichi in Japan, allows us enormous reuse. So, we look at capital intensity, how much capital do we want to spend for the amount of bit growth, knowing our install capacity of tools. So, we watch for that. So, when we went from 96-Layer to the 112-Layer, we are the lowest capital intensity 3D NAND node disk. Meaning, for a fixed percentage of bit growth, the amount of capital we spend is the lowest. So, when you know how much cost reduction you need and how much bit growth you need, we specifically optimize on that capital intensity. Others would come back and give you on why have a 128-Layer, whey have a 176-Layer. That does not matter to us. For us, cost, capital intensity, bit growth, what is the lowest to go get that. We achieve that between Yokkaichi and us by leveraging, a, our existing install base, b, philosophically making sure that we shrink laterally. If you shrink laterally, when you multiply by the big number of a 112-Layer, I get a lot more bits per volume. People don't talk about it. People just talk about I have more layers. In our mind, if you end up with more layers, you lost. Higher number of layers is bad news. Every time, they said, hey, we introduced a 128 or 176, I said, oh my God, poor guys, what happened to them. You want to get the lowest number of layers to get to the same bit growth and cost reduction you want. You can see we will get the lowest cost reduction every time due to the lowest layers. So currently, the 112-layer is ramping. I think as of this week or so, we are -- the majority, the virality of the bits coming out of the 112-Layer, bit 5 is higher. We have cost over it. And we will be ramping that into full spectrum across all product lines throughout calendar '22. Second half of calendar '22, we will introduce our own 162-Layer. And that will get into meaningful volumes very late 2022 and into 2023 is when we will be commercially more important. So, every 18 months or so, we continue to introduce the next node. We don't want to introduce very shorter node. That does not benefit anybody, because you don't get the yield improvements and the DPPM quality improvements and give the customers the chance to absorb the different qualifications. So, trying to do nodes shorter than 18 months is usually not productive over the long term. You can see some of our competitor saying, oh, I'm going to do within 1 year, I'm going to do here, I'm going to do in 14 months, is usually not the right thing to do. But about the 18 months is the right time frame to introduce the next node. And you've got to ramp literally vertically, very, very fast. So, we have talked about going from 96 to 112 to 162 to 200 plus layer, which will come 18 months later. So, this roadmap looks very strong for the future.

Very helpful, very helpful. So, I think when you look at that technology and congrats on the crossover, I know that you guys had mentioned that you were going to get it done in the second half. So, congrats on hitting that mark. When you look at the technology overall, you talked about longer nodes. Can you talk about why layer count, sometimes is used liberally, and that's not always the best reason? But something that's interesting that I think investors will focus on is, from an end market perspective, the difference is in technology, where one end market like an enterprise end market may need something very different than a guy in the cloud end market would want. So, can you talk about what are different pieces of technology? And how do they apply to those 2 end markets? Let's just use enterprise and cloud for an instance, what is very different about those 2 end markets? And how do you meet your technology given the length of your node to match where those products end up?

We have refined that process to signs over the number of nodes we have done. As you would expect, mobile requires end user reliability at one end with relatively low power with very few die because the densities are lower. On the flip side, enterprise, very large capacities, meaning, you are talking about 2, 4, 6, 8, 16 terabyte in, so which means you're going to use 64, 128, sometimes as much as 1,000 die into a drive. On retail, which could be extremely -- a microSD is smaller than your thumb nail and you need to get something into it. You want to accomplish all this with a minimum number of die in the fab because yields are the most important. You want to get to the highest yields as quickly as possible. During the ramp-up, when the first wafers are coming out, when you're starting to do an internal qualification and you are very rapidly ramping up, during that time period, you ramp with 3K, because that's the product that we do the call, we shift into the marketplace very, very quickly. Immediately afterwards, we take it into client SSD. Client SSD market is a very large market. We will have about 30% market share in client SSD, highly segmented all the way from channel entry-level products to OEM high-density products. We are able to take the same product, qualify early, ship into the channel, et cetera. Mobile runs in parallel, so you customize, you trim the bits, you trim the die for that characteristic, very high reliability, low power and then comes enterprise. Enterprise, because it is a large number of die, you worry about it in an aggregate. I want to make sure when I ship and drive with 500 die in it, how do I make sure I can manage any 1 bit failure in one place or the other. How do I take care of the error correction across the board. You want very mature die to be able to do it. So, those qualifications and features -- in the vertical integration, there are 3 major links; the memory die, the controller and the firmware. The 3 have to work extremely close to each other. The memory die produces the basic storage, the controller acts as the engine that directs the traffic. That engine has to be well crafted to having the PCIe Gen 4, PCIe Gen 5, and knowing what's happening, this and that, we need to have all of those in the controller. The firmware in the end is that traffic cop. It decides that bit is getting worn out, I want to redo it, and that needs to be very, very mature for an enterprise. And so enterprises are the trading part of the trailing node. It is not -- retail is in the leading part of the new node, enterprise a trailing part of that node when we are ramping. That's the way you think of the entire marketplace.

Very helpful. Very helpful. So, we've touched upon on the technology side to start. I was curious if we could spend some time talking about the health of the end markets today. So, you've looked from a semi cap perspective, you've seen this massive investment over the past year, particularly on the Flash side. But what do you think the right way to think about long-term growth is? You've seen all this technology investment. You hear the semi cap guys talking about a level of investment that's required as you advance in technology. Can you help me figure out the balance there? How much is additional investment is needed as you advance in these technologies? And is bit growth going to consistently stay at this higher level? Or will you see some rationalization longer-term just given the technology complexity?

So, let me start with stating the publicly available obvious data. All of the Flash manufacturers at this point are behaving rationally. Nobody is going out to grab market share at the expense of prices, trying -- nobody is doing that, which is a good thing. The market is behaving rationally at this point. Having said that, there are 2 or 3 things to think about. One is, if someone is adding greenfield capacity as opposed to conversion, we are 95% conversion, 5% new wafers. You take an existing capacity and convert it to the next node. That's the reuse I was talking about, we are very extremely efficient about. There are others who don't have the existing capacity, they end up adding greenfield capacity. That's much more expensive. You'll see larger CapEx layout from them because of that expense. We achieve the bit growth primarily through technology, use the technology to achieve the bit growth that is needed without having to add 2 million new wafers. That's the second part of the equation that is different for different numbers. The third is what we talked about earlier. Some of us have a little bit more capital-efficient than others are. And there the technology when it drives higher capital expenses, you end up seeing that. And you can see from our semi cap suppliers as well. The ones that have a higher exposure to memory, you can see how they talk about it. It's not a uniform message. In general, capital intensity is growing. So, to get that additional 1% bit, you now need to spend a little bit more capital. However, there are better ways to address it and that's where the technology makes the difference. And we are focused on making sure that we minimize capital spending to produce the same bit growth. Looking for the longer term, last year, so calendar '21 was built on a very low calendar '20. So, our bit growth rate was a little bit higher. But going into calendar '22, we expect us to grow at the same rate as the marketplace, somewhere in the low to mid-30s.

Very helpful. Very helpful. So, why don't we switch side to the house here back to the drive side. So, you've seen this very strong year. I think you had a couple of unique circumstances, which helped drive bit growth, but you also have seen this real tailwind from nearline drives and hyperscalers moving to -- moving not only to spend more, but also to use more storage in the drive side of the business. So, could you just talk about the health of the drive business? Do you think that this -- this trend that you're seeing in cloud is a sustainable one? And then from a mix perspective, obviously, you have this rising tide with nearline. You have this commercial PC market that's rationalizing on the other end. Can you just talk to that trade-off? And is that a healthy one that you can kind of overcome? Is the good better than the bad, I guess?

Yes. So, let me start with the overall picture of the hard drive business. Think of 3 large segments. One is the client PC segment where Flash has been consistently replacing hard drive in that segment. That segment is in a secular decline over time. But on that same form factor is smart video, very large growing segment that offsets some of the decline in the PC because of replacement from Flash. Then the data center large capacity, capacity enterprise drives. That in the last, say 6 years or so have come to dominate the overall mix in how much bits are being shipped. We have always led in that segment. And despite our small misstep in 16-terabyte, we're back to being where we are. We are the #1 volume supplier in the capacity enterprise segment. In this background, that capacity enterprise drive growth was often subsidized by the decreasing client for a long time. No, longer the case. We as much as we we've have moved, we've already moved. So, from this point on, any expansion has to be done by capacity addition and capital investment in the drive capacity itself or capacity enterprise. So, that's on the supply side. On the demand side, as you said, just as we were coming out of our earnings call, you saw all of the major cloud players calling out that they are increasing capacity and increasing capital spending. The metaverse is happening. So consistently, demand for hard drives is absolutely a unstoppable phenomenon in terms of its growth. Given that the hard drives are coming down in cost in 2-digit percent year-over-year, Flash is also coming up in mid-teens. The 2 are going to live together for decades more. It's not something that is going to get -- there is not -- especially in the data center, this is not a replacement market. Different parts of the demand, the warmer and hotter part of the data and the cooler and the colder part of the data are distinctly going to be supplied by these 2 that will in the long term, continue to exist for a long time more. And so overall, the drive business is enormously healthy because of this big growth engine and the smart video. Both are driving our positive momentum. Having said that, the cloud titans are very big. Each one of them is a very large buyer of capacity enterprise drives. Each one of them have their own unique dynamic. There's always a little bit of averaging out and lumpiness that go with it. We have had a cloud titan with who we have a disproportionately high share. When they have a semiconductor shortage issue where they can't build out their boards and et cetera, yes, you're going to have a little bit of lumpiness in that quarter. But that's not perishable demand, it will come back. It will come back as the semiconductor shortages ease up and they are big players, they now how to get their semiconductors, it'll come around. But you will see that a little bit of choppiness in the short-term here and there, but not -- in the longer-term look of it, capacity enterprises doing 2 things right. The demand is growing very, very nicely. On the supply side, we are consistently doing cost reduction in the 2-digit numbers. Between the 2 of them, that business is very, very healthy.

Very helpful. Yes, I just -- I think we have time for one more here, but I think we walked through the drive business a bit. I think in your preamble about things you were excited about, you mentioned OptiNAND. Could you talk about what do you think is the most innovative bit -- innovative trend or development that Western Digital is looking at into next year? Is it OptiNAND or you can talk to something on the drive side? But what one area of technology are you most excited about that could impact this business in the coming 1 or 2 years?

Yes. So, I actually want to go slightly away from you. We don't want it to be one thing. That's the philosophy of Western Digital. This -- take a homerun swing and strike out is not us. We have a cost reduction roadmap that consists of 3 or 4 extremely important things based on one basic underlying philosophy. We want to be the Areal Density leader. We are the Areal Density leader. We will continue to be the Areal Density leader. Areal Density is, how much bits can you fit in a platter. Whatever I can do to make sure I'm pushing that I will do it. So, there are 4 things I will tell you that we have done very well. ePMR, energy-enhanced PMR that we are the leaders. We have shipped over 100-exabytes of ePMR, leaders in the industry, that provides us the ability to put more bits on -- with energy-assisted recording. Second is triple level stage -- triple-stage actuators. On the mechanical side of it, to be able to do a shoulder-elbow risk strategy of making sure they can get more and more precise. The third, SMR, Shingled Magnetic Recording. The read track and the write track are slightly different. The write track is wider, aha, I can now go back and put 2 read tracks by singling them one on top other. It gives us a 15% or so advantage Areal Density uplift because of SMR. Fourth is OptiNAND. OptiNAND, where we take our expertise from NAND and come back and say, there is a lot of drive level metadata. For instance, when I read it, many, many times I write on the same thing that track wears out, how do I re-map it. How do I keep that information originally used to record it back on the drive. We don't need to. We have another non-volatile memory source. So, you can very quickly, very rapidly reduce the amount of DRAM, put it on our own Flash, increase Areal Density. These 4 will give me a roadmap all the way to 30-terabyte when hammer will be ready.

Well, I appreciate the color. It's very helpful to dive into the technology. Thank you for joining us. Both Peter and Dr. Siva, thank you so much for your time. Have a great day and have a great year guys.

Appreciate having me here Tom. Thank you.

Thank you very much.

You're welcome.