Western Digital Corporation (WDC) Earnings Call Transcript & Summary

Good afternoon. Thank you again for joining us at the BofA Tech Conference, day 3. We're down to the last few sessions of the day. But we're super delighted to welcome Western Digital now. There's been tremendous amounts of interest in this space from investor clients. And so we're really, really delighted to have the management team here. We have Dr. Siva Sivaram, who's President, Technologies and Strategy; and we have the CFO, Bob Eulau. So welcome to both of you. Thank you so much for joining us at our conference today.

Thanks, Wamsi. Thank you for being here.

Thank you. I know, Bob, you wanted to make some opening comments forward-looking statement disclosures. So I'll allow you to do that.

Yes. I just want to quickly remind everybody, we'll be making some forward-looking statements, and I ask you to refer to our SEC filings for the risks associated with these statements. We will also be making references to non-GAAP financials and a reconciliation of our GAAP and non-GAAP results can be found on our website. So with that, we're ready to go, Wamsi.

Awesome. All right. Great. Well, Bob, let's start with an overview of the demand environment. Can you just briefly talk about what you're seeing from a demand perspective?

Yes. I mean, we had expected 2021 to be a really good year for us, and that's the way it's started out. Really pleased with how it's progressing. If you think about the markets from a content solutions standpoint, which is really our consumer business, it's progressing very nicely, probably a little bit ahead of seasonal norms. And we're really excited about how things are going there. We've got over 350,000 outlets around the world. Most countries we participate in, we're in a #1 market share position and continues to be a real strength of our company. If you think about the Client Devices business, again, it's going very well. Strong PC demand and demand on the mobile phone side is really helping drive a lot of volume in that market. And it's across both our flash and our hard drive areas. So that continues to be really good this year. And then lastly, if you think about our Data Center business, we went through and digestion cycle at the end of last year, but we expected this year to come back strong, particularly with the cloud, and that's what we're seeing. Things are relatively stable on the enterprise side. We're hopeful that they'll begin to pick up towards the end of the year as we start to see more and more companies bring employees back into the office. So I'd say fundamentally, business is really strong across the board.

Okay. That's great. So Bob, maybe -- you addressed on the last call, some long-term agreements with some of your customers. Can you elaborate on that a little bit, please?

Yes. We did a couple of long-term agreements that we talked about in our last earnings call. And for us, long term, really means 2 to 3 quarter agreements. And we really think those are important. We think it's a good change for the industry. It had been an industry where pricing was negotiated on a quarterly basis. And we really wanted to get away with that and get away from that. And at the same time, we wanted to provide assurance of supply for our customers. So I think we struck a deal that makes sense for everyone. And as we move forward, we really think the industry is beginning to transition to one where we need to have longer-term agreements with each other. The reality is we went through many years when we were transitioning client resources and client capacity to the capacity enterprise side of the business, and we really don't have that available to do anymore. So we need to make sure that we understand where demand is and what kind of capacity we need to have in place.

Yes. No, that's super interesting. As you point out, that longer-term trajectory between excess capacity versus the utilization rates today is starting to change. Maybe, Bob, you can address, your largest competitor, a couple of days ago, preannounced positively on some upside in demand. Can you address sort of are you seeing similar trends? It seemed like they called it an industry-wide phenomenon, and also your quick thoughts maybe on Chia as a demand driver?

Yes. I mean, happy to do that, and we've gotten a lot of questions regarding Chia. I think the important thing to, again, keep in mind is, overall, we believe this is a strong year. We expect it to be strong, and that's the way it's playing out. We view Chia as a new vertical market, one that we're monitoring very closely, really trying to understand, is it a bubble, how sustainable it is and how we make sure that we can service it and our traditional markets. The reality is it came on very quickly, actually probably about a week before our earnings call and ended up with very -- basically taking a lot of the inventory that was in the channel. And our channel inventory continues to be very low at this point in time. So we've got our factories operating at 100%. We're working to try to get our inventory levels back up. But it's a pleasant problem, but it's a challenge nonetheless and we are seeing pricing going up in the more transactional markets. As we just talked about, some of the other markets are negotiated over either a quarter or even in some cases, 2 or 3 quarters. So we won't see immediate changes there, but we're pretty optimistic about how things are going.

Would you say that the strength with respect to Chia is insulated in the HDD side? Or would you say that it's extending into the SSD side? Because some of it is actually being deployed on the SSD side from a plotting perspective. Can you maybe just talk about what you're seeing on the SSD side as well?

Yes. Yes. No, that's exactly right. We're seeing incremental demand, both on the hard drive side and on the flash side, and it is plotting where solid-state drives are being used. And what we saw was initially the higher capacity points we're selling out in the channel and then worked their way down to -- customers worked their way down to lower capacity points. By the way, that's the same thing we saw in the hard drive side. So higher capacity, hard drive is sold out first and then worked their way down into other capacity points. So it's good for both businesses. And again, the real key is what's the longevity of that market and what kind of impact it has on the market overall.

And would you say that it's too early to assess that, Bob? Or would you say that we're -- I mean, as you -- whatever -- like things we can monitor externally like Chia in that space, for instance, continues to grow. I mean that's sort of almost astounding to see this new demand vector or exceed 20 exabytes in a matter of months. And so when you think about sort of capacity planning, are you baking into sustained sort of demand from this? Or is it too early to do that?

Yes. I mean, it's too early to do that. We're not changing our capacity at all. Well, again, it was always going to be a fairly tight year in terms of supply, and this makes it a little bit more of a challenge, but we're not going to change our plans. Even if we were to change our plans, it would be a number of months or quarters until it really had an impact on our supply.

I wanted to just talk back about sort of away from Chia, enterprise and cloud sort of demand, right? As the economy is reopening, where are you seeing sort of the best demand trends? And can you give us some sense of what you're seeing across the different capacity points, what -- how demand is playing out?

Yes. So again, as we anticipate, the demand on the cloud side is very strong this year. We had a digestion cycle at the end of last year. Now as we expected, things are quite strong. We're very fortunate. We have a very strong product lineup right now. We just introduced our new 16, 18 and 20-terabyte families. The flagship product there is really our 18-terabyte drive. If you looked at the March quarter, I would say our exabytes were roughly balanced between 14, 16 and 18 terabyte. You look at the June quarter, it's going to be dominated by the 18-terabyte drives. So making very good progress in the adoption there. And that will definitely be a strong product for us for a number of quarters. I anticipate it will be a flagship product in the industry. So the demand is really good in that respect. On the enterprise side, I'd say it's fairly stable. We're also hopeful, as you noted, as people start to come back into the workplace, as you see new server chip come out, that there could be a catalyst for some incremental demand for the on-premise business.

Okay. That's great. I do want to ask you one more question, Bob, before I pivot to Siva to ask him some NAND questions, but this is a NAND question just on NAND price trajectory and what you're seeing, what your expectations are? And what are your expectations around margin trajectory for that too?

Yes. So we've been -- again, we've been expecting things to be fairly tight this year. Our forecast is for demand growth of around 35% in bits and we expect the supply growth to be a little bit behind that, setting up as a pretty constructive year. We actually saw this trend start back in October in the more transactional markets, which for us are the retail markets and the commercial distribution markets. Now we've seen that transition into the negotiated markets. We think that, that trend will continue, and we think that margins will be very good this year on the Flash business.

Okay. That's great. Maybe turn a bit to Siva over here on the NAND side. Siva, can you talk about sort of the differentiation at a product level perspective? What is it that differentiates WD NAND versus Samsung? And what's different in the process?

Yes. First and foremost, it's our breadth of products, right? The fact that we have -- we had a full service, full spectrum storage provider all the way from HDD into flash, makes us -- our interactions with our customers much more closer. We understand industry trends in a level of depth, whether it is the PC OEMs, whether it is enterprise players, whether it is retail players. We have that depth of knowledge as we define our products. That's number one. The second is the fact that we are vertically integrated. We think about the product all the way from what's happening in the fab architecture to what happens in the firmware architecture. So when we are able to trade-off where we take the risk that allows us to think through that architecture very, very well. So that -- the idea of having the customer knowledge and the ability to do it vertically makes the second thing. The third is predictability in our scale because of Kioxia, we have about 40% or so of the 9 bit share together with Kioxia. We are able to look at what's happening in the marketplace and our customers in a very, very wider lens. So that scale helps. The last and equally important is our development philosophy, where we come back and focus on capital efficiency and performance. We want to make sure for the same dollar invested in capital, how do we maximize bit output, while keeping performance to be on top. And you can see that consistently in the metrics we talk about, the number of layers you need to produce the same number of bits. Or a single plane performance in either read or write, we lead the industry in all the business. So you can see there is a coherent story in the way we develop and differentiate ourselves in the marketplace.

Thanks, Siva. So you noted about layer counts. What's your technology roadmap look like? Is there a BiCS7, 8, like how far can you push this technology?

Yes. We -- you remember when -- you've been in the market long time, Wamsi, when we got the 2D NAND when we are coming from 19 to 17 NAND to even 15, oh, my God, am I going to go -- there was a cliff in front of us. There is no such thing right now in front of 3D NAND, and we are in our sweet spot. We can see for the next 5 years. Yes, you said BiCS5, BiCS6, BiCS7, BiCS8 and 9. There is that progression standing right behind. Each one is tougher. There's no question, each generation is going to be much harder to do. But it is a known set of things that we need to get done. There is not one where we are going to be staring at a cliff to see, oh my God, what am I going to do next further. So we are looking at that 15% cost reduction and 35% bit growth. Can I continue to achieve that? It looks like it is possible to be able to do that in 3D NAND. There is still sufficient runway left.

Okay. Okay. That's great. Can you talk a little bit about what percentage of your NAND shipments are BiCS5? And when you see crossover with BiCS6?

Yes. So the BiCS4, BiCS5, BiCS6 transition has been an interesting thing to watch. BiCS4 was an extraordinarily successful node for us. Almost 90% of our bits got to BiCS4. We usually don't get that high because such a successful node. We are now starting the BiCS5 ramp in earnest a quarter ago. By end of this year, by -- in this calendar year, we'll have -- in the second half of this year, we will be having the bit crossover from BiCS4 to BiCS5 in terms of -- and BiCS5 will then reach its height, when all the conversion is converted to go from the crossover to maximizing the bits out of the fab on BiCS5. Then BiCS6 will start. I expect meaningful BiCS6 production at year [ hence ].

Okay. Okay. That's helpful. Yes, which of your customers are still choosing to buy 2D NAND? Why is it important to keep making 2D NAND?

That's an interesting question. There are certain long-term customers who do not transition very fast. So a very small part of our capacity remains 2D NAND. For example, an automotive customer that qualifies for the next 5 years. They don't transition very, very fast. We try not to do a lot. So we try not to keep a lot of capacity and trailing nodes. They have to be extraordinarily profitable segments for us to keep that trailing node on for a long time.

I see. Okay. That makes sense. What about this cost reduction trajectory that you spoke about? So you mentioned the 15% cost reduction. Why is 15% the right number to target? Why can't it be more than 15% in terms of cost. And I think some quarters, you have actually shown that you can do more than that. So what is the gating factors? And what are the main factors to think about in the limitation of that?

Wamsi, you sounded like my CFO, Bob, [indiscernible]. It does get harder as you go along in the technology. So we have seen this movie play out a few times from BiCS3 to BiCS4 to BiCS5. We have -- yes, there may be a quarter where I get better a quarter, a bit lower, but on average, 15% in the 3D NAND era seems to be where we are settling down. There are 3 or 4 factors that go simultaneously. How fast can I convert to the next node? How capital-intensive is that node? How much bit growth do I get in that node? So when you put all that together, per bit cost reduction, we target that 15% plus/minus. You're absolutely right. Last couple of quarters, we have done a lot better than 15%. But there have been times ahead of that, we have done worse than 15%. So we want to get to that consistent delivery of 15% year-over-year.

Okay. That's helpful. Can you talk about the evaluation of QLC and PLC? And what share of the overall flash market can they get to in the next 5 years?

Yes. So we have seen the transition from SLC to MLC to TLC and the QLC change has already begun. So you can start to see some market starting to convert. Maybe in some client SSD markets and some retail markets, we are starting to see QLC come over. The gain, as you can see, any growth from TLC to QLC, QLC to 5 bits per cell, the gain goes down every generation as a percentage. We think the QLC across the broad segment will happen in the BiCS6 generation. Our BiCS6 generation will be the time frame that a majority of the bits will switch over to QLC in the marketplace. That's when you get the mass adoption across all segments, enterprise, client, maybe mobile and retail will get QLC adoption. So QLC, we think will be in the next 2 years, plus we are going to see the rapid acceleration of QLC adoption. Five bits per cell on the other hand, PLC, if I -- many, many different words, the incremental gain is not quite as much, where you're going from 4 to 5 bits on the same cell, so you're getting 20%-or-so gain. To get that gain, you're sacrificing a lot. You need additional redundancy, additional ECC. So the net gain as opposed to the performance loss may not be quite as desirable. So I expect that transition will be lower, slower. So maybe in the second half of this decade, we are going to see some segments starting to get 5 bits per cell.

Okay. There's been a lot of talk and news around bringing manufacturing back to the U.S. and especially semiconductor manufacturing back to the U.S. Can you share your thoughts around sort of on that? Like is WD going to be participating in any of that? Is there an interest to do that? Are you in any discussions to do that?

We explore all options all the time. Our partner and us have looked at this many, many times. We have certain advantages, the economies of scale when we have it in Japan, where we have today. So our development and production are in the same site. So inside the manufacturing facility is our development line. So we don't have the transfer time. When someone has to develop in the U.S. and transfer to Singapore, our developer in Korea transfer to China, you will lose a couple, 3 quarters in that transfer. We don't do it. We get that scale and the transfer because of that. So we have had a relatively concentrated footprint for fab production. Now remember, semiconductor production does not stop at the fab. You still need to have assembly and test and qualification, et cetera. And we have them around the world. We expect that we'll be evaluating it all the time but I don't think we have any immediate plan to put a factory in the U.S. yet.

On the hard drive side, we do have our hedge fabs in the U.S. We have 2 fabs in the U.S. already.

Right. Okay. I mean would government subsidies help in making that decision differently?

Well, I think you're always going to make economic decisions. Like Siva said, we need to continue to evaluate all alternatives, and we'll do what we think makes economic sense for the long term.

Okay. And I assume, I did want to ask you about your thoughts around the competitive landscape as it pertains to China. There are a lot of investors who think NAND is a tough place to invest given what they have seen in other industries, whether it be solar or LCD or whatever, where China starts to, like, get into the market and the economics of that market get disrupted. What are your thoughts around that? How can you give confidence to the investor base around where you are in technology relative to what the application set is, where the market is going? Can you just talk about that?

Yes. First and foremost, in a rapidly changing, rapidly evolving technological industry, it is hard for someone to catch. We have a -- you see the scale of R&D investment and the scale of customer applications we talked about. So we are going from 112 layers to 162 layers to 200-plus layers. And these things are coming at a breakneck speed. And then that's not the end of it. I still have to figure out to convert a raw NAND die into a customer-specific product solution. The firmware said that is needed that is adapted to this medium. You cannot take one person's NAND and take somebody else's firmware and make it work. It is custom developed the firmware, the defect modes, the speeds, the ability to adapt the die. Those are completely specific, even in a customer when I change generations, it's tough. For someone else to come in, that's not an easy thing to do. So you are absolutely right that our competitors' coming up from China. But you also have really -- this has been going on for a long time. We survive because we move at that rapid pace of technological innovation and product innovation. The ability to work with the customer, use a firmware, a controller, a custom controller that works with our die, that's going to be hard to reproduce for a long-time mode.

Yes, that makes a lot of sense. You guys see a lot of corner cases that are hard to for a customer to make sure that they don't want to do any of those error situations when they use something different. In your case, however, does it make it easy for you on the opposite side when you go from sort of a BiCS3 to BiCS4 to BiCS5 to BiCS6? I mean is it easier and easier relatively to qualify these at a customer?

Two sides to that equation. On one side, yes, the customers -- we know the customer usage models very, very well. We reproduce the customers' qualification in our own site. We are able to do it a lot better generation to generation. Also because the firmware code, the base firmware code remains the same, so you are adding on top of it. So incrementally modifying that helps. However, having said that, the workloads are evolving so fast that the customers do need the time to take this firmware and this die and requalify. And that does take time, and whether it -- especially in the enterprise case and sometimes in the mobile and the client case where they have to qualify. So we know what to do. We are doing it more efficiently, but the demand is also correspondingly increasing. So we end up having to requalify every time.

Okay. Can you talk a little bit about sort of the trade-offs on -- from a customer perspective as they look at HDD versus NAND, right? There are certain applications where clearly, I mean, you think about laptops, the HDDs are out and SSDs are in. When you think about certain applications that are like machine learning, and AI-driven applications, you could see why you need that performance and that clearly plays into SSD's favor, whereas some cloud applications, maybe you're doing YouTube videos or Facebook feeds, and you don't require quite the same level of response times. And HDD might be the better solution. Is there a point that you see where the market gravitates entirely towards flash-based solutions away from HDD entirely? And what would prevent that?

Yes. So there are 2 or 3 segments of markets that behave differently. Clearly, there was substitution in the PC case, where you had, as the NAND cost came down, you go into a substitution mode for that kind of densities, the 512-gigabyte kind of densities. On the flip side, as you said, in the enterprise, large data sets, very large, medium, HDD will continue to and will remain the dominant force for a long, long, long time. There is no possibility that a NAND can go replace 18, 20, 24, 30-terabyte drive in terms of capacity and cost easily. What is important, however, for NAND to grow is not to think only in terms of HDD, but to look at entirely new applications, which we have done many, many times in the past. None of us ever thought that digital photography had anything to do with NAND, and that market was created or with audio. So my expectation is that there is going to be a very, very large segment of the market that will be HDD for, and it is going to be growth for a long time mode. That 35-plus percent growth -- bit growth in HDD will be served by capacity enterprise HDD. There are going to be entirely new markets that flash will find. When you just saw Chia, Chia came out of nowhere, a new vertical that grows. That's the way storage has always grown. And I think the future of HDD is HDD, the future of flash is flash. Those 2 will have definite finite roles to play for the long term.

I wanted to ask you, Bob, about CapEx, right? Clearly, I mean, Siva pointed out, the challenges and the difficulty in technology to sort of really drive NAND bit growth, and there is significant investment needed. Can you just talk about -- just give investors a context of like how the next 5 years of CapEx should look relative to the last 5?

Yes, 5 years is a long time. Maybe just talk a little bit about this year and next year. I mean, this year, when we think about CapEx, we think about gross CapEx. So it's investments that we're making in the JV as well as investments that we're making -- that show up on our own financial statements. And it's about $3 billion in gross CapEx this year. We think next year will be in that same ZIP code of around $3 billion. We're going through our budgeting process right now. More to say about this in July. One of the challenges that Siva and his team work on, I think he alluded to already, is constantly trying to be as capital efficient as we can. And I think we've done a really nice job on that, 112 layer is very capital efficient, and we'll continue to make sure we make trade-offs such that we invest what we need to, but we're not overinvesting.

I know we're coming up to the top of our time over here. So Bob, I just wanted to give you a chance to talk about maybe 1 or 2 things that investors might be underappreciating about the Western Dig story. I mean to me, as an analyst, it feels like if you've put a set multiple on your HDD business and look at what the NAND business is being valued at, it seems just very, very inexpensive. Just purely from that, just not even taking into account what the future on the HDD side could do. So love to hear your thoughts and where there might be an investor misperception?

Yes. Well, I mean, we're fundamentally in great markets. And I think we have great routes to those markets, whether it's through retail or through cloud service providers, OEMs, or through commercial distribution. I mean, we really have a big opportunity there to continue to reach customers very efficiently. From a flash standpoint, I think that the joint venture is probably underappreciated. It gives us tremendous technology and cost leadership it really allows us to get economies to scale, both from an R&D standpoint as well as from a production standpoint. And I don't think that, that's fully appreciated, and it's part of what allows us to continue those 15% year-over-year cost declines. And from a hard-drive standpoint, it's really a very good market. It's fundamental technology to the cloud compute environment. And we're in a really strong position there. Customers really need us in order to continue to grow their service offerings. So I think overall, I agree. I mean, we probably are a little underappreciated, but I think the team is executing really well. We are excited about the new structure we put in place within the company, and we think we can create a lot of value here.

Excellent. Unfortunately, we're just about out of time. So I really appreciate you both coming on and participating in our conference. Thank you so much, Siva. Thank you, Bob. We really appreciate your time.

Thanks, Wamsi. We appreciate the opportunity.

Thank you. Talk to you soon.