Western Digital Corporation (NASDAQ:WDC) Wells Fargo 6th Annual TMT Summit November 29, 2022 4:50 PM ET
Company Participants
Peter Andrew – Vice President, Investor Relations
Robert Soderbery – Executive Vice President and General Manager, Flash Business
Conference Call Participants
Aaron Rakers – Wells Fargo
Aaron Rakers
Perfect. So Rob, why don’t we start — I got to start with Peter. Why don’t you start with the safe harbor there.
Peter Andrew
Yes. I have got to read the safe harbor statement here. So please bear with me for a quick second. During today’s presentation, we may make forward-looking statements, and I would ask that you refer to our SEC filings for the various risks and uncertainties behind these statements. In addition, we may also be making references to non-GAAP financials and a reconciliation between the GAAP and the non-GAAP financial measures can be found on the Investor Relations section of our Web site. So with that, let me turn it back the mic.
Aaron Rakers
You nailed it. Good job. So Rob, why don’t we just start for those in the audience that aren’t familiar with — I think you joined, what, in late 2020. So maybe give us just a quick overview of your role, your responsibilities and maybe background and then we’ll dive into a ton of questions.
Robert Soderbery
Yes. We’re in south of about two and half years now. So by way of background, I think I had a unique background to come to Western Digital, early semiconductor experience, storage experience, systems at Cisco, running enterprise at Cisco, had the good fortune of getting a call from David Goeckeler, about two and half years ago, sharing with me the basic thesis that Western Digital was actually a wonderful asset with amazing technology and amazing market position that for a variety of reasons the value wasn’t being realized and persuasive enough to have me join. That was coincident with his decision to move us to a BU structure, creating a flash business and a HDD business getting focused around it with myself and flash and my colleague, Ashley on the HDD business.
Question-and-Answer Session
Q – Aaron Rakers
That’s perfect. So look, I’m going to start right out of the gate. Why don’t we just kind of hit the current demand environment, just the current trends that you’re seeing. I think all of us probably know things in the memory, the NAND market can move rather quickly. And so maybe just give us your thoughts on what you currently see in the environment in the midst of kind of, I would argue, unprecedented declines in bit shipments year-on-year right now?
Robert Soderbery
Yes. I guess I would start by framing the way we think about performers of the company is what all the things that we can do under our control and particularly is how we measure our performance or relative to the industry. And my own personal MBOs were joking about last night are relative to the industry, kind of generate alpha relative to the industry. That said, we do play in a cyclical market, and there’s only so much gravitational pull you can resist from that cycle. So we’re definitely in the middle of the downturn here. The downturn was characterized by essentially the different sectors rolling sequentially into the downturn. I think it will be — the coming out will be characterized by those sectors sequentially rolling out of the downturn. I can say pretty strong confidence. We feel like our consumer business is first in, has now stabilized at new demand levels and supply demand balance and is in a stable spot.
We have a leadership position in client PCs and our client OEM or SSD products. Their OEMs are dealing with significant inventory conditions and they’re burning down that inventory in a relatively methodical fashion. So we see ourselves — we’re shipping in at lower than the replacement rate while they burn that inventory down. Couple of quarters out, we’ll get through that inventory burn and we’ll return to the replacement rate. Big picture, we think PC demands come from about 350 million units in the kind of COVID peak, will be down to the low 300s, maybe 295 as more of a post-COVID sustainable level. And then last to the party was cloud. There, we definitely have seen — it’s really been over the 60, 90 days here that you’ve seen cloud giants come to grips the fact that they had overplanned and overpurchased relative to the underlying demand and move to much more conservative postures. And so we’re really still in the middle of that. We talked about cloud digestion, like before the digestion process is the indigestion process what we’re in right now as we look to sort out what the real level of demand is, what the inventory situation is. But it’s clearly going to be a couple of quarters to sort cloud out.
Aaron Rakers
And in the context of everything you just mentioned, which is very consistent with other companies and other things we’ve seen over this last earnings cycle or so. Remind us again of what Western Digital, what your flash business is doing as far as your decisions to modulate the supply side of the equation and from a bit perspective in NAND versus we’ve seen Micron cut [50%], your partner cut [30%].
Robert Soderbery
So there are basically three levers on the control on the supply side. There’s a capital lever, there’s an inventory lever and there’s a utilization lever. Consistently, everybody has pulled back as hard as possible on the capital lever. And I think it’s important because you’ll see something the paper about so and so is reduce CapEx by 30% or 40%. All that means is that they already had 60% committed. So they have set discretionary CapEx to zero, and they’re decelerating as fast as they can. We had — I think there was some confusion as to whether this was going to be consistent across all the NAND vendors and there would be some clarifying statements that clearly, everybody has pulled way back on that capital lever. And so I think that’s going to be the big picture that characterizes the recovery, which is supply growth is going to go to zero as you come down off this capital curve and then demand will catch up. The second lever is inventory. And here, we’re seeing some changes in market dynamics. Earlier in flash, the underlying cost structure and flash pricing was falling so fast that you really could hold inventory at your apparel. Now there is a greater opportunity with lower cost downs, there’s a great opportunity to use inventory as a lever in terms of smoothing out the market. And we’re seeing use of that lever, again, consistently across the vendors.
Where things have diverged so far is in factory utilization and looking to save variable cash costs by underutilizing the factory. Our partner, Kioxia, has made a statement to the marketplace. While we make all our capital decisions jointly with Kioxia, they have discretion over what they do with their share of the fab. So they’ve made statements. I think Micron has made statements. Like our position is underutilization to be good for the industry and will accelerate recovery, we don’t think unilateral underutilization is the correct move for us. And so we’ll watch what the industry players will clearly do. We have a lot of outlets for bits. So we don’t have any risk of obsolescence in terms of where to place product over time. So we feel like we’re in a position where we can continue to watch kind of the industry dynamic around fab utilization and don’t have to make a decision immediately.
Aaron Rakers
And do you have a current view on bit supply versus demand as we start to think about 2023 in your mind?
Robert Soderbery
Well, the supply picture is easier to describe because it’s this phenomenon over the first two quarters of ’23, you’ll see the last of the capital go in and then you’ll see supply level off. On the demand side, the hard forecast is the elasticity of demand. Flash is a very elastic business. Obviously, if I’m willing to sell to you 5, 12 gig for same price is 256 and then your phone, you’re probably going to take it. So it’s a very elastic industry with a 30%-ish price down, you should expect 40% to 50% bit growth. Now there’s some — there’s a time delay. It takes time for that to propagate through the system and the industry is in a place where we can really hit the gas right now. But as we get into mid ’23, I would expect to see the elasticity is really what’s going to drive us out of the cycle and on to the next curve.
Aaron Rakers
And you just touched there 30% price compression. Is that what you’re seeing on a year-over-year basis or…
Robert Soderbery
Yes, I just sort of think aggregate this whole thing up, it’s going to be significant, not to pin myself down to a specific number.
Aaron Rakers
Understand. On the client SSD side, you mentioned earlier that you have number one, number two position in client SSDs in the market between you and I think your closest competitor Samsung. So I think last quarter, and Peter certainly is going to correct me if I’m wrong, you started to imply that maybe you’ve seen some normalization show up in client PCs that maybe some of those end customers are starting to maybe show signs that they work through inventory, or how do I take those…
Robert Soderbery
I’d say predictability is returned to the market. Customers are accurately forecasting, they understand their inventory position, they have a methodical view of driving that down. In certain cases, they’re running low on individual SKUs and so they need to go ahead, and there’s a little bit of upside coming through from that regard. So there’s still — we’re in the digestion, there’s still a couple of quarters before everybody is through that, it’s hurt the consumer guys more than the client, more than the commercial teams. So it varies a bit by the different PC OEMs.
Aaron Rakers
Yes, got you. So earlier this year, you guys had an Analyst Day, pretty great updates. You talked about kind of in the flash business a longer term revenue CAGR of around 10% to 12%. Obviously, cycles come and go and that’s a longer term framework. How does — remind us again of how you guys think about the underlying bit demand growth supportive of that 10% to 12% CAGR.
Robert Soderbery
Yes. So I think the dynamics of the flash industry are starting to change. Flash has been this magical technology where we’ve been able to create the huge increases in bit growth, huge reductions in cost and then rely on elasticity to wash this whole thing out. If I think about the longer term picture, 30% bit growth, 15% cost down, 15% revenue growth kind of an equation is really where the industry was centered. I think our view now is that you’re going to see bit growth come down on a more sustained basis. So that sort of 30% number, I don’t think it’s going to be consistent with what we see going forward. So you see a lower 20s type number. But I also think you’re going to see price declines moderate as we come through this cycle as well.
Aaron Rakers
And the cost equation, cost curve down still you believe 15% for now?
Robert Soderbery
Yes. I think 15% is kind of the curve that we see sort of here in the near term. Clearly, the capital — NAND capital intensity is increasing, which is going to cut into that cost down number over the longer term. Ultimately, we believe that’s a healthier trend for the industry, forces more disciplined capital decisions and will be a good thing, but it’s going to start to change the base of mathematics of NAND.
Aaron Rakers
Yes. So I’m going to shift gears a little bit and talk maybe on the technology side of the equation. And Western Digital with your partner, Kioxia, has consistently moved forward in NAND flash, but it’s been at a different pace than some of the competitors, right? So I tend to get a lot of questions of Western Digital at 112 layer versus 128, or now moving to 162 versus 230-plus layer stack. So maybe update us on how you think about the cost competitiveness, how you get confidence, I guess, the cost curve that you’re on with different layer stacks relative to competition?
Robert Soderbery
Well, I think it starts by actually measuring the cost curve and not measuring the layer count. Layers — increased layer count just means longer cycle times, more expensive dry etch machines, more complexity in our process. Like all you really care about is bit production versus cost reduction, that’s the core equation of NAND. You have an XY scaling problem, how aggressively can you XY scale, in terms of semiconductor scaling, scaling your layers on top of that. I think our belief is that by keeping capital efficiency as the number one goal and keeping that bit production versus cost reduction ratio is our number one goal, we get a better outcome. And we don’t feel pressure to necessarily lead on layers as a vanity metric. We don’t feel pressure to lead on the vanity metric, we will lead on metrics. I think it’s pretty clear, I think, some of the folks that have been most vocal about their layer strategy or having some of those challenges dealing with the downturn. So I feel somewhat we’ve validated in our communication to the market.
Aaron Rakers
And on BiCS6, 162 layer I think the company has talked about a 40% smaller die size. That I think 112 was pretty noted as capital efficient. I guess, do I keep consistent thinking that BiCS6 drives 15% cost curve down, or do you see any kind of changes in the quarter.
Robert Soderbery
Yes. So BiCS5 was extraordinarily efficient. It was a very, very efficient node. BiCS6 allows us to go to QLC’s, or mainstream product. We’ve had a performance advantage in NAND and that performance advantage hasn’t turned into benefit to the marketplace prior to QLC, but a combination of faster performance with QLC essentially are getting 33% more bits per die. So that is a little mid light kicker in terms of keeping us on the more aggressive cost down schedule.
Aaron Rakers
And that’s due to the transition you moved then effectively to a CMOS under array or periphery under array…
Robert Soderbery
Yes, that has an architectural benefit. But the real story around BiCS6 is the fact that we were able to deliver an enterprise class QLC or 4 bit per cell technology, and you’re literally just putting 4 bits in the same cell you used to put 3 bits. So that’s material productivity.
Aaron Rakers
Perfect. So I’m going to keep bouncing around here a little bit on you, Rob. So I’m going to shift over to gross margins on the flash business, it’s I mean, let’s just be honest, difficult to sometimes model gross margin. So how do you — a simple question, how do you think about the reasonable through cycle gross margin in order to get the appropriate return on capital for the flash business that you run?
Robert Soderbery
Yes. I mean we think that the business needs a 30% through cycle to be viable. We could keep going at 30%. I would not be — this conversation isn’t very interesting at 30% gross margin and so we’re out in May talking about 35% to 37% as our gross margin target across both HDD and flash. And I think underlying that are two separate thesis. One is that we can generate some alpha that WD can perform better, and we’re actually — I think we validated that. We’re ahead on ’22 and ’26 on the HDD front, we’re executing well on our eSSD roadmaps. Consumer, we currently have share and price leadership like higher price and higher shares. So there’s a lot of data points to say we’re doing better at executing. The second part of the thesis is that we can better manage the industry such that the overall industry average is a sufficient level to support that 35% to 37%. Again, that plays in both HDD and flash, it’s really about the structure of the industry. We are, I think, in a better position now in flash. We’ve seen the market essentially go from a seven player market to a five player market in a relatively short amount of time. We have a great relationship with our partner. We make common capital decisions with our partners. So if you stand back in squint, it looks more like a four player market. So there’s reason to believe we can start behaving more rationally as an industry.
Aaron Rakers
And mix within the flash business as we move forward, I think at the Analyst Day, you talked about commodity products and some of the mainstream products and maybe that mix goes from whatever, 75% to 50%. What should I be thinking about as far as the roadmap that really starts to drive that going forward?
Robert Soderbery
Well, first of all, the big picture on mix is the shift from the whole flash industry from mobile centric to SSD centric. That’s not just a cloud comment, that’s consumer, that is client and that is in the cloud. In the consumer business, there’s still 100 exabytes of consumer of HDD, client HDD going to consumer. So that is essentially ours for the taking because of our brand and position there. So we’re pursuing that SSD strategy across all three of those sectors, and we expect that drives us to 80% plus SSD product volume by end of our three year forecast cycle. So a very significant shift to an SSD centric company. Really, if you think about who is WD, right? WD is a drive maker and we’re really turning the whole business back into being a drive maker.
Aaron Rakers
Okay. Very helpful. So maybe in the context of that portfolio thing and maybe for the audience, just give us a quick reminder how the JV flash ventures Kioxia actually operate. Like how decisions are made, what happens coming out of the fabs and just that whole process as we think about that?
Robert Soderbery
Yes. So it’s financially complicated, but operationally quite simple. We had a chance to have a summit here, a post COVID summit in September a nice spa in Northern Tokyo, with the top WD and Kioxia leadership and really reaffirmed what we’re doing together. So we make all of our capital decisions in unison or in synchrony. So I say we’re essentially one planning entity. So we do BiCS5 investment, BiCS6, BiCS8, a new facility. Those are all done in synchrony. The fab is run as one entity. And then over the top of that, there’s a financial construct that’s supplied where the wafers are split 40% and 60%. We can do — we have a 40%. We can do whatever we want within those 40% in terms of product mix, if you want to underutilize whatever we can play within our 40, they play within their 60. But all of the fundamentals of the fab are run as a single combined team. And then post that, post wafer sort, we go our own way. We build our products, they build their products. The actual trim or the micro firmware that goes on the die, we develop independently.
Aaron Rakers
Yes. That’s helpful. So — and I’ve always — I’ve asked Peter you this question. When we think about Western Digital and the stock does what the stock does. But remind us again how much equipment, the total cost of what you have sitting in the fab at Kitakami and Yokkaichi.
Robert Soderbery
So it’s $16 billion to $18 billion in the fab. Now book value is materially less than that, but all that equipment is there. The reuse rate is very, very high. 80%, 90% of the equipment is there, it’s productive, it’s doing its thing. And when we talk about being capital efficient, it’s as much how you reuse your gear, how much brownfield you have, how efficient your factory is, all these things are a big deal and that contribute to cost advantages.
Aaron Rakers
And then final on the structure of the business. When you look at the Western Digital business, and you mentioned you joined the company at a point in time when David was making decisions between hard drive and structural flash businesses separately. Remind us again about the go to market alignment between the two companies and how that feeds off each other.
Robert Soderbery
Yes, there’s material synergy in go to market alignment. There are 70% of our customers are common, and these are customers that buy a lot of both product, tens of millions of dollars of both products. And all that’s covered via 100% common team, single executive on each cloud customer and so on. Then we have some unique things, some mobile is unique, some of the video surveillance is unique. But the bulk and an important part of our GTM is fully common, so synergy there. We run the flash and HDB is quite separate really for focus and performance. And then within operations, it’s kind of a hybrid part. We use common folks of operations where we get — we have synergy and then have quite large independent entities as well.
Aaron Rakers
And then the other topic of the discussion I have listed here is that the enterprise SSD market, to me, represents probably the biggest incremental demand elasticity vertical for your business. you guys have — and I think you’re 6%, 7% market share today, you have aspirations of being 16% market share over time. Can you just remind us of where we are at as far as the enterprise SSD traction for Western Digital?
Robert Soderbery
Yes. So the enterprise SSD is — and cloud is going to be the largest segment growing fast. About a third of that growth is unit driven. So when you look at PC server demand that’s kind of correlated with units and drive units, but two thirds of that is just content capacity growth. So you’re kind of getting a multiplicative effect, bigger drives and more drives. So that contributes to the share of flash going into eSSD, it means we need to have a very competitive portfolio to play. We have made tremendous progress. Two and half years ago when I joined, like every day my phone would ring with another escalation coming from a cloud customer who’s trying to use one of our drives and it was not working. And that was with our BiCS4 AeroDrive. In May at the Analyst Day, we launched, we had just shipped drives into qualification, we’ve completed our first, our major hyperscale qual, where [indiscernible] to the second, we’re halfway through the third, zero escalations, all gone super smooth. So we’ve proven that we now have a scalable eSSD engineering capability. And that’s not easy. Very, very difficult. And so it’s an intermediate point in the way that share goal, that share goal is a three year goal. So I’m not expecting that you can be able to just linearly plot every quarter along the way. But clearly, being able to execute from a technology perspective on the portfolio is step one.
Aaron Rakers
And so you’ve got three hyperscale cloud guys that you’re either qualified or actually getting qualified with — obviously, that 16% goal reflects not just BiCS5, but going to BiCS6 BiCS6 plus as well.
Robert Soderbery
Yes. And so step one was to get into market with a good solid product which we did with BiCS4. Step two was to follow it up with BiCS5. In the BiCS6 era, we have three platforms coming out, gives us broader TAM exposure. In some of the swim lanes that we’re currently serving, customers have committed up to 40% of their TAM. We’re just not serving all of the — all of the market. So our SAM to TAM ratio isn’t high enough to really go after those kinds of the [16%] number yet.
Aaron Rakers
So BiCS6 really kind of completes out the broadening of the product portfolio…
Robert Soderbery
That’s correct. That’s right…
Aaron Rakers
Exiting calendar ’23, or is it…
Robert Soderbery
Correct.
Aaron Rakers
Okay. And I think BiCS6, as you mentioned it earlier, also entails QLC or quad level cell NAND. How does — I mean, when I look at the enterprise storage market, my math would still show that 80% of the storage shipped capacity wise is hard disk drive base. Western Digital is unique, right? You have an SSD business and a hard disk drive business. How do you see QLC playing a role in driving maybe more bits to move off drives to SSDs or do you not see that?
Robert Soderbery
There is some interest in that. I would say it’s more a question about where there are workloads that could benefit. So object storage is a good example. A lot of object storage is not performance oriented, but some is. And so you’re seeing cloud architects start to think about where I should be using more performance. So this is not really a cost crossover question. This is much more of a question about do the use cases need the performance that cost gap is still very, very significant. The HDD growth rate is still quite high. It would take hundreds and hundreds of billions of dollars of flash capital investment to try and replace that HDD expand HDD business. So that’s not happening anytime soon.
Aaron Rakers
Yes. In a minute or so I’ve got left, just any — Peter, for you, too, is there anything that we I should have asked you that I haven’t asked you or any kind of topics you seem to get today in the meetings that we didn’t really touch upon. Nothing from a broad diverse Q&A perspective, but do you want to make a quick comment…
Robert Soderbery
Yes. I mean, I would say, just to frame where the market — the wobble in the market inserts a lot of noise relative to where we are in the journey of transforming the company. David came in two and half years ago with the goal of transforming the company. We’ve made huge strides, right? Product leadership on HDD, product leadership on NAND, we didn’t talk a lot about the next BiCS beyond BiCS6, but we’re extremely pleased with the NAND roadmap in development. We think that’s going to be a much smoother node than the current node we’re on. So from an execution standpoint, a lot of confidence in the company. We feel like we got to get through the market cycle and then we’re really positioned to thrive in the — as the market turns.
Aaron Rakers
And on the geopolitical side, I’m going to slip one last one in. Obviously, the discussions around US-China restrictions and the kind of the fighting to goes always with YMTC and whether or not you would see them competitively in the market is what you’ve seen over the last month or so really how do you feel about their even ability to compete.
Robert Soderbery
It’s challenging. We have Yokkaichi, we have a thousand employees of our vendors in our fab every day working to keep that gear running. So if you’re running a NAND fab and you don’t have access to the international community to help you run that fab, that’s super, super challenging.
Aaron Rakers
Perfect. I think we’ll wrap it up there. Thank you so much, guys. Appreciate it.
Robert Soderbery
Thank you.
Be the first to comment