There we go. I guess we're live. Welcome back, everybody. I'm Joe Moore from Morgan Stanley Semiconductor Research. I'm very happy to have with us today the management team of Western Digital, David Goeckeler, and Wissam Jabre. I think, Wissam, you wanted to read a safe harbor, and then we'll go into some Q&A.
Thanks, Joe. Happy to be here. We'll be making forward-looking statements in today's discussion based on management's current assumptions and expectations, including with respect to our product portfolio, business plans and performance, market trends and dynamics, and future results. These forward-looking statements are subject to risks and uncertainties. Please refer to our most recent financial report on Form 10-K and our other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ materially from expectations. 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. Thank you.
Okay, great. Whoa, that's loud. Thank you for that. So there's been quite a bit of news on WD the last few days, including this morning, which, to be honest, I've been on stage, so I haven't read yet, but, yeah, I got the gist of it. But maybe you could, b efore we go into that, you know, maybe we could just kind of start on the business environment, and you know, how you guys are seeing things nowadays.
So Joe, first of all, thanks for having us. We really enjoy being here. So yeah, let's talk about the current business environment. I think we came into the quarter. If you go back to last quarter, we were very happy with the way the business performed. We clearly saw both businesses turning. I think if you look at each business, on the NAND business, what we've been trying to do is really put together a best-in-class cost position on our fundamental NAND roadmap. Of course, we've developed with our partner, Kioxia, and then tie that to a world-class portfolio, you know, where we have a lot of optionality across consumer business, across gaming, across client SSD, across mobile, IoT, and of course, an emerging business and enterprise SSD.
And then, of course, mix across that portfolio to get the best economic result we can, and also dampen volatility. I think we did that. I think through the downturn, we've delivered best-in-class gross margin in our NAND business by a significant amount, quite frankly. I think it's really the combination of those two things, where we're the best core technology, given the scale we have in R&D with Kioxia, and then margin tied to the best portfolio. Then in the HDD business, of course, we saw the big hyperscale players in the U.S. coming back, more robust conversations there. We saw good growth in China. We had over 100% sequential growth in China in the HDD business.
So we came into the quarter with a lot of strength and a lot of momentum in the business, which brings us to the current quarter. I'll give a little bit of an update on what the environment looks like right now. I think from when we guided, the business is stronger than that at this point. If we look at the guidance ranges we provided, we expect to be near the high end of guidance, of the range we've provided. But each business is performing better from a profitability point of view. Couple points of additional gross margin in each business. All the rest of the numbers are within guidance ranges, so we expect this quarter, we expect on an EPS basis to exceed the high end of our guidance range.
Okay, great. That's good to hear. Maybe you could parse that a little bit. So a couple points of gross margin improvement, maybe start with NAND. You mentioned how good the numbers were last year, and I think that maybe set up an expectation thing where other people who had burned a lot of inventory at very low prices saw a bigger rebound versus you guys, which you took quick action to reduce supply, kept the mix pretty good, and then guided for some improvement, but maybe less than some of the peers. You know, how are you seeing that progress now, and what's the general tenor of supply-demand for NAND?
Yeah, supply-demand is coming back into balance. We're not all the way there yet, but it... You're right. What we're trying to do is dampen volatility and deliver better through-cycle profitability. That's what we're trying to do. So hopefully, you don't see as much, pricing acceleration and deceleration in our business, and just, you see a more sustained higher level of pricing. I mean, we look at it as absolute pricing versus pricing change, and when you look at that way, we believe we're the best, we're best-in-class NAND business. If you look at what's going on now in the business, you know, this quarter, bits will be down, sequentially. Going into FQ4, we expect bits to be flat. So any change in the business in NAND is gonna be all pricing driven.
Okay, and so I'll come back to some of that, but maybe just on the quarter being better, you know, talk similar on the hard drive side, where are you seeing that gross margin improvement? Is that all price, or is there a cost element?
The hard drive business is driven by both better pricing and higher shipments. So we're still in an underutilized state. We expect to still have some underutilization this quarter, but we're seeing a little better volume. And of course, when we have underutilization, we're gonna see a margin impact of that, but we're seeing better volume and better pricing in the drive business. And as is... You know, I think if you roll both of these forward into Q4, we're gonna see... Again, in Q4, or FQ4, which is our June quarter, again, on the flash business, you're gonna see bits flat, so we'll see some modest pricing-driven-
Mm-hmm
Improvement there. And then we'll see, we'll see some modest growth in exabytes and pricing in the drive business. So I would expect some modest sequential growth going into the FQ4.
Yeah.
But we're also- I wanna balance it with one thing. We're gonna see more variable comp headwinds-
Okay
Given the outperformance of the business in Q4.
Yeah. Okay, that makes sense. And then, you know, you guide one quarter at a time, which I appreciate. But, you know, how, any sense on, you know, the durability of the good things that you're seeing beyond the current quarter?
Well, if I look at the drive business, we're getting, you know, we're coming back. We did a lot of work during the downturn, where we just structurally removed capacity from the system.
Yeah.
I think the drive industry is an industry that has been going through this client-to-cloud transition for literally 15 years. You know, the client business is just a lot more unit capacity that in the system than is required in a nearline business-dominated business. We've just removed that-
Mm
From the system. I think we've signaled that to our customers. We don't have as much capacity as we had in the past, and in return, they're giving us more visibility into what their ordering looks like throughout the rest of the year. So, in the drive business, we came into the fiscal year believing we were going to see sequential growth throughout the fiscal year. Certainly in the first half of the year, you know, first three quarters of the year, we're seeing that, and at last earnings call, we extended that to this calendar year. So we feel good about that business and a return to, you know, kind of a reversion to the mean of the 20%-
Yeah
20%-25% through cycle exabyte growth, we're getting back to that. And in the flash business, you know, we expect fab-out supply this calendar year to be mid-single digits.
Mm.
So we've seen, you know, the profitability of the business has been significantly impaired, and so there's been no CapEx going back into the business. And I think if you look at the wafer equipment spending in NAND-
Yeah
It's been low for quite some time. And, from our perspective, we're going to need to see ourselves... You know, we expect that 35%-37% gross margin as our target on a through cycle basis. We've been below that for quite some time-
Yeah
S o we expect to be above it-
Yeah
For quite some time before we're going to see CapEx coming back into the business. We're just now, you know, although we've taken a few good steps here, the first couple steps have been very, very good. Business is going in the right direction, but we've got a ways to go before we're back into our model, and we see, and we see CapEx spending come back into either business.
Okay, great. And then the announcement you made this morning, you've given us some further data on the separation. You're going to be taking the lead for the flash business, and Irving Tan will be the head of HDDs. Maybe can you talk a little bit about that decision? You know, what made you decide to go to the NAND? I, I agree with it, but that's because I'm a NAND guy, not a drive guy. But, you know, what, what drove that decision?
You convinced me, Joe.
Yeah, that's right. And, you know, can you just kind of talk about what makes Irving the right person on the drive side?
Look, these are the, A couple of comments before I get into that. These are both great franchises-
Mm-hmm
And it's very difficult, you know, to be forced to be in a position where you're going to pick one of them, because I, I think that they're both tremendous assets. I think that they're- they have both have tremendous futures. And also, I think that there are multiple executives inside the company that were candidates to run either one of them.
Mm-hmm.
I think one of the things we've done over the past four years is really build a very, very capable executive team, of course. You know, Wissam here is, of course, a huge part of that. If you look at my choice, the reality is, I think the NAND business is, you know, the more complex business with the JV, the relationships you have to manage across, you know, cross-border relationships. I've spent a lot of time building those. I'm very excited about the business, quite frankly. I think it's underappreciated. I think especially the Western Digital NAND business is underappreciated with our consumer business. I think our consumer business is a total gem.
I think there's a lot of value in that, and I look forward to all the work we can do to really fully value that asset.
Yeah.
And, you know, on Irving, Irving is a tremendous executive. I mean, Irving is a world-class executive. I worked with him for many years. He joined Western Digital a couple of years ago now and, you know, just immediately had a huge impact on the business. We have a massive Asia presence. He's based in Singapore. He's got an enormous amount of operational experience. If you look at all of the changes we've made when we talk about our HDD business and the amount of capacity we've taken out of it and how we've structured that business, that's Irving's been driving a huge part of that. And so he's just very, very well positioned to take that business forward, and I'm very, very optimistic about the future of the HDD business.
Great. So we've done, we've made some effort to parse out the earnings of the two businesses over the last seven years to try to triangulate the value. And obviously, there's some guesswork in there because you haven't, you know, talked about the OpEx splits and things like that. But I guess the one thing that we need to incorporate into that is the sort of inverse synergy of spinning this into two companies. How much incremental cost do you expect that this takes to build two separate companies? And if I look at the relative profitability, for example, of, you know, the hard drive business versus Seagate, how do I think about bringing in incremental cost into the business?
Yeah, and so look, it's still early days. We're still working through a lot of these details. While I can't provide necessarily financial numbers, what I can tell you is both businesses from a cost structure perspective are in great shape and ready to be independent companies. What we've done over the last several quarters is we've right-sized both businesses. We've taken a lot of cost out. You've seen us take a lot of actions on OpEx and CapEx, taking it down. And so we've created quite a bit of room for any potential dyssynergies. There will naturally be some dyssynergies as you mentioned, as we create the two companies. But with the actions we've taken-
Yeah
We have room to absorb those.
Yeah, I mean, maybe this is an unfair question, but, you know, we try to think of the hard drive business, we discount it 10% to the EBITDA sales of your closest competitor. And that's largely because I don't know this question, right? That there's an incremental cost that comes in. You know, but maybe over the long-term, any reason why you shouldn't have the same kind of ROI profile that they have?
I don't see any reason why in the long-term, the business won't be at or better, best in class.
I mean, I completely agree. This is the whole thesis behind the split.
Yeah.
I mean, I kind of keep hearing things like this, and it drives me a little bit crazy. Like, why should it be discounted?
Yeah.
It's more-
It makes a big difference 'cause we assume the debt goes there, so if it's not a discount, it's worth, like, you know, several dollars more per share.
Yeah. I mean, it's, like, more profitable and growing faster, other than that. So, yeah, I mean, look, this is. We think there's fundamentally different investor bases for these two businesses, and I think that we've done a ton of work over the last four years. Like, when I came into the company, there's no excuse for not running the business as best as it possibly can be run, and I think over the especially through the downturn, it really showed up. These businesses are best-in-class assets, both of them.
Yeah.
They’re ready to stand on their own. We’ve built an executive team that can run both of them independently. I think they’re both world-class assets, and I don’t think they’re gonna be discounted to anything in the market. I think they’re-
Yeah
They're really premium assets, and we're gonna-
Well, if year one in profitability and growth is better, then it'll trade richer like that.
Yeah.
That's how it works.
That's how it works.
The stuff I say now matters very little in the long run. Anyway, I appreciate that. So then the other announcement on Monday was the JV. Maybe you could talk to that.
You want to take it?
Yeah. So, basically, we announced this JV with JCET in IC manufacturing, and a technology service provider, very well known in the industry. We've partnered with them for many years. So here, what we're doing is, basically, we're selling a factory in our backend assembly and test factory that's flash-based in Shanghai. Basically, we sold equity, 80% of our equity interest, and we're retaining 20%. What this gets us is, it gives us flexibility and to become more effective at managing through the industry cycles. It gives us a bit of better free cash flow and working capital going forward, and we get to monetize the asset for $624 million.
What it gets JCET is a great manufacturing asset with a highly specialized team that has run this asset for many years. We do expect the transaction to close in the third calendar quarter of this year.
Okay. Okay, great. Well, thanks for that. So on that note, of the balance sheet, you know, how do you feel at this point? You know, you've sort of gotten through the convert, you know, replacement that you had to do. How do you feel about the balance sheet, where we are today?
Yeah, I feel much, much more comfortable than where we were, but we're definitely not where we need to be.
Mm-hmm.
So from a liability management, as you mentioned, we've sort of refinanced the convert. We've paid down half of the delayed draw term loan that we drew down last year. There's another $300 million due by the end of June. We've done a lot of work on working capital. You can see over this last few quarters, we continued to improve our cash conversion cycles. But, you know, leading up to the cyclical downturn, we had spent quite a bit of time also paying down debt.
Yeah.
And so that financial discipline is still there, and as the business continues to recover, we expect to continue to improve our balance sheet going forward.
Yeah. Okay. Well, it all makes me feel much better when there's a positive EPS associated with all of that.
That's all. The world is a happier place.
That certainly helps.
Um, there's, uh...
And can you talk about the net CapEx that you guys are looking at this year? You know, as the business improves on the NAND side, do you start thinking about investing more in the business, or are you waiting for certain profitability targets?
So, for this year, we're still looking at a significant decline relative to last year. For the fiscal 2024, basically, we still see a much lower, significantly lower CapEx spend relative to fiscal 2023. With respect to NAND CapEx, look, it all depends on the profitability of our business. We're looking to... We are aiming to achieve that through-cycle 35%-37%. And so what that means, we would like to see the profitability of our business improve above those levels before we're ready to invest more capital.
Yeah.
I think this is a really important point. We've been getting this question a lot, especially in the drive business, as demand starts to return, how fast can we add capacity back? That's really not our focus.
Yeah.
I mean, our focus is getting to supply-demand balance. I mean, clearly, in NAND, we've got a ways to go till we get there. We've made a couple of good steps now-
Yeah we're coming from, you know, what? A 20, 30-year low as far as the down cycle.
Yeah.
So the recovery is gonna look very different than anything than it has before. And we've got a ways to go before we see... And, you know, we wanna see that through cycle model-
Mm-hmm
R eally have conviction in that going forward before we put capital back in the business.
Yeah. Okay. I mean, that idea that on the hard drive side, your focus isn't on, you know, supply growth. What if demand comes back? I mean, it seems like it could be a very good backdrop for drives in the sense of, you know, what's your ability to respond to a pickup in demand? And because it's such a complicated supply chain with so many parts, and it's been bad for so long, frankly, that I would think some of your subcomponent suppliers have limited flexibility as well.
Yeah. Well, remember that in the drive business, even—you're coming back to more exabytes, but units-
Yeah
are still going down.
Yeah.
Right? So we've sized the business for what we think the number of units we need going forward that'll keep the market balanced, and that's where we want to stay because, you know, every quarter you go forward, the mix moves forward to higher capacity drives. You're adding exabytes through more capacity per unit.
Mm-hmm.
But I think the drive business has been persistently oversupplied for what? Maybe 15 years, 20 years.
Yeah.
I mean, everybody talks about the Thailand floods. Like, it's like lore.
Yeah, yeah.
I mean, it's like the only time where we had supplied, you know, more demand than supply in the drive business was when we had a flood in Thailand. And I was just there, like, three weeks ago, and they still have the thing on the wall that tells you how high the water went.
Yeah.
But I don't think the drive industry should be any different than any other industry in the data center, right? You can't just show up in a quarter, you need something, and you get it.
Yeah.
That's the way the business has been run for a very long time because of this persistent oversupply. You know, I talked about earlier, we've now... The downturn was, you know, a brutal downturn.
Yeah.
It had impact on, quite frankly, a lot of people's families that work in those places that we are being underemployed. And so, you know, we want to set the supply at a certain level, and then when we get visibility and conviction on sustained demand, not demand that just shows up for one quarter or two quarters or three quarters, I mean, sustained demand for a long time, then we can start talking about-
Mm-hmm
If we put supply back in the system. But we're not at that point yet. We'll see how quickly we get there. I do think in the next several quarters, we'll get back to supply-demand balance.
Yeah
In the drive industry for maybe the first time in a long time.
Yeah. Okay, and then talk about the roadmap. I mean, you said you're growing faster. You know, the 26 TB and 28 TB UltraSMR roadmap seems pretty sound. The bear case always seems to come back to HAMR, and being early on HAMR for its own sake. So I ask you about that every year, and the answer is kind of always the same, but, like, can you just talk to that concern that people have?
Yeah, look, I mean, HAMR is an extremely important technology to the drive industry because what HAMR says is the drive industry has got a long future of continued TCO improvements. And what we're in now, in the drive industry, again, we're this is not a software business. This is, like, the furthest thing from that, where it's like you're actually dealing with material science and physics and all kinds of things that, you know, are not necessarily quite as predictable as far as when they're gonna. You know, big advances will emerge. So HAMR's been in development. You know, I was in Japan, I don't know, it was the end of last year, and I met the gentleman that wrote the original paper on HAMR, and that was published in 2002, and which is fantastic, right?
And now here we are, and we're talking, and, and it's been under development since that time. We've been investing in it for, for years and years and years, yeah, well over a decade, two decades. And so what we're seeing right now is the end game of that development process. And HAMR is going to be technology when we look 5 years forward, where there's a lot more deployment of it. The question is just: When does the transition happen? How does it happen? At what pace? And I think our team, I don't think, I know, our team made a very difficult decision about 5-7 years ago, where they had to bet when HAMR was gonna be available.
They basically decided that, like, this mid-20-TB-per-disk, per-unit timeframe, they didn't really think HAMR was gonna be ready, right? And we're gonna need to get up into that 30-40-TB range. So they started a bunch of other projects, ePMR, OptiNAND, UltraSMR, and over the last 5 years, we've commercialized all of these technologies. And each one of these technologies gives us a little more capacity in our architecture. You know, ePMR has more areal density than PMR, right? That's why we commercialized it. You know, we've now shipped hundreds and hundreds, I think more than 700 EB of ePMR. So it's a commercialized, highly available, highly reliable technology that the largest, most sophisticated data center operators in the world bet their business on.
You know, next came UltraSMR, and SMR is a technology that's been around a long time, right? SMR is not anything new. The issue is it had never been adopted at the high end of the market because it was an additional 10% of capacity, but you have to do work on the host side, so the customer has to do work in their infrastructure to adopt it. And why would you do that if you just thought you could wait another year and get a bigger drive anyway? But about 2 years ago, 2 and a half years ago, the most sophisticated data center operators in the world, like, a couple things happened. One is we launched a version of SMR that was 20% more capacity, not just 10% more.
So now that gets pretty interesting when you're talking about a 20 TB drive. The biggest, most sophisticated data center operators in the world looked at this whole roadmap and the whole technology landscape of hard drives, and they came to the conclusion that we need to adopt SMR because that's going to be the next leg of growth in the data center, and that's turned out to be true.
Mm-hmm.
You know, the reality is, maybe for the first time in a very long time, there's architectural differences in the roadmaps of the largest providers in this market.
Mm.
We've taken a path that is, you know, predictable. We've commercialized it. We can produce it at scale, and that's gonna give us a nice transition to HAMR when we get to the 40 TB range or the 4 TB per platter range, when it economically makes sense. Other people have made other choices. That's okay. We're very, very confident with the choices we've made. It's very clear that the biggest operators in the world have looked at the same set of technologies and come to the same conclusion. Otherwise, they wouldn't be adopting UltraSMR, and they are.
Mm.
So we're very comfortable with our roadmap. We're very comfortable with our HAMR development as well, and we will fold that into the roadmap when we get closer to that 4 TB per platter, and it makes economic sense. Until then, HAMR adds cost per unit.
Yeah.
And you absolutely, you know, in any technology business.
Those costs
You don't want to add more cost to your product unless you can get more value from it.
Yeah.
You know, so on a like-for-like basis, we think we have a cost advantage with the roadmap we have, which is a nice bonus to the technology strategy we've taken.
I appreciate the thoughtful answer there. In terms of the environment for nearline, you know, you talked about drive units, kind of soft exabytes growing. It seems like we've deferred a lot of the technology upgrade business in cloud to make room for more AI spending and stuff like that. Any indication that that could change? Any indication that people start to invest more in storage in the cloud?
Look, I really believe that the recovery right now is really inventory driven.
Mm.
I don't think it's as much as this squeeze-out of other spend-
Yeah
Because storage is important in the cloud.
Yeah.
Right? And I think that there had just been so much over-procurement in the drive industry, that there was a lot of inventory to work through, especially at some of the big cloud vendors. So I think, you know, that long-term, 20-25% exabyte growth still holds. We're significantly below that, and now we're having this kind of reversion to the mean in that ordering as all this inventory has been worked through. I think we're gonna have that return to normal ordering over the next several quarters, and while we're going through that, we're gonna see what impact AI has on our business, which we think is going to be substantial.
We think all of this, to your point, all of the spend that's going on to build out the compute infrastructure for AI so that all of us can be enabled by this cloud-driven technology, is gonna lead to more content creation. The vast majority of that is gonna be in the cloud, is gonna be stored on hard drives, and of course, on the device, it's gonna be stored on NAND. So we think we have those tailwinds behind both of these businesses.
Great. So on, on the flash side, can you talk about your technology roadmap there? You know, when do we see BiCS8 starting to ramp?
BiCS8, BiCS8 is a great node. You know, we can talk some about wafer bonding. It's a big innovation that allows us to really build a high-quality node. We will ramp that node at some point next year. We still have BiCS6 to get through, right? Most of our portfolio is on BiCS5 right now. BiCS5 is the highest-yielding node in the BiCS roadmap, so cost downs continue to be still very good, even though we're not rolling nodes forward, which usually what drives cost down. You'll see our technology roadmap. Traditionally, we have moved the entire portfolio from node to node, right? All of the portfolio is on BiCS4, then all the portfolio is on BiCS5. We're gonna change that now. Only part of the portfolio will go to BiCS6, mainly QLC-driven.
We'll have a QLC, client SSD coming out, which I think people are gonna be very impressed with the performance of that.
Yeah.
And then we'll take parts of the portfolio onto BiCS6, and then we'll move other parts of it on to BiCS8 as we move into next year.
That should eliminate some of the volatility that you saw in, like, BiCS5, where initially you're not qualified with some vendors, and you end up selling to the lower-cost customers.
Yeah, that's a big, that's a big issue, I think, in the NAND business that's underappreciated-
Yeah
Is, like, which-
The mix
- which mix-
Yeah
Is on the leading node, which mix is on-
Yeah
The trailing part of the node. A lot of it has to do with you have to build a controller or not. Usually, you tend to see the mobile providers on the leading edge of the node. You tend to see enterprise SSD on the trailing side of the node because the controller and the qualification is complicated. So yeah, this will make all of that a little more, a little more of a natural progression.
Then on the NAND cycle, I mean, there's no CapEx. Talking to 3 equipment suppliers in the last 2 days, that doesn't seem likely to change this year. It's gonna, you know, NAND spending will be up at most, very modestly, which is like maintenance level to me. There's very little supply growth coming from that. The only anxiety I have is utilization. You know, you took down utilization, others did as well. You know, people are in various stages of acknowledging that they're going back to full utilization, but at these margins, it seems like it's pretty much just math, that you run the factory full. So how do you think... You know, do you have any trepidation about that utilization coming back? And clearly, it's offset by very little underlying supply growth.
No, we have all that modeled out in our own internal models, and quite frankly, you're gonna need that.
Yeah.
I mean, one of the things I'm excited about going, you know, leading a NAND business-
Yeah
It is it's a great business. I mean, it's a great market. There's elasticity, there's growth. Like, if you can continue to drive the cost down, you can, you know, there's more bits that are required every single year. It's a very dynamic market. And so, you know, we see fab-out bit supply mid-single digits, and so obviously the thing, the buffer on that is inventory and utilization. And so we're working through the inventory now. We expect utilization to come back up and—but the real story is CapEx.
Mm-hmm.
Because that's whether it's. And usually the CapEx is for nodal transitions, right? It's not for greenfield starts. It's just when do the nodal transitions happen. And so we see very little CapEx, and, you know, it takes quite a while from when you start spending that CapEx to get, you get the output from it.
Yeah, I mean, it's very clear what's happening to me is that, you know, you have a lot of investing in specialty DRAM in areas like AI, and it's siphoning away CapEx from NAND. If the two companies who aren't in DRAM also aren't spending very much, which you aren't, and your partner isn't, it seems, it seems like a very good cyclical backdrop to me.
I think that this downturn, you know, I think, showed a lot of things about the NAND market that people believed going into it were not quite true.
Yeah.
Right? And so, a lot about, you know, if you weren't in the enterprise SSD market, you couldn't have a great NAND business. Turned out that's the worst part of the market.
Mm-hmm. Mm.
This, I, you know. There was just a whole bunch of stuff about the way the NAND, you know, that we could just, like, continue to put the CapEx in, and the elasticity would take care of demand.
Yeah.
Like, I think there's been, you know, the formula of R&D in NAND, where you're focused on how much bit output you're gonna get, how much cost down with how much CapEx, right? That- Like, how you solve that equation, I think, is going to change going forward. It's not just about how many bits I can get. You know, a lot of the discussion going into the downturn was whoever has the most layers is ahead on technology. Well, that didn't work out very well, 'cause that just means you have a lot of CapEx going in to produce a lot of bits that can't be absorbed by the market. You need to focus on cost downs as well, and how much CapEx you're putting in to get those bits.
So I think the industry, and certainly we, we think about this very, very deeply, about the economics of NAND in the 3D era and how they're changing, given this equation about how much CapEx you're gonna provide, how much bits you're gonna get out, and how much cost down you're gonna get out of the portfolio, and how you think about that going forward, I think is changing.
Yeah. I mean, if you look over 20 years, SanDisk gross margins, you know, tended to bottom in the mid-teens, peaked in the mid-50s, which gets you through cycle to about the numbers that you're talking about as your long-term model. The last cycle, it was below that, both the last peak and the last trough, but we still, the 7 years since the acquisition, we modeled out the EPS, you still have... You know, again, we'll find out where you're putting OpEx, but we still have $4 of EPS on average from NAND over the last 7 years, even with a bad cycle, relative to the 4 prior cycles. So, you know, what do you think happened in this last cycle, that it was a little bit weaker, and what gives you confidence that the next cycle will look more traditional?
Well, the last cycle, we got to kind of a mid-cycle level-
Yeah
O f, like, and then the pandemic hit and, like, all heck broke loose and, like, something we've never seen before. And we had this huge, you know, demand-driven downturn instead of supply-driven. And so, you know, I think there were also some people ramping very large nodes right into that downturn, which ended up with just a flood of bits that couldn't be-
Yeah
Absorbed. I think we've always. Look, I think the real key to our business is the fact that, as I said earlier, you gotta get two things right. You gotta have the best fundamental technology. Like, the wafer coming out of the fab has got to be at the best cost point and the best product out there. You can't make up for that with a great portfolio. But then you've got to marry that to a really, really strong portfolio, where you have the most optionality possible to mix into where you're gonna get the highest return. Because NAND is, like, a bunch of different markets that all require different amounts of investment.
I mean, literally, you can take the wafer out of the fab, and you can just sell the wafer, or you can, like, cut the wafer up into components and sell it into the mobile, where you don't have to, you don't have to build the controller-
Mm-hmm
So there's no OpEx involved. Or you can get into the enterprise SSD market, where you have to have a very large engineering team that are building controllers that are extremely, extremely sophisticated. Each of those markets has a different growth rate, a different profitability, a different OpEx profile. How you solve that equation is extremely important, and what we're able to do is have a portfolio that's very diverse with an anchor of the consumer business, which you said earlier, right? You talked about that going way back.
Yeah.
Like, that consumer business is a real gem of the portfolio-
Yeah
Because the brands are very strong.
Yeah.
I think everybody knows the SanDisk brand.
Yeah.
SanDisk Professional, WD_BLACK in gaming now, is a hugely valuable brand for us that we've built over the last three years. So you have that best-in-class core technology position. Why is it best-in-class? Because we invest with Kioxia.
Mm.
Between the two of us, we're the largest provider.
Mm-hmm.
At the very high level, the amount of engineers you can invest in something is commensurate to the market share you have, right? So we have really, really bright engineers who've been doing this for 24 years, and we have them at scale. And then on the other side of it, we have a very unique portfolio with a lot of optionality, and a big part of the business, which, you know, the consumer business, which provides through cycle, let's say, call it resilience, or it's better profitability, less volatility. And we have the ability to build brands in that, like in any consumer business, that, you know, continue to accentuate that, that attribute.
Great. So I think we have time for one question from the audience, if there are any. And if there's not, you know, a lot of focus on DRAM and AI because of HBM and some of the specialty technologies. But, you know, we're also hearing a fair amount of interest in NAND. You know, low latency inference at the edge requires a lot of SSDs, and there's a lot of investment going on there. Can you just talk to, you know, whether or not AI can become a driver for your business?
I don't think there's any doubt that AI will become a driver for our business, right? I mean, we're already-- we're seeing data points of it all over the place, whether it's the, you know, the corporate-- the traditional corporate PC, which is, has the least amount of NAND on it of any PC that's sold-
Mm-hmm
Like, being respec'd with twice as much NAND going forward. You know, we talked to device manufacturers. This is a consistent theme. In an AI-driven world, we're gonna have more NAND per device. So you kinda got that another driver of that elasticity lever in the NAND business. So 80% of the NAND TAM is devices, right? And that's gotten more concentrated in the downturn. And so we see, you know, the classic edge-driven, device-driven refresh from AI coming to the business.
Great. Well, I'm losing my voice, so we'll just wrap it up there. Thank you, Dave and Wissam.
All right, thank you.
Thanks.
Appreciate it.
Thanks, guys.
Thank you for being here.
Thank you. Thanks. That's great.
Thank you.
Thank you. Thank you so much again.