Okay, we're going to start. I'm Tim Arcuri, I'm the semiconductor analyst here at UBS. Very happy to have Seagate with us. And we have Dave Mosley with us. Thanks, Dave. Dave is the CEO of Seagate.
Thanks, Tim. I've got some just forward-looking statements. Check out our SEC filings on our website. I want to talk about an 8-K that came out this morning. Just to clarify a couple of points, and hopefully that'll inspire some questions that you can come up with. First on HAMR qualification, which is our new product introduction. Think 30 + TB, 3 TB per disk. We've completed our first qualification with our lead CSP. We had a few other calls done already, but the lead CSP has been a long ordeal for us. Part of it's because if you think about cloud qualification, the cloud drives are going to last for a long time, seven years, 24/7 operations. It's a tough environment. The industry's done a really nice job at responding to that.
We're proud to have gotten through this point, and we're going to start shipping initial volumes as well in the coming weeks. Happy about that. On the flip side, we also, in the 8-K, said that we've got some production issues. This got to be a little bit explanatory about what exactly is going on. We're not adding capacity, but we're bringing capacity back online. That's the way to think about it. We have been for the last nine months. As we've done that, we had some issues just getting enough labor and the right tools on and getting them qualified and so on. It's not going to hit this quarter, this fiscal Q2. It's going to hit fiscal Q3. In the 8-K, we said, "Let's bracket that." We said it could be as much as a $200 million hit.
Now we're going to continue to work it. It does not affect HAMR. It affects all legacy products. I shouldn't say legacy because we use that word differently, but all the PMR products from prior. But we've prioritized to make sure that the build-to-order commitments that we've made to certain customers are protected, and we're still working it. We've got a few weeks before that quarter starts, and then we've got the quarter to continue to work it. Again, there's no impact this quarter. And probably most of the impact is going to end up being after we get through all the mixes. There was a little bit of nearline, but it'll probably be some of the things that we're seasonally already going to be down anyway, which we talked about on our last earnings call. So that pretty much walks through the 8-K.
Well, let me just ask about that. So I chatted with a couple of folks. I think there's still some confusion just in the sense that I would think that that would cost you incremental capacity, not from your current volume. So can you walk through how it takes your current volumes down versus your incremental down?
Yeah, bigger supply chain. So if you think about the issue, we think is largely behind us. We still have to get more tools back ramped, but it's kind of like the egg going through the snake right now. We've got to make sure we get these tools ramped to make sure that that capacity bubble, if you will, gets processed properly. But the time that we already lost is already gone for good, and that's what's really going to impact in Q3. We think we can mitigate it for anything any further out than that. There were some questions people would ask about how we manage inventory and so on. So we can actually pull inventory from the inventory pool, if you will, and we've depleted that inventory pool somewhat, which is why it didn't impact this quarter and does impact next quarter.
Is that because you had to take some of the existing capacity down because it interlinks with these new tools you have to bring back on? I'm just trying to click on why the decline.
Oh, sorry. The capacity was offline before. What really happened is we just weren't getting the tools back online fast enough, and we looked at it and said, "We're not making it," but we have enough for this quarter, but we don't have enough for next quarter and so on. Being really transparent about it.
Got it. But you're not saying off of what because you had never guided March.
That's right. We didn't guide March. We're just trying to bracket what the impact could be. Like I said, we'll still work it. We've still got some time. Operationally, we'll mix as much as we can towards the products that people need and where the market is and so on and so forth.
And then how to think about what the gross margin impact would be from that? I mean, should we just think like normal drop-through?
Yeah, we haven't guided it again. We'll probably end up chasing product with the limited upside capacity, which for the cloud, it's going to be almost zero. But with the limited upside capacity that we can actually eke out right now, we'll just chase the best product we can.
Okay. So I guess that leads into how would you characterize sort of the demand picture right now by all the different end markets?
I think this quarter has been fairly predictable, like we said on the earnings call. There's a little bit of trepidation in early next year, but generally speaking, I think the cloud is feeling like growth, so I think the 25% growth number that we've been pointing to feels pretty good right now, long term. We have to get through these little tactical issues. But what we're focused on is getting our products right, so getting 3 TB per platter ramped and then 4 TB per platter qualifications done and so on and so forth so that we have the right products for when that growth ultimately kicks off.
Great. This is just sort of a big picture question. Things are really changing a lot in the hard drive industry with build-to-order. And just can you talk about, and I ask you the same questions on these earnings calls, there's a point at which customers just don't know what they're going to need a year from today. And they say, "Okay, you know what? Fine. If you tell me you're not going to start a drive until I order the drive, that's fine. I'm just going to order the drive." So how is the dynamic, the discussion with the customers, and how do you, to the extent you can, ensure that they're not just placing orders just because you're telling them that they have to place an order for you to start the drive?
Very carefully, so I'd say prior behavior was more we forecast and say, "Does this demand look right?" and some of that was our influence as well, and of course, where we were as an industry, we've always had overcapacity since the peak of client server like 10 years ago, so now that that capacity has been taken offline, and we're going through these aggressive product transitions, we can steer at customers and say, "Okay, do you really want us to build that for you?" and so I would say we're going for predictability right now. We're not going for big upsides. We're not installing any additional capacity. Again, we're bringing back online some of the tools we had taken offline, so the peak exabyte capacity is still going to be roughly where it was when we started and when we had the peak of the pandemic.
From my perspective, that's kind of back to normal health, and the industry is not back to that normal yet, but once we get back there, then we should have this thing where predictability is an outcome, and instead of seeing demand that we are forecasting or the customer could actually overdrive, hopefully we'll get into situations where that's not happening. There's not a lot of buffering or stockpiling or anything like that going on. I know there's a lot of people that service the cloud industry that are demanding this kind of predictability just because the cyclicality is so poor for your investment thesis. It doesn't work for anybody. It doesn't work for them. It doesn't work for us, so we've got to try to smooth it together.
If when the time comes, if in a year they place an order with you and you go out and you build the heads that you need, and when it comes time to deliver the drive, if they say, "I don't really need it," what happens then?
We have long-term agreements that say they'll still continue to pull it, so that'll give us long visibility into they'll say they don't need it today or tomorrow, but then ultimately their signal is going to come down. Hopefully, the world is diverse enough at that time, and hopefully also, we've got our products qualified against fewer families, so we don't have pockets of inventory all over the place, but we can actually be a lot more nimble against that kind of world, but like I said, what kind of got our industry into trouble was the cyclicality of the pandemic was we just had way too much capacity, and to some extent, when the demand was really hot, we all built a little too much and had that inertia coming out of our supply chain.
Customers got, and when they stopped, they stopped and they got great pricing on the way down. The economic magnification was huge.
So your total exabyte capacity, I think you were doing a little more than 160 exabytes last time. And really, capacity is not an exabyte thing. It's a unit thing. So your drive capacity is not really growing, but you can obviously grow exabytes through density. So exabytes will grow from here. But the question really is more around absolute drive capacity because you were bringing some of these decommissioned tools back online. And sort of what's your mental thought around how much that's going to add in terms of drives, of your capacity to actually produce drives?
Probably not much at the drive level. So to your point, there's heads, there's head wafer capacity, there's media capacity, there's drive capacity. Probably not much as far as drives is concerned. But each drive has got a lot of heads and media in it. And as those heads and media become richer and richer and richer in capacity capabilities, then we can grow exabytes, if you will. Drive capacity, though, is from an ability to build drives. It's pretty easy for us to do. It doesn't take a lot. And once upon a time, we did over 50 million drives in a quarter. So we still have clean rooms, and we could get that capacity back online. We could up that part of the capacity if we wanted to. Test is big. It gets a lot of airplay. So test is something we'd have to go manage.
There's great strategies around that. We could output more drives, but right now, we're not planning to. We're planning to hold the line even on that. The really expensive and long lead time stuff is way better than media.
Most of what you were bringing back online relates to pushing density per drive?
That's right. That's right. Yeah. Tools that we had taken offline intentionally to manage the financial situation we were in about a year ago.
And it's encouraging to hear that you're finally qualified with this customer. Can you talk about sort of now that you're qualified, how is that going to ramp? Because there's a concern, I think, that, well, this has taken so long that the customer is now seeing that maybe they have other ways to get the density they need and not adopt HAMR as rapidly as they had originally planned.
I think there's going to be a desire to continue to ramp because it's the best value out there on the market. I mean, if you think about you're building a data center and you're building at mid-20s capacity and you have the opportunity to go into the 30s, then you do so, right? You're going to run that data center for seven years. It's just way more efficient power space, total capacity inside the data center. So there will be a push. It's a different type of economics. I said this on the last earnings call. Different than it was maybe a year ago. A year ago, when our margins were so depressed, we had to push to HAMR. Now, we're not going to incentivize the transition as heavily because some of our products are doing quite well right now.
So we're going to make sure that we get paid for it as well. But there will be a ramp. And I said in the prepared remarks, if you will, that we'd start shipping in the coming weeks.
Then, just with respect to share, I know you don't manage for share. Share is more an outcome than an input. How do you assess? I mean, now that you have qualified HAMR, I mean, I think it probably stands to reason that you've been trying to qual 24 and 28T alongside HAMR, and your competitor was just focused on 24 and 28T. So they weren't focused on as many things as you were. So maybe that is to sort of explain the recent gap in the exabyte shipments.
Yeah. I think when we were so depressed as an industry a year ago, we made some conscious decisions that we were going to do this build-to-order. I'm not going after share at all right now. I need predictability. And to some extent, our industry is still in that mindset. We need predictability. We need to be able to get paid for what we do. If we're going to extend hundreds of millions of dollars' worth of material out against POs that will ultimately come, we need to know those POs are going to materialize. And so I think the industry is still there. We're still working through a bunch of working capital issues for the next couple of quarters on that front as well. So yeah, share is an outcome of having great products, but also supply and demand management.
Right now, finally, the supply and demand is finally coming back into balance, but we still have some patching up to do of what we've just been through. So I'm not really focused on market share.
And do you think that there is any potential? Back to my question about, well, why not just place the orders? Is there any stockpiling that could be happening? Or would you say, look, supply is still pretty tight, so there's not much of an ability to even stockpile even if you wanted to?
Yeah. We don't see any evidence of it. I mean, there was probably evidence that we didn't see two years ago, three years ago when it was happening, but we don't see any evidence of it now. We're looking really hard. But the other thing is, as an industry, we were also kind of pursuing that behavior by having too much capacity and pushing it. Right now, the customers don't order it. We just don't build it for them. And so we're trying to instill that discipline, at least for a while, until we can get all our through these product transitions and all of our products right where we want them.
And so ESSD is taking a lot of incremental demand around AI. I guess it's not really demand that would have gone to HDD anyway. It's just incremental demand that goes to ESSD. But how do you think about your leverage to AI?
Big, big discussion topic. There's a lot of new applications coming in, and some of the applications are what I would call small data applications. It's perfect for memory, and so the SSD, great technology. It's very applicable there. When I think big data and big data centers, I usually think big unstructured data or video or something like that, and the application space, the AI application space, you could always say that some of the video properties today are already using a bunch of AI. Fine, but I do believe there are new applications coming as well. That's the watch item for us is to see if we break north of the 25%. There's a lot of small data applications as well, so it's going to be a healthy space for SSD. I view them as complementary technologies in a lot of cases.
There's some places where an application will not need mass capacity. There's other applications where big data pools, data lakes, they're going to need a lot of mass capacity, especially when you can process that data with AI much more efficiently. So we're excited about it. It's still very early days, I think, relative to mass capacity in AI.
And how, I mean, do you have concrete examples of, well, anything that drives more data is ultimately good for you because hot data becomes cold data eventually? So are there any concrete examples where you see, if you're monitoring data, you see an increase in cold data because of what's going on with these large language models, or is that still out there in front of us?
I pay attention to all these hobbyists that you can just watch their YouTube channels and see all the cool stuff they're doing. And a lot of people don't want to pay for tokens up in the cloud. And there's all this AI tourism. They're trying to find what the best model is for their workflow. Ultimately, they might say, "I like that workflow, but I want to do it in a way that's much more cost-efficient for me, so I'm going to do it on-prem somehow." And in that case, there is an opportunity for a lot of on-prem data. As far as what's going on inside the existing cloud applications, you'd have to talk to the cloud service providers. I believe large language models are gateways to much more efficient use of data. And so that's the thing I'm really focused on, not the text itself.
I mean, the text is usually pretty small data, frankly. So it's going to be fast, but it's going to be pretty small data.
I want to go back to HAMR and the ramp. And when you were going to get qualified, I think it was last spring maybe, and you had an expectation that things were going to go vertical in the back half of the year. And I think you were talking about a million units during the year. Can you speak to the dynamics around sort of from the customer perspective? Might they just stick with taking 24T SMR and 28T PMR rather than moving to HAMR?
No, I think ultimately everybody's going to move to higher capacities. I mean, they'll go in the 30s and the 40s as soon as we get them. And I think there's still a big push to be able to make your data centers more efficient, space efficient, power efficient, and so on. As we talked about in the earnings call, we've shipped Qual units to a bunch of people right now, and they're very hungry for them and very active in the Qual. So I'm not really that worried about it. It's a different dynamic economically than we were in a year ago where HAMR was kind of our way out. Now we have a lot of products represented quite well, doing well, making money. We're a lot healthier as an industry, so I'm happy with that.
But by the same token, I think we'll continue to push for higher capacity points. And that pull will be there from our customers as well.
Given where margins are now, is HAMR accretive to where margins are now? I mean, I know initially it was supposed to be, but now, even so now.
Yeah, yeah. I mean, John Lucas answered this on the last earnings call the same way. I mean, I think the margins come up in the industry, so that's a high bar to jump over, but the HAMR still can be accretive, and we'll continue to work the cost down and the technology transitions and so on and so forth to make it even better as we grow.
Can you talk about just the competitive balance? I know that you don't want to gain share, but if HAMR really allows you to scale and you're out in front on HAMR, there's a lot of investors that think, well, okay, it's not your desire to gain share, but the natural course of things would be that if you have a solution that can scale density better and is a better solution, then the natural course of things is that you will gain share.
Yeah. There's a little bit of a drug in our industry, has been for a long time of full factories gets you margin, right? And so if you build bigger factories, then you have to be able to fill them up. And that's part of the thing that keeps on taking away, right, rather than keeps on giving back. So I think what we have to do is we have to get our product set to the point where we like the margins, and then we'll make the right investment for them and not incentivize with a margin drop, right? So we have to make sure that the supply and demand balance is proper. It's not there yet, but we will get there. And then make sure we have the right products against it, and then we can continue to make money.
If that translates into market share, it's because the customers demand our products. That's the way I look at it.
Yeah. Was there anything in the Form 10 from your competitor that sort of more clearly defined what the financials are going to look like for the standalone HDD asset? Was there anything, or just the general concept that they are separating the two businesses? If it was such a great thing to put them together in the first place, then you could argue there's a synergy to rip them apart. I'm not saying this is my view, but there's a view that, well, on paper, that makes the drive asset maybe not as strong as it was when it was together with the NAND asset. Do you see anything changing in the competitive dynamics of the industry since they're separating those two businesses?
Not really. And I think it's been a long journey. So I won't make any comments about where things were five years ago or three years ago because I think that's ancient history now. But it's been such a long journey. I think they've just got to get through whatever they've got they're going through. And there's been twists and turns too. So I don't know that we know it. We know what's going to happen. We'll just wait and see, I think. And it sounds like we'll have a solid competitor that's ripe for the fight and willing to invest in hard drive technology. I will say maybe a little bit about seven years ago, eight years ago, is that we said hard drives aren't going away, and they said we're going to flash. And hard drives aren't going away.
I mean, they're going to be here in 10 years. They're going to be healthy and storing your data and everything else. So that feels good. Now we got to make the right investments, get paid back for those investments, and keep the industry going.
How do you view your cost curve relative to NAND? Because NAND has been, it's been in the mid-teens. It depends on how much capacity they're going to upgrade now because there's a lot of stranded old stuff. So maybe they can maintain mid-teens, maybe. But how do you feel about your cost curve relative to NAND?
Yeah. I mean, there's a lot of people that say six to one, seven to one. What's the right math? I would say at levels like that, it's virtually infinite when you talk about mass capacity in big data centers. I mean, there's not enough capacity to go replacing hard drives, and you just wouldn't need end to it very much. You might buy more flash because you need flash for new applications, which is great. Some of those applications might store colder data, to your point, right? So I look at it as a balancing act, not as an either/or. And I think the people who are investing in capacity and so on and so forth would be wise to do the same.
Got it. Can I ask about some gross margin puts and takes? I know that John Lucas is not on stage. But can I ask about some of the gross margin puts and takes? September was a bit of a tough comp. You had good mix with higher mass capacity, and then the underutilization went away. So that helped gross margin. So what's the right baseline and sort of some of the puts and takes on gross margin as we exit Q4 and we go into next year?
Yeah. I think we're still in the period of putting the industry back together. So we'll still look at that customer. Rather than say no, we may say yes to a certain price or as part of an LTA. We may say yes to a certain price that we like. We're not really trying to drive margin up, but we are very conscious of needing to get paid back. And the revenue is still too low relative to total dollars for us to get paid back for what we need to do. We'd like to escape this environment as quickly as we could, right? But until we get the revenue up another 10% or 20%, I think we're still going to be on a plodding path. And we have some debt to pay down. And so we have a lot of needs for that cash.
We're going to demand that. Otherwise, we're not going to build it because we just can't get into the same game again of putting hundreds of millions of dollars at risk and then not getting or getting played for it on the backside of that. After we're through this period, we get the products where we want them to be, then we can have the discussion about what's next with customers and so on.
So where do you think you could take, and I'm not asking you for a target, but I'm just asking you generically, is there any reason why for somebody. I mean, you have these massive companies that are $100 billion companies globally that have almost infinite financial resources that are highly dependent on a number of infrastructure providers. And in a pretty important part of the market, there's two. And the margins for those two suppliers are below any other supplier that they have that is as critical as you and your competitor are. So why can't gross margins ultimately be 40% to 45%? I mean, why not?
Yeah. Ultimately, they can. If the customers come to those kind of epiphanies, like you just said, if they decide that they're on a different strategy, then they might try something different. But for now, the demand is not significantly higher than what the supply. Actually, the industry has been able to respond pretty well to the demand swing up in the last six months. So from our perspective, we're saying, "We'll only build it for you at this economics." They're saying yes to that economics, but not much more than that at this point. If data needs to be stored for longer, archived for longer, if some of these new applications take off, then you could see more demand, not just exabyte demand, but more demand for high-value products, and then they start paying more for it.
Our goal is to go get through those new product transitions so that we can make as much margin as we can underneath that.
So I guess I always think, if I'm in your seat, I push pricing as far as I can until the customer pushes back and says, "No more." And I get the feeling when I talk to your competitor that we're nowhere close to that, that they're not getting any pushback. But it sounds like you feel like actually the customer is beginning to say, "Hey.
Well, I certainly think six months ago, I mean, when we first started into some of this build-to-order behavior, people said, "There's no way you can do it because supply is way above demand, and the customers have gotten all this. They're accustomed to this." But they do need predictability. To your earlier point, they do need predictability as they're running data centers, and they need product across the entire swath of what we do. And then if you have a great value proposition, you're out in front on technology and so on and so forth, then that'll command a premium as well. So that's how we're looking at it, and that's exactly the future we're trying to build. We're just not there yet. I mean, like I said, the industry has responded pretty well to the upside. Demand is not way, way, way above supply. It may go there, but we'll have to wait and see.
Right. So you'd say that the customers are not pushing back necessarily, but it's still a discussion. They're not just happy to pay higher prices.
Yeah, that's right. That's right.
Got it. And then maybe just a last question. You have a big slug of debt due in January 2025, and I think you plan to pay that out of cash. Maybe can you just talk about the balance sheet and sort of debt reduction versus share? I know you want to bring down debt prior to resuming the repo.
Work our working capital and bring down debt. We've set targets so that tranche in 2025, the maturity, would take it down to 5.2 or something like that. We were pretty transparent on this on the last earnings call or the one before it. Yeah, I think, are we happy with the debt at that level? Maybe it could, it depends on what the revenue ultimately gets to, but maybe it could use a little bit more trimming. But once we get through that period, we've upped the dividend recently, so we're showing that we're getting back to being who we want to be. We don't feel like we need to grow OPEX a lot to be able to hit significant upsides. We don't feel like we need to grow CAPEX as a percentage of revenue either.
If the revenue starts to grow, then we can then make investments underneath our 4% to 6%. So I think optimistically, once we get through all these working capital challenges that we've talked about, then I think we get back to being the person we want to be, and some of that includes share purchase.
Great. Well, we're out of time. Thank you, Dave.
Thanks, Tim.