Seagate Technology Holdings plc (STX) Earnings Call Transcript & Summary

I'm Tim Arcuri. I am the semiconductor and semi-equipment analyst here at UBS. And for this next session, we're very happy to have Seagate and happy to have Dave Mosley, who is the CEO of Seagate. So Dave, I don't know if you have any Safe Harbor that you want to put up or if you want to just go right into questions.

Yes, I think, Tim, we'll just start with questions, keep everybody in mind that we will be making forward-looking statements and all the relevant documentations are on our website www.seagate.com.

Perfect. Thank you, Dave. So first, I just wanted to talk about the transition between 16 and 18T on the HDD. You did a great job on 16T and you sort of held serve certainly for 18T. You're shipping about the same in terms of exabytes as your peer but your December quarter guidance is much better, maybe due in part to your China exposure. So I guess the question is, can you talk about your competitive position and maybe how you see yourself more broadly positioned with the cloud customers in particular?

Sure. We started our journey on 16T. Gosh, it seems like 3 years ago now, really pivoting towards that platform, which we call the common platform, allowed us to extend to 18 terabytes and 20 terabytes and beyond without having to change tooling every time. And a lot of commonality between our fundamental critical components, heads and media that served us really well. And it's what has done the most for us has allowed us to have long-term discussions with customers about, all right, how many 16s do you need against this quarter, this quarter, this quarter? How many 18s do you need about this quarter, this quarter, this quarter, so on without some of those typical transition conundrums that you get yourself into. And so now as we march towards 20s and beyond, we'll still be able to have those conversations. The platform continues to serve us really well. It's true at many cloud service providers around the world to your point. So we've got pretty good visibility, and we're having long-term discussions with people about what they need out of the platform.

Great. And can you just talk maybe about just if you kind of zoom out, talk about the competitive position? Should we expect just this volley back and forth like we've seen between 14 and 16 and 18. Is that sort of how you think things are going to go for like in exabyte shipments over time? And maybe just zoom out and talk about your competitive position just in the marketplace.

Yes. I don't really fixate on those kind of puts and takes that way from a market share perspective or anything else. I mean, we -- again, we sit down with our customers and we say, what kind of build-out are you going to be doing over the next year? What do you need when, where? And then we'll have a negotiation with them about what we start, therefore, in our factories. And that's why we're getting some predictability of demand. It's when you actually get into that quarter and you see these little puts and takes, I don't put a whole lot of stock into it. The way we're trying to run the business is as predictably as we can with as much forethought into those future quarters as we can get, and it's working out pretty well for us.

Great. Can you just talk about the disparity, as I said, obviously, between you and your peer for December? Is there any of the disparity we see between your December quarter outlook relative to your peers maybe as a function of inventory build in China? Can you talk about that? Is that something you see? Is it something that you're concerned about?

A couple of points I would make. First on inventory in various chains, things like distribution channel. Back when we were a heavy client server business 10 years ago, you could probably read something into what was going in the inventory in various channels around the world. The channel's become such a de minimis part of the overall business with the rise of cloud data centers and things like that, that I would probably say don't draw too much an inference on what's going on in channel. So from our perspective, our inventory was pretty good ending last quarter, going into seasonally what's a fairly strong quarter for us. And I also would say that there's a lot of people who are having supply chain issues out there. It's palpable. Freight logistics issues as well going through a strong quarter, right? So it doesn't surprise me that people were having these little temporary issues. We see it through our customer base quite a bit on semiconductors and various piece parts that everybody else is struggling with. So -- and we'll get through all this together. I don't think it really changes the long-term dynamics relative to demand in our industry though.

And then do you get the sense just from that view that is there a significant part of demand that -- not that you can't need it, but more because the customer is not pulling the drive because they can't get other parts, how prevalent is that?

Very small. But there's certainly some of it to say it doesn't happen, it would not be right either. We've seen some seasonality, if you will, especially in some of the consumer markets and even the VIA markets part, what we call the smart city, smart factories, VIA markets that are really part of mass capacity drives for us. There's some natural seasonality. We've been forecasting that all along. But -- and some of that may be impacted by, to your point, you can't get parts right now. So do I really buy against that or buy something else. But long term, the growth of data, the growth of exabytes and so on, I think that's a minor blip this year against what's happening out in the world.

Maybe we can talk about cloud. I mean, obviously, the demand signals in cloud have been quite strong. Maybe can you just talk broadly about the demand tone you hear from your cloud customers obviously on the nearline side and really particularly I'm curious about their willingness to engage in discussions around longer-term fragments.

Sure. I think it's really a good discussion point when this unfortunate pandemic hit us all at the front end of it, I don't think people really understood how to react. And a lot of the way I see it, at least a lot of the CIOs out there in the world pushed a lot of workloads right into the cloud and said, "I can't get people on-prem so let's do it anyway. And that put a lot of strain on cloud service providers to meet their service level agreements on compute, network and memory and other things like that. Not so much storage because the storage usually comes later. So -- but those -- the pushing of those workloads to the cloud will create a situation where there'll be a long protracted build-out of mass capacity on the back of it, now that everyone is relying on that. I think there's also -- the good thing about us is that we leverage our investments that we make into those type of products for instances that are even closer to the edge, if you will. So if there are mass capacity instances at the edge, we get to use the same hard drive that's in a cloud data center as we do in an edge location for that. And so we get to leverage. As that edge continues to grow because of latency or cost or data sovereignty or whatever other issue is, we'll take advantage of that as well. So I think we're seeing a compounding of that over the long run. But basically, the shift to the cloud that created all this momentum that we see is really -- it has to be followed up by the investments in data in the cloud.

Can you just walk through maybe double click for us on what these LTAs will look like. Are they -- do they have teeth in them? Is it more like, okay, over the next 6, 9, 12 months, I'll take -- my target is to take X exabytes and you give me Y price. If I upside that then I have to pay market price. And if they don't take that product, I mean, you're not going to put that product to them, but at least that gives you some kind of visibility kind of in terms of what they need. Is that roughly how these are working?

I'd say I wouldn't say the word teeth because we're talking about customers here. We're trying to do what they need out in the world. And we -- and this is a true co-planning exercise. You're helping them. They're helping you. We're not all super aggressive. We're basically saying what we need to book together to make sure that you have a certain amount of the drive type that you'll need. The drive needs to be through qualification and up the ramp and all the other metrics that it needs to meet. And just giving visibility to people, I think on both sides, it helps us run clean operations and it helps them be very predictable from what they need to do as well. So I think that's a win-win for the customers, it's a win-win for all the value propositions everywhere. That's what -- really do for us. They're not about -- if anything, they're probably a minimum balance, if you will. So there are upsides that come. But if there's downsides, we'll work it through with customers. They're just a way to drive us more predictively.

Got it, got it. And do you have any statistics or maybe even anecdotes, even qualitatively, where the average duration of any conversation that you're having with these cloud customers, the average duration used to be 3 months. It was very transactional, where they would just negotiate out 3 months. And now the average time frame is 6 to 9 months. Anything sort of even qualitative that you can give us on that?

I would validate qualitatively what you just said. The -- if I go back 5 years, so we were still had a lot of our capacity pointed at client server. And so -- and the cloud really hadn't come of age to the same level that it is today, not only on the replacement cycles in the cloud but on new build of data centers as well. So I think those dynamics changing over the last 5 years have really changed things like you just described. People could afford, 5 years ago, to be a little bit more transactional at the end of the quarter. And now if you're going to make these massive data center investments, you need to make sure that parts are coming in consistent with that, so the 2 can service what you've got in the data centers and build out according to what your business plans are as well. So it's providing more visibility.

And I guess to that end, Dave, so the HDD business reminds me a bit of the DRAM business from maybe 4 to 5 years ago. And certainly, in that business, we've seen higher highs and higher lows in terms of gross margin for the industry. But the HDD business, until only very, very recently has not really been able to drive balance in terms of supply and demand. But customers are now replacing old 4T capacity with now 18T. And you're kind of on the cusp of a CapEx cycle. So it seems to me that you can really start to sustainably drive better gross margins over time. Can you just talk about the changing dynamics of the business and sort of what that means for gross margin over time?

Yes. It is interesting to reflect on that history and think about so a 4T might have had 5 or 6 disks in it and an 18 has a lot more than that. So relatively speaking, the higher cloud demand is driving a lot more of the critical components, heads and disks, into our business. If you think about the drives 6, 7, 8 years ago, that was typically a 1-disk drive, maybe a 2-disk drive was your average capacity point. Now you're talking about the average drive having over 5 or 6 disks in it. So that means that what's going on in the fabs that are making recording heads or recording media are a lot more relevant into the build materials, a lot more relevant to the supply chain planning and so on. It's been an enormous shift from client server to the mass capacity in the factories around our supply chain, not just for ourselves but our supply chain, but it's also created a little bit different dynamic, like you described, in other businesses where you need long-term predictability based on what you're starting. The cycle times have increased as well, so there's a little bit of parallel and some of the memory businesses there. And all of that's translating into the same kind of predictability we've been discussing.

And do you think from a gross margin perspective, when you were laying out the benchmarks that you think that the company can achieve in the next, say, 3 to 5 years, is it relevant to look at some of these other industries and you say, "Well, look, we're still in the low 30s in terms of gross margin." Some of these other industries that have similar structural dynamics are in the mid-40s, some of them, and some of them are even higher than that. So is that a relevant, like when you kind of sit and you sort of dream about where gross margin can go for the company and you think about what could be a road map for that, is that a relevant comp where you'd think at some point, there's no reason to believe that gross margin could begin before?

Fundamentally, it's all about supply/demand matching to the point we just had. And so we've got more and more of the critical components inside each drive that are longer and longer lead time. And then if you get out there and you say, the mass capacity continues to grow at 35% CAGR, so on and so forth, these numbers get big over time. And there will be cyclicality in all of them, of course, but the way we look at it is we're going to go through a kind of plotting investment thesis. We're going to develop as much areal density capacity as we can, as quickly as we can. But still, if the supply is well balanced to demand, then we should -- or if demand is a little bit stronger, we should be able to grow. And we'll continue to look at that model as we make our investments and make sure that we're balanced as properly as we can to grow. That's why we -- in the last Analyst Day that we had last year, we revised the model a little bit, and we'll look at it over time and see if it's appropriate in the next one.

Got it. So the next question is on HDD cost road map. And there continues to be a perception that the HDD business is on a pretty steep decline. And you can hear there's terms either a melting ice cube or a rapidly melting ice cube. And maybe as I heard in another presentation, maybe it's a slowly melting iceberg. But A, I'm just sort of curious, if you look at the nearline applications, there's still an 8 to 10x price delta. I mean, it's huge. And even if you were to get the price parity, just the amount of CapEx that the NAND industry would have to put in the ground to displace all of those exabytes would just take such a long time. So you've got like 2 factors that make it seem like maybe this is a slowly melting iceberg. And so I'm wondering if you can talk about that, and then I guess maybe dovetail into what you think the sustained cost curve is for your business.

Yes. I think if I go back 5 or 6 years, and I understand why this was happening, there were a lot of people who had small businesses that were generating only a little bit of revenue and they'd say, "I'm going to put our drives out of business." It was just the math didn't make sense then. It doesn't make sense now, to your point. But I think this noise is still out there. I think as we've changed to mass capacity largely, I think the economics equation out there where these drives are being used, massive data centers and things like that, have kind of put that argument to rest. The idea that, exactly to your point, that you spend a lot of money to -- or invest a lot of money to then supplant that business over time makes no sense if the math breaks. From our perspective, we're trying to create value, so we do have an investment thesis. CapEx as a percentage of revenue in hard drives is not particularly huge. And so therefore, we're able to free cash flow a lot. We can reinvest in ourselves and our areal density curves. We can invest in CapEx again. We can return value to shareholders. I think we've been good stewards of our cash over the last few years as things have been tight. But right now, with the amount of data growth that we see and the S-curve that's coming in areal density, we think we're going to be sitting in good shape for the next 5 to 10 years. And there will be 50-terabyte drives out in the world. We said someday they'll have a great TCO proposition. I don't think there's going to be any technology that's really close to them at that time. I don't foresee why the architectures and data centers around mass capacity would change very much. I always register this point, that's not a negative statement on NAND. It's a great technology. It actually has many frontiers that it can open up as well where hard drives probably won't play. So therefore, I think the architectural balance around mass capacity versus what NAND is being used for more in memory side or mobility side or something like that is great for them as well. I think it's a win-win for everyone. There's not this overlap that we have to really fixate on and I don't.

To that end, I mean, is the NAND business something that you think about having an interest in? Is it something where you look at it and say, "Well, I would much rather be in the HDD business because it's a much more consolidated business and there has been structural elements to it?" Or maybe if NAND became more consolidated, is that what would sort of keep you from entering that business? I'm just kind of curious as to whether that's even a discussion point for you.

Yes. We've been very diligent lately last few years about how we use our brand. And we can move some NAND products, some SSD products in consumer, in enterprise and so on, with partners, largely. I think we've been very smart about how we deploy. To grow scale then, there's other problems with that and there's some scale in NAND. Of course, it's just absolutely massive if you have a fab. So we're probably not going to have a fab and we're probably not going to ever get to those scales. We'll just be smart with how we want to serve customers with our brand and with our technology portfolio. I don't really foresee that, that's going to change too much. We'll see. It could be a crazy world, but as I look forward to other opportunities at the edge and in the data centers and around other ways that we can service our existing customer base, there are other ways as well. So I'm not, say, overly fixated on NAND versus HDD.

Got it, perfect. Can we just talk about HAMR for just a moment? It doesn't make sense at 20T due to the higher cost per bit, but it certainly enables higher capacities that would make sense from a cost perspective probably in 30T, which you've talked about a lot. So can you just talk about the timing for ramping HAMR? And maybe how much of an impact HAMR will have on capacity growth and on your cost curve? Does it inflect the cost curve in a positive direction or does it keep it in the mid-teens roughly?

Well, I think ultimately, areal density is growing. Areal density is the way we get cost per terabyte down, so yes, it will ultimately affect it. There's -- any time you introduce any new technology or any new technology platform, there're start-up costs, if you will, and we'll work to get those as low as we possibly can. I have confidence that HAMR is not going to be a relatively expensive solution compared to other solutions we've used over time. I think the team is making great progress. I'm not going to do any product announcements today. But I'll just say every time that we speak externally, we get more and more bullish on the technology getting to those 30 terabytes, but not just at 30, 40, 50-terabyte capacity points. The technology is very, very difficult out there. But it's -- but in the labs, we see it happening and we're really excited about it. There are literally hundreds of people engaged. We've been talking about product development teams being engaged for quite some time. We've got to go dot the Is and cross the Ts like we always do. I think the customers are really interested in the products as well. And they'll get samples when -- of those big capacity points when we think we're ready for it.

But you would still say that HAMR is not going to be mainstream until 30T?

Sorry?

You would still say that HAMR -- don't think about HAMR as being the majority of your mix until you get out into the 30T?

I think we have good solutions today that are non-HAMR but we also have solutions that are HAMR today. So how we decide to ultimately deploy it, there will be high capacities at 30 and 40 that are inaccessible with a main CMR technology that we have today. But -- and we'll probably use HAMR back into 20s and 15s and things like that when we can because that's the way you get costs out of those capacity points as well by taking heads and disks out. But it's all about when is the technology ready? When are the teams done with their products? When are we confident in the ramps and things like that? Showing more confidence all the time. And so that's why we have product development teams engaged.

So can we just maybe talk about what is sort of a typical March sort of in terms of seasonality? How do you think about a normal baseline for the March quarter in terms of what a normal seasonal March would be? And is that the right starting point to sort of think about benchmarking the company into next year? I mean, yes, it's going to be strong but it's sort of a normal seasonal Q1, which I think is flat to down to be a smidge. Sort of how do you think about all that? Because you are doing very, very well in December. And so some people would argue what that creates this comp that's sort of hard.

I think there's naturally seasonality. Obviously, consumer market, you go back a year, COVID was changing a lot of the work-from-home, game-from-home, those kinds of dynamics so the year-over-year compares are really tough in that. And then if you go back 5 years, we were so much more exposed to PCs than we are now that seasonality was actually a really different equation. Right now, the mass capacity does have some seasonality that's in the VIA markets around government investments in smart cities and things like that, that usually time out at the end of December. There's a subtle other seasonality, but this is something we've been preparing for. And if you look at our entire year guide from 6 months ago and then what we said about getting to low double digits or something like that relative to this year, you can see that we strengthened a little bit in the back half of the fiscal year what our exposure was. And a lot of that was based on what we see going on in the cloud, right? So as we grow as part of mass capacity, then all those other seasonality impacts become de minimis over time. And so I think we're having -- starting to have better visibility not only in Q3 but also Q4 and what the cloud service providers are going to need globally. And we're also transitioning to our 20-terabyte, ramping that very aggressively through that period. So we're pretty happy about exabyte growth.

Great. So just in summary, so net, you consider kind of normal seasonal for March to be just down a smidge?

Yes, I won't quantify it here. We'll get through this quarter and then give a good guide. But there is some normal seasonality. And then I expect the exabyte growth that's coming out of the bulk of the VIA markets to just keep going through next year.

Got it. Can we talk about channel inventory? I think you did alluded to this before, but how would you characterize the channel? And I'm particularly curious about China because I hear -- we hear such conflicting things about China and how much inventory there is in China. Do you think that there is stockpiling happening in China?

No. Coming out -- I'll talk about last quarter now. Coming out of last quarter, it was particularly high. I mean, the entire supply chain has been so whipsawed, I think, over the last couple of years that there may be points where you say you've got 5 or 6 weeks or something like that, maybe that's getting a little towards the high end. But generally speaking, I think we did the right things coming out of the quarter. So I don't foresee -- I don't really see any particular channel inventory. If you talk about individual customers, they're all doing build-outs according to some plan. There are some subtle changes that happen in various customers, but other customers are completely on the plan that they were on 3 months ago. So I wouldn't say that there's any cause for concern there.

Do you think that given the potential for export restrictions that would extend beyond the 1 large customer that the U.S. government went after, do you think that the potential for export restriction, Have you seen signs where it's causing any demand to get moving forward where other customers say, "Well, we have better gear up again to use kind of a better word stockpile of drives, that if the same thing happened to other customers that at least we would have some drives on hand to actually pull from?"

And it's hard to speak very broadly, but I would say that against the backdrop globally of everyone worried about supply chains, there are some people who are making sure they have enough inventory for their critical build-outs. I don't think anyone really has the money to over-stockpile, if you will, for that kind of thing anywhere globally, frankly. I mean, you're managing supply chains just in time around the world or relatively speaking, just in time I think -- and that's how I think about it.

Got it. And then maybe I know that you guys think about what the longer-term domain is for HDD, but as you think about metaverse and it's really a very nebulous term these days and everyone's going to have their own metaverse. But clearly, it doesn't take a rocket scientist to see that there's a lot of compute that is going to be -- and storage necessary to bring this to life. And so the question is, have you thought about sort of what this means for your business and what it means for HDD demand over time?

Yes. Actually, a lot. I've thought about it a lot. I'm really excited by the trends. And I'm not going to pick favorites of which particular ecosystem is going to go big, maybe more than one. I think about Seagate for 20 years, we've been involved in the gaming business, whether it's console gaming or PC gaming. It's been a really strong business for us and it means a lot of data pretty close to the end user. And so I think to your point, if whatever verse you're talking about wants to grow, it's probably going to need a lot more data at people's fingertips and a lot closer to the end user for latency expectations and things like that. So we're quite excited about that. The gaming ecosystem actually around the content creators themselves is fantastic. And in some sense, they're already their own little universes that they create for people. I think the big tipping point for us, not only on compute network but on storage in particular, is when you start seeing applications that become business applications where people need the data close to them to make decisions because they're input/output with that data can be totally different. That kind of transformation is really exciting and I think will drive a lot of edge deployment, not just cloud deployment but edge deployment as well. So yes, we'd like to stay really close to it and provide the best cost proposition for lots of data at the edge so that people can use those applications to take advantage, make better business decisions. So fun, but also productivity gains would be great to see. And I can't wait to see whichever ecosystem develops those.

Got it. You've alluded to the common platform before in the 16T to 18T transition. But can you just talk about common platform and how much of a game changer that is when you're talking to customers? I mean, how big of an impediment was it in the discussions where you were making a platform transition and a customer may be balked at how much they would want and the timing on that? I mean, was that really like a big thing in the past such that common platform really takes that off the table?

You bet. I mean if I think about 2 terabytes, 3 terabytes, 4 terabytes, 6 terabytes, 8 terabytes, I could keep going, there were almost always a new drive. And not just a new set of mechanical parts either, you've got heads and media changes going on throughout that. We decided a few years ago now, maybe 5 years ago in the development ranks, to start to get our hands on this to be more predictable ourselves. And we came out with that common platform that was extensible. We could add one more disk to it, it's not a problem. We actually did 18 terabytes with the same heads of media that we might be using for 16s and 1 and 20s for the other. So we have that kind of leverage capability off this platform. And depending on the customer, some customers have deep histories with these product transitions. They know what an easy product transition is based on the technology. They know what a hard product transition is. And so they'll move faster or slower depending on what they see coming, and that definitely helps us.

So you think that this really greases the skids for the 20T transition?

Yes, yes. I think the 20T is a highly leveraged product of that platform and we'll continue to leverage. Yes, as a matter of fact, we've said many times before that, that platform is even HAMR-ready, right? So it's designed not to be a big lift to put the technology constituents in to be able to get to HAMR drives as well.

Got it. Can we talk about balance sheet? You bought back about $2 billion worth of stock in fiscal '21. You bought back more than $400 million last quarter. Quite solid support from the stock. The question is, what is the target cash or leverage ratio? And sort of how do you think more broadly about capital return?

Yes. We do think about liquidity, especially in these times. It's only been a year since the major swings that were associated with the pandemic at the front end, right, demand swings and supply swings and everything else, so you have to -- every good business has to watch their liquidity targets. Relative to shareholder return, we've been very, very consistent on this for a decade now about what our strategy is. And frankly speaking, we invest in ourselves. And then when we have excess cash, and we are generating quite a bit of cash, we'll give that back to shareholders if we feel like we don't need it. And there's cyclicality to that as well, as you've seen. But from my perspective, Seagate's a good investment. Whether we're investing in our R&D, whether we're investing in our CapEx, the datasphere is just continuing to grow. There's lots of opportunities for us as well. And I'm glad that everyone shares that optimism now. It was 5 years ago that people were saying, exactly to your points earlier, something different, but I think people see a different future right now. And so we're going to be aggressive against it.

And Dave, have you given any sense just back to your -- back to the discussion about common platform? Have you given any sense as to how much of your nearline shipments are on a common platform? Is it the majority of your nearline shipments now?

We have what I'll call a midrange nearline as well, so I think 10 terabytes and under. So usually, when we talk about common platform, we're talking about 16s, 18s. There might be some 14s in there and things like that, but that's 16, 18, 20 going up, we're talking about that drive family, yes. There is a mid-cap nearline platform as well, a different one. And it's going through its own product transitions over the next year or 2 as well. So we'll get cost out of it. And there are certain customers, I think this is what a lot of people don't understand. There are certain customers that just can't move from 8 terabytes up to 18 terabytes. They -- because of the nature of their architecture, how many drives they have and how responsive they need that architecture to be to whatever application they have, they just need to stay at 8 terabytes, and that's fine. We'll continue to service them there.

And how much of an issue has being able to pass on input cost increase has been? I mean, you've done pretty good -- pretty well on gross margin so it certainly seems that you've been able to pass on most of the input cost. But sort of how does the discussion go with your customers on as you're sort of working out the next 6 months and you say, "Hey, Mr. Customer, sorry, you're going to have to pay more because my input costs are going up".

I think Gianluca talked about this a little bit last week, in particular, with freight logistics costs around the world changing quite a bit. On a monthly basis, there are some surprises that can happen. I think these are things that we service the customers by being as predictable as we can. These are unpredictable costs inside of us. So we're going to have to go out there and to your point about these long-term agreements, when we start negotiating 6 months or 9 months, we're going to have to say, hey, these are the realities as we see it. This is what we need to be able to predict. And I think the customers have been pretty good with us there. It's not something that you react a week later after you see a change inside the system to them. So there is some time lag to it.

And just on the 20T transition, so you're beginning to ship 20T while you ship 20T PMR drives in the third quarter. So that conversion, when do you think we would reach some sort of crossover point?

The volume of 16s and 18s is pretty big so that crossover, it's not like when we did the 16s. But the 20T is going to be a very aggressive ramp, potentially even more aggressive than the 16s were at the time because customers now have figured out that they get that extra little benefit, whether it's 2 terabytes or 4 terabytes in their -- in perpetuity, they're going to be using these drives for 5, 6, 7 years in their data center. That's a huge TCO proposition for them. And they know the platform, to our earlier discussion, about common platform. They know the platform well. So they've qualified it. They know all the nuances of it. They can see the predictability of it. They're ready to go.

And just from a longer-term cost, I don't think we sort of talked fully about the cost curve. But is it reasonable to continue to expect to be able to cost down mid-teens? Or do you think in a world where HAMR becomes more mainstream, that you can do better than down 15 because it's just that much more scalable?

I don't really think about cost takedown that way. I think more about areal density and then we'll price for what the market needs out there at the time, keep our factories full, so it's more of a supply-demand balancing perspective. Obviously, on a cost per terabyte basis, getting that HAMR point out, getting 30 terabytes and 40 terabytes out affords us the opportunity to get to a better cost per terabyte proposition, but then our customers can use to go drive more exabyte demand up in the world.

Got it. And then just in terms of some of the input cost headwinds you're seeing. Can you just talk about like how long those will be persistent and when you can see those rolling off?

This is one where if I looked back a year ago, I probably would have gotten some of this wrong. And I think a lot of us who have tried to been seers in the market here relative to all things related to the pandemic. And make no mistake, in my mind, it still is related to the pandemic. I mean, we still have factories who are going through various kinds of shutdowns inside supply chains in the last -- even in this quarter. There's been all kinds of disruptions to people. So we got to get everybody back and healthy. I do think that there's some positive signs there next year -- or sorry, next quarter. Obviously, with the pauses that we see after Christmas and Chinese New Year and so on, there will be a little bit of recovery inside of supply chains, but we're all hoping that the pandemic doesn't shift into a different phase at the time. So I think relative to our supply picture, we expect it to be a little bit better and we've been saying that. We'll have to just react to whatever is going on in the world.

Got it. I'm getting some questions from investors so maybe I'll ask a few of those. So your peer is talking about that at 22T, that they simply add a platter to go to basically 10 platters in a very simple transition. You were already at 10 platters at 20T. So how -- so like looking beyond 20T, how do you go from 20 to 22?

I think this is where a lot of people are maybe drawing the wrong conclusions off of how many disks you have and how many heads you have. Look, it's all about what's running in your factory. Those negotiations that you've had with customers, what capacity points they want and how well you can yield things, I have no doubt that we can get to 22 terabytes. We'll have lots of different options to be able to get there. We'll be able to do it at good cost, good margins and without too much of a scrap build at the end because we're trying to push the component technology too hard. So I'm not really into this 9-disk tender. We started shipping a 9-disk HAMR drive -- 20-terabyte HAMR drive a year ago. It's not the one we took into volume. There may be other solutions that we take to volume. And actually, there already are 20-terabyte 9 disks that we take the volume, too. So we said we have multiple flavors of 20s. I think we're going to have multiple flavors of 22s as well. And so I think overly fixating on whether you have 1 disk or 1 disk more or something like that in 1 platform is probably not the right way to think about it. And just take it to a logical extreme. What if you had 30 disks and somebody has said, why, I have 31. It wouldn't matter. What matters is can you yield the thing? Do you have predictable customer relationships? Are you through their calls in time? That's all the stuff that really matters.

Great. And then I had another follow-up that whether you signing more of these LTAs, is that maybe a headwind to pricing in the near term and maybe some near-term margin expansion? So you're sort of trading some longer-term demand visibility for maybe a little bit less near-term pricing and margin leverage.

Yes. I see your point. I mean, definitely, what we want to do is run the business as predictably as we can. And so when we have confidence in one of these capacity points and our ability to yield it against the millions and millions of components that we need, exactly to the last question, then we'll go out there and have a good conversation with customers and say, "Do we want to book it?" And we won't wait until the end and say, did somebody else trip and fall and maybe we can raise prices at the last minute? I mean, that's -- to me, that's not the way to be predictable. So I think this is a long protracted grind rather than these opportunistic swings, exactly to your point. So yes, I think maybe we do leave a little bit on the table, but that's what we are trying to do with our customers, too, is give them a predictable model. They know they can count on our supply, they can count on the economics around it and they're good, too.

Got it. And then I just wanted to clarify some comments that were made at a competing conference last week. And the suggestion was that there are incremental logistics costs in December that sort of weren't anticipated in the gross margin. Can you just clarify that?

I think it's been a very dynamic quarter for everyone. Inside of our supply chain, inside of other supply chains, we can definitely see this with our customers. And so we're just mindful of the fact that moving parts around the world right now is really tough. Things get stuck in various places, and you have to break them free and get to the front of the line in order for customers to get exactly what you've committed to them. And that is a very dynamic situation, and that's all we're doing. It's just being really mindful of that. Some of the costs are exact -- not exactly what we had forecasted. We're doing our best to handle it. I think everyone is going through the same thing.

Got it. And then maybe the sort of -- this is part and parcel with a question I've asked before about the seasonal trends into March. But right now, you're more supply-constrained than you are demand-constrained in terms of what you ship. So does seasonality not really -- I mean, you already said that yes, you will see the resumption of some seasonal trends during the March and June quarters. But when does -- do you become demand-limited versus supply-limited? And when you do, do you worry about some inventory adjustment?

Not too much. I mean, so we -- to your point, we predict the seasonality, if you will, a quarter or 2 ahead of time. It's not like we get into the quarter and then realize we have that seasonality picture. So we're predicting it and doing the wafer starts that we need, the media starts that we need against that prediction. So as long as things are in balance relative to what we said, we can be a little bit flexible. That's where you can run out of parts if you haven't balanced the supply appropriately against that demand. I'm not too worried about that because mass capacity is -- they're so much off the common platform, I'll say it now, that we have a lot of flexibility even up until the last minute. But it will be more about the seasonality of VIA markets or the seasonality of consumer or something like that, where we had planned a certain thing, and if there was more pronounced seasonality than what we even forecast, then there will be issues. Again, long term, with mass capacity continuing to grow, the secular demand of data really being on high-capacity points, I think this is becoming less and less of an issue.

Got it. And then we have time for maybe 1 more question, which I -- I'm just going to ask the last 1 again on HAMR. So when the HAMR conversion does actually happen, are there any other issues to consider such as software ability? Do you -- do the customers have to rewrite any code, anything like that?

No. And that's the good news. I think a lot of people have been talking about that for a long time, but the way we're staging the technology is we want to make this as plug and play as possible. There are always -- I don't care if you're going from 10 to 12 terabytes or 12 to 14 terabytes, there are always little nuances that some customers have to deal with to make sure the performance doesn't change too much or the responsiveness of the drive doesn't change the entire software stack, if you will. But -- and obviously, going from 20 to 30s or something if the customer is making that transition, that might be a relatively bigger jump. But inside of the drive, how it works, whether it plugs in and operates, it's the same old disk drive. That's the way we want it. I will say that optimistically, there's also -- it's not -- our industry is not just about HAMR. It's not just about areal density. That is a key driver. But the responsiveness of the drive, say, for example, we've come up with dual-actuator drives because people can't just put more and more capacity behind one actually. They need that drive to be -- that capacity that they put online to be more relatively responsive. There will be more of a trend towards dual actuator over time. We're really happy with how that product is going now and we'll continue to refresh that product. There's all kinds of other features that make the thing more valuable in a data center. We're really excited about some of those features: better reliability, better long-term stickiness inside the data center, better use cases that change over time. So there's a lot of optimism out in how we're dealing with this -- with our customers. There are tech -- so just to be clear, there are technology transitions going on associated with features over time, but it's not completely gated by HAMR at all. It's opportunistically being developed with our customers.

Got it. Well, we've run out of time. But Dave, thank you so much for your time.

Yes, Tim, take care and good holidays to everyone. I appreciate your time as well.

Thank you.