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

Perfect. Why don't we go ahead and get started. We'll stay on schedule here. So I'm Aaron Rakers with Wells Fargo IT hardware and semiconductor analyst. And pleased to have with us for a quick discussion Gianluca Romano, Executive Vice President and Chief Financial Officer of Seagate. So first of all, Gianluca, thank you so much for joining us this morning.

Maybe before we go into my list of questions, I want to -- yesterday, there was an 8-K announcement out. I'd like to -- I've gotten a handful of questions from some investors. Can you level set us what the message is? I think David had made a comment about some trepidation maybe in terms of dynamics into the next quarter. But just level set us on the key points of that announcement?

Yes. Thank you, Aaron, for inviting us here. And before we start, I will be making forward-looking statements today, and you can learn more about the risk associated to those statements on our website. Yes. So yesterday, we issued the 8-K with 2 news. One is a very good news and refers to our qualification of the HAMR product in the cloud segment. It took a little bit longer than what we wanted, but we now have a very, very strong configuration of the product that can be used in the cloud. And we, of course, are going to qualify more and more customers between now and, let's say, mid of calendar '25, so over the next couple of quarters and then ramping a good volume of HAMR. So today, we are the only company in the world with this technology. We are the only company in the world shipping today 30 terabyte drives. So we are very pleased with this achievement. You may remember, we actually qualified HAMR in different segments in prior quarters. We started with video and image application and then we qualified customers in the on-prem data center segment. And finally, we had the right configuration also for the cloud. So a major achievement for the industry and, of course, for Seagate in particular. At the same time, we also talk a little bit about a manufacturing issue we had recently that is impacting our volume production in the March quarter, not in the current quarter, not in the future quarters. But for the March quarter, we have a lower WIP coming out, so it's reducing our opportunity to have a certain level of volume that we were previously expecting. So those are the 2 news that we discussed yesterday.

Right. So you -- just to maybe double click on that a little bit. So the $200 million impact, are you guiding March quarter down? Or is $200 million, that's a transitory issue. There's no follow-on effect into the June quarter. Just...

Yes. There is no follow impact in future quarters. We are not -- no, we have not guided March quarter. I would say, if you start from the current quarter, we have guided $2.3 billion. Usually, when you go into the March quarter, you have 2 different trends. One is on the near line that is growing. As we now discussed in the past, we continue to see that growth. You also have a low seasonality for video and image application and the legacy business. So now you can estimate how much is the growth in near line and how much is lower seasonality in video and image application and the legacy is 2 trends tend to offset each other, you will then have $200 million of impact from this lower production.

Yes. Okay. And again, what that reflects is basically bringing equipment back online. So you've done a very good job over these last several quarters of implementing a build-to-order dynamic in the business model that's very different than what we would have talked about a couple of years ago. But basically not having the ability to supply is more of a discussion about the upside opportunity -- being able to capitalize on upside and just to characterize that?

Yes. Yesterday, Dave said, no, we have enough volume to support what we have already committed with the build-to-order. Unfortunately, we will not be able to get the normal upside that we get in the quarter. So there is a fairly substantial volume that is part of the build-to-order. But every quarter, we had a little bit of an upside, but actually normally at higher price. So specifically into the March quarter, we will not be able to support that upside.

So on the flip side of this is that the actual maybe positive underlying factor is that it's more of a supply side discussion. It's not a demand discussion at all. It's just being able to fulfill you'll be tight. The industry is tight.

It's not demand. I guess some of the demand will shift to further quarters out in the future. But again, demand is continuing to grow. This is why we have a little bit of this problem, like if you look at near line space, September quarter a year ago, we were shipping 55 exabyte in a quarter. Last September, we shipped 110. So we doubled. It's not so easy to put back that capacity in place. And we are chasing volume every quarter. This quarter, that segment will be even higher. Our guidance in revenue is higher and all the upside is basically coming from near line. So more volume this quarter compared to September. So we are still chasing volume. And so we have anticipated all our WIP and accelerated as much as possible to go to finished good and support demand. But now putting back in line equipment has taken a little bit longer than what we are anticipating. So we have this little bit lower WIP that will impact than the March quarter.

That's -- so it's read -- its heads and wafer stuff. Okay. So moving past that, and maybe somewhat tied to it, but this characterization of demand durability, visibility in terms of near line, persistent questions always like, well, how is this different, right? Like do you really have good line of sight in terms of these drives that are obviously recovering from a demand perspective over the last couple of quarters are getting deployed. And are we not concerned at all that there could be an inventory build that you have another correction in front of us? Just talk about that.

So we have this build-to-order methodology. And we have orders for, let's say, in the near line space, I would say, covering all the calendar year '25. So we have good visibility and that is a base volume. As I said before, there's always a certain part of the volume that get added during the current quarter. So it's upside during the quarter. So I think the industry will have a cycle in the future. We see the technology industry as any technology industry, you will have cycles. But it's very important to us to have visibility and able to anticipate a little bit of the cycle. So with the build-to-order, we should have that visibility and a certain point, we should see build-to-order start to decline in terms of volume, so decrease the volume. And we can anticipate in our manufacturing, how to reduce our cost, keep a very high level of profitability and then go through the down cycle, but hopefully will be not as deep as it was in the past, because we can anticipate so we don't need to move all the WIP we already have to our customers and pushing that extra inventory again to them. We can anticipate the cycle and then so reduce the impact of the down cycle and then enjoy again the up cycle. So will not be a trend up forever, but data is growing forever. So the long-term growth is there forever. It's just that will not be linear. We'll have to go through cycles.

So to be -- just to be clear, so you've got very good demand visibility through calendar '25 based on your build-to-order. You're not necessarily saying we won't go through some cycles at some point in time. It's just how you manage through that...

Absolutely.

Has structurally changed. How much of your near line business has split between cloud versus maybe traditional system enterprise OEMs?

That has changed in the last maybe 7, 8 quarters. So before -- even before, I would say, before going into the down cycle, the 2 segments were fairly similar. So we were saying it's about 50-50. Not every quarter is the same. But in general, they were fairly similar. Right now, because cloud recovered faster than enterprise OEM, we're probably around 70-30. We don't know exactly how this will play out in the near future. But in general, when you go longer, we expect cloud to continue to grow faster than any other segment in the storage business.

Okay. So back on the supply side of the equation, maybe first, I'll just put this out there. Remind us again of the lead times of a near line drive. I think people may be underappreciated how complex these drives actually are as far as production cycle. And basically, what I'm going to ask you is like, with 109 exabytes of capacity shift in this most recent quarter on, I believe, exabytes, you're getting back to peak -- I mean, so how are you -- remind us again how much you can supply today, and I think there's always this push and pull around bringing capacity on. Is that a good thing or a bad thing perception-wise, from an investor perspective. So how do you think about that supply-demand balance?

Yes. I'll say, bringing capacity back. So meaning equipment that we already have is probably good, because it's fairly cheap capacity. So you can have a very good financial return from those equipments that you are putting back. You already spent the money in the past. After that, we probably don't want to put more capacity. So we want to use what we have, but we don't want to go above that level. So we think we can go back to the level of exabyte we had before the down cycle, so around maybe 150, between 150 and 160 max. That is the right level, with the current mix. Ow producing this -- because what we produce is ads and media. So with the current mix, we can produce ads and media for a total of 150, 160 exabytes. And then what is the mix. So when we transition more and more from the current PMR technology to the HAMR technology that we have just qualified with the same number of ads and media, we can generate more exabyte. So our manufacturing footprint can remain exactly the same, but we can start to generate more exabytes. So it's just a matter of mix, not a matter of adding capacity, more equipment, greenfield, more people, we don't need that. We just need to move more and more of our mix to HAMR. So in terms of near line, I think our peak was around 120 exabytes. Last quarter, we were not very far from that number. This quarter, as we guided higher, and we said this increase in revenue is coming from the near line, we will be, of course, higher in volume. So I would say we are already at the level of that peak. But the industry has changed. As you know, this industry is always moving out from legacy and into the near line space. So we continue to transition the same 150, 160 exabyte out from the legacy more into the near line space.

That's perfect. And so the qualification on the HAMR drive just to go down that path a little bit more. So the 30Ts, the 3 terabyte per platter drives qualified. I know that we entered this year, and I think everybody appreciates the qual took longer than what we thought. I think you entered the year with kind of the view that you'd ship 1 million HAMR drives first half of this year. Do I think about 1 million drives first half of '25? Is that fair? Or is it now that we've got the qual behind us? Is that the volume ramp we should think?

I would say, we have, of course, a certain level of inventory of HAMR because we have produced not during this time. We have produced different configurations. So we have some volume in bid configuration for the cloud, but of course, we are starting already to ship this quarter and that we will ship much more next quarter. We also have different configurations that we didn't qualify for the cloud, but we qualified for other segments. So we will move into the video and image application, on-prem data center. Those customers are consuming a much lower volume. So that part of the inventory will take a little bit of time to go out, but it will go out over the next several quarters. For the cloud, we need to produce more because we need to produce this configuration. And so we will sequentially increase the volume in the March quarter, in the June quarter and after that. Most important for us is to qualify more customers because at a certain point, we will not have enough volume of HAMR for all the cloud demand. But if it's only one customer, no, we are limited to bad customer demand. So we will sequentially increase demand and around mid of calendar '25, when we will have a certain number of cloud customers qualified, we should see a good step up in rent because now we have a point, we will have several customers, so we can ramp much more and they will, of course, be very interested in buying highest capacity drive.

So with this qual now announced, it was consistent with what you said on the last earnings call, which is the volume is midyear calendar '25. So not to put you on spot, but 1 million drives, let's not throw out a number at this point. It would....

Yes, we don't give numbers for several reasons, even for competitive reasons. But as I said before, we are shipping today 3-terabyte per disk, 30-terabyte drive, only company in the world doing this. And next quarter, the volume will be much higher. June quarter will be higher. And after that, we will have a step-up in volume, perfect step up, depending on how many customers will be qualified. And how much volume we can produce of that configuration that is good for the cloud.

Yes. Again, congrats on getting the lead guy qualified. Now all of a sudden, you're going to have the perpetual question of like what about cloud number 2, 3, 4, et cetera? What's the path to those guys? Is the learning curve now behind you? And these other guys happen in short order? Is it the next 2 quarters where maybe cloud #2 comes on? Any thoughts around that?

Oh, more than one. No, we have -- we said in the past, we started qualification with the majority of the cloud customers in U.S. and China already in the September quarter. So this is why I'm saying between the September quarter when we started and probably mid of kind of '25, I expect a good number of customers to be qualified. And therefore, we can ramp much higher volume of HAMR.

So I'm a financial analyst, a good number is something more than 2.

Absolutely.

Okay. Fair enough. The 3 TB per platter drives, as we think about the scaling of HAMR, we think about not just 3T because we don't stop there, we go to 4-terabytes per platter. Where -- how do we -- how soon do we think about 4? And do you always stick with 10 platter stacks? Or does Seagate have a road map that has 11 platter stack?

No. So we don't need to go to 11 or 12 was is an additional complication and also in terms of yield and manufacturing. So we think that 10 is a good number for us at least with HAMR because we can continue to increase the density of each disk. And so you will have after this 30 terabyte drive, so 3 terabyte per disk. The new product, 4 terabyte per disk will be in about a year from now. So this time next year, we will have the product available. But we always talk about 3-plus terabyte per disk because between now and the 4 terabyte, you will see an evolution of the current product. So today it's 30 terabyte. Every quarter, you will have some units that are higher capacity. So we are continuously increasing the capacity per disk. So when we are ready to go to 4 terabyte per disk will not be a jump from 3 to 4, it will be maybe at that point 3.6 to 4. So we -- and you will have a mix of what we sell. Now we will have some units at 3 terabytes, some units at 3.2 terabytes, some units at 3.4 terabytes depending how much we can produce. But it's a continuous evolution of the improvement of the aerial density per disk.

And that's a function of leveraging shingled magnetic recording, or SMR, or is that different?

No, that's different. That's a choice from the customer. So we will increase the aerial density per disk, so the capacity per this in both versions, CMR and SMR. HAMR has no different than the prior technology. You can sell an hard disk in the CMR version or SMR is the customer preference. If you want to run many different applications with a certain performance, you probably prefer the CMR. If you go on certain kind of application, you care more about the size of the capacity and less about the performance, you go with SMR. But the product is exactly the same. You basically just run the software at the end of the process to go from CMR to SMR. Now we have sold millions of SMR drives in our history is -- I mean it's something bad now is nothing new actually. SMR is 10 years old, is a choice. So whatever the customer prefers, we will change the drive from CMR to SMR if that's what they want.

So now that we're talking about HAMR and a volume ramp, the question that comes up almost every call. I'm sure you guys prep a ton for this as HAMR's accretion to gross margin. And so is it accretive at 3T per platter? Is it -- do we wait to maybe more of an inflection at whatever 3.6. How do we think about the guide path of that accretion in gross margin that HAMR could bring?

It is accretive right now, and it will be more and more impactful when we go up in capacity because the bill of material will not change. So the 3 terabyte per disk, 30 terabytes per unit as the same below material of a 4 terabyte per disk 40 terabytes per unit. So of course, there is a huge cost per terabyte reduction going from 30% to 40%. So if you're already -- if the cost per terabyte is already better compared to PMR today at 30 terabytes or 3 terabytes per disk, just imagine how much better will be at 4 and then 5. So of course, we'll be much better in the future. But already right now is accretive to gross margin. It has to be up in volume. So we need to have a certain number of drives sold to see that in the P&L. Now if you just sell 50,000 drives doesn't really make a change out of the 12 million drive that we sell or 12 million or 13 million that we sell. So it has to go -- we need to give the time to ramp the volume and sell it, but it is positive to gross margin.

So back to that point of midyear '25, kind of that starts to be the inflection point on the accretion?

I think so. I mean I think every quarter, you will see some impact. But of course, the higher they are as the volume, the bigger is the impact.

Okay. So I've been covering the hard drive industry for a long time, and I can I continue to tell a lot of people like I've never seen structural price increases. We've seen transitory price increases, I guess, one time in my career covering the space. But that -- how are you thinking about price increases and resetting, obviously, the structural gross margin level, not just Seagate, but it feels like the industry.

I think the industry is focusing on the right things, that is improved profitability. It's not volume. It's not market share, it's working on producing the best product at the lowest cost and pricing it very well in order to receive a fair profitability. A year ago, I think on September last year, our gross margin was 20%. It's not fair for this kind of product. And I think the industry moved in a different direction, but we stop pushing on moving volume and start requiring a certain return from what we produce. And in 4 or 5 quarters, we increased our gross margin from 20 to 33. And the hard disk is actually higher than that, it's the mid-30s. So it's a good trend. It's still not where we think it should be. And this strategy is working well. That means we -- every quarter or every time we renegotiate build-to-order, we increased a little bit our price the like-for-like pricing for the same product, is going up. And this is giving us and the industry in general, a much better profitability. But we are still not even close to where we think we should be. And so we continue this strategy. Demand is strong. Supply-demand balance is very good. So we continue. And I don't think we are in any way disruptive to our customers because the industry strategy was to increase pricing not in a very aggressive way. Now it's just a few percentage points every time you renegotiate a contract or every quarter when you can. It's not disruptive where relatively small part of the CapEx of our customers. So this increase is -- I don't think is disruptive to them.

Just so I can kind of think about my model that 2%, who knows what 1.5%, 2%, 3%, whatever number is that you just referenced. But that's on a per unit basis, not a per capacity basis. So the per capacity will come down.

The mix? No, when you mix up in capacity, in general, you have better gross margin, lower cost and a little bit lower price. So -- but it's a good thing for the industry to scale up in capacity per unit because gross margin tend to improve.

And in that envelope of price up, gradual, it sounds like -- I mean, correct me if I'm wrong, it sounds like there's durability of that to continue to play out over the course of a handful of quarters still.

Absolutely. The cost per terabyte is declining. And it will decline even more when we move more production to her. And as I said before, the cost per bit is -- cost per terabyte is going the direction we like, that is when we renegotiate the contract, the like-for-like pricing is actually a little bit better.

Well, it feels to me like if you can execute on the cost down curve consistent, right? The price of the delta between the 2 is really the incremental margin. That's a fair model exercise?

I think it's a good one.

Okay. So I'm going to shift gears real quick. We talked a lot about gross margin. We can go try and think about where that ultimately goes. OpEx level, you have some variable comp that's coming back into the model. But remind me again where we should think about that playing out?

Yes. I think OpEx, first of all, you need to look at resources, so headcount. I think we have the right level of headcount. We decreased our head count a lot during the down cycle, mainly because we were able to, at that point, focus only on HAMR technology. In the past, we had basically 2 groups working on the PMR technology and HAMR technology. We don't have new product on PMR technology anymore. So we just focus on HAMR. So we have reduced our head count in R&D and focus only on HAMR. So I think we are at the right level of headcount. In the past, we were as high as maybe $400 million a quarter in OpEx. We went down to about $250 million during the down cycle. That $250 million, of course, was not including any variable compensation because during the down cycle, of course, we were not giving any variable compensation to employees. Right now, we have variable compensation, and we are at about $285 million. So I think this is a good level for our OpEx. When you go out in the future, I would say the headcount will probably remain substantially the same. Of course, we need to consider the annual increase or the inflation impact, but not increasing resources. So I would say more or less this level should...

That $285 million a quarter. Okay. So we can all kind of think about what that means on an op margin EBITDA basis as well. At that level, with 3%, 4%, 5% CapEx to revenue, company is going to generate a good amount of free cash flow. Remind us again on -- it sounds like take the debt a little bit down. But to me, if I paint the scenario out, it looks like you're going to be in a pretty good position to kind of reengage on a capital return strategy.

Yes. We have not changed at all our shareholder return strategy. We have always focused on dividend first and then share buyback. And for the dividend, we have increased our dividend during the up cycle and then protected our dividend during the down cycle. So we actually just raised the dividend after the October board, and we protected our dividend during the last couple of years of the down cycle. We also want to reduce our debt, as you say, we are now at $5.7 billion of debt. I would like to go to something around 5%. We have a maturity in January for about $500 million that we will repay. And then we will address the other about $200 million later in the next few quarters. After that, with the free cash flow available, we will start again on the share buyback. So not this fiscal year, but I think sometimes maybe during the next fiscal year.

Next fiscal year after June. So when we think about that, when you think about what you just laid out, what's the right mindset of your thought of like how much cash you need to run the company? Like what's the right level of operating cash or liquidity, if you will?

I'll say, in terms of liquidity, we said in the past about $2 billion of liquidity, I think, is a good level for us. Of course, this includes the revolver, cash and the revolver.

Yes. Okay. With all of those dynamics in mind, I wanted to just ask you kind of on the news that we saw last night, not from you guys, but there was a system vendor, we've talked a little bit about it earlier that announced a hyperscale win in flash. How do you think about the long tail of hard drive versus flash encroachment I think I tend to agree that there's clearly a growth dynamic for both in the enterprise market. But how do you think about that?

I agree with you. I think there is a great opportunity for both technology to continue to grow. They are used differently in the cloud. So the storage is on hard disk. And when you need to run the applications, so the compute part, analytics part, the AI part, you move the data from the hard disk into the flash. And in a different environment where you have the DRAM, the HBM, the CPU, the GPU, you run your application. When you have created something new with that application, if you like it and you want to save it, you send it back to hard disk. So in particular for AI, there are 2 major benefits for hard disk in particular. The first one is you need the more data you can access, the better results you have using AI. So the retention period is becoming longer before people and companies have some strategies in terms of the retention period for data that is becoming longer. That means more data stay on hard disk. And second, people and companies and government are just starting now to use AI. So when AI will start to produce a lot of new data for them and that new data is valuable, they will want to store it. So back to our hard disk. So this is why AI is very positive to us. So there are 2 different impacts, but are both very helpful to us.

So the final thing I was just going to ask you is, it's been a while since you guys had an Analyst Day, right? I think it's February '21. You guys set long-term margin targets at that point, correct me, 31%, 33%, I think. And you made some changes, right? You accelerated some -- or extended some depreciation and stuff like that. But is the thought that it's time to kind of reset, recalibrate that long-term model? Is there an Analyst Day that...

We will probably have a new model at the next Analyst Day. I would say right now, we don't really have a target. I think we have an opportunity to continue to improve our profitability. We think it's fair for us to receive a certain level of return from the huge investment that we have done in the past, and we are continuing to do it. Now this product is very complex. It's very high technology and is very much needed in the data storage and in the cloud space. So we are running our strategy, reducing our cost, try to get the best price possible in the market, always being a good partner to our customers. So always not being disruptive. And this is working fairly well. So no, we are continuing to go in that direction. But at the next Analyst Day, probably we will have a full modeling, not only in terms of gross margin, but also in revenue growth expectation, operating margin, and of course, finally, EPS, which is the most important.

Okay. Gianluca, thank you so much for joining us. Appreciate it. Thank you.

Thank you very much. Thank you.