All right. Good morning, everyone, but early afternoon. Welcome to Citi's 2024 Global TMT Conference. My name is Asiya Merchant. I'm part of Citi's technology, hardware, and tech supply chain coverage. Very pleased to have Gianluca Romano from Seagate here. He's the CFO there. This session is obviously only for Citi clients. We have a set of questions that we're gonna go through. Before that, I'd like to turn it over, see if there's some opening remarks from Seagate's management here, and we'll jump right into it. We also have members of the IR team here. Shanye Hudson right here in the audience as well. Thank you.
Thank you for inviting us, and before we start, let me remind everyone that we'll be making forward-looking statements today, and you can learn more about the risks associated with those statements on our website.
All right. Welcome. Thank you again. We've been asking all our companies, you know, how are they thinking about end demand here. You guys are sitting in calendar third quarter, you know, after the holidays. Maybe you can just remind us, you know, how end demand's looking for you guys for calendar 3 Q.
Yeah, of course, now we are in a, in a good part of the upcycle, especially for the cloud segment. I would say we are, we are pleased with, with how the current quarter is shaping up. We guided the revenue sequentially up more than 10% at this point in the quarter. Now, we are, we are fairly comfortable with this guidance, and maybe it will be better. More importantly for us is profitability. We guided the quarter, sequentially now way higher, from $1.05 to $1.40 as a midpoint. I think we can be in the high part of the range, so, better than the midpoint that we guided. This is coming mainly from the mix, moving more into the 24-TB CMR, 28-TB SMR, and pricing.
Pricing is also a little bit better than what we were expecting. On the revenue side, the sequential improvement is coming from two major segments. Now, one is cloud, as I said before, is continuing to improve every quarter since September last year, and finally, for the first time, I'll say, after the downcycle in the current quarter, we see enterprise OEM segment starting to improve, so I would say the enterprise OEM and then the Video and Image Applications, possibly in the next few quarters, will probably complete this upcycle. Right now, in the last four or five quarters, the upcycle has been driven mainly by cloud, so a good upcycle, but still limited to one market segment. Now, we're starting to see other market segments too starting to improve and join the upcycle.
Just as you talk about the key demand drivers here across your various end markets, you know, is it just normalizing post the downtick, or are there actual incremental demand drivers that you're seeing that takes you beyond just, you know, mean reversion to normal trend line?
Yeah, it's always difficult to define what is normal. I would say in the storage business, there is no normal. No data is going to grow-
Right
... every year, sequentially every year. Data storage will continue to grow. This is driven by many applications. Of course, AI is one of those application, but is not the only one. There are many reason why data will continue to grow, many applications that will drive, growth in the creation of data and then in the storage of data. So I would not say that, what the level we had a couple of years ago was normal, was just part of a cycle. As we understand now, some of that volume went to, into our customer inventory, and, this is something that, of course, we need to try to limit in order to reduce the impact of the downcycle. And to do this, we need to be very disciplined with, capacity.
Our focus is on profitability in long-term demand, is not in taking no short-term volume upside. And capacity, of course, will be always kept very well under control.
Mm-hmm.
I think if the industry moves to continue to move in this direction, even if you will always have some cycles in the industry, because in technology, you always have some upcycle and some downcycle, but you can limit the impact of the downcycle.
Mm-hmm. Okay, and maybe just, you know, you talked about AI, Gianluca. Do you see AI as an incremental demand driver for HDDs in particular?
It will be. Now, I would say right now, it's still fairly limited. I would say this first part of the upcycle is more driven by what we call traditional demand. But no, AI is very important. I would say, I would see AI in two phases. Now, the first phase is the build-up of the infrastructure to use AI. And to do that, you need to create an infrastructure with, now, a lot of GPUs and HBM and DRAM and NAND.
Mm-hmm.
So to run the compute analytics part of AI. This is important because you need to create data with AI, so this enables the creation of data. Storage, the benefit to storage come from two things. One is AI needs a lot of data to give you a valuable result, and therefore, the retention period of data is now longer. So people, businesses, government, will not delete data because they want to keep the data to use AI.
Mm-hmm.
So this is the first benefit. The second benefit is when AI start to access the data and generate valuable data to you, as an individual or as a business, you want to save that data, and that data is saved mainly in the cloud or on-prem data center, and is saved on hard disk. So this second benefit is just starting now.
Mm-hmm.
So you will see much more of that benefit in the next, you know, several quarters and years. We think, you know, AI is not a two or three quarters event. Will be with us for many, many years.
Mm-hmm.
You will have new applications. Now, every couple of years, you have a new, very important application that will drive the creation of more data, and humanity, you know, is now used to have more data in, you know, in everyone's life. This is very important because finally, all the data end up in a hard disk.
Yes. Okay. Your other competitor earlier today commented about, you know, how they're so pleased about these structural changes that are happening in the HDD industry, and you guys have also talking about build to order-
Mm-hmm
... for your cloud customers, which are one of your largest customer base. Maybe if you can just talk about this, you know, visibility, what's changed, and why your cloud customers are open to sharing, may perhaps increased visibility into their demand?
Oh, for sure, there is a change in the industry. In my opinion, this change happened already before the down cycle. If you look the trend of the HDD industry in calendar 2021, and the first part of calendar 2022, was actually fairly similar to what is happening today. Or the gross margin, at least for Seagate, now in maybe four quarters, moved from mid-20% into 32%-33%. So a great improvement. The industry was. And I think now, even for competitors, was fairly similar. I think the industry moved the focus from volume market share into profitability. Now, in this industry, you can have little bit better market share when you have a little bit better product, in term mainly on capacity product, but after a quarter or two, things get realigned.
So really, fighting on market share was not the best solution, I think, for the industry, after the big consolidations that happened now in the prior years. Right now, the focus is on profitability and trying to keep those cycles as smooth as possible. So now discipline on capacity, focus on profitability, focus on getting the best possible product to market and time to market. But as I said, I think that happened already before the down cycle. Then during the down cycle, that was a bit surprising in terms of timing and how deep it was. Of course, now the industry had some inventory already produced, so we had to move that inventory. That had a little bit of an impact on the pricing situation.
But as I said before, already a year ago, even if demand at that point was still weak, but the inventory on our side was clean. We're starting to increase pricing. No, of course, we didn't produce any drive that didn't have an order, and so we moved to this build to order, so customers, let's say, if they want to get units from us, they need to give us an order well in advance, and we produce what is ordered. We don't produce speculating what they will need for the quarter, and then try to move the volume at no, but that, of course, has a negative impact to pricing, so this is not happening anymore. Even during the, you know, deep part of the down cycle, we were not doing it, and we will not do in the future, so it's a change.
Customers, I hope now understand how this industry wants to operate.
Mm-hmm.
I think it's fair. Now, if you have a demand, you need to give that demand on time. The cycle time to produce a hard disk is very long.
Mm-hmm.
So it's difficult for us to, you know, anticipate the cash flow and produce a product if you're not sure you're going to sell that product at a, you know, at a price that has been agreed. So I say this is a big change in the industry, and now hopefully, we can continue to go in this direction.
Okay. And if you can just remind investors where you are with your visibility right now, with your cloud customers?
Yeah, I think as the earnings release, we said, we have already, you know, a certain number of quarters in, in certain markets that, no, we don't have more volume anymore. So-
Yeah
... we can create more volume through mix change. Now, with the same units, if you move your mix to higher capacity drive, now you can have more exabyte.
Mm-hmm.
Finally, what we sell is exabyte. We don't really sell units, we sell exabytes.
Right.
So mix is what will continue to bring some increase in exabyte quarter after quarter. Now, the number of units right now, I would say, is at a, probably already fairly well optimized.
Yeah. Okay. Given, you know, one of the questions you get is, given that there is stretched lead times, if you may, and you do have visibility, how do you make sure that your cloud, that as it relates to your cloud customers, they're not over-ordering because they realize, you know, it takes, I don't know, nine months or whatever, to produce an HDD part? So how do you make sure these cloud customers are not over-ordering-
Mm.
from you or your peers?
Well, this is very difficult, right? It's always try to understand if demand is demand for a real short-term need, or is a demand for inventory to make some, you know, safety stock. Especially in hard disk is even more complicated because when there is a new data center, and all the hard disk are installed in the new data center, at day one, utilization rate is zero, because there is no storage for that building. Now, inside that location, there is no storage yet. It's a preparation for future storage.
So there is always, you know, a certain number of hard disk that really have a very low utilization, and you need to wait with the time that location will start to increase the, the storage level and, and get to a certain level when they need to refresh the location in term of getting more capacity in the same location or build a new data center. But there is always part of the hard disk that have low utilization rate, so it's not easy to keep, you know, this, let's say, inventory under control. I would say the best way is not to increase capacity until nobody is a very strong visibility or a very long-term change in demand. And I would say the industry is able to increase exabyte capacity through the product transition.
Now, we have seen how through the history of this industry, how the industry moved from 1-TB hard disk drive to today, 24-TB hard disk drive. So there is always that opportunity, but, not increasing the number of heads or the number of disks or what we produce, we think is probably the best way to keep this, possible additional inventory under control.
Okay. Maybe switching to, you know, I'll get to your product and technology roadmap, and I know there's lots of questions around that. But before we do, maybe just on the enterprise side, you said, you know, you seem to see some green shoots there and demand there. Just if you can remind investors where you are with those enterprise OEMs, you know, what, what's driving the demand there? Is it again, normalization, or do you see incremental demand, from other applications there?
At this point, I think is more traditional demand. No, there is an increase in server demand. I would say AI will be also a part of on-prem data center demand increase in the future, but right now it's still fairly low. Now, it will be a great opportunity for, you know, the next several quarters. But right now it's more traditional demand. Enterprise OEM, during the down cycle, I would say they reduce their volume less than cloud-
Mm-hmm.
But they took longer to start to increase again. So it's just a different shape of the cycle. But now in the current quarter, we start to see the increase, and of course, our expectation is that that increase will continue. And I said before, even other part of the mass capacity business will start to increase. Now, this is why we are fairly confident in the next few quarters. We discussed about December during our earnings release, and I would say despite the September quarter is coming out a little bit better than what we were expecting, especially in profitability, we still believe December will be another sequential improvement, both in revenue and EPS. So I think we are going in the right direction.
Okay. When it comes to switching back to cloud customers, I think there is some concern, just given, you know, when we look at what you guys are producing in terms of revenues, EBs, compared to, you know, your competitor out there. There is always this question I get from investors about Seagate's market share.
Mm.
You know, what gives you confidence that you're not losing marquee applications or marquee workloads, and not losing share to your, you know, in this very important market for you?
As I said before, we don't really focus on market share. We think market share is driven by, you know, the product that you have in the market. As you know, we were very focused on, you know, the first part of the HAMR call during our March and June quarter, so we were maybe a little bit late with our 28-TB SMR and the 24-TB CMR, so we were not first to market with those two important products, so for a couple of quarters, possibly we lost some market share in the nearline space. We qualified a lot of customers during the June quarter. We are selling the product this quarter. We will ramp even more volume in the December quarter and in the following quarter.
So my opinion was very temporary, and again, and then when we move to HAMR, I think we will have the opposite situation, where we will have the better product in the market with the highest capacity, and that will maybe bring some more volume to us. But again, it's not our focus. No, I think the industry has a certain level of capacity, and I think that capacity will be sold.
Okay. HAMR, since you brought it up already, remind us where you are with your HAMR qualifications with your lead customer. We'll start with that from first. Yeah.
So HAMR, we qualified the first customer in March, as you may remember. That was a customer in the Video and Image Applications market segment. Just yesterday, we complete another call in the enterprise OEM segment, so on-prem data center. I think this is a very important step to move up in another market segment. And as you know, we are still working with our cloud partner to get a call into the cloud segment. Of course, the cloud segment put more, especially in the call environment, put a lot of stress on the product. So the call that takes a little bit longer is a bit more difficult. But no, we are making good progress. Every week, now we have a review.
I would say there are still some few adjustments, fine-tunings to the product, but the product is progressing well in terms of qual. I think the timing could extend into the next quarter for the cloud space. But I think it's very important that we have achieved this other qual in the on-prem data center, because that is also a segment that you now will consume a lot of units, especially when AI demand will be increasing the volume into the on-prem data center in the next several quarters.
Okay. And then, outside of this key cloud customer, maybe if you can talk about, you know, what's the reception been to qualifying at these other cloud customers?
Yes
... outside of this one customer?
Absolutely.
Yeah.
Our confidence in the product is very, very high. We already discussed at our earnings release, we are starting multiple cloud customer calls in the current quarter. So we have decided, because now this first cloud qualification is taking a little bit longer than what we were expecting, but our confidence is very high. So we are starting the calls with almost all the cloud customers in U.S. and outside U.S., now mainly, our Chinese cloud customers. It will probably take about three quarters to get the qualification complete, and then we will ramp volume. So as we said at our earnings release, we expect volume to start significantly increase around mid-calendar 2025.
Mm-hmm.
That is when you will start to see, now, an impact to our P&L. Until that time, we will have HAMR volume that now has accretive gross margin that is positive to the company, but is still limited in volume. So you need to have a certain number of customers and a certain volume to really see the change into the P&L.
Mm-hmm. And, outside of these cloud customers, you talked a little bit about the video applications as well as the enterprise OEM for HAMR.
Yes.
If you can talk about, you know, qualifications, with a broader set of group outside of-
Yes
... cloud. Yeah.
Yeah. What we will do in the future will be to expand the volume of HAMR into different market segments. As a beginning, because the volume now will not be enough for all the segments, the focus will be mainly on cloud and some of the enterprise OEM customers. But in general, now, we already announced our 4-TB per disk, so at high capacity will be a 40-TB through the end of calendar 2025. At that point, you can start producing lower capacity drive, like 20-TB, 24-TB drive, with only five or six disks and 10 or 12 heads. That is a major change in the bill of material compared to the current 20-TB or 24-TB.
You can really extract good profitability, entering into those kind of market segments with products with lower bill of material, and so lower cost per terabyte, and generating good profit.
Mm-hmm.
As I said before, of course, we need to have enough volume to satisfy the cloud demand, and also moving into those other segments.
Yeah. Just talking about cost per terabyte, maybe, you can remind investors, like, you know, where are we excluding HAMR? How does HAMR influence that cost of terabyte decline?
Yeah. The cost per terabyte now in the industry has always declined.
Right
... fairly strongly, mainly because the mix was moving from Legacy product, 1- to 4 -TB per drive, into at that time was 16, 18, and now is 24, 28 -TB. So a big change in capacity. But to increase capacity, we also had to increase bill of material. So a 2- TB per drive has one disk and two heads. A 24 -TB has 10 disks and 20 heads. So there is more cost, but because the capacity was, of course, increasing a lot, you always show some decline in cost per terabyte. HAMR will be even better, 'cause HAMR will not need to increase the bill of material to grow in capacity. All the capacity growth is coming from areal density. So a 30 -TB HAMR drive has 10 disks and 20 heads.
So very similar to what we have for a 20-TB PMR or 24-TB PMR. When we move to the 40-TB HAMR, it's still based on 10 disks and 20 heads. So the increase of the capacity is all coming from higher density in each disk. So instead of 3- TB per disk, you will have a 4- TB per disk, and then you will have a 5- TB per disk. And now our R&D is already working on much higher capacity. So this is where the cost per terabyte decline will be even better than in the past.
Mm-hmm.
-because it's, it's basically the same unit cost-
Mm
or same Bill of Material, but much higher capacity.
Okay. The initial HAMRs, which are more in the 30- TB, those do have higher costs, though, relative to the PMR?
I would say they have the same number of disks, the same number of heads.
Right.
Of course, there is an increased cost of the laser.
Right.
Each has one additional laser. I would say the substrate, the glass substrate, is still glass. We started to use glass from our 16-TB drive.
Mm-hmm
... so many years ago. But this substrate is a different glass, so it's a bit more expensive.
Mm-hmm.
But again, you move from 20 -TB to 30- TB, so there is a cost decline. We always said we think our HAMR drive, even the first generation of HAMR drive, will be accretive to gross margin. And of course, when you then move from 30 to the mid-30s to the 40 -TB to the 50 -TB, that will continue to show, now, a lower cost per terabyte. Now, of course, you, the other part is always what is the price?
Yes.
So assuming we are still in this kind of environment where now there is a certain level of price, of course, we will have a much higher profitability from those drives.
Okay. All right, I'm going to just open it up to investors here. If you do have a question, please raise your hand so we can bring the mic. Any questions from the audience? Okay, we have one here. Do you mind bringing the mic, please?
Can you give more color on incorporating HAMR in your production? Like, how long does it take to upgrade a production line, and how many months in advance before you start executing HAMR capacity expansion?
Yeah. For manufacturing now, the tools that we are using for PMR are generally the same that we will using for HAMR. So there is a lot of compatibility between the two technology. Of course, there are some tools that are only for HAMR. But in general, I would say, no, we don't need to have a specific line for HAMR and a specific line for PMR. No, in many cases, it's the same line where we produce the two different technologies. So not a lot of complications from a capacity standpoint. You don't need to separate the two lines in term of, you know, how much volume you need to produce on HAMR and, you know, bring all the tools for that capacity, and then do a different estimate on PMR.
Now, you can estimate the total, and then you have more or less the same tools to produce a full volume that you need. We have capacity. As you know, we actually reduced capacity during the down cycle. As I said before, we can increase the exabyte capacity, not number of heads and number of media, but exabyte, just moving the mix to higher capacity drives. And this is what we are doing every quarter. Now, right now, it's more on the PMR side, now moving from 18- and 20-TB into now 24- and 28-TB. And now in the near future, we'll be now increasing the volume of HAMR, as a beginning with a 30-TB and then moving to the 40-TB. But I would say manufacturing, now, we don't see a lot of problems in defining one capacity versus the other capacity.
In fact, right now, we are able to move now the mix between HAMR and PMR as we need during the quarter.
Mm-hmm. Okay, just, you know, when you look at mass capacity, HDD exabyte growth, I think people-
Yeah
... industry generally kind of pegs it at 25% CAGR. How do you think about the ability to act for the suppliers, yourself and your peers, to meet that demand, given you already talked about mix, obviously, HAMR comes in there as well. In other words, are we at a point where we, industry needs to raise capacity, or there is enough from that mix shift that can still meet the demand?
I would say no, the mix shift will take a big part of that growth.
Mm-hmm.
... I also think the 25% CAGR is probably not including the full impact of AI-
Correct
which we will see in the next, you know, quarters and years. So I think in the mass capacity, especially in the nearline space, could be higher, and, now, if that really happens, we will, now we will see what is, what is the real volume that is needed. But we are not, no, we are not looking at, increasing capacity in term of units, in term of heads, in term of disk. We focus a lot on optimizing our production and bringing to the market higher capacity drives.
Right. So is the market, you know, if it's undersupplied, is there a point in your forecast, whether it's in the next twelve months, where you see the market being more balanced and not tight? And then within that context, how should we think about gross margins?
Right now, we don't really see a change. No, and in fact, we don't want to modify our current strategy.
Mm-hmm.
that is, quarter after quarter, or whenever you renegotiate a long-term contract with a customer, is to increase a little bit our pricing. I would say our strategy has been very consistent. Now, every time we discuss a new contract or, you know, every quarter, depending from which segment we are working on, now we increase our pricing, a fairly low percentage, but very consistent. So we don't have a strategy that is very disruptive, I think, to our customers. It's, I think, very predictable, and as I said before, it's not a huge change from one quarter to the next. So they can, I think, predict what is happening and put that in their plan, in their forecast. We are very big customers. I think that level of price increase is not at all disruptive to them.
It's very different from what other components in the cloud have done. But the benefit for us is, I think you can keep this for much longer.
Mm-hmm. Okay, great. Any other questions from the audience here? Oh, we have one more here.
Just as you guys progress through the qualification of HAMR with the big cloud customer, right, you've said that the technology is fairly interchangeable with some of the PMR drives. So just not getting specific, if that gets delayed, are they just you know, is it pretty easy just to replace it with other products, or do some of those market share dynamics you spoke about earlier come into play a little bit?
Yeah, absolutely. Now, as you have seen, even in the last quarter or the current quarter, now we have achieved even going to be better of what we guided, just because we can switch from HAMR to PMR. And, now, as I said before, we have, you know, some customers will get qualified. Now, it's an enterprise OEM space, so there is some volume that will go, HAMR volume that will go into that space. The cloud, if it takes a little bit longer, we will sell more PMR product. We are just ramping our 24- TB CMR, 28- TB SMR version, so we can ramp that more, and until now, the qual is there, and we can start moving more HAMR.
But I said, no, I said before, I think December will be another improvement in terms of revenue and also in terms of profitability, in terms of now, EPS growth.
Okay. SSD cannibalization.
Mm-hmm.
You know, that's something that investors talk about a lot, and I'm piqued. My interest is piqued because you talked about enterprise OEM showing interest in the HAMR drives. And so we hear a lot about QLC NAND and SSDs more actually on the enterprise side as well. So just how you guys think about, you know, Q SSD cannibalization across your various end markets?
Yes, I would say in the cloud is not existent.
Correct.
Now, the cloud and even the big on-prem data center are structured in a way where storage is on hard disk, but you need a lot of NAND also to do the compute and analytic part of the application. So you move data from hard disk to NAND, you perform the part of the application that now you want to perform, and then when you save the result of that, goes back to hard disk. So there is a transfer of data, but all the storage, 90% of the storage, is done on hard disk. Now, you can see that even during the down cycle, when the NAND price was very, very low, it didn't change the structure of the cloud. It didn't change the structure of the on-prem data center.
When the NAND price went much higher, because the increase, the quarter increase of NAND were way higher than what we did in the hard disk industry, still, the architecture of the cloud did not change. Those are two components of a big infrastructure. When you go to low-capacity drive, it's different. Even in the enterprise space, now, if you look at mission-critical, those are drives of one, 2- TB, and then the client, so desktop, laptop, some of the consumer. Low-capacity drives, those are moving more into the NAND space.
Mm-hmm.
... and I think now this has been evident for the last many, many years. Even if in the last few quarters, our legacy business, where we now include mission-critical and client and consumer, has been fairly stable. I expect in the future to, you know, see some reduction and now that volume will move into NAND. But today, 90% of our exabyte are sold into mass capacity.
Yeah.
So the legacy is only 10% of the volume, so it's less material to the Seagate business. The gross margin is actually lower than what we get in mass capacity. We are, no, we are still serving these markets mainly because of free cash flow. Even if the gross margin is lower, there is basically no CapEx, there is basically no OpEx, so free cash flow is fairly good.
Mm-hmm.
So we are happy to serve that volume until it's there, but it's becoming less and less material to us.
Right. And again, is that main volume pretty fungible to the higher capacity drives?
Yeah.
Okay.
No, we have done that for many years.
Okay.
If you now in probably ten years ago, Legacy was 80% of the volume.
Mm-hmm.
Now, it's 10%. And now, the Mass Capacity was very small and continued to grow.
Okay.
So of course, we have transitioned that capacity from legacy into high-capacity drives. I would say this is also a reason why the industry is different now. Now, through that transition, we always had oversupply available because the legacy business, that was a big business, was declining.
Mm-hmm.
We had that oversupply always available. Now that 90% of the exabytes are produced and then sold into Mass Capacity segment, you don't have that oversupply.
Oversupply situation.
So we start from a very different point now. The industry is not sharing volume because supply is available. Now, supply is actually all consumed, and so it's a very different situation.
Okay.
And there's a lot of consolidation also-
Yes
... that happened in the industry.
Yes. Maybe just as we wrap up here, we have one minute to go, Gianluca. Just a little bit about your capital allocation. You guys are now, you know, in a envious position, generating free cash flow. How do you guys think about your balance sheet, and just return to shareholders?
Okay.
Yeah.
Yes. Well, first of all, even during the, you know, the deep part of the down cycle, we have always protected our dividends.
Mm-hmm.
So that is a priority for the company, you know, a big part of our shareholder return strategy. In the past, we have also done a lot of share buyback.
Mm-hmm.
I would say in the short term, we will be more focused on reducing our debt. We enter into the down cycle with about $6.2 billion of debt. We are now at $5.7 billion, so we already reduced a good part of the debt, but we want to go lower. I think I discussed as the earnings release, I want to go to about $5 billion, and after that, now we will, we'll probably restart the share buyback as our normal practice of maximizing shareholder return.
Okay. All right, and then, you know, as we wrap it up... Oh, we are actually out of time already. Look at that. All right.
Thank you.
Thank you very much, Gianluca.
Thank you very much.
Appreciate it.
Thank you.
I lose track-