I'm Krish Sankar with TD Cowen, and the next company presenting is Seagate. We're fortunate enough to have Gianluca Romano, the CFO. Gianluca, thank you for your time.
Really appreciate it. Last week you guys had a very good Analyst Investor Day here in New York. Just wanted to quickly, like, is there any couple of things you want to highlight from the Analyst Day, that key takeaways that investors should remember?
Yeah, I would say, you know, the main messages that we wanted to give to our investors and also to the analysts is this industry is growing very well. The demand is strong, especially in the nearline space. We are developing a new technology called HAMR. We are already, you know, qualified three main customers on HAMR product. We are quickly moving to, you know, second generation of product and qualifying more customers in the next, you know, several months. The trend out of, you know, the prior down cycle is continuing. The industry is very consolidated. The focus, at least on our side, is on continuing to improve not only our revenue but also our profitability and continuing to focus on what is the most important things in this industry, that is technology and technology transition.
We think if we execute well on, on those items, you know, we will have very good results in the next several years.
Gotcha. And, you know, on that, you know, one of the questions that keeps coming up is obviously sustainability of the cycle. If you go back, I think, this cycle, or like the past cycle, troughed around September of 2023 and has been, like, growing pretty nicely since then. But if you compare it to prior nearline cycles, you know, five, six quarters into it, the cycle starts to wobble. Kind of curious, how do you think about this cycle? And is there a way to compare and contrast versus prior cycles? Because people just worry about the sustainability of this cycle.
Yes, of course, every cycle is a bit different. This is a technology industry, so, you know, we expect from time to time to have different cycles. There are differences in the industry in general. Now, if you look, for example, how we model the expectation for exabyte growth and revenue growth, you can see they are much closer right now compared to what they were in the past. If you look at the exabyte trend, even in the last 10 years, really, exabyte grew for every year except for the down cycle we had in calendar 2022, 2023.
Mm-hmm.
Really, exabyte always grew. The revenue was a little bit different, a little bit more bumpy, but it was not a matter of exabyte. It was a matter of how you manage that exabyte growth with your pricing. Now, as I said before, the industry is more consolidated. Demand is above supply. There is not a lot of CapEx going into this industry. I think the industry is very disciplined on how CapEx is managed and how units are added. I would say the increase in exabyte volume is coming from technology. They're not coming from more units. All this, I think, will help in several aspects. One is to extend the cycle. The industry is not taking all the demand that is available today. Some of the demand will continue to shift out in future, keeping the cycle going.
Of course, you know, this has also an implication on how many cycles eventually we will have, you know, how frequent are those cycles. I think it's a great improvement, not only for us but for the industry in general. Of course, we try to leverage as much as we can on our new technology, so try to generate more exabyte from the units that we have and try to grow with the demand. It's not easy because, as you know, demand is growing very rapidly. I think, you know, if we execute well our transition to the future product based on the HAMR technology, we can be close to that growth and keep benefiting from the right level of supply-demand balance.
Mm-hmm.
From the growth generally that will help our revenue.
Gotcha. You know, I think one other thing that's kind of interesting in this cycle, I guess, is maybe on top of, like, you know, your hyperscalers adding capacity, there's probably an AI angle to it too.
There is some view that incremental AI side is helping, the longevity of the cycle. I also heard the other side of the story, which is that once hyperscalers rationalize their storage needs, then it's going to come down pretty sharply, and that might be a negative.
Yeah, I would say there are different aspects from AI, and there are different benefits to this industry from AI. I would say the first benefit is already present today, is already part of this up cycle, that is retention period. People, companies, government are not deleting data anymore.
Mm-hmm.
They keep all the possible data because they want to use AI, and they know that AI will give you a better result if you can access a bigger pool of data. This is already part of, you know, today's growth. What is still not part of the growth, I think, is the vast volume of data that will be generated by AI itself. Not by humans anymore, but by AI application. The transition from data in the form of text to data in the form of video, that, of course, consumes a much higher level of exabyte. I think, you know, will happen in the near future. It's still a bit difficult to estimate how much it will be. We have our estimate of the exabyte growth for the next two or three years, but AI could be a big variable.
There could be another upside to that, to that forecast.
Gotcha. I just had a couple of more, like, cycle-related questions. Another one is when you look at where your cloud, customers' hard drive inventories are today versus, say, three or six months ago, where do you think they are, you know, compared to one or two quarters ago?
Yeah. In this industry, inventory is always a bit complicated because, you know, hard disks usually are installed in a data center. The difficult part is to understand how much they are utilized in the data center. You know, if the data center has a low utilization rate, of course, that could be considered like inventory. It's actually storage capacity that is available.
Mm-hmm.
For the future. It's not easy, in this industry, to completely understand inventory. That is not only what, you know, customers eventually have in the warehouse, but it's usually a very small number of drives. What we know is demand is strong.
Mm-hmm.
When we have a little bit of extra volume, customers want to buy that extra volume at a higher price. So we do not have any indication that inventory is growing.
Mm-hmm.
The bill-to-orders are actually growing in volume. In January, when, as the earnings release in January, when we discussed a little bit about the calendar 2025, we said based on the bill-to-order, we actually see revenue increasing sequentially through the calendar year, and that will help us with even a better profitability. This has not changed.
Mm-hmm.
So far, we don't have any evidence that customers are reducing their demand or they are increasing their inventory. I think there is still a lot of appetite for storage.
Got it. I mean, in a hypothetical scenario, let's just say, you know, cloud demand does decline next year, 10% or 20%, how easy is it for you to scale down OPEX and any how do you think about the COGS implication?
I would say the bill-to-order should help us to have an early visibility on the eventual cycle, you know, because customers have to place a bill-to-order. If they have too much inventory or for any reason they need less data storage, they should start to put that lower volume into the future bill-to-order. As I said, until now, we have not seen it. If we have visibility on what they want, you know, three, four quarters from now out in time, we can, of course, prepare our internal structure to support that volume, much better than what happened in the past when, you know, we were a little bit surprised by the change in orders. You know, one quarter was very high, and the next quarter suddenly went very low.
Mm-hmm.
We had a lot of WIP, so a lot of inventory inside our manufacturing that, you know, you need to finish, and then you need to sell to someone that actually does not really want that volume. That has a big impact on pricing. This is also why bill-to-order is important to us, is, you know, if there are cycles, the ability to manage that cycle in advance will help us to have the right structure, have the right internal inventory, and go through the cycle with, you know, a much lower impact than what we had in the past.
Gotcha. You know, Doug, I mean, when you look at the nearline cycle or the nearline demand, I do not know, maybe 30-40% of that is probably enterprise versus the majority being hyperscalers. You know, the bull argument is that, you know, the enterprise spending cycle has really not started, so there is a long runway to that down the road. The bear argument is eventually enterprise demand for hard drives could probably go away. Enterprise might be more SSD, and mostly hard drive would be done for the hyperscalers. I am kind of curious, like, where you stack between those two. You know, is there a risk that enterprise demand for hard drives goes away?
Yeah, we do not see that happening. Of course, we have bill-to-order also with enterprise OEM, so we do not see that happening. Honestly, in general, for us, it is not so important who is the customer. You know, we produce the same hard disk drive independently from who is the customer. So if data is stored into on-prem data center, so with enterprise OEM, or it is stored in the public cloud, no real difference.
Mm-hmm.
Right now, we do not see this change in the enterprise OEM. I think AI will have some impact on that. There is maybe a theory that is saying more sensitive data will be on-prem.
Mm-hmm.
Favoring the, you know, the increase in the enterprise OEM business. Another theory that is saying, you know, public cloud because there's, you know, a higher level of compute and more sophisticated cloud actually will finally attract also that part of the business. But again, for us, there's not really any difference, no.
Mm-hmm.
For us, the important is what is the exabyte growth year after year, and wherever we have to sell it, we will sell it.
Gotcha. Then maybe on HAMR, you know, I think last week at the analyst day, you mentioned that three of the top eight CSPs have qualified HAMR. It also felt like you have confidence that the remaining five would probably qualify by, I don't know, maybe by next year. How should we think about HAMR going mainstream? You know, do you think once all eight of them qualify, or is there another metric like a million units a quarter? Like, how to think about HAMR and when it goes mainstream?
Yes. In terms of call, of course, we announced the two new customer qualifications. So we have now three in the cloud space. Dave said we will qualify all the big cloud customers in the next 12 months.
Mm-hmm.
We also gave an indication of how HAMR volume is going to grow. Our fiscal year is starting next quarter. We said by the end of next fiscal year, four quarters from now, we will have 40% of the exabytes sold in nearline sold through HAMR product. It is a fairly good part of the initial ramp. We also said in the following four quarters, we will go from 40% to 70%.
Mm-hmm.
Let's say between now and, you know, the next eight quarters, you will see a fairly important ramp of what we are going to sell, in terms of HAMR products.
Got it. I think Gianluca, like, I asked this question in the past too, you know, it's kind of more to do with, like, failure rates for HAMR. How to think about it vis-à-vis PMR, right? You know, obviously, your customers look at PMR failure rate, and then maybe HAMR has to be similar to that. Is there a way to quantify where HAMR failure rate is today versus relative to PMR? Along the same path, you know, as you go mainstream, you know, is there any, have you taken any accounting for reserves or things like that in case the failure rate on the field ends up being higher than expected?
Yeah, we don't expect any higher failure rate. I would say this is why we do qualification.
Mm-hmm.
You qualified. If you have a failure rate that is higher than what it should be, customers are not qualifying the drive. We had an extended call with the first customer and a very quick call with the two following customers.
Mm-hmm.
We think we have a very good configuration also for the cloud, and this has been qualified by, you know, three big customers. We do not expect any difference between PMR or HAMR. It is not a matter of the technology inside the drive.
Got it. And you do not feel like you do not have to worry about having any reserves set aside or any such thing?
No. I mean, for every product that we sell, there is not a reserve, but there is no real difference between the two.
Got it. Got it. And, you know, the other one that is kind of interesting is that, you know, I understand you said in the past HAMR is gross margin accretive. You are vertically integrated, so in theory, you should have better margin profile. Yet, when you compare it to WD, Western Digital, you're still, like, a few hundred basis points below. Now, I understand, obviously, you've been ramping HAMR, which probably has a negative impact on yield. Is that the main reason and now that, you know, HAMR goes mainstream, or are there also any other things you had to worry about, nuances that, you know, early call of a customer for HAMR, maybe some sweetheart deals were given or things like that?
Obviously, I don't have all the details about the gross margin of our competitor. They're obviously doing a very good job on their part. I would say we were very focused on HAMR, and HAMR took longer than what we were thinking in terms of going through the fourth call. We lost a little bit of time. If we had HAMR a year ago, you know, probably we were already at a different level of gross margin also. We lost a little bit of time. We probably lost a little bit also focus on the last PMR product. The 24TB CMR and 28TB SMR, we were clearly a little bit late to the market. We have those products now, and we are transitioning to HAMR.
As we said as an analyst day, you know, our profitability is going to increase and, fairly rapidly.
Gotcha. Maybe I'll just pause to see if anyone in the audience had any questions. If not, I'll just continue. I think one of the—oh, go ahead. You have a question. I think one of the things you said last week was, you do expect to reach 40% gross margin in a few quarters, and then, the incremental 50% gross margin you hit, like, $2.6 billion in revenues. In that assumption of that incremental gross margin 50%, is there any HAMR mix we had to assume, or is it just a more straight-up thing that, you know, just think the drop-through is going to happen when you start getting to those revenue levels?
No, that is HAMR, for sure. HAMR is one of the contributor factors of the continual increase in gross margin. Of course, the volume of HAMR is the same that we communicated in term of the ramp.
Mm-hmm.
You know, 40% of the exabytes sold in the next four quarters and then continue to grow from there. Yeah, HAMR is one of HAMR in terms of transition to continue to transition to higher capacity.
Mm-hmm.
That was not different than what we did in the past. Of course, HAMR is giving us the opportunity to continue to do it and to do maybe even faster than what happened before.
Gotcha. You know, if you go back a couple of quarters ago when you looked at your exabyte shipment versus your competitors, it kind of felt like the share was more like 60-40 kind of a thing. You were probably more.
Mm-hmm.
High 30s, low 40% market share versus them being much higher. Is it fair to assume that probably goes in your favor with HAMR, or do you think that given that you have large customers like the hyperscalers, they will try to manage a rationale duopoly versus giving you a lion's share of the market, even though WD is technically maybe two years or even longer behind you?
We don't know how customer will allocate their exabyte.
Mm-hmm.
I would say we are not increasing our units of what we produce, that is, as a media. But those as a media will generate a much higher level of exabyte because of the transition to, you know, higher capacity product. We will have an increase in the exabyte sold. I do not know how, you know, how competition will do and what they will do in the same time. If they have the same growth, I guess we will have a similar market share in terms of exabyte. If they grow less, they will have probably a little bit less market share. It is not really important, no? Demand is covering supply from, I think, the entire industry at this point.
Mm-hmm.
Customer will have to buy product from us and product from competition, and they will manage as, you know, as is, their need in the specific time frame that they are assessing.
Gotcha. You know, last week, I thought Dave kind of made an interesting point. I think he said for every one exabyte deployed.
Mm-hmm.
It's roughly 42,000 units of PMR or 25,000 units of HAMR. As your mix shift starts going towards HAMR, it feels like you probably don't need any additional CapEx given what your install capacity is for a pretty long time. Is that a fair assumption, or do you think at some point you need to think about adding more capacity?
No, as I said before, we right now, we are not planning to really produce more units. We think that moving our manufacturing from lower capacity drive to higher HAMR capacity drives will give us enough exabyte growth.
Mm-hmm.
To stay very close to demand. You know, the demand is growing, you know, at a certain CAGR that we indicated at Analyst Day. We think through technology transition, we can increase our exabyte fairly close to that number.
Mm-hmm.
CapEx that we indicated into that range between 4% and 6% of revenue is, of course, to replace equipment that are becoming old and, and to help with the technology transition, not really to increase the number of units that we will produce.
Gotcha. And, if you do, you know, as you scale up HAMR, is there any other, because I remember during the down cycle, both you and your competitor brought on the capacity, but it looks like there's also some personnel layoffs that were done in regards to personnel and things like that. Would that need to scale up, or do you think you don't envision that happening, i.e., you don't need to hire more folks to scale up manufacturing?
I would say in terms of units, we are fairly full at this point.
Mm-hmm.
We rehired a certain number of people in manufacturing, in the last maybe seven, eight quarters. Now, I do not think we will have to hire many more at this point. You know, the factories are fairly full, so we are good. In OpEx, the reduction was mainly coming from going from, you know, developing two technology to one technology. More than a year ago, we, you know, announced that we were stopping the development of PMR products, and we just focus on HAMR.
Mm-hmm.
At that point, we had two teams, one working on PMR, one working on HAMR. We moved some people, of course, from PMR into the HAMR team, but we also reduced.
Mm-hmm.
The overall, you know, level of R&D, and we also did, you know, a few reductions in SG&A. At this point, we think we have the right level of resources. I think we are fairly good.
Gotcha. You know, when I look at your manufacturing footprint, if I remember right, I think you have Minnesota, Ireland, Thailand, and Singapore. You know, there's more like a tariff-related question because I know, I think the discs and heads are probably done in Singapore. Different, they're spread out.
Yeah.
How easy or fungible is it manufacturing if you had to shift it around in case, you know, tariff things get out of hand?
I would say we have several manufacturing sites, but they mainly do different things.
Mm-hmm.
So we have two sites producing heads. One is in the U.S., one is in Northern Ireland. We have one site for the media that is in Singapore, and we have one site for substrate that is actually in Malaysia. We have two sites for assembly and final test, one in Thailand that is a big volume, and one that is smaller that is in Wuxi, China.
Mm-hmm.
That we use for what we sell in China. So it's, it's not that we can really move different parts of manufacturing into the, you know, the sites that we already have. That is probably not, not happening. With the right time, we can move some of the manufacturing in different locations if our locations, for some reason, get penalized in a disproportional way on the tariff.
Gotcha. On China, how is the VR market in China? You know, if I remember, I think several years ago, the prior, the peak used to be close to, I think, like a billion-dollar run rate on a quarterly basis or something from China.
Mm-hmm.
Do you think you get back to those levels, or how do you think how is that market evolving today?
Is starting to recover. I would say if I look back a few quarters, September, December, they were fairly good quarters. March is seasonally low.
Mm-hmm.
As you know, those companies work on projects, and the budget for those projects is usually approved in the fourth quarter of the calendar year and then get deployed through the rest of the year. We see maybe an improvement already in this quarter compared to March. Usually, the strong quarters for video and image application are September and December. We, you know, we will probably have some upside from VR in the near future.
Is the way to think about that business, is it more a function of China macro? Like, you know, as China macro gets better, the VR demand in China is going to get better too, or is there some other nuance for that business?
I would say China macro is very important. I would say also, you know, worldwide macroeconomy.
Mm-hmm.
You know, many projects that are developed and deployed by those customers are not in China.
Mm-hmm.
They are actually in other parts of the world. So depend also, you know, how many projects are approved in different cities, in, you know, different companies that use VR for quality control or other manufacturing reason. In general, it tends to trend more with how the economy is doing.
Mm-hmm.
particularly in China, but not only.
Gotcha. And, you know, I think a couple of months ago, you guys announced or did the acquisition of Intevac, which is very interesting, though I should say not completely surprising because you bought, like, 10-12 years ago, XyraTex on the testing side. So there is a history of vertical integration. If I remember correctly, I think Intevac does the iron-platinum coatings via granular HAMR heads. How is that integration going? And I think your friends at WD also use that. So how do you think about their, you know, what's going to happen there?
It's a fairly easy integration. It's a fairly small company.
Mm-hmm.
You know, there are not many people, so it's much easier to integrate. The main location is Singapore, where we are already present. So not, not a difficult acquisition. Seagate was more than 90% of the revenue.
Mm-hmm.
Of Intevac. It was making sense to us to integrate that team with the rest of our manufacturing, and it's progressing fairly well. There are other suppliers that can do what Intevac is doing, so I'm sure competition will have other opportunities.
Gotcha. I mean, because it seems like, you know, the other suppliers over there in Japan or whatever it is, do not seem to have the same—they have not invested as much in these iron-platinum thin film compared to Intevac, so.
I don't know. I cannot comment.
Gotcha.
I'm not expert enough in that. What I know is, you know, there are competitors, and I think they have a very similar product.
Gotcha. Maybe one other thing. I think, last week, you kind of mentioned that you're going to start doing buybacks over the next few quarters. In the past, the goal was always like, you know, you want to bring down debt to $5 billion. Is that still the goal, or do you think that, you know, come to the point where it might actually start doing the buybacks?
No, that's still the goal, and we are very close.
Mm-hmm.
We did a treasury transaction this quarter. We are fairly close to that goal. We have also taken the time in the last three, four quarters to rebuild a little bit our working capital that during the down cycle was a little bit stretched. You know, we think in a few quarters, we will generate, you know, very good free cash flow, and we can possibly restart on share buyback.
Got it. All right. I think we did it. I think we're out of time. Thank you very much, Gianluca .
Thank you.
Appreciate it. Thank you.
Thank you.