First day here again this year in San Francisco. I'm Wamsi Mohan. I cover IT hardware and supply chain for the bank. Delighted to welcome Seagate again to our conference again this year. We have EVP CFO, Gianluca Romano, who most of you are already familiar with, if not all of you. Thank you so much, Gianluca, for taking the time to do this.
Thank you very much, Wamsi. It's always nice to be here with you.
Fantastic. Yeah, likewise. You just hosted a pretty bullish analyst day, and I would say, you know, looking at some of your targets, you know, they were materially above the last time you had spoken about any of these numbers. Maybe for context, it'd be helpful to hear from you what has changed in the industry from your perspective that is enabling this kind of financial profile.
I would say in general, the change is a better alignment between exabyte growth and then revenue growth, something that we have not seen in the past. In the past, we have always seen a good increase in exabyte. We had a few slides indicating the past growth. Every year, exabytes were growing, but we did not have the same increase in revenue. The main reason was an oversupply situation in the industry, moving from a peak level of client business into a very important cloud business, but starting from a very small exabyte number. It took a certain number of years. Of course, when you have oversupply, all the benefit that you get from your technology transition, generating more terabytes per unit, is actually mainly moving to customers. It is not really staying with the company.
A few years ago, we had this crossover between the cloud business and the client, but client was still big, was still 50% of our business. Right now, data center is above 80% of the exabyte, and we do not have oversupply. The industry does not have oversupply. I would say demand is strong. Demand is above supply at this point in time, and this allows us to better align what is exabyte growth and revenue growth for the future.
No, that's great. As you've noted, right, a significant part of your bids are now coming from Nearline, and you noted sort of this demand dynamic that continues to remain very strong. A lot of investors we speak to sort of talk about how long this has persisted. Maybe can you give us some context on, you know, how many quarters you've seen this? Is that even the right way to look at this? Because you're also coming out of an unprecedented downturn that we've had. Maybe just some context to help people think through the sustainability of demand.
I would say the business started to improve maybe nine quarters ago. It has been a long improvement, a long cycle. The main reason is now demand in the data center continues to grow. As I said before, because demand is above supply, the industry is pushing some of this demand out into the future. This is keeping the cycle going longer. We have started the build order now two years ago. This is giving us very good visibility for now maybe three, four quarters. The build orders are strong. We said already in January, based on what we already have in our orders, we see an increase in revenue, and we see an improvement in our profitability for the entire calendar 2025. Of course, now that we are almost in the middle of calendar 2025, we are already negotiating contract for next calendar year.
Everything is progressing as we were expecting and as our customers were telling us. Build orders are good, and we continue to move our product roadmap through higher capacity drive, and this will continue to improve our profitability.
Oh, that's great. In the past, also, I think outside of the oversupply situation, there was also maybe not just from an exabyte, but also from a unit perspective, there has been more cyclicality, so to speak. Can you help us just think through over the next couple of years the increase in to keep up with the increase in demand? What is your thinking around how that supply keeps up with that level of demand growth?
Yeah, I would say if you look at the number of drives that we sell in the last several quarters, it has been fairly stable. What is changing is the content per unit, how we increase the terabyte per unit. This is mainly coming from HAMR, especially now in the next several quarters. We think that moving from the last product of PMR that was a 24 TB in the CMR version to the 30 TB HAMR, to the 36 TB HAMR, which is intermediate between the 3 TB per disk and the 4 TB per disk, and then the 40 TB that we start qualifying next quarter, this technology transition will be enough, at least on our side, to generate that increase in exabyte that keeps up with the increase in demand. We don't see an increase in unit for the industry.
That's why I know when they ask me, "Oh, talk to us about market share." We are very different market share. There is unit market share, there is exabyte market share, there is revenue market share. Now, I think because demand is strong, all the units will be sold in the industry. The exabyte generated by those units can be different depending on which company is moving to which level of terabyte per unit. Depending on what product you have developed, what product is qualified, you can sell a little bit more or a little bit less of exabyte. The exabyte market share can be a little bit different than the unit market share, but I still believe all the units will be sold. Revenue is more related to exabyte. That part depends more from the technology transition and the product development.
The unit market share, I've not seen changes in the past. I don't see changes in the future.
Okay, great. If you think about the exabyte growth that you highlighted at your analyst day, can you maybe help us just think through decomposing that a little bit? Like how much of that would you say is what has been historically traditional demand versus maybe incremental coming from AI or any other way that you want to construct that?
AI will be a big part of the increase. It's difficult for us to time exactly when the big volume will happen. I often refer to what we were talking several years ago about autonomous driving. Many people were saying, "Oh, in five years, no one is going to drive." It takes a little bit longer, but it's true. In the future, probably we will use an autonomous car. AI is the same. It's probably taking a little bit longer in terms of adoption compared to what other people are thinking. They believe in one quarter, everything can happen, the full transition of AI. It will take longer, but it's huge, and it's going to change how we work, how we live, how businesses are running their business in general. It will be a huge part of the increase.
We are maybe a little bit conservative in what we have in our model right now because we want to see more evidence. It is going to happen, and it is going to be a huge part of our exabyte sold in the future.
Maybe just to digress for a quick second on AI, like how is Seagate actually deploying AI internally? Any color you can share there?
Yeah, we use AI in many parts of our business. We use AI for quality control. We use AI in manufacturing. We use AI in other parts of the organization. Investor relation is using AI. Accounting is using AI. Of course, in our R&D, all our coding, there is a lot of AI. This is still relatively small compared to what will be in a few years from now, where the vast majority of our business, all our jobs, will have a part of AI. It will not be full, but a big part will be AI. Simplification, going faster, being more accurate. We need to give the time to AI to improve, to do the learning for if you really want to use them and trust the AI results.
Great. Now maybe just coming back to the business and thinking about the margin guidance you gave around incrementals of hitting 50% incremental margins when your revenue goes over $2.6 billion. What are some of the underlying dynamics that create that incremental margin for you?
I would say the two main drivers will be the transition to higher capacity mix. Continue to move up in the product that we sell in terms of capacity per unit. When we go up in capacity, our cost per terabyte of cost declines, and we keep a good part of the cost decline to improve our profitability. That is one of the main drivers. The other one is pricing. We are starting a certain pricing strategy about two years ago, and it has worked well. As you know, we did not really increase pricing too much. It was a very reasonable increase. Every time we renegotiate the contract with a customer for the same product, we ask a little bit higher price. It is very predictable.
I know we don't think it's disrupting at all for our customers, but it's giving us a little bit more profit every time we go into the negotiation. Those two drivers are the more important. In the past, of course, when the volume was very small, the growth of gross margin was even faster, but we had a lot of unused capacity, underutilization charges. When the volume goes up, you have that improvement in the gross margin percentage. Of course, we don't want to go back there. That's not a good place to be, but give you as a growth in gross margin, give you a little bit of boost. Right now, of course, we are full, so we don't have that benefit anymore, but we still have the benefit of the technology transition and of the pricing strategy.
Okay, that's great. Just to step back for a second, right? I mean, the demand is largely coming from cloud. You're saying that you have good visibility because of BTO. How does this change the way that you will manage or the industry can manage through any deceleration in demand patterns?
Yeah, no, we are a technology industry, and there are cycles in the industry, so we always think about that. The important thing is how you can manage the cycle. If you are surprised by the cycle, it's very complicated because you enter into the down cycle with a lot of volume in your inventory that you are producing, expecting a certain demand that finally doesn't realize. When you need to move that volume to customers that they don't really need it, of course, you have a pricing impact. This is the second benefit of build order. Now, we will have visibility of when those orders will start to decline in terms of volume. We can reduce our manufacturing and get ready.
If you enter into the down cycle with a lean organization and a lower level of inventory, of course, you will have the impact of the volume decline, but at least you will avoid the impact of the price decline. That is a much better way for the industry to manage the cycles that every technology industry will see in the future.
Maybe just on the build order. This has been ongoing now for several quarters. You also noted some visibility even as far out, almost a year from today. When you think about the pricing around for the visibility that's being provided out, how are you thinking about that pricing trajectory? Maybe you can also comment on for HAMR in particular, as you ramp HAMR units, how should investors think about the dollars per terabyte for HAMR?
Yeah, in our earning release in January, we talked a little bit about the calendar 2025, and we said now based on the build order that we already have, we see revenue increasing sequentially every quarter, and we see profitability to improve every quarter. We have confirmed that in any meeting after that. We will talk about calendar 2026 in the near future. Of course, now the build order are usually two, three, four quarters, and the majority are between three and four quarters. Of course, we start to have visibility on the first part of the calendar 2026.
HAMR will not be different than other products. Now, for our customer, finally, the benefit comes from buying one unit with more terabyte because that is a TCO improvement for them. They need to spend some money and put some resources in qualifying a new product.
Generally, we give a little bit of the cost per terabyte discount to them. We incentivize customers to move to the next product. The vast majority of that saving, that is coming from our development and our improvement and our investment, it has to remain with Seagate.
As you think about the relative sort of margin profile across your nearline drives, across PMR and HAMR, and obviously, your HAMR is also progressing from 30 to 36 to 40 over time. When you think about the relative margin profiles between these, how accretive can HAMR be? You just mentioned TCO. Can you give us some indication of how much of that sort of cost down can you actually share with your customers?
Yeah, we had a slide at our investor day showing actually the cost per terabyte decline going from our last PMR product to the 30 TB HAMR product to the 40, 50, 60. You can look how big will be the decline, especially when you move from the 30 to 40 and the 50. It's already a price decline, sorry, a cost decline, cost per terabyte decline going from PMR to the 30 TB. The improvement comes actually much bigger later when you go from 30 to 40. You have basically the same drive with the same 10 disks, with the same number of disks, with the same number of heads. Basically, the same bill of material in terms of units, but you add 10 TB per unit. That is 30% more capacity per unit.
When you go from 40 to 50, now you have the same, and you have a 25% improvement. There is, of course, a huge cost decline. Of course, not all the mix is moving immediately from PMR to 30 TB and from 30 TB to 40 TB and 50 terabyte. No, but it's always a different mix on what we sell. Every segment has a different requirement in terms of capacity. That is now the cost decline we can get as a top with a top product. As I said before, you need to incentivize it a bit, customers to move and to spend the money for the core. We will keep a very good part of that cost decline to improve our profitability.
That's great. One of the questions we often get is just as you think about, and you commented a little bit about share, but given sort of where you are with HAMR relative to your primary competitor, shouldn't investors think that you should have materially higher exabyte share as you go through the next year and a half or so, supposedly at the time when maybe your competitor also has a 40 TB HAMR drive? Why would be a reason that that would not happen?
I think demand is strong. Therefore, I think all the units will be sold. I do not see a big change in the unit share. We will have the product with a higher terabyte per unit. Those additional terabyte that we can produce and put in our units will probably generate an improvement in the exabyte share and therefore in the revenue share. Our competitor, our main competitor, is saying they will have a similar product in the next maybe couple of years. Again, at a certain point, maybe that share will realign. It is absolutely not a problem for us or for the industry, I think, because demand is above supply. All the units will be sold anyway. Of course, it is better to have a higher capacity. We strongly compete on product development.
Everyone wants to have the best product, the product with the highest capacity per unit because that gives you a little bit of boost in terms of revenue and exabyte. It is not really impacting the result of the individual company, I think.
What would you say around just sort of the sustainability of going back to cloud demand because it is such a material part of the story? How can you get comfortable that there isn't a lot of inventory that is sitting somewhere that we're going to get caught surprised by?
As I said before, the build order should be the main way for us to look if customers are needing more volume or less volume. As I said before, the build orders through this calendar 2025 are growing. We see a need for more volume, not less. We do not have the impression they are building inventory. When in the quarter we eventually have a little bit more production available and we put it in the market, that production is sold very, very quickly, even if actually the price per terabyte is higher than what we have in the build order. If there was inventory, I do not think our customers were buying that volume at a higher price. They were maybe sticking with a build order because it is an order, but they were not buying more.
Every time we have a little bit more, they buy and they spend more money to buy that volume.
How should we think about, from an OpEx perspective, you noted that in your longer-term model or your three-year-out model, 10% of revs. And when we think about where you are today, what is it that gets you? Is it absolute OpEx dollars in certain areas that you're targeting? What are the areas where you're investing maybe? How does that make shifting over the next few years in terms of to get you to that? I understand the mathematics of it. Obviously, it's a lower % of sales and higher revenue, but anything else that you're doing under the covers?
I think in general, in terms of resources, we are where we want to be. We went through a restructuring during the down cycle, not only because it was a down cycle, but also because we were already at the end of the PMR technology. We already had developed the last product in PMR, and we had developed the HAMR technology. For many years, we had basically two teams, one working on PMR, one working on HAMR. When we decided to stop doing the PMR, we moved some people from the PMR group into the HAMR group, but we also reduced the size overall of our R&D. We have what we need. We are now, we do not need to develop a new technology. We just need to develop a new product with the same technology, very similar to what we were doing in the past.
I think we have the right resources in R&D. I think we have the right resources in general for our SG&A. In terms of number of people, in terms of headcount, I think we are good. Of course, every year, you will have the annual salary increase. Every year, the company will perform in a different level compared to the expectation we have internally. The variable compensation can be higher or lower. In terms of, let's say, the base headcount, I don't see a need to increase.
Okay. Okay. That's really helpful. What about on the CapEx side of things? Do you foresee any changes in there?
Not really. No, we have this target of 4%-6% of revenue since many years. We have been in that range always, sometimes even a little bit lower. I do not think we need more than that. No, we need to replace some of the old equipment that every year, they go end of life, and we need to replace with new equipment. We do not need more units, as we said before. No, we are happy with that. We produce heads and media, so the disk. We are happy with our capacity. We do not think we need more. No, we think we can produce more exabyte because those heads and media moving into HAMR and moving into the 3 TB per disk, 4 TB per disk, 5 TB per disk will generate more exabyte with the same number of disks and heads.
Can you just also talk about, I think at your analyst day, you shared a lot of detail about aerial density improvements, what goes into that, and some of the technology improvements that you're making underway. How far along are you around those technology improvements? Is there potential? What is the potential to do incrementally more around some of integration of some of those underlying components? Can that be also a margin driver incrementally?
Yes. No, we will have improvement in the future as we have done with PMR. Now, there are some more theoretical assumptions on what will happen. There is the reality. The reality is people learn more, and AI will help. Even with the same manufacturing processes, you always find a way to be a little bit more efficient. The yield can be a little bit better. Maybe the number of steps can start to reduce. There are improvements that now are not in our forecast, but I think eventually will happen in the future. One of the improvements that we have recently done was the internal production of the laser. Now, John, our CTO, explained how technically it is done on the wafers. It is very complicated. We are happy with that progress.
We will have a good part of our HAMR drives that we adopted. We also have an external source for the laser. It is good to have two sources. We will have a mix of production, some using our internal laser, some using the external laser. I'll say, of course, our internal production has a lower cost. From a financial standpoint, no, I'm happy when we sell a drive with our laser. It is also very important to have supply stability. Being also in charge of supply chain, I'm very happy and very focused on keeping supply stability and business continuity.
Can you maybe just also talk a little bit about, in the context of what you mentioned about downturns earlier, you said hopefully the volatility around that comes down a little bit because you have a better line of sight and you can adjust your output earlier. Could you just talk about this from a financial model standpoint too? Because one of the things that I think people are worried about is just it's still a cyclical industry. When you go into a downturn with some leverage, both operating and financial leverage, it creates a lot of anxiety for investors. What can you do to sort of eliminate at least, I mean, it sounds like at least on the operational part of leverage, you've got more confidence in handling that. What about the financial side of leverage?
What's the right sort of debt levels for you over time?
As I said before, almost every technology industry has cycles. We expect some cycle also in our disk, possibly not as bad cycles as we had in the past because of what we discussed before. I think for a long-term investor, it should not be a problem. You stay through the cycle. You know that every cycle you get better. That is a reward for our long-term investors. If you are a short-term investor, you need to choose well when you enter and when you want to exit. It's your choice. I don't think it's a great solution, but you can always do it. We will manage better because we will have visibility. Therefore, we can avoid some of the additional costs we had in the past and some of the additional pricing pressure we had in the past. We are very focused.
I think to be successful, you always need an industry going in the same direction, having no similar focus. I would say in the last several years, this has been the focus of the industry. We compete very strongly on the product. I think we have a focus on improving profitability for the entire industry.
No, that's great. I know we're almost coming up on time on these sessions this year. Only 30 minutes. It is really not a lot of time to get into the weeds on a lot of stuff. Maybe if you could just, John, look at it to wrap it up, right? As we think about your guidance and we map that and say, all right, let's say you grow mid-teens and then you put the margin profile that you have given us, you can get to $16 plus of earnings power by looking at you and your new higher tax rate. As investors think about that and the fact that these cycles are diminished, you really have historically traded at a pretty inexpensive multiple, so to speak. Given all that, why shouldn't you?
It feels to me that Seagate is an extremely attractive investment at these levels. Why wouldn't you lever up maybe a little bit or even not lever up, but do a little bit more of buybacks and sort of do that ahead of when you get all the realization of all this benefit over the next few years?
We will do share buyback. Now, as we said at our investor day, we have an authorization, a new authorization, so up to $5 billion that we can use whenever we think is the right time to use it.
Now, right?
I also said, no, we need to take the opportunity right now also to strengthen our balance sheet. We are almost there. Now, we wanted to reduce the debt. We entered into the prior cycle with more than $6 billion of debt. It was a little bit too much. We are now fairly close to our target of $5 billion. We did a couple of transactions in the quarter. We should be very close to the $5 billion to where we want to be by the end of the quarter. During the prior cycle, we also had to stretch a little bit our working capital. We took the opportunity in the last three or four quarters to rebuild our working capital. I think we are almost there.
I'll say, as I said one week ago, 10 days ago, I think fairly soon is probably the right time to restart with the share buyback.
Okay. Amazing. We are just about out of time. John, Gianluca, thank you so much for all your time and insights. Really appreciate it.
Thank you very much, Michael.