Seagate Aktie

Okay. Good morning, everybody. Welcome to the Goldman Sachs Communacopia Technology Conference. My name is Jim Schneider. I'm the semiconductor analyst here at Goldman Sachs. It's my pleasure to welcome Seagate Technology and CFO, Gianluca Romano, to be here with us today. Thanks, Gianluca, for being here.

Thank you very much.

Maybe start off at the highest level. Help us understand where do you think we are in the buildout of mass storage for AI infrastructure. And where do you see the biggest opportunities specifically for Seagate over the next, say, 12 to 18 months?

Yes. Before we start, let me inform everyone that I will be making forward-looking statements today, and you can learn more about the risks associated with those statements on our website. Well, it's a very good question. I would say, if you look at the CapEx of our customers, the growth in the CapEx is signaling still being in the first part of the phase of AI investment and therefore, for what we are concerned on investment in storage, in data storage, especially in the big public cloud. And more recently, I would say, even more on-prem data center with a fairly good growth on enterprise OEM.

Okay. Great. With your competitors, you've been able to significantly increase the amount of exabytes being brought to the market without really significantly increasing unit production and by just doing the capacity additions rather than unit volume. So the industry has been prudent on supply additions. It seems the strategy has really paid off for most of the players, yourself included. How do you think about striking that balance into next year and beyond?

In general, I would say it's important for this industry to be disciplined with CapEx and with capacity addition. I think we have been very well aligned to this discipline for 2 or 3 years at this point in time. The business is growing because of technology, not because of more units. We move up in technology with products that have more content inside the box, and that is generating at the end of the quarter, more exabytes. And finally, this industry is selling exabytes, not units. So what we need to focus is, of course, always looking at supply and demand and keeping a good balance between the two and also strongly, strongly pushing on our technology to continue to grow in exabyte year after year.

I mean what is your view on sort of the long-term exabyte growth for the industry? You've talked about 20% long-term growth as a target. Clearly, you've been outstripping that for a little while now. Is that still long term -- the right correct long-term target? Or do you think we can kind of sustainably ship above that given the sort of current supply balance we have?

Well, every quarter is different. Every year is different. We gave this mid-20% CAGR for a fairly long period of time, we said about 3 or 4 years. Of course, you need to look at the real exabyte growth not only in percentage, but also the real number in absolute number.

And of course, when the base start to grow, you add a lot of exabytes maybe in percentage is not exactly the same of what you were doing 2 or 3 years before. But actually in exabyte is more. So that increased number of exabyte is actually what is impacting positively your revenue growth. So when the base change percentage, it could be a little bit misleading.

So I always look more at the absolute numbers and percentages. But this industry and Seagate in particularly, I think, is growing very strongly in exabytes. And as we said, this will continue because of the applications that are today and in the future impacting the need for more storage. If I look today where storage is, there is a lot of traditional application and then there is the new impact from video AI, from starting robotic AI.

If you are here in San Francisco, you see a lot of autonomous driving. All those cars have a lot of camera and all the data that they collect all day long get stored. If you multiply San Francisco for the rest of the world, you can just imagine, imagine how much data autonomous driving will generate and the need for storage. And this is just one application.

Even robotic AI is just the beginning. There is a lot of quality control that is done connected to AI so that you get an input from AI on how to identify an eventual problem and eventually how to fix the eventual problem. Some of that is in the form of video and video consume a lot of exabytes.

So there are a lot of new applications, but the common theme of those application is everything is based on data. And that data is very, very valuable. When you have a valuable data, the cost of storage is minimal comparing to recompute that data. And that's why storage is growing at the pace that you have seen in the last 2 or 3 years.

Yes. I mean you mentioned this idea of absolute exabyte growth for the industry. That's been a very interesting notion. I don't think I remember you raising that before. Curious, is there a way to frame the industry growth in terms of absolute exabytes?

No, I think you can look at a trend, but because we are growing fairly rapidly in terms of exabytes, the base is really growing. If you look at growing, I don't know, 25% or 30% this year comparing to 3 years ago, that is maybe 50-plus percent if you look at the number of exabytes. So when you look at the revenue growth, you need to look at the exabytes that we sell and what will be the price for those exabytes more than a percentage because I guess the base is very different, could be a little bit misleading.

Yes. Okay. And then -- so if you think about kind of all the things you said that are driving demand for data, you've also been driving a lot of your exabyte growth through technology transitions as well, especially with HAMR, which we'll get to. So do you think you can continue to support all of the exabyte growth you expect through tech transitions alone?

Well, today, this is our view. This is our objective to go as fast as we can on technology. So technology can give us a fairly good growth in exabyte. When we go from a 30-terabyte drive to a 40-terabyte drive, we can have 33% more content. So that, of course, is implying a huge increase in exabyte even if, of course, we don't just produce one drive, already is a big mix. But when you can have that kind of growth in your -- through your technology, you can generate a good growth in exabyte overall, that is what finally we said, we sell exabyte. And of course, growing exabyte and keeping a good balance between supply and demand is what is making this industry very successful.

And so we don't see any reason today to change this strategy has given us huge improvement together with our pricing strategy. The exabyte growth through technology and the pricing strategy that we have applied for the last 3 years, allow us to almost triple our gross margin without really impacting our customers too much.

Now we are low to middle single digit of their CapEx. If we are reasonable with the pricing strategy that will allow us to apply this strategy for a very, very long period of time and continue to improve our performance from a financial standpoint. We have done that for 12, 13 quarters based on our PO that we already have in place for the entire of our fiscal year, we are now at the beginning of the fiscal year. We said we see every quarter of this fiscal year having higher revenue and higher margin. So this strategy is really working well, and we don't see any reason today to change it.

Right. Okay. And so to that point on -- you raised the topic of pricing. So maybe I wanted to think about that for a second and zoom out for a second. Several years ago, we were talking about mid-single-digit declines in price per exabyte, then we went to flat, then we went to mid-single-digit growth. Last quarter, you talked about 11% year-over-year price per exabyte increase. So we're comfortably in the double digits now. How should investors think about sort of the cadence of pricing growth over the next few quarters? And should last quarter be kind of like a good ballpark to use modeling forward for this fiscal year?

We changed the pricing strategy about 3 years ago. I would say there are 2 reasons why this strategy is working well. The first one is, of course, the supply-demand balance is different than what was 4 or 5 years ago. The second is the value of storage is much higher. As I said before, when you have data and that data is a value for you, you want to keep it. You don't want to recompute later because recomputing later will cost you more. So there is always that double check between keeping the data or the cost of recreating that data later. And of course, there is -- at least today, is a huge difference between the cost of storage and storing the data or recomputing data.

Got it. Okay. You often referenced total cost of ownership as a way to sort of approach the economics of higher capacity drives, at least from a customer perspective. Now how receptive have customers been to sort of these higher per drive and per exabyte prices? And is there really any different from prior cycles in terms of the way they perceive CPO?

I would say the value of data is, for sure, growing. And our customers, of course, are pushing more and more on optimizing their structure. That means optimizing the storage and the compute and how much storage to support that compute. And of course, they realize that the value of data is not the same today that was 10 years ago or 5 years ago. And that, of course, is giving us the opportunity to also optimize our own business and driving through technology to generate a little bit more exabytes every quarter and of course, try to get the targets that we discussed before for the CAGR, but also extract some more value for us. And of course, increasing pricing is part of that strategy. And as I said before, this is a huge contribution to our gross margin and operating margin at the end.

Okay. I mean it may seem really hard to imagine in the current environment. But let's say we find ourselves in a situation some place down the line where the industry has overshipped in demand, customers cut orders. What kind of risk in the downside scenario does Seagate see in that kind of eventuality? And what measures has the company put in place from a manufacturing perspective to sort of derisk that?

Yes. Today, we don't see that situation happening. As I said before, we already have purchase orders in place for the next 4 or 5 quarters. And as I said before, based on those purchase orders, we actually see revenue increase and profitability improving. So we don't see that possible change in trend happening for the next many quarters. I would say being disciplined with CapEx is, of course, something that will help you at a certain point if the trends change or if for some reason, our customer, despite having a very high demand, maybe they cannot build all the data center that they would like to because of some constraint or can be a power constraint, can be a delay in building permit, can be some components that is not fully available.

So there are many reasons why eventually, even if demand is very high, our customer could have a need to slow down a little bit, the building of the new data center and pushing that demand out in time is not going away, just pushing that out in time. But if that happens, I think the fact that we are very disciplined with the CapEx and how we increase capacity through technology instead of through units will allow us to eventually manage much better a possible change in cycle.

Yes. Fair. So I want to actually to the point, shift to the technology and product side for a moment. Following the qualification of Mozaic 4 with 2 hyperscalers you talked about last quarter, what further progress have made with additional customers beyond those 2? And what feedback have they given you on Mozaic 4?

Very good. Now I would say HAMR is a fantastic technology. It is allowing us to really have very good results, not only in terms of more exabyte volume, but also in terms of financial performance. And you will see more -- the more we scale up in capacity per unit, so going from 30 terabytes that we are already selling to all the hyperscalers in the world, the big guys to the 40 terabytes that we're already selling to the 2 biggest hyperscalers in the world and that we are qualifying with a few more.

And so we will be able to qualify fairly soon. And then we discussed at our earnings release already the next product that will be a 50 terabyte. So you can see how quickly and how much we can grow in terms of content per unit. And now the 50 terabyte will be more calendar '27 through the end of calendar year '27. So very good progress.

We are very happy with HAMR.

I'll say it took a long time to develop the technology, many years. And it took a little bit of time to qualify the first product because it was new, new technology, new product. It has to work. We had to find the right configuration to work in the cloud. After that, everything went really well. And we qualified the top 8, 10 customers in basically a year on the 30-terabyte drive, and we are already qualifying on the 40 terabyte, a good number of customers, both in U.S. and in Asia, and we will do the same with the 50 fairly soon.

Yes. You've also talked about getting volume crossover of HAMR by the end of this calendar year. Has there anything changed that would allow you to even pull that in? Or is there anything that would delay it all?

.

No, I think our plan is really solid. I think by December, so just a few months from now, we will be at that level for the 40-terabyte drive and for HAMR in general. I think in a couple of years from now, probably in the data center environment for Seagate, 80%, 90% of the volume will be sold through HAMR. It's just a matter of time. There's no way that the PMR technology can keep up with HAMR.

There's no way. You can put how many disk you want inside the box. You will not have a 50-terabyte drive. You will not have a 60-terabyte drive, you will not have a 70-terabyte drive. So that's the reality. So because in that environment, customers have a huge benefit from higher capacity per unit. Customers will want to buy those big drives and those big drives can only come from HAMR.

Yes. Now how are you thinking about the cadence of HAMR adoption by your customers? Do you have any visibility into sort of any kind of lumpiness that might get treated in customer demand? And -- or should we just expect that sort of HAMR ramp to be very gradual in terms of mix throughout 2027?

Well, as I said before, all our customers, all the big customers are already buying HAMR, everyone. So everyone has qualified at least the fourth-generation HAMR. Some have qualified a second-generation HAMR. So they're all buying hundreds of thousands of units every quarter.

So I would say technology is not even a discussion with them anymore. The discussion is the capacity per unit. How you go from 30 to 40 to 50, how fast you can go and then what happened after that and et cetera, et cetera. So I would say technology has been proven with the first drive, with the first generation drive. Right now, it's just a product development. It's not anymore a real technology development. For us, it is a product development.

Got it. And you've also pointed to some of the advantages you see in terms of your vertically integrated supply chain, specifically lasers as the key kind of innovation that you have. So as you ramp up HAMR production, sort of do you see margin tailwinds from an internal production of lasers on the higher capacity drives?

Well, partially, yes. Of course, the cost for us to produce the laser together with the head is much lower than buying the laser outside and then attached to the head laser. So there are a lot of benefits from a cost perspective, from a process perspective and of course, also derisk a little bit the supply chain. So it's with one of the improvements that we have done in the last few quarters.

Yes. Can you help us understand the dual source or second source strategy for the laser specifically? I mean, is there any kind of like supply chain impact or margin impact from any of that?

Well, the first generation of HAMR was only using external laser. When we transition to the second generation, we started to have a mix. Some of the products are built with our internal lasers. Some are still using external laser. I think more and more we will go out in the future, more the mix will go into the internal laser. And as I said before, there is a cost differential, of course, is one component out of a certain number of components inside the box, but it's, of course, one component where we can generate a little bit more profit.

Yes. And then you talked about HAMR obviously getting bigger and bigger in terms of mix. Longer term, how quickly, if at all, do you expect PMR to sort of rotate out of your product offering? Is there still place for PMR in your product portfolio for outside of legacy and sort of client type applications?

As I said before, I think in a couple of years from now, at least for us, the vast, vast majority of products will be based on HAMR. There are different capacity for different segments inside the business. We always talk about the big drive, but don't forget that we sell hard disk drive from 2 terabytes to 40 terabyte today.

And in the future, will be maybe not 2, maybe 4 terabyte to 50, 60, 70. So EMR is a very good technology for the lower capacity. But when you go into the public cloud and a certain point in time also for on-prem data center, when you go above the 30 terabyte drive, I think HAMR is a clear better solution that is available.

Yes. And you talked about -- we talked a lot about the data center market and AI, and we think we all know that's growing very strongly for you. Maybe talk about sort of what's happening in some of your other segments, whether that be the client business or even other kind of like legacy video applications and things like that. What's happening in those segments? And do you expect those to kind of remain stable over time? Or like do you expect those to sort of like fade a little bit?

Yes, it's very interesting. Our business has 2 major segments, data center and what we call edge, edge IoT. Data center is what we have discussed until now, big drive. Bigger and bigger and bigger drives give you a better revenue, a better financial result. This is today 80% of our revenue. 20% is lower capacity drive, so what we call edge.

This is serving consumer, client, some part of the video surveillance. So those are drives that are between, let's say, 2 terabytes and 10, 12, maybe 14. In data center, the architecture of the data center separate storage from compute. Storage is hard disk, compute starts with an SSD. So in a sophisticated data center, all the storage is hard disk and then the data is physically moved from an hard disk into an SSD when you need to do the compute or the analytics of the data and then the new data goes back into the hard disk.

When you go to different segments like consumer or some of the video surveillance, capacity is much lower. So there is an overlap between hard disk and SSD. And therefore, the volume and the price also depend from the situation into the NAND. Today, the NAND price is fairly high, and this is giving the opportunity to hard disk also in those low capacity hard disk drive segment to actually have the opportunity to increase our price and actually to keep the volume fairly consistent.

I would say even in that -- in those segments, demand is above supply, but we need to allocate because demand is above supply in every segment today. So we give priority to our big customers, I would say, and the data center because it's 80% of our business. We also allocated a certain volume to consumer clients, video surveillance.

But we have recently in the last couple of quarters, we have changed a little bit our pricing strategy there because we don't have purchase orders in place. We don't have LTAs. We don't have specific big customers. So it's a different kind of business, only 20% of our revenue, so not huge, but today is giving us actually a very good result also from a financial standpoint.

So what will happen in the future? I would say, if you look at the past, that business has declined a little bit in terms of volume, and was also kind of declining in terms of pricing. Today, I would say the volume is fairly stable because we cannot allocate more. Otherwise, it could be even higher, but pricing is much better. So let's see what happen in the future. Again, it's not so material to our business in total because it's not the majority, but it's important.

Yes. On the financial picture, I think it's fair to say that as we stand here, you've already pretty materially outstripped the financial targets you provided just last year at your Investor Day. But maybe I want to sort of understand like where you think could be going on that front. You recently kind of like upgraded your long-term revenue outlook for the company. Maybe talk about like that target, how sustainable and over what duration you kind of can maintain that?

Yes. I would say the situation from our Investor Day has continued to improve. Demand is much, much stronger than what we were seeing at that time. So we are doing all what we can through technology to increase the exabyte and to support a good part of that demand. But of course, this is also giving us the opportunity to raise pricing a little bit more than what we were discussing at the Investor Day, and this is resulting in better gross margin, better operating margin, of course, much stronger EPS.

This is continuing. I don't see a reason why it should really change at this point. Situation is -- the trend is really going in one direction today. And the strategy is strong, is giving us very good results. So we are applying the strategy with a little bit more maybe emphasis on pricing. And we are performing well. And I think you will see our results continue to improve in the next several quarters.

Yes. And then in terms of margins, I mean, I think you previously laid out the idea of incremental margins of 50%. You've obviously been dramatically outstripping that in recent quarters, as you just said. I think the 50% kind of baked in a number of things that were true at the time. For example, the start-up cost of your HAMR ramp, which was very early days before, that's probably no longer a big headwind for you. And as you mentioned, pricing has been a big tailwind. So is there a new construct you can kind of share in terms of how we should think about incremental margins in the business?

Well, I say, if you look at the last several quarters, our incremental gross margin, as you said, has been way higher than 50%, I would say, in the 70-plus percent. The trend is, as I said before, the trend is continuing. So today, I don't see a reason why that should change. So for the next few quarters, I think that is a good way to look at the business.

Yes. Then on balance sheet and capital, you've basically been carrying a few notes with relatively high coupons. And you've talked about sort of retiring already $1 billion or more of those recently and then sort of getting back to a more normal cadence of capital return. After you kind of get done with the debt retirement, how are you thinking about the sort of how this plays out? And how do you sort of weigh the differences between -- or the trade-offs between increasing the dividend versus share repurchase?

Yes, a very good question. I think we have reduced our debt already by a lot in the last 1 or 2 years including this quarter. We are also doing a decent level of share buyback this quarter. After this quarter, we probably still have one note that I would like to repurchase that has a very high interest rate and probably we will do next quarter.

So starting calendar year '27, I would say you will see a higher level of share buyback. Not that we are not doing today, we are doing it today, and we will do next quarter. But I think when we are done -- completely done with the debt, we will have even more free cash flow available for share buyback.

Dividend, usually, we review internally with our CEO around October, November time frame on what to do in terms of possible dividend increase. Generally, in a strong period of time like this one, we will probably increase the dividend. But I don't think dividend is actually the focus of our investors at this point.

So we want, of course, to provide a good shareholder return and part of that is dividend. A much bigger part will be share buyback. And of course, the focus is on growing the company, growing the top line and continue to improve the profitability.

Excellent. Maybe just to close on one last question before we go. If we're sitting here on stage 5 years from now, and we look back, what do you think is going to be the one thing that really surprises investors about what Seagate has done over, say, the past 5 years, which are the next 5?

Well, I hope they will be surprised by our financial results. I think we have all the opportunity to do it. As we discussed before, demand will be there. You can have temporary changes, but the underlying demand because of the new applications that are coming out in the world are all based on data and storage has a lot of value, has a lot of value in the data center business.

Demand will be there. I think we will be able to grow through technology at a very good exabyte CAGR. And you know what is our view on pricing at this point. So I think there are a lot of opportunities for this company to grow to become even more important into the data center ecosystem in general and therefore, justify better results.

Okay. Excellent. I think we'll leave it there. Thanks so much for joining Gianluca. We appreciate it.

Thank you very much.

Good morning. I've been told it's still the morning. I haven't stepped out, so I'm assuming it's still morning of day 2. We've had a lot of meetings and so just -- I can't believe, like I said, it's still the morning.

So welcome to day 2. Asiya Merchant. I lead Citi's tech hardware and tech research here at Citi. I am delighted to have Seagate's CFO, Gianluca Romano, here with me. I also have Shanye here in the audience, she's part of -- she heads Seagate's IR team.

So this obviously is fireside interactive. I do have some prepared questions. I do leave time at the end for investors. So if you have burning questions, please make sure -- we're short on time here always, but do raise your hand, please. We will bring the mic to you so that folks on the web can also hear.

I'll first turn it over to Gianluca for some prepared comments.

Thank you, Asiya, and thank you, everyone, for being here today. As always, I will be making forward-looking statements today, and you can learn more about the risks associated with those statements on our website.

All right. Great. Gianluca, I've been asking a lot of these companies. Obviously, you guys are seeing great growth here. Investors hit back, there's always going to be skeptics worried about, is this cycle really different? Why is it different? So maybe you can opine on that. Why do you think the cycle is different for the HDD space and for Seagate in particular?

Well, I would say every cycle is different drivers. And I would say we are very happy on the situation today. It's not for sure the beginning of the cycle because it's already more than 3 years, but we are improving our revenue and our profitability. And we discussed at our last earnings release, based on the PO that we have in place, we see improvement also for the entire fiscal '27, fiscal year '27. So every quarter, we see improvement in both revenue and profitability.

So it is for sure a strong cycle. We see demand growing even faster than what we were thinking. So the gap between supply/demand is not decreasing, it's actually a little bit increasing. We are trying to grow our exabyte as fast as we can through technology transition and moving from our 30-terabyte drive HAMR to our 40-terabyte drive HAMR. We also discussed about the next drive, that will be a 50-terabyte drive, that we will start to qualify about in a year from now. So 4 quarters from now, we will be already in qual with a new drive.

So this is where we really focus, is technology transition, drive more exabyte as a company, moving the mix up through the technology. Of course, we are not increasing our units, but we are increasing our exabyte and we are targeting that at least 25% CAGR in the next 2 or 3 years.

Yes. And like I said, there's always folks who are going to say, like, when do you think they can digest this capacity? As if the build is complete. But obviously, we're seeing CapEx numbers go up. Citi themselves raised their AI CapEx numbers as well. But just as you think about the build that's happening and the exabytes that are getting deployed for storage, in particular, HDD storage, do you foresee, at least over the next, I don't know, 2 to 3 years, that there's going to be some digestion of this storage media that's been deployed?

Well, based on our LTAs, I would not say so. We see volume growing every year. We have LTAs in place for the next 2, 3 years, and we see those volumes going up. We have purchase orders in place for the next 4 or 5 quarters, and that are very specific, POs with the right mix, the right price, with the right time to deliver. But then we have also longer-term LTA where we commit on a certain exabyte volume that our customers need to plan their data center growth. And we see those volumes actually increasing year-over-year. So we don't see today any change in the trend that we have seen in the last 2 or 3 years.

Okay. And then obviously, as the industry is transitioning from training, now we're seeing a lot more inferencing. Inferencing tends to be just that much more data being generated. Just how do you think about that 25% exabyte CAGR sort of that you've laid out? And I know that's supply versus demand, the demand probably greater than that. But when you think about the incremental HDD capacity that your customers are deploying, like is there any way for the investors to think about, okay, this is going towards training, this was -- or towards inferencing? Or this is just more traditional HDD demand that was getting replaced?

Well, there are a lot of drivers for demand. Of course, AI today is a huge part of that demand. There are a lot of traditional applications that require a lot of data. And on top of that, you have AI, and more recently, you have video AI that is consuming even more exabytes.

But there are also other applications that you will see in the next 2 or 3 years. Now we are looking at robotic AI, how to use more robotic automation in manufacturing and also outside manufacturing for more normal life, that require a lot of video cameras, a lot of monitoring, a lot of data storage.

You will see also, I think, a strong improvement in autonomous driving. And all those cars, if you are in San Francisco or if you are in Phoenix, you see a lot of cars going around, all those cars have a lot of cameras and they collect data all day long, all night long, and all the data gets stored, to improve their performance and also for compliance. So when those new applications will start to evolve in different places and in different parts of the world, that is a lot, a lot of data storage that will be required.

So we see a lot of reasons why data storage will continue to grow. And so when we put all the data together, the new applications, what our customers are telling us, and the other component side, the data center, when you put all together, we don't see today a reason why this trend should not continue.

Okay. And then KV caching, I know Dave Mosley on the call talked about KV caching. And then typically, when you're KV caching, you think of a very high-performance storage, which tends to be more flash. How does KV caching, sorry, affect HDD demand? Like how are you thinking about demand coming through for you guys?

I would say there is, for sure, already a lot of demand for our normal storage. So actually, demand is already above supply. So it's not that we are really looking for more demand. But I would say if you look at data center, you have storage and you have where you run the application. And so you move the data from storage, that is our disk, into a NAND when you run the application.

Now in reality, there is another tier that is, let's say, in the middle, that is the warm storage that you called KV cache. Inside KV cache, you have also different tiers. I would say for sure, there are some tiers inside KV cache that can use our disk drive. So I would say that is the only part where you can see a little bit of overlap between NAND and our disk. Otherwise, the normal storage is our disk, the running of the application is on NAND. There is something in the middle that could go, let's say, in both directions. And I think that our disk can, for sure, be part of that warm storage.

In the future. I would say today, again, demand is well above supply. So it's not that we have a lot of opportunities, but could be.

Okay. And if that were to materialize and as -- or as you see that opportunity materialize, in order to participate in that space, do you guys need to do some development here? Is there some partnership with flash? Or is it really just the technology innovation within HDDs?

Yes, I don't think we need to partner with flash. I will say, depending from the tier inside the KV cache, there are probably some requirements for performance of the hard disk. So we will have to meet those requirements. Now I think we have for sure the technology to do it, and then we need to see what is the capacity per drive that it will be required. But we have a lot of opportunities I think to develop the right performance or whatever storage level is required.

Right. Okay. And then one of the things, we had some of our enterprise OEMs just report earnings just before Labor Day, NetApp, Dell. Very, very strong growth there as well. And I know you've talked a little bit it's a smaller part of your portfolio. But just what are you seeing there? And as you talk to your customers, your enterprise OEM customers, what are they seeing on storage from the enterprise side? And how much do you think is durability here?

It's very interesting. I would say in the first part of the last 3 years. The majority of the increase in demand was coming from public cloud. And that has continued. But at a certain point, we started to see our enterprise OEMs, so more on-prem data center demand, to start to grow again.

They buy a little bit different kind of drives. If you look at the capacity of the drives, every segment has a different capacity. The capacity actually goes up every year, but they don't all buy the highest capacity drive. Public clouds buy the highest-capacity drive. On-prem data center, they buy a little bit lower capacity.

So today, if we sell a 40-terabyte drive to a public cloud, we probably sell a 28, 30-terabyte drive to on-prem data center. Probably in 2 years from now, on-prem data center will consume a 40-terabyte drive. But they don't have the same structure. They need to evolve their architecture to use a bigger drive.

Now public clouds, they already have a very sophisticated architectures so they can get all the capacity and the highest-capacity drive. When you start to go in different segments, you see a different kind of capacity. So this is why we produce actually between 2 terabytes to 40 terabytes, because every segment has a different demand.

That is very good. If they are growing, and I think they will be very important to this business. Also depends where this new application will become important in terms of storage. Some of the applications could be more on-prem. So maybe a certain point, on-prem will grow faster than today. I'm sure public cloud will continue to grow very rapidly. But there are some applications that maybe companies prefer to keep on-prem or hybrid. Some they go on-prem, some they go on public cloud. So we serve all our customers in the same way and it's a very good business already today.

Okay. All right. And then demand, obviously, consistently outpacing supply here. It's obviously reflected in your own pricing. It's gone from declines to stable to much better even in the June quarter, and I think you just talked about further pricing improvements. So just help -- investors always push back. Well, a lot of their capacity is already allocated and build-to-order, those agreements like you talked about. Where is this incremental pricing? Like I'm always surprised, well, okay, pricing was much better. Like where is that coming from? Is it this enterprise that we just talked about?

No, I would say it's in every segment. As we were discussing before, demand is above supply and is actually growing a little bit faster than what we were expecting. So the gap is -- eventually, it could be bigger. And when demand is above supply, you have opportunity to increase your price. And as you know, we do this in a way that is not super aggressive. I think it's very reasonable. We don't want to create a problem to our customers and we want to keep this sequential improvement going for a long period of time.

We have already done this for 3 years. And based on our orders that we have in place, we said at earnings release, every quarter of this fiscal year will actually see an improvement in both revenue and profitability. And based on the LTA, I can say, I think this will go on even after. So we want to keep this for a very, very long time. And to do that, you need to do it in a way that is sustainable. So we are doing that. But of course, we are increasing price as we have done in the past.

Okay. And when you talk about units, right, I mean I know you're trying to maintain your units, do the areal density to get your exabyte going. As you think about unit growth, like are you seeing any change, whether it's from your peers out there, who are also stressing units being kind of flattish, but any change as you're thinking about the fact that, okay, there is a lot of upside here in terms of meeting demand because, clearly, supply is undershooting demand?

Well, we think the best way to address the growth is through technology. So growing the content of the drive, not the units, not the number of the units that we sell. We have a very strong technology. And with this technology, with HAMR, we can grow without the need to increase the units that would change a bit the dynamic between supply and demand. Of course, we want to keep a good balance between supply and demand, and so we like the situation today. It helps us to continue to improve every quarter.

Technology is where we focus. We have developed HAMR for many years. We are starting to sell HAMR product almost 2 years ago and we are already on second generation HAMR. We discussed at our earnings release, we will start third-generation HAMR qual in just a few quarters from now.

So we are progressing very well. When you go from 30 terabytes to 40 terabytes, you increase 33% your capacity. So it's a very good increase in exabyte even if the unit remains the same. Then going from 40 to 50 is another 25%. So this is where we focus.

Of course, on the PMR part of the business, so the lower-capacity drive, to increase capacity per unit, you need to increase the bill of materials. So you need to have 1 more disk and 2 more heads. Now if you want to go from an 18-terabyte drive to a 20-terabyte drive, you need to add 1 disk and 2 heads. And as I said before, we sell from 2 terabytes to 40. HAMR is only on the top of the capacity, so 30 terabytes and up. Between 2 terabytes and 28 is basically PMR.

So on that part, to increase the capacity, you need to have a little bit more heads and more media because you need to increase the components inside the box. With HAMR, you really focus just on the technology. So we have 2 dynamics going on.

Maybe in future, when we go high enough in capacity, even HAMR can have 11 disks or 12 disks, but there's still space inside the box that we could use at a certain point. Today it's less important because when you go from 30 terabytes to 40 terabytes, you have a 33% increase. If you go from 10 disks to 11 disks, you only gain 10%. So today it's not so interesting for us. Maybe in the future it will be interesting to add another disk and maybe 2 disks.

Yes. Okay. And then you talk about HDDs, I think there's still a very small percentage of data centers, but a critical component. You look at alternatives that are out there to storage, and obviously the pricing there is significantly higher on a per terabyte basis. When you think about the infrastructure spending that is continuing, storage being a very critical component, I think the question we always get is, how much more pricing flexibility is there for HDDs? I mean just given the alternative, NAND, is just significantly much higher priced.

You're talking about the neocloud space or, generally, the data center...

No. AI. Just infrastructure spending on the clouds, like as you think about how much more pricing flexibility is there as you continue to -- as spending continues?

Well, I think there is no reason why we should not get a similar result in the future than what we got in the past, because demand is actually higher than -- the gap between demand and supply is actually a bit higher. And again, we are not a big part of our customer CapEx, so it's not that we are impacting so much their spending overall even with our increase. And again, we don't want to be super aggressive, we want to be reasonable. But we want to keep this very, very long.

And every quarter is different for many reasons. Some quarters you have more volume because you have prepared your manufacturing for a new product, so we have a lot of new products coming out. Other quarters, you have a lot of new contracts that are renegotiated, so you have good pricing, sequential improvement. Depends also what you did a year before because that can impact your comparables.

So every quarter is different, there are a lot of variables. But the result is the same. Revenue will go up and profitability will go up.

And as we think about these transitions, I mean, you talked a little bit about density goes up as you go from 30 to 40 and then eventually the 50-terabyte there as well. Just on the cost per bit reductions, like now that you're in your second-generation HAMR, obviously, yields could be better, how should we think about that cost per terabyte decline that you are on?

It's very good. I would say we produce internally where the technology is, so on the heads and on the media. Of course, the cost to produce a head or a media, the unit cost actually is not changing, if it is a 3-terabyte or a 4-terabyte or a 5-terabyte. So we have a very good cost per terabyte decline on what we produce.

What we buy externally, it depends. There are some years where the component cost is going down, so that adds to the cost decline. There are some years where it is a mixed bag, some components are increasing, some components are declining. In some years, if you think about the COVID period of time, all the components were actually going up and there was a little bit of inflation.

So that part every year is different and it's less under our control. But on what we control, of course, we have a very good cost decline.

And then when we're thinking about this qualifications for the 50 terabytes, you said about a year from now, are they just -- are your customers just getting maybe faster at qualifying this, now they've played around with HAMR, now second generation? Or is it the same length of qualification cycles?

I would say the first-generation HAMR was a little bit more difficult. Well, for sure, with the first customer, because it took longer to find the right configuration of the drive to work as it is used in a big data set.

But I would say for every customer, they had a little bit more -- they took a little bit more time to test. It's not that was different. They were just testing for longer, just to be sure that the change into the technology was not resulting in some unexpected result.

Second generation is going very, very fast. I would say, right now, they don't even think about technology, it's just a new product on something that they already know very well. So I don't think it will be different HAMR compared to what was PMR.

Right. Okay. And then each time you do a ramp as you're transitioning from 40 to 50, for example, there are some inefficiencies that do creep in, whether it's shipping products just for qualifications and so you can't recognize that as revenue. Just as you think about and you're getting faster and learning from your prior transitions, how should we think about those manufacturing inefficiencies as you transition?

I would say, well, there is always a little bit of transition cost when we go from one product to the other. But I would say it's normal for the business. It's always been similar. So we always have some transition, even with PMR products.

I would say because now we grow more, you go from 40 to 50, so now you will see eventually more variability in the exabyte. When you grow from 18 to 20, if the units are similar, you grow by 10% or less, 5%. Now with HAMR, you could have a quarter where the exabyte growth is very variable. When you prepare the line for the transition, you don't grow so much. But when you start to get the output, you have a quarter where maybe you jump a little bit. So you will have a little bit more variability on exabyte volume, but the CAGR will be about 25%.

Yes. All right. Just going to turn around, see if there's any questions from the audience. Please do raise your hands. Okay. I'll continue.

Nearline capacity, obviously, again, you said you have LTAs, you're extending quite -- and yet your CapEx is just 4% to 6% of your revenue. You want to stick to that range. At what point, I know investors constantly ask, like at what point -- what do you have to see to maybe just step up that CapEx? Or do you see -- again, whether it's transitions to higher terabytes, I don't know if that causes a step-up in that sort of CapEx as a percentage of revenue?

No, I would say we see a strong demand already today. Already demand is above supply. I would say what this industry needs is 2 things. One is when you are on HAMR, especially the heads, they require a little bit longer cycle time. Basically a head is produced on a wafer. This wafer has to go through all the PMR equipment like before and then has to go through some specific HAMR equipment. So cycle time is longer. So if you want to extract the same volume, we need a little bit more space. But we need to start more wafer and get to the same output.

So of course, a little bit more space is part of what we need. And especially on the PMR, you need a little bit more head and more media because you grow through more components, not through technology. So this is what we do. We increase the number of heads, we increase the number of media. We increase a little bit the space that we need. Everything in our CapEx range, so between 4% and 6%. So it's very, very reasonable. And all this, when you put all of this together, you will see the same number of units, but about 25% more exabytes.

Okay. And then just on gross margins, I mean, again, very strong incremental margins. On top of that, you have OpEx leverage, of course, in the model as well. When you think about the margin expansion, I know there's various variables. There's pricing, cost per terabyte declines, there is some other on the OpEx side as well that you have flexibility in terms of OpEx leverage. So help investors understand like which are the biggest drivers on that margin expansion story.

Yes, I would say pricing -- of course, pricing and costs are both very important. I would say, if you look at our last quarter, I think we had a very good performance on both, maybe even more on pricing than cost.

I'd say, we were discussing before, every quarter is a bit different. We had a huge increase in revenue in the June quarter. That was coming also from a very good volume. That was coming from our 40 terabytes, it was a new product. You go into September, we guided another very significant increase in revenue and in profitability. Of course, this also means that, in this quarter, we have a good level of new contracts with high volume, so there is a good pricing support in the specific quarter.

We are preparing more 40-terabyte volume coming out because we will have other customers that will be qualified fairly soon. So again, some variability between volume and pricing in terms of revenue, and of course, pricing and cost in terms of what we produce that will generate, finally, the improvement in the gross margins.

Okay. And without talking about necessarily units or CapEx growing, I mean, there's just so many opportunities. We already talked about physical AI, robotic AI, KV caching, of course, AI inferencing demand continues to grow. Just on the OpEx side, I mean, you've kept an OpEx-to-revenue target there. If these things speed up at the pace that we think it could speed up to, how are you thinking about R&D spending and sales -- SG&A spending?

Yes. I think in terms of headcount, we don't see a reason why we should really increase our headcount. I think we have all the talents that we need to work on HAMR. And the business is not changing from a sales perspective. The number of customers are so far the same, so we need more or less the same people to support.

So I think we will be fairly stable in terms of headcount. And OpEx, the majority of the cost is headcount, with some variability in R&D materials because there are some periods where you run a little bit more materials in R&D, some periods could be less. But I would say, for this fiscal year, we guided our OpEx fairly stable at around $300 million per quarter.

So I would say, for the next 3, 4 quarters, we don't see any reason for change. And then if we see future opportunity maybe in technology to do something different, maybe we will take the opportunity. But so far, we don't see it.

Okay. All right. And then these LTAs that you're signing with your customers, they're now extending beyond -- I know some of them are just volume beyond a certain distinct. Maybe how have the volume assumptions changed? Can you help investors understand, within this agreement? They're growing bigger, obviously. But how -- when you talk about how much capacity they need, just within what's written in the agreement, are you guys building a lot more flexibility in these agreements beyond the next 4-quarter period?

No, not a lot of flexibility. No, we try to allocate all what we think we can produce. Of course, we don't allocate more than what we are sure we can produce. So we don't want to run into that situation where we promise something and we cannot deliver. So we promise what we think we can deliver now assuming everything goes well, and we transition in a certain way product after product.

So as we said before, we see more volume, so higher volume for all of the customers that are into those LTAs. This is why we feel comfortable that it's not a short-term cycle. It's actually a very different trend. I would say every time we rediscuss the agreement, our customers are asking for more volume, not less volume. So this gives us even more confidence.

And of course, we try to go as fast as we can to the next product, the next product, the next product. Now when we talk about 3, 4 years out in time, will be a capacity that is way higher than 50 terabytes. So we need to keep all that into consideration and try to commit to a number that we can achieve, that can at least satisfy a good part of the demand that our customers have today.

Okay. And then capital structure, Gianluca, I mean, that's been important for you. You've been working on bringing that net leverage down. I think now you are like well below your target levels, right? So there is a lot of that convert debt that you've been retiring as well. How do you think about that optimal capital structure for Seagate? How much cash do you want to keep? How much do you want to return?

Yes. I think in terms of debt, we are almost done. So all the activities we wanted to do this quarter are at this point already done. There is still one note with high interest rate that we want to address, possibly next quarter. After that, I think we are good. We have -- we will have a very low level of debt. We will generate a very strong free cash flow. We are already generating a very strong free cash flow. And therefore, we will move to know what we have done very consistently in the past, so higher share buyback and still focusing on a good return in terms of the dividend.

Okay. And just remind us, like free cash flow margin, have you shared a target?

We didn't share a target. Our cash flow has improved a lot, especially in the last couple of quarters. I think last quarter was already above the $1 billion, was $1.1 billion. We said, of course, increasing revenue, increasing profitability will result also in higher free cash flow. So we will generate a strong free cash flow this quarter and next.

After that, all the debt that we wanted to repay will be repaid. And so we will -- we are already doing a good level of share buyback this quarter, and we will do also next quarter. But after that, we will probably do even more.

Yes. Fair enough. Right. Let me ask the audience for questions here. I'll continue then.

You have all these other various opportunities. You talked about hyperscalers, enterprise, robotics, physics AI. Which is the one that you are sort of most excited about or maybe investors don't appreciate it. I mean hyperscaler story has been growing. They have been growing. Enterprise seems to be now something that's coming up as well. But when you think about what investors are under-appreciating about the growth opportunity for HDDs and for storage, what are investors missing?

Well, I think everything that is based on video. So robotics, for example, there's a lot of video attached to the robot. So all that part is what will consume more exabytes. So you say, what are you more interested in? It's whatever is a video attached because we consume more exabytes in terms of storage. So autonomous driving, I think, will be huge. Robotics, especially when it's applied into manufacturing, will be huge, or quality control that is based on video.

So all those parts, I think, will be extremely important to our disk. Of course, everything is then linked to AI because then AI improves the result and give back a tangible improvement to people and companies that are using AI. And this is why it's so important to us. It's a benefit for everyone. There is a cost, of course, but it's very important to everyone.

And sovereign AI, I mean, we talk a lot about hyperscalers, where does sovereign AI come into? Is this just another end market that you're addressing through your hyperscaler customer...

Today, yes. I would say there is a lot of sovereign AI. I mean there are some data centers that need to be in a certain country and somehow segregated from external access. But they're still built today by the same big cloud guys. So they contract manufacturers, they're a big data center. And then very often, you see employees are only from a certain country and there is some way to separate that data center from maybe having visibility externally.

But they're still built by the same big public cloud companies. So we sell our disks to them and then they install into those specific data centers.

But in future, it could be different. It could be other companies that specialize only on sovereign data center. But today, I would say, it's mainly through the big public clouds.

Right. And the same for the neoclouds, it seems like that's the end market, but they're fulfilled through the...

Yes. Neocloud today, they basically run the compute application. And they do it for the big public clouds. So we basically have the same customers. We sell storage, they sell compute. And we sell storage directly to the public cloud.

So the neoclouds access the data that is in our disk into the public cloud. They import into their AI data center. They run the application. As a result, gets stored again into the public cloud storage. So the storage is centralized into the public cloud. It's huge and it's centralized.

In future, if data centers want to have different customers, they cannot access the public cloud storage anymore because they don't work for them. So they work for someone else. So they need to have -- they will need to have their storage. So at that point, you could see some of what we -- today we sell to the public cloud, will be sold to the neocloud. And they will start to have their own complete data center, including storage.

I think it's good from a customer diversification. In terms of volume, probably not very different because today it's probably a bigger volume centralized with a certain number of customers. Tomorrow, some of that volume will go to the neoclouds. But will be positive from a customer diversification.

We're up on time, Gianluca, but just maybe why should investors still be holding or buying more rather of Seagate stock?

Absolutely. No, I would say basically what we said at the beginning, demand is very strong. There are a lot of new applications that will drive the need for more storage. There is no replacement in data center for our disk storage. So the only way to store data and use data into a big public cloud or on-prem data center is our disk. So demand will continue to grow.

And we will continue to increase our exabyte, at the level that we discussed. And we are continuing our strategy in terms of pricing and customer support, that has resulted in better revenue and better profitability for already 3 years, and we see this continuing.

All right.

Thank you.

Well, thank you very much.

Welcome to the Seagate Technology Fiscal Fourth Quarter and Fiscal Year 2026 Conference Call. [Operator Instructions] Please note, this event is being recorded.

I would now like to turn the conference over to Shanye Hudson, Senior Vice President, Investor Relations. Please go ahead.

Thank you. Hello, everyone, and welcome to today's call. Joining me are Dave Mosley, Seagate's Chair and Chief Executive Officer; and Gianluca Romano, our Chief Financial Officer.

We've posted our earnings press release and detailed supplemental information for our Q4 and fiscal 2026 year-end results on the Investors section of our website. During today's call, we'll refer to GAAP and non-GAAP measures. Non-GAAP figures are reconciled to GAAP figures in the earnings press release posted on our website and also included on our Form 8-K.

We've not reconciled certain non-GAAP outlook measures because material items that may impact these measures are out of our control and/or cannot be reasonably predicted. Therefore, a reconciliation to the corresponding GAAP measures is not available without unreasonable efforts.

Before we begin, I'd like to remind you that today's call contains forward-looking statements that reflect management's current views and assumptions based on information available to us as of today and should not be relied upon as of any subsequent date. Actual results may differ materially from those contained in or implied by these forward-looking statements as they're subject to risks and uncertainties associated with our business. To learn more about the risks, uncertainties and other factors that may affect our future business results, please refer to the press release issued today and our SEC filings, including our most recent annual report on Form 10-K and quarterly report on Form 10-Q as well as the supplemental information, all of which may be found on the Investors section of our website.

Following our prepared remarks, we'll open the call up for questions to provide all analysts with the opportunity to participate, we thank you in advance for asking one primary question and then reentering the queue. With that, I'll hand the call over to you, Dave.

Thanks, Shaney, and hello, everyone. Seagate delivered a very strong finish to an outstanding fiscal 2026. Our June quarter results outperformed our expectations for both revenue and non-GAAP EPS and and we expanded our non-GAAP gross margin for the 13th consecutive quarter. Our performance led to free cash flow margins of 31%, which totaled more than $1.1 billion, our strongest quarter in over a decade. .

Our impressive fiscal 2026 financial performance was underpinned by the 3 structural growth pillars that I outlined last quarter: sustainable market demand, our differentiated technology road map, and disciplined operational execution.

First, sustainable market demand. As our results show, demand for mass capacity storage is strong and growing. We delivered fiscal year revenue growth of 34%, led by cloud customers' demand for data storage solutions amplified by the adoption of AI-enhanced applications. Given our momentum and the improved visibility we have into demand, we expect fiscal 2027 revenue growth to outpace our performance in fiscal 2026.

Second, we are executing our differentiated technology road map, anchored by our HAMR-based Mosaic platform. HAMR enables us to increase aerial density and store more data on each disc. As a result, we can scale exabyte shipments to meet rising demand in a highly capital-efficient manner to capture more value per drive. Exiting the year, HAMR-based products represented approximately 40% of our nearline exabyte shipment run rate, and we continue to invest in HAMR capabilities to support our mid-20% exabyte growth target while further enhancing profitability and capital efficiency.

Our third pillar centers on translating demand strength and technology advancements into profitable growth. In fiscal '26, we increased non-GAAP gross margin 10 percentage points, grew non-GAAP EPS more than 90% and generated record free cash flow of $3.1 billion. Looking ahead to fiscal '27, we expect to deliver sequential margin and cash generation growth throughout the year. Our confidence is supported by the scale, quality and duration of our data center customer commitments in a strengthening demand environment.

Data center demand now represents approximately 90% of our exabyte shipments. Based on the long-term supply agreements in place today the vast majority of our nearline exabytes are now allocated into calendar 2028. Importantly, we are not seeing customers pull back on planning horizons. As our strategic relationships deepen, many are actively seeking to extend planning horizons through 2029 and beyond, which we believe reflects growing confidence in their own long-term infrastructure needs.

These engagements reinforce our view of demand durability while providing customers greater supply assurance and support for the key technology transitions. We remain disciplined in securing orders from these customers prior to initiating drive production with contracts that define both product configuration and pricing terms covering the entirety of calendar 2027. We continue to execute our value-based pricing strategy, balancing a stronger demand environment with our objective of supporting sustainable, profitable growth over the long term.

Cloud customers remain the largest driver for nearline demand today with 3 years of sequential quarterly exabyte growth, no evidence of a slowdown as AI adoption now builds on demand for traditional data-intensive applications, including video. We continue to benefit from cloud infrastructure deployments, which fuel the need for scalable, cost-efficient and reliable storage. At the same time, we believe storage demand will improve durable through investment cycles.

First, new data is constantly being created across existing cloud and enterprise infrastructure. And second, customers are retaining and reusing more of that data over time as its value extends beyond its initial use. AI is reinforcing these trends and illustrating how data is not only growing it is compounding. With the transition from AI model training to inference to Agentic applications, more data is generated and retained for historical context, compliance and future reviews.

As these data center environments become larger and more complex, customers must balance performance, energy consumption and cost across distributed infrastructures. Cloud providers have long addressed these challenges through tiered storage architectures that combine high-performance memory and SSDs with mass capacity hard drives to optimize performance and economics at scale.

Our recent white paper with SK Hynix illustrates the importance of tiered storage for inference and agentic AI workloads, which show a direct benefit to hard drive storage. These workloads rely on persistent context across user interactions and key value or KV Cache is used to retain and reduce that context efficiently. KV Cache can expand significantly as the number of users increase, and interactions become longer and more sophisticated. Our research found that by extending KB Cache data across memory, SSD and hard drive tiers, organizations can retain more context and avoid recomputing previously generated data.

This drives the need for increased hard drive storage and reduces GPU usage during the most compute-intensive phases of an agentic application. As a result, GPU resources are available for additional revenue-generating workloads. Additionally, we are now seeing the relevance of tiered storage extend beyond large cloud data centers into enterprise deployments. As enterprises increasingly operate across public cloud, private cloud and on-prem environments, they must manage growing volumes of structured and unstructured data while facing similar performance, cost and energy consumption trade-offs that hyperscalers have addressed for years.

We see this broadening of demand in our business. Enterprise nearline revenue increased for a fifth consecutive quarter in June and we are engaging more frequently with Neo cloud operators and leading model developers. As our data management needs scale, these providers are starting to adopt modern tier storage architectures where hard drives provide a trusted mass capacity foundation.

Looking ahead, we believe physical AI applications, such as robotics and autonomous vehicles will drive the next step function expansion in data creation and retention at the edge. These applications rely on world models trained on millions of hours of historical and synthetic video content in order to understand and reliably interact with the physical world. Taken together, these trends reinforce a structural change in storage demand.

Applications are creating, retaining and reusing more data across cloud and enterprise environments than ever before, extending the role of mass capacity storage across modern tiered architectures and creating additional opportunities for Seagate over time.

Our technology road map plays an integral role in Seagate's ability to capitalize on growing storage demand. Advancing aerial density is our North Star. We believe that increasing the amount of data stored on every desk we produce is the fastest, most capital-efficient path to supporting long-term exabyte growth while maintaining relatively stable hard drive unit output. Our expertise across material science, precision manufacturing, advanced photonics and nanoscale wafer production has enabled us to pioneer HAMR technology and the Mosaic platform, which has increased storage density per disc and per drive. We continue to push the boundaries of innovation as demonstrated by our vertically integrated laser manufacturing capabilities, which yielded tens of millions of edge-emitting lasers last quarter.

Our team's achievements underscore the scale, maturity and supply chain resilience behind our Mosaic platform. These innovations are improving the total cost of ownership for our customers. while expanding our exabyte output and enhancing efficiency across our operations and supply chain. We ended fiscal 2026 on schedule with our HAMR based product ramp. Our Mosaic 3 products are now qualified and operating in production environments across all major cloud customers.

Our second-generation Mosaic core platform capable of supporting up to 44 terabytes per drive continues to ramp with the 2 largest global CSPs and additional customer qualifications are underway. We expect to achieve our next ramp milestone by exiting calendar '26 with 50% of our HAMR exabytes on our Mosaic 4 platform. Looking further ahead, Mosaic 5, our 5-plus terabyte for disk platform remains on track for qualification shipments in late calendar 2027.

Wrapping up, we delivered across the board in fiscal 2026 with each quarter building on the momentum of the last, and we expect that momentum to continue through fiscal 2027. The growth in data creation, retention and utilization continues to elevate the importance of hard drive storage and modern data architectures. Together, our demand outlook, differentiated technology strategy and disciplined execution position Seagate to capture the significant opportunities ahead and create long-term value for our stakeholders.

I'll close by thanking our global team for another year of outstanding execution. I'd also like to thank our customers, suppliers, partners and shareholders for their continued support and trust in the company.

With that, I'll turn it over to Gianluca.

Thank you, Dave. We capped fiscal 2026 delivering strong sequential double-digit top and bottom line growth in the June quarter, supported by disciplined operational execution and both revenue and gross margin expansion across every end market we serve. June quarter revenue was $3.6 billion, up 17% sequentially and up 48% year-over-year exceeding the high end of our guidance range. We achieved record profitability levels across gross margin, operating margin and earnings per share. Non-GAAP gross margin came in at 52.7%, up 570 basis points sequentially. Non-GAAP operating margin increased 710 basis points sequentially to 44.6% and non-GAAP EPS was $5.71, up 39% quarter-over-quarter and 121% year-over-year, exceeding the high end of our guidance range by a wide margin.

As Dave noted earlier, we generated free cash flow of more than $1.1 billion, rounding out our best quarterly performance in over a decade. Sustained data center demand continue to outpace broader company growth. In the June quarter, we shipped a total of 218 exabytes, up 34% year-over-year, with data centers representing 89% of the total. We shipped 195 exabytes into the data center market, up 11% sequentially and 43% year-over-year. With data center revenue coming in at $2.9 billion, up 17% sequentially and 57% year-over-year.

Global Cloud customers are driving the vast majority of data center revenue and exabyte demand. At the same time, demand trends in the enterprise OEM data center market has strengthened, reflecting growing storage requirements across a broader set of customers and workloads, many of which Dave highlighted earlier.

In the June quarter, we delivered strong double-digit year-over-year growth across both revenue and exabyte shipments into the enterprise OEM markets. To support long-term demand growth, we continue to expand the deployment of HAMR technology across our product portfolio. Our goal is to transition an increasing portion of production to HAMR-based products, first to address cloud customers and over time to broaden adoption across enterprise deployments.

As we make this transition, we are strategically investing in additional tools and technology to support the manufacturing of our HAMR products. These investments enable us to maintain relatively stable drive unit output as customers mix up to higher capacity drives and manufacturing cycle time increase. We believe this action will enable us to deliver nearline exabyte growth in the mid-20% range over the next few years.

Beyond the data center, our IoT market made up 19% of revenue at $697 million, up 14% sequentially and 20% year-over-year. Due in part to ongoing tight supply conditions and increasing NAND pricing.

Moving on to the rest of the income statement. Non-GAAP gross profit increased significantly to $1.9 billion, up 31% quarter-over-quarter and more than doubling year-over-year. Non-GAAP gross margin expanded to 52.7%, up from 47% in the prior period. This improvement reflects continued execution of our long-term pricing strategy and a stronger product mix. We expect this trend to remain favorable, underpinned by strong demand.

Non-GAAP operating expenses were $293 million or 80% of revenue, reflecting our discipline in cost management. Non-GAAP operating profit increased 39% sequentially to $1.6 billion, representing 44.6% of revenue and underscoring the scalability of our financial model. Continued areal density innovation, supply discipline and pricing strategy execution.

In the June quarter, other income and expenses were $58 million and we project [indiscernible] to decrease further in the September quarter to approximately $45 million, reflecting the benefit from lower interest expense as we continue to reduce our outstanding debt balance.

Non-GAAP net income grew to $1.3 billion with corresponding non-GAAP EPS of $5.71 per share. Based on tax expense of $242 million and a diluted share count of approximately 231 million shares, including the net impact of our 2028 convertible notes.

Turning now to the cash flow and the balance sheet. In the June quarter, we invested $187 million of capital expenditures with total fiscal 2026 CapEx, representing 4.7% of revenue. Looking ahead, we expect capital expenditures for fiscal 2027 will remain well within our target range of 4% to 6% of revenue. Free cash flow generation expanded to $1.1 billion, up 17% from the prior quarter. We expect cash generation to further improve throughout fiscal 2027 supported by sustained demand, operational efficiencies and CapEx investment discipline.

During the June quarter, we returned approximately $283 million to shareholders through dividends and share repurchases. Strengthening the balance sheet was a key objective for fiscal 2026 and we delivered on our plans. We ended the year with cash and cash equivalents of $1.7 billion and strong liquidity of $3 billion, including our undrawn revolving credit facility.

Our gross debt balance was approximately $3.6 billion exiting fiscal '26, down $1.4 billion year-over-year, including $300 million that we retired in the June quarter. Our resulting net leverage ratio improved to 0.4x based on adjusted EBITDA of $1.7 billion for the June quarter, up 37% quarter-over-quarter and 142% year-over-year.

During the September quarter, we are retiring an additional $1.2 billion in debt. We have already extinguished $1 billion in high-yield senior notes in July and plan to retire the remaining balance on our convertible notes in September.

Turning now to the September quarter outlook. Visibility from our BTO model reinforces our confidence in sustained demand for high-capacity nearline drives as AI adoption accelerates. We see continued revenue and profitability expansion in the September quarter, supported by our Mosaic rent and pricing strategy. We expect September quarter revenue to be in the range of $4.1 billion, plus or minus $100 million, which represents a 56% year-over-year improvement at the midpoint.

Non-GAAP operating expenses are expected to be approximately $300 million. Based on the midpoint of our revenue guidance, non-GAAP operating margin is expected to be around 50%. Non-GAAP EPS is expected to be $7.30 plus or minus $0.20, [indiscernible] tax rate of about 16% and non-GAAP diluted share count of 231 million shares, including estimated dilution from our 2028 convertible notes of approximately 2 million shares.

To close, Seagate's financial results and outlook demonstrate our ability to deliver profitable growth, expand margins and drive significant cash generation. We remain confident in delivering sequential revenue growth and margin expansion through fiscal 2027, while creating long-term value for customers and shareholders.

Operator, let's open the call up for questions.

[Operator Instructions] Our first question comes from Aaron Rakers with Wells Fargo.

I want to dig a little bit deeper into the gross margin. I guess given the guidance that you've outlined it looks like your guide is implying like a mid-57% or so gross margin into this next quarter. So I guess my question is, one, is that kind of the guidance that you're providing? And two, how do you think about the cost down execution as we move through Mosaic 3 to Mosaic 4. you've been operating at like the mid-teens kind of cost down per year on a per terabyte basis. Do you think that's sustainable? Or how should we think about modeling that over the longer term?

Aaron, I'll let Gianluca answer this quantitatively, but the way we're thinking about these product transitions is -- and I think you know this well, we have to actually put our factories on pause to go through the product and transition a little bit. So as we're moving product from 3 terabytes to 4 terabytes of 5 terabytes and then there's yield issues. And as we out-execute our plan, what happens is we have opportunity for costs to drive better costs than we thought. So that answers the second part of your question. Gianluca? .

Yes. Aaron. So yes, very good result, I'll say, on gross margin already in fiscal Q4. So strong improvement sequentially and we are guiding out again. So our pricing strategy is continuing. As we have discussed now for several quarters. I would say we have updated this strategy more than 12, 13 quarters ago. So we are continuing in that direction. Every quarter is a bit different, but the strategy is the same. The mix is helping. We are moving more and more into the high capacity nearline product.

You have seen another strong increase in nearline exabyte in the quarter. So everything is continuing in the same direction that we have driven the company for many quarters or they've discussed about the cost, of course, moving the mix from 3 terabyte to 4 terabyte per disk is, of course, giving us another boost in terms of profitability.

The next question is from Ben Reitzes with Melius.

It's great to be covering you again. I wanted to talk about 2 longer term, to ask you about 2 longer-term demand drivers potentially. I mean, this key value cash use of the HDD peer at hyperscalers. How much is that helping right now? Is it on the comp? And do you expect it to kind of ease into your exabytes? And how do we look at that? And then -- and when do you think physical AI really starts helping your exabytes as well?

Right. Ben, I think that's good. Both very, very early days. I would say that the agentic flows that we're seeing are actually what -- that's the reason we pointed to the KV Cache discussion. I think the key word here is context when you set up these agents, you really need to give them context and sometimes that's a very broad set of rules across your business or your problem set or whatever. And as you do that, then you don't want to have to redo that context every time. You don't want to have to recompute all that context every time. So that's what's driving storage, but still very early days. .

Fiscal AI, we are quite excited about. I think I hear a lot about robotics. We've all seen autonomous vehicles, people focus very much on the end product, the robot, if you will, but I actually think it's a lot more of a data play. These are -- these robust have sensors on them, they're sensor networks in order for them to learn. The data actually comes back up into a local cloud or a bigger cloud. And so when people say physical AI to me, I think it's a lot more about the data, the data processing, what kind of learning you're getting from that and exactly to your question, how much you have to store to make sure you have that context long term. So we think these are both great opportunities.

In particular, the fiscal AI stuff is largely more about video. So it's a very unstructured type of data coming. It's not like the days of old where you had spreadsheets or tech list to fill out that were complete structured data. This is very unstructured data that the machines are learning from and they might want to learn again and again and again, but you don't want to have to repromote that into the memory tier. So we think that's a great opportunity for us.

The next question is from Erik Woodring with Morgan Stanley.

And congrats on the really nice results and guide. Gianluca, for a number of quarters, you've been quite steadfast that price per exabyte growth would be kind of mid- to high single digits year-over-year. You just reported 10% year-over-year price per exabyte growth in June. I think the September quarter guide implies pricing growth closer to maybe 20% year-over-year or even above that. Can you maybe just provide an update for us on how we should be thinking about pricing looking forward? Why this trend we're seeing in the September quarter shouldn't sustain or maybe even accelerate just given supply-demand imbalance, customer demand strength, delivering more value to customers, et cetera. .

Yes, correct. As I said before, it's not that we are changing our strategy. But for sure, the gap between supply and demand is now a little bit bigger than a few quarters ago. And our volume was a little bit higher in fiscal Q4. Now we think it can be maybe a little bit of output available in fiscal Q1. And of course, we are pricing that increased output at a very good price right now.

So I'd say not really change in our strategy, but a very good execution and with demand being particularly strong right now, we take a little bit more pricing benefit. Now of course, every quarter is different. We will see in the following few quarters how the pricing will evolve, but I think we were very clear both Dave and I -- in our prepared remarks, we see every quarter revenue to improve. And every quarter, gross margin and profitability in general to increase. So of course, pricing is a part of the sequential improvement through the fiscal year.

The next question is from Asiya Merchant with Citigroup.

Great -- great results here. And if I may, just on the CAGR growth, I think you guys reiterated sort of this mid 20%. You guys obviously have been executing too much greater than that. I think I heard about some investments that you're doing. Just help us understand like this above 30% exabyte growth rate, could that sustain as you enter fiscal '27 especially as you're migrating more towards your second-generation HAMR and then you're ramping into -- further out into your Mosaic 5.0. So if you could just help us understand why exit by CAGR could or could not sustain at this 30% as we look into fiscal '27?

Thanks, Asiya. We -- as we've said before, we're not really increasing the box count. We are working really hard to get the heads and media inside the boxes to be able to go up in the technology capability to get exabytes out. And exactly to your point, what's the ultimate CAGR? It's how fast we can do that, how successfully we can do that. We are going through product transitions. I mentioned this earlier. So as you do that, there's a little bit of inefficiency in your factories. But long term, you actually get many, many more exabytes out as we go from 3 to 4 to 5. And that's the way we're focused what you've seen so far is the transition largely the 3, we are ramping the 4 right now and the 5 is coming and so on.

How we play that is depending upon how we see end customer demand what the qualification schedules like for those customers, and we need visibility because that's 3 to 4 quarters out from when we do wafer start. But also, we get a little bit better yields. We can add a few more exabytes here and there, and the team has been doing fantastic on that front. So all of these dynamics are how we actually have to predict the next few years. And that's one of the reasons we say mid-20s could we execute a little bit beter not conceivable but there's a lot of invention required still.

Next question is from C.J. Muse with Cantor Fitzgerald.

Follow-up on pricing. I'm curious if you could kind of speak to like-for-like versus the benefit of newer products. And then moreover, if you could speak to how we should be thinking about contracts rolling off. renegotiating of existing contracts and how that is impacting the relative kind of year-over-year pricing particularly as it relates to that strong 20-plus percent number embedded in the September guide? And how to think about the moving parts into December and beyond.

Thanks, CJ. We are -- as we said in the prepared remarks, we're trying to be predictable for at least the next year because that's what we have visibility to in our factories. And that's when we know the exact configuration and we determine pricing with our customers. But as various customers are rolling through those periods, interesting things happen. First, there's product qualification, like-for-like is kind of tough because we are moving products so quickly through transitionary periods that benefits us and also benefits them. If they're building a data center, they get a better TCO out of those products. And then there's other architectural reasons that may slow down or speed up their ability to ingest these things. So it is fairly complicated. .

What we're seeing over time is not only what we lock in for that period of a year is we also see our ability to execute a little bit better and that's usually 2 or 3 quarters out. That's not in quarter, but we execute a little bit better. We have more exabytes to give and we determine how hungry the market really is for those exabytes. And usually, they'll pay more than that contract price, if you will, for those exabytes. And so that's why you see these step functions. And therefore, I think we're all confident about the demand that we're seeing, both ourselves and our customers, but also they're voting every day with that by signing up to pricing that is even higher than some of the contractual stuff that we had done together.

Yes. CJ, you were asking about the new orders that we are negotiating in the new LTAs that we are negotiating. The trend is the same as the past. So we see strong demand, and of course, opportunity for us to continue to push on our pricing and pricing strategy and continuing in the same direction. Every quarter is a bit different, depends how many contract, we negotiate what is the volume for different customers, what is upside volume, if any, what is available and what is the price. So it's not a straight line. But I would say the trend is clear. I'd say we are really performing the same way for more than 3 years at this point is right now, it's particularly strong in terms of supply demand and pricing, but it's going to continue, we're already discussing about the next 2 or 3 quarters going in the same direction. .

Next question is from Tom O'Malley with Barclays. .

I just had a 2-parter here. So you guys have previously said 70% of nearline expects will be hammered by June of 2027. How are you tracking to that? And then I saw in the preamble that you specifically called out as you're transitioning there, you're strategically investing additional tools, technology and manufacturing. Is that just something that you would normally put the frame it stood out a little bit on me. Could you maybe be more specific on what you're investing in there to help you get to that percentage of the total mix.

Yes. Tom, we're still pushing HAMR well, and it's reacting exactly what we thought it would a couple of years ago. There we have -- as time has Marched on, we have pushed maybe PMR a little bit harder than we thought. So transitionary I think we're still on the -- generally the same trends and everything is going well. Relative to the investment, most of the tools that we're investing in are directly contributing to those heads and media that are driving those technology transitions. I'm very happy with how the team has executed on that front, 3 to 4 to 5, like we've talked about before. very optimistic about it. And I think, ultimately, HAMR is going to completely take over the portfolio because of it. we're learning more and more about the tools all the time, and that's part of how we do aerial density development is to get the latest tools on, learn how to run them and see what we can do with them. And I would say long term, I think there's probably more favorability for aerial density than I had thought a couple of years ago.

Yes. In terms of percentage of exabytes, we have actually just achieved our first milestone, which was to achieve 40% of nearline exabyte sold on Amerive by June. So we just did that. And so I'll say we are we are on track to achieve the future goals. Actually, on the new investment, there is, of course, a big difference between components and are these drive units. Now if you -- for example, if you look at our last year and if you look at the number of disk and the number of heads in sizable. They probably grew between 15% and 20%, and the units were absolutely flat.

So there is always a mix up of drives going more and more into the near line and going more and more close to the 10 years and 20. But of course, there is strong shift year-over-year-over-year. This happened for the last 10, 20 years. So it's normal that even with flat and these drive units, we need to increase as a media through time. That's the best normal part of the business.

Sorry, one more thing on that point. because I think a lot of people still don't understand this. So the thing that's actually driving factory complexity is not just drive numbers over heads numbers or media numbers, it's actually the product transition.

These new products, say, 4 terabytes going to 5 terabytes -- there will be more time in the tools, more time through the tools. Sometimes it has to touch the tool multiple times. The factory complexity is what's driving a lot of the investment that you made reference at the time.

The next question is from Mark Newman with Bernstein.

Congrats on another good quarter. Lots of questions on the pricing. So I wanted to talk more about the technology and the cost. Could you update us on the HAMR portion of your shipments. I think you've guided before 40% exiting the fiscal year on HAM and 50% exiting calendar year '26. Is -- are we on target for that or tracking ahead of that guidance you can say on that? And then also Mosaic 4, you still ramping to global CSPs. So is that going to I think you said it's a very small portion of revenue in the previous quarter, but it's going to become more significant in the first -- in the September quarter. Just want to clarify that.

And then given all that, -- should we expect that cost declines should be potentially accelerating given this upcoming ramp of MSCI and now you have almost now the shock is part of the score the hammer adoption. So just don't want to see if anything you can comment on that would be really appreciated.

Thanks, Mark. Yes, we're on target for all the metrics that you talked about. I would say relative to Mosaic 4, it was pretty consequential last quarter even, and it's ramping quite nicely. We intentionally have maybe throttled the Mosaic 4 ramp because of qualification cycles and everything else other customers as they qualify, they're full of the last generation product as well. So things are fairly complicated in the supply chain. -- but Mosaic 4 is quite successful out in the market. It will continue to ramp over the course of this next fiscal year.

Yes. No, just to clarify. No, we start shipping Mosaic 4 in March. So March quarter volume was pretty low, but June quarter was a good ramp-up is also a strong contributor to our financial performance and will be even better in the September quarter. And then of course, we are already all focusing on the next steps that will be the 5 terabyte per disk and the 50 terabyte drive in next calendar year.

The next question is from Wamsi Mohan with Bank of America. .

I was wondering if you could maybe just clarify on how much of your fiscal '27, fiscal '28 exabyte view is locked in via build-to-order versus maybe not under LTAs. And as you think about the pricing uplift in September, part of that is coming from one of your initial HAMR customers that had more favorable pricing rolling off. So should we still expect the price momentum to continue at those levels for the rest of for the rest of the fiscal year? I know you said that you would see revenue and margin increase every quarter, but any thoughts on sort of the magnitude of either sequential or year-on-year given those comments around the initial customer? .

Yes, Wamsi, you're right. On the second point, I think it's important to realize that as we do roll through, there's different phases that different customers are under and that renegotiation occurs. We are pretty predictable, I think, through FY '27. So we have a good line of sight, but we are also getting a little bit more product out as we continue these product ramps because we're working the yields and scrap really aggressively on new products. So to the extent that we can, since there's such strong exabyte demand out there, we'll offer that to people out in the market that are showing us that opportunity.

Yes, relatively to that particular customer, the volume that was sold at a preferential price in June was minimal. So September, we will not have any. So there is a little bit of positive impact from there. But it's not the major reason why pricing is a little bit better in September than June, of course, it's more overall demand and those are customers that are chasing a bit more volume right now. .

The next question is from Joseph Cardoso with JPMorgan.

Maybe just a clarification from my end, I'm not really seeing or trying to increase the box count here. Does that encompass the visibility into nearline allocations into 2028 and the planning that you're seeing extending into 2019? And then how should we think about visibility into pricing in those outer years as well?

Thanks, Joseph. Yes. I mean you're right. Thanks for the question because many investors are new to stock, so we'll try to explain this again. So our strategy coming out of the last down cycle has been to keep the number of drives flat, and we're still on that path. Inside the drives, however, there's all these critical complements heads and media, and that is rising slightly. I think John Lukamade reference to this before. So that puts a strain on our internal heads and media fabs, which are under our control. And we're going through these aggressive transitions, that's the big story. The big story is the process content of manufacturing complexity as we move from 3 terabytes to 4 terabytes to 5 terabytes the routes get more complicated and the technology transitions are putting a large strain on these -- on our internal complements. .

But with the curves that we're on, we believe that this is the best way to bring more exabytes out in the world is to stay focused this way. And so that's our strategy.

Next question is from Karl Ackerman with BNP Paribas.

Dave, you spoke about qualifications on Mosaic among hyperscalers. But how should we think about CA growing exposure to Neoclouds and foundational model companies. I was hoping you could parse between demand from traditional hyperscale neoCloud and maybe on-prem and list in your September outlook.

Thanks, Karl. Yes, I -- 2 years ago, I would have said NeoClad is probably largely compute-based -- but we're starting to see that even some of the large neo clouds, they need a lot of data coming at them. And where did they get that data in the past, they might have got that from traditional hyperscalers but there are some places where neoClouds are saying, I need instances close to me. By the way, I do not think that's necessarily competitive with the hyperscalers. I mean, there are so many different applications specific reasons for people to have an exabyte or 2 sitting around, especially training various types of training for applications.

So -- we are starting to have exactly the dialogues that you talked about. Everyone knows the hyperscaler architectures and the efficiency of the hyperscale architecture because they know that they want that same efficiency. And in some cases, we're talking about systems level discussions with these customers. In other cases, we're just talking about drive. They also -- since they're going to be running this gear for a long time, they want to be on the cutting edge of technology transition. Sometimes that's hard because of feature sets and they may not be as robust as some of the other people said they need help with that, and it is a very complicated qualification space.

The next question is from Amit Daryanani with Evercore.

I guess, Dave, as you look at the LTAs and the visibility on exabyte demand that you have for '28 and net income to '29. Can you meet all the exabyte demand that's [indiscernible] '28, '29 entirely through air density gains -- and maybe just touch on how secure do you think your own upstream supply chain is for specialized components, especially as HAMR to scale up. I'd love to just kind of understand the component side from your perspective, -- and John, look, I'd love to understand where you're going to get to 80% gross margins, if you want to oblige and answer that.

Thanks, Amit. So on the supply chain piece, working with our supply partners that have been through a lot, like we were a few years ago, as you know. We're making sure that everybody is kind of lined up and it's well orchestrated. That's an important part of our supply chain. We cannot have people kind of individually doing investments and then someone else not doing the investment, it not be well orchestrated because that drives cost the wrong way. How confident are we in demand long term, very confident.

I do think that there's A lot of people out trying to understand all these new applications that are coming at us and saying, what does that mean for the storage tier. I also think that there's well traction to applications already in the market, whether they're pre AI, which was huge, right? And some of the video applications we've talked about before or whether they're now AI enabled that are driving the storage tier even higher.

And so the forecasting, especially for some of those new applications is relatively harder. I think there's a lot of optimism around it. But I also think that the existing data sphere, if you will, inside of these called service providers is growing at a certain large clip anyway. I do not think that probably our aerial density transitions are going to be sufficient. But I do think that they are strong, and I think that allows people to plan their business as well, and that's 1 of the reasons we're having good conversations out in that time frame.

On the gross margin, Amit, I would say our incremental gross margin has been very strong for the last several quarters. Overall, gross margin is improving sequentially very well. So we don't have a specific target. We will continue to improve based on the business situation and -- and we know already that for the rest of the fiscal year, we will have sequential improvement every quarter. And then we will see at a certain point where we are, but we don't have a specific number that we are trying to achieve. .

The next question is from Steven Fox with Fox Advisors.

I just wanted to ask a free cash flow question, if I could. So off of a 10-year high. Like can you talk about, I guess, the dynamics that drive from just a manufacturing standpoint, a higher free cash flow margin in the future? Because as the aerial density increases, the increases as a percentage are smaller. So I don't know if that helps or Dave, to your other point about passing through the same equipment, makes it more capital intensive, et cetera. And just as a follow-up to that, just can you maybe talk about where -- remind us where you want to get debt levels to and when buybacks could start?

Yes. Thanks, Steve. So we'll stay within our capital model. I mean we talked about 5% -- 4% to 6% of revenue is our range for CapEx. The tools we're buying are modern tools, and we're refreshing part of the fleet, even with that 5% of CapEx to 5% of revenue as CapEx. We're refreshing the fleet and doing quite well with that. So we'll turn all that into aerial density, and that's what we're really excited about.

I think relative to free cash flow, from an OpEx perspective, we don't see the need to add a bunch of OpEx. We feel our team is doing really well on all the innovation vectors, whether it's a sort of mechanical vector or it's quantum device vector and the recording fabs or lasers now or whatever. I mean, we think that the team is funded well and doing well. And so we think we have visibility to continue aerial density without raising CapEx too much. So I think all of that translates into free cash flow that's growing like we talked about.

On the debt part, we ended our fiscal '26 with $3.6 billion in debt. That is already a huge reduction from about $5 billion that we had at the beginning of the fiscal year. we will reduce debt even more during the quarter, actually already done a good step in the month of July, but we will probably end fiscal Q1 at $2.4 billion in debt. And we still have 1 note that has a fairly high interest rate that I would like to address in the near future, maybe next quarter, maybe the following quarter. But we are doing already more share buyback than what we have done in the prior quarter. So this quarter, we are having -- we are executing a higher level of share buyback, and we will continue in the next several quarters. .

The next question is from Vijay Rakesh with Mizuho.

Congratulations, Dave and Luca. Just 2 quick questions. One, when you look at the -- how does it drive the nearline attach rate? Is there a way to look at how the attach rate has changed on the GPU ASIC side per WAC with agent AI or KV Cache picking up? And how does -- how is that trend looking this year versus last year, let's say, when there was no agentic AI. And then look, on the margin side, should we expect margins to kind of get to the 60% plus? Or if you can give us what the incremental margins are on HAMR 4 versus prior? Or is there a way to look at mix of HAMR 4, I guess?

I'll take the attach rate discussion. I know there are people out there in the world trying to model this, and it's a noble effort, but I think it all comes down to application space. So there are certain applications where you may need a lot more context and there are certain applications where maybe you don't need as much. And so depending on the application pickup on this Agentic AI, we talked about KV Cache in the prepared remarks, it could be a little. I think we're still trying to factor that in. And that's some of the stuff that -- as we get into modeling '29 and '30 and beyond, I think we're going to have to work study with our customers to watch those applications carefully, pretty excited about it. And that's not the discussion about physical AI, either that's just on what I would call more enterprise-type applications.

Yes. On the margin, we -- last quarter, we were at almost 53%. We are guiding higher in September. Our incremental gross margin is well above the 60% that you were indicating. So I'm not guiding for the future, but now the trend is, of course, to have a stronger and stronger gross margin and we will see what we will achieve in the next few quarters. But as you know, we are guiding something that is not too far from that number already in September.

The next question is from Ananda Baruah with Loop Capital.

Dave, maybe just sort of dovetailing off your comments about application type. Is there an easy way or a simple way to think about currently what you guys see as the more prominent applications driving demand right now. And before you get to physical, maybe how you see those meaningful application types manifesting over the next couple of years?

Thanks, Ananda. Yes, the way I think about it, and I've been around for a long time, we know so you have to be careful with me. But taking small blocks of text, whether it's forms that somebody filled out or an ERP or something like that in the -- and I say small blocks of text kind of jokingly because that could still be terabytes worth of tax. I think that's probably not what we're talking about.

But when you start to have a lot of unstructured data like video data or multiple sources of unstructured data, sometimes may be sensor data -- it may actually still be taxed, but it's just necessarily coming in from all kinds of different sources. Those are the applications that I think are going to require a lot more processing power, and you don't want to redo that processing power over and over again.

So -- this is happening in enterprises. It's not just happening in hyperscalers. It's happening at the extreme edge as well. And I think these trends are very favorable for us before we ever get into something like physical AI.

This concludes our question-and-answer session. I would like to turn the conference back over to management for any closing remarks.

Thanks, Gary, and thanks, everyone, for joining us today. Fiscal 2026 was an outstanding year for Seagate, reflecting strong execution by our global team and deep engagement with our customers. As we move into fiscal 2027, we are well positioned to address the opportunities in front of us. We remain focused on executing our technology road map, capturing profitable revenue growth and delivering long-term value creation for all of our stakeholders. We look forward to updating our progress with you in the quarters ahead.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

Thanks for joining us here again, day 1 of Bank of America Global Technology Conference. I see a lot of familiar faces. Glad you could all make it. I'm Wamsi Mohan. I cover IT hardware and supply chain here at Bank of America. Today, we're delighted to welcome Seagate to our fireside. We have EVP and CFO, not ex-VP, but EVP and CFO, Gianluca Romano. So Gianluca, welcome. Thank you so much for taking the time to be with us here today.

Thank you for inviting us.

I feel like every year when we talk, there is something new to talk about. But at the same time, I feel like your message has been very consistent. And in some ways, that consistency has really played out very well for you guys.

Maybe to start, right, I think one of the questions that we get often is just on sustainability of trends and where we are in the cycle because this industry historically has been quite cyclical. So I would love to get your perspective on -- is this an elongated cycle? How do you think about it? You have a lot of things that have changed from a historical perspective in terms of market exposure, in terms of your customer base, a lot of things have changed. So maybe just to put some perspective around where we are in the cycle question.

Yes, very good question. Before I answer, let me remind everyone that I will be making forward-looking statements today, and you can learn more about the risk associated with those statements on our website.

Well, I would say, for sure, a different cycle than what we have seen in the past is now 13 quarters of continued growth. Every quarter, we increased revenue. Every quarter, we have improved profitability. And we just discussed at our earnings release based on the orders that we have in place already, we see this happening continuously and sequentially for the next 4 to 5 quarters. That is exactly the time of our purchase orders.

So I would say the trend is not changing. Demand is probably higher than what we were expecting a year ago or 6 months ago. So maybe that is a little bit of a change in the trend, but still same trend in terms of improvement for both pricing and profitability and revenue. So I'm not saying that we will not have a cycle anymore in this industry because, of course, there are a lot of factors that could drive a cycle, probably mainly externally to our business, more macroeconomic cycle, not data storage cycle. The world is going more and more into digital applications that are driven by data that need a lot of storage. But of course, you can have a macroeconomic cycle.

So we don't see it happening right now based on the orders that we have. Every quarter, we have bigger orders with more revenue, more profit. And so far, we don't see it.

Yes. So maybe just on that visibility point, right? So you guys have obviously been working with your customer base to get improved visibility. How would you characterize that? What are some of the maybe parameters of that visibility in terms of capacity, in terms of the SKUs you're shipping, in terms of how much forward-looking capacity people are committing to and the variability of that?

So for the next 4 to 5 quarters, we have orders in place and an order has a precise mix, precise exabyte volume, precise price and time to deliver. So this is why now we have high confidence on what is going to happen into that period of time. When you go longer, so year 2, year 3, customers are very interested and they need to know how many exabytes we can allocate to them because they decide today what data centers they will need 2 years from now. They need to start today to build it and then get all the infrastructure in place. So they need to know what will be the exabyte that we can allocate to them.

So the discussion is a bit different. It's not so precise in terms of the mix, in terms of the price and the exact quarter of delivery. It's more an overall exabyte number. So we had that discussion with basically all our big customers. And then when we get closer to those 3, 4, 5 quarters, we basically translate this exabyte commitment into a very precise order. At that point, we know what is our ramp, what is the product that they are qualified on, so we can have a detailed look.

Okay. So if you think about pricing, and we get this question often, right, like we're shipping to what seems like a higher demand level than maybe what people anticipated even 6, 9 months ago. And in a lot of adjacent industries, I would say, right, like if you think about memory, if you think about optical in these data center exposed places, pricing has gone up a lot. And by a lot, I mean, like 100% quarter-on-quarter, right?

So when you think about that in relation to what is happening in the hard disk drive world, we are seeing improved pricing. It's much more measured. How are you thinking about the pricing trajectory as you think about the next several years? And why is it not going up at the rate at which some of these other ones are moving? Is that by design? Is it like could it if you wanted to?

Short answer is yes. I think this industry is now very disciplined. It's very disciplined in how we add exabyte capacity and not adding units. It's disciplined on how we approach customers on continuous improvement on pricing, but not in an aggressive way, not in a way that cannot be sustained or that is strongly impacting our customers even as a surprise without giving them the time to absorb that. So we have been way more disciplined and the result has been fantastic. So I don't see any reason why we should change our strategy in terms of pricing, in terms of exabyte growth without adding units and focusing all our energy on the product road map and moving up into capacity per unit and ramping as much volume as we can of those higher capacity units.

So I think it has been extremely positive for Seagate and maybe even for the industry in general. If you look at the last 3 years, 12, 13 quarters, every quarter, higher revenue, every quarter, higher profitability. And we want to do this for a long period of time. And as I said, based on the order that we have in place, we see this continuing to happening again, at least for those 4 or 5 quarters. And I'm sure that will happen also for a much longer time.

Yes. So can you talk a little bit about the drivers of that, right? So obviously, there is -- what you're charging on a dollar per TB basis today on average is in the $13, $14 range. When you look at some of the spot markets, I mean, those are transacting at 2x that. So clearly, there is room for this pricing to move in the long term much higher. So as we think about your margins and profitability comment, pricing is one lever, cost is another and your factory utilization and other is a lot of factors at play. So what are the -- how would you rank order those and your confidence in sort of being able to drive higher gross margins over time? What's the right like way to think about that?

Yes. We focus on all those items. In terms of utilization, probably we are already a little bit maxed out at this point. So our factories are full. But our product road map is very strong. So we will continue to move up in capacity per unit. So with the same number of units, we actually generate 25% more exabyte every year for a long period of time. Pricing, of course, pricing finally is always a matter of supply and demand. Demand is way above supply. Demand is actually stronger than what we were expecting. So of course, pricing will be -- continue to be better, absolutely. But again, with the same strategy that we have applied in the past.

And the cost per terabyte is another positive variable, especially when you can move from a mix that was based on a PMR technology of 20, 25 terabyte per unit to an MR technology that started 30. Now we're qualified on 40 terabyte. We will be qualified on the 50 terabyte. We discussed that on our earnings release. So we will start the call for the 50 terabyte in a few quarters from now before the end of calendar '27.

So we are progressing very well, and this is how we want to extract more exabyte from the same footprint, getting more revenue because we have more exabyte, getting more revenue because we have higher price, fairly similar to what we've done in the past, I would say.

Yes. So if we think about this gross margin trajectory, you're incremental gross margins are extremely strong. And as you just said, right, you're continuing to take a little bit of price. There's higher demand, so exabytes are also growing at the same time. Is there any reason that these incremental margins should not be as strong as they are?

No, really. No, we gave an indication a year ago during our Analyst Day. We did better every quarter. We did better than what we indicated because demand was a little bit stronger than what we were expecting. And therefore, with better demand, you get a little bit better pricing also on the part that is not really committed on the order because we have 80% of our business that is data center. On data center, we commit a lot of that volume. But every quarter, we try to get a little bit more out of our manufacturing. So we commit what we are sure we can produce, then hopefully, we can produce a little bit more. That extra volume get a different price. So there are different economics on that extra volume. So every quarter is different. Sometimes we don't have more volume. Sometimes we have a few exabytes, and we can sell those very quickly and at a higher price.

And then we have the other 20% of the business that we call edge. That part of the business is low capacity drive. So it's not the 24, 30, 40 terabytes. It's like the 2, 4, 8 terabyte drive. That is the only part of the business where we overlap with NAND. In the data center, the 2 components are used very differently. storage, hard disk, when you need to do the compute, you move the data from the hard disk into the NAND, you do the compute and then you store it back into the hard disk has been the structure for 15 years. It's going to be the structure for the next 15 years.

But when you go to low capacity, for example, in the consumer business, if you need to have an external storage device, you go to a store and you see both. You have an SSD and you have an hard disk. The price is very different. SSD price today is very high. So in that part of the business where we don't have orders, we can actually increase price also looking at what other components are doing. And so that part of the business is, for sure, another upside to our results in the March quarter, a little bit in the December quarter, but say more in the March quarter, possibly this quarter.

So now we have both segments that are very strong before it was mainly data center, data center, data center, the other part was not so profitable. Now even the other part gets good. And again, because it's not the part of the business where we have LTAs or build to order, we can be more opportunistic or more than opportunist, we can act faster on changes.

Yes. So on that part of the business, if we think about -- I mean, it just doesn't seem like we're going to get any reset on at least NAND for many, many quarters to come. So we should be expecting the profitability on that side of the business to continue to improve from here in the foreseeable future.

Absolutely.

Okay. Maybe just going back to the fundamental premise around demand, right? Like it seems as though we're going from like 1.5 zettabytes to close to 2 zettabytes of HDD shipments. When you think about it from that perspective, I mean, those are huge numbers to be able to deliver that kind of growth and sustain that kind of growth on sort of like at least mid-20s or so seems pretty huge. So what is the underlying things that are fundamentally happening that you see which is supporting that kind of a viewpoint on such a strong incremental exabyte demand?

Yes. I think when you look at percentages and when you look at absolute exabyte number, the exabyte number is really impressive because the base is growing and growing and growing. So the same 25% that we were growing 5 years ago and what we're growing today in exabyte today is double. So it's hugely different. We focus all on technology. We are happy with our technology. We are happy with having qualified all the important customers on HAMR technology in the last year. And now we start qualifying customers on the second generation, and then we will have third generation. So it becomes a normal technology for us and for customers.

This is how we generate more exabyte. And of course, moving with HAMR, you can move higher, faster. With the past technology, we are not growing 2 terabyte for each new product. And we were having a new product basically every year. Now we have a new product every 18 months, maybe 24 months, but the growth is much bigger. You go from 30 terabyte to 40 terabytes. So this is how we try to keep up at 25% in exabyte is now a much bigger number. So that is the main focus.

Yes. So just talking about this areal density transition, right? So I think there is -- like some people try to draw an analogy between NAND going from 2D NAND to 3D NAND, that was initially like a shortage and then you just got a flood of like bits in the market. How would you say HDDs are different in the sense of as you -- how do you prevent the same thing of now you give the ability to make like significantly higher amounts of bits, if the demand were to taper off in some way, how do you manage that?

Well, the main difference is this industry is not adding units. We are not building new factories. The other components have decided to go into different strategies and adding factories and factories spending a lot of CapEx to add those factories because they were assuming that was needed and was good for them to do it. I think for this industry, the best solution is to add exabyte, not to add factories and units. And I think we can do it, and we are actually demonstrating that we can do. We have increased exabyte even more than 25% in the last several quarters.

So that is the right way to do it. And limiting the number of units reduce the risk of going to an oversupply situation. I also say this industry has been in an oversupply situation for a long period of time. All the transition from the client business into the cloud business was done with an oversupply situation. Cloud was very small 10 years ago. So we had to give the time to grow and absorb that supply that was created for the client business that was going away. So we know what does oversupply means to the business. We don't like it. So we will try to avoid to go into that situation as much as we can.

Yes. You got a name for that, right, legacy products at some point in time. But if you think about where we are from just the sustainability point of view, like what you're noting about this growth that is compounding at such a high rate off of such a high base. Hyperscale obviously is driving a lot of this. But if we think about hyperscale CapEx eventually at some point, maybe flattening out, not growing anymore. Maybe it's $1 trillion is not very far away now, but you spend $1 trillion each year. How does that impact the trajectory of growth for you?

Well, no, CapEx is a big number to buy a lot of different things. Hard disk, good for this industry. Hard disk is probably low to mid-single digit of that CapEx. So that help us also on the pricing discussion. The impact of an increase on hard disk is very limited to our customers' CapEx overall because we are a small part of the CapEx. So again, the CapEx trajectory is important. But then, of course, because we are not a big part of the CapEx, depends exactly how much of that is allocated to storage. And CapEx can go up and down, maybe it's not 100% correlated to where storage is going, depend also from the pricing of other components.

For example, for a period of time where GPUs were fairly expensive, our percentage in terms of CapEx went down. They were not buying less hard disk. It was just they were spending more money on something else. And maybe now it's going to a different direction because we are also increasing our price, and we do that sequentially for a long time.

So let's see. But today, all the indicators are for a demand that is well above supply. So again, even if that CapEx could a certain point stabilize or even reduce, I don't think will impact our situation.

So as your customers come to you and say, hey, like my demand, like I got to grow my business. I have need for incremental supply. And you're saying demand is well ahead of supply. So there must be a lot of pressure on like, hey, what can Seagate do to deliver more exabytes here? What's your response to that?

Well, our response is we do all what we can in terms of product mix. And of course, they need to do their part that is qualifying as fast as possible new products because new products have higher capacity. The number of units is the same. So that is a way for them to get more exabyte in total is to move their mix to higher capacity drive. That is helping us because higher capacity drive has a lower cost per terabyte. So good for us, but it's also very good for our customers because they get more exabyte. And very importantly, in the data center, you have a lot of physical slot. And every slot has a cost, but it is the real estate cost, the power cost, the compute cost, the hard disk cost, people that are working into the data center.

So everything is brought to a slot cost. That slot also has a revenue. The revenue is how many terabytes our customer can sell from that slot. So if in a physical slot, you put 20 terabyte, there is a cost for the slot and there is a revenue that is equal to the 20 terabyte that they sell to their customer. If you can put 40 terabyte drive for the same slot cost, you double the revenue. This is why for our customers, it's so important to get a bigger drive, bigger drive, bigger drive because for basically the same slot cost, they get much bigger return. This is why they want bigger drives. And of course, they want more exabyte.

Yes. Yes, the TCO value proposition just goes up tremendously with these larger drives. Maybe just on that, right, like how much of the demand that you're seeing today is coming from replacement demand versus new demand in these data centers?

Well, today, demand is really strong. So they tend to use the vast majority of what they buy for new data center because they can keep the exabyte in the old data center and add 100% of what they buy in the new data center. But at a certain point, the drives, they don't last forever. So our guarantee is 5 years. So they can use 5 years, maybe 6 years, maybe 7 years, but then they need to refresh.

So there is a huge benefit also in doing a refreshment because as we said before, they just spend the money to buy new drive, but they replace 20-terabyte drive with 40-terabyte drive. So the return from that small RD spending is huge for them. So there is a lot of benefit also in doing refreshment for our customers. So not only because they drive they have a limited useful life, but also because of the financial return. But today, we see the majority of the drives going to new data center because they try to keep those exabytes that they already have intact and not taking an offset.

Can you just talk a little bit about sort of the gross margin trajectory in terms of -- as you're going to the subsequent versions of HAMR, you already demonstrated that in going to 4 TB per platter, like your qualification time has come down relative to sort of the speed of qualifying has come down relative to what it took on the first generation.

So as we think about subsequent generations of HAMR, should we be thinking that we should be seeing some kind of a margin inflection because just as you said, like what you can charge on a dollar per TB basis is going to be so much more attractive to your customers relative to what you're delivering even today. And so shouldn't we be seeing sort of even stronger incremental margins in some ways as we go to next and future generations of HAMR?

Well, generally, if you just look at the last product, of course, that is a very good product to us. Now we generate a lot of revenue and profit. Of course, when you look at the entire company, you need to assess the volume. And of course, the volume is starting fairly slow and going up quarter after quarter after quarter, so getting more impactful later until we go to the next product and do that again.

So yes, now with this new technology, as you know, we don't need to increase the number of disk and the number of heads inside the box. So we have even a better opportunity to drive a certain level of cost reduction, assuming everything else remains the same. And of course, the price is now a price per terabyte. So it's kind of independent from the product. So the 2 things combined should give us a good opportunity to continue to improve.

Now as I said before, we have orders in place. So we know what is the price. We know what we will produce. So if we execute our plan as we have committed to our customer, we will see higher revenue, higher profit.

Yes. Yes. Maybe just to think about sort of the amount of cash that you guys are generating now, right? It has been like amazing to watch sort of over the last 3, 4 years, the change in the cash flows. How are you thinking about prioritizing cash flows when you're throwing off this amount of cash?

Yes. We have always focused on shareholder return. So we have always returned the vast majority of our free cash flow through dividend and through share buyback. And we will continue to do it. We are -- we have increased our dividend fairly recently. I think the first payment with increased dividend was January, so just a few months ago, and we will do that again. I think every year, we will do it. We are doing share buyback. We are doing less than what we will do in the near future because we are also paying down our debt. In particular, we wanted to reduce the convertible.

Now to me, the convertible is very similar to a share buyback. If you don't do it and share price continue to appreciate, you continue to have a bigger dilution, bigger dilution, bigger dilution, then those shares need to go out and buy them back at higher price. So to me, is in this environment, it's probably good to anticipate the repurchase of the convertible. We are almost done with that. We still have about $200 million outstanding that we will repurchase next quarter. We will reduce our debt even outside of the convertible a little bit more. But you will see sequentially more share buyback. And fairly soon, we will be done with the debt repurchase. So we will focus the vast majority of free cash flow between dividend and share buyback.

What's the amount that you need to sort of cash balance that you need to have that you feel comfortable to run the company?

Well, I think what we have today is good. And we also have a revolver that we are not utilizing, but it's always available in case we have a short-term need. So I think where we are today between probably around $1.5 billion in cash is a good level for now. Then we see the business it becomes much bigger, we also need to manage the working capital. So we could change a little bit, but probably not much.

Okay. I know we're coming up on time, unfortunately, these are only 30-minute sessions and a lot to talk about over here. But Gianluca, maybe just to close it out, right, like what do you think investors should be most focused on over the shorter or medium-term horizon as you think about where Seagate is positioned today?

I think the trend. The trend is very indicative. The last 12, 13 quarters, I think, are very indicative in terms of profitability improvement of what is going to happen, I think, for the next 4 or 5 quarters. Now I can talk about that period because we have orders. It's not only those 4 or 5 quarters, it will be longer than that. But at least we know what is going to happen in the next 4 or 5 quarters. So I think the trend is very clear. Demand is only getting stronger. So the trend will not change. and we have opportunity to price even better through time, of course, not on the orders that we already have, but on new orders. So that is what I think you should focus on is a very good industry right now.

Amazing. Well, thank you so much for your time, Gianluca. I really appreciate it. Thank you very much.

Thank you.

All right. Good morning, everyone. I'm Krish Sankar from TD Cowen. I'm the analyst covering Seagate, where we are fortunate enough to have Gianluca, the CFO, and also Shanye from the IR team here. Seagate, obviously, as you know, one of the leaders in hard drives.

Gianluca, thank you very much for your time. And while I'm on it, I also tell every investor on this, like please do vote for TD Cowen in [ AXL and Fannies ].

So anyway, with that, let's start. I think -- Gianluca, I think what is kind of interesting is that I think there seems to have been some confusion last week where some folks assumed that there's capacity addition. But I think -- can you just clarify what's going on? Because I think what has been pretty like consistent with what you have spoken in the past is mid-20% exabyte growth, probably not real unit capacity additions. Has anything changed? Or is still that the narrative today?

Yes. Thank you, Krish. Always nice to meet with you. Before we start, let me say that I will be making forward-looking statements today, and you can learn more about the risks associated with those statements on our website.

So short answer to your question is no, we are not adding any unit capacity. We think our technology road map is really strong. So we can generate the exabytes that we need through technology transition, going from first-generation HAMR to second-generation HAMR. We already discussed at our earnings release about third-generation HAMR that is 5 terabytes per disk, so a 50-terabyte drive that we will start qualifying at the end of next calendar year. So our road map is really good. We think we generate about 25% increase in exabyte year-over-year through technology road map. So we don't need more units.

Got it. And then what about head capacity addition? Is that a slightly different story compared to units or...

No, really. I would say we produce heads and media. So the head to read and write the data and the media is a disk. We also buy a lot of other components externally. But the technology is in those 2 components, heads and media. So depending how you assemble the drive, if you have a 5-disk drive or 8-disk drive or 10-disk drive, for the same number of units, you can need a little bit different number of heads and media. So it depends how -- where the mix is going.

But generally, I would say our focus is keeping the units very stable and, of course, increase the exabyte that we generate through this change in mix, change in technology and transition every 18 months, 24 months to a new generation of HAMR that give us -- in this case, from first-generation HAMR to second-generation HAMR give us more than 30% increase in exabyte. And then from the second to the third, another 25%. So we can grow through technology, we don't need more units.

Got it. And also along the same path, when you look at it, clearly, demand is strong. I think last time we spoke about kind of -- if I remember, right, kind of sold out for all of 2027. Is that still the case? What is your visibility into 2028? And how to think about some of the LTAs that you signed?

Yes. So we have 2 different kind of agreements. For the next 4 to 5 quarters, we have orders. So an order has a specific product, specific volume, specific price and time to deliver the product. So we always want to cover the next 4 to 5 quarters because this is the time we need to produce a hard disk drive. To produce an HAMR hard disk drive, you need about 3 quarters. So we want 3, 4, 5 quarters that are fully defined. When we start the product in our manufacturing, we know exactly who will buy it and at what price.

After that, we have LTAs that are agreement based on exabyte. So for our customers, it's very important to know what kind of storage they can get, 2 years out in time, 3 years out in time, even longer because they need to plan their new data centers. So the price is less important. The mix is less important. They don't even know what product they will be qualified 3 years out in time. But they want to know how many exabytes we allocate to them. So they know how many data centers they can build.

So we have agreement on exabyte. But then when we arrive into that 4 to 5 quarters range, we translate the exabyte LTA into an order. At that point, they know what is the product that they are qualified. We know how much we can ramp up the product. And so we have -- usually, we have a list of products that they buy not only one. We define the price, and we define exactly when we ship.

But historically, if I remember right, the pricing was negotiated on a quarterly basis, but now has that changed to an annual basis?

Yes, depending from the customer and the duration of the orders is 3, 4, 5 quarters.

Got you. The other interesting thing I remember is that in the past, you've mentioned that 70% of hard drive demand comes from new data center openings. How do you track that? Because obviously, there are a lot of things that goes into a new data center opening. Obviously, you had to deal with like permitting process, power supply, things like that. So how much visibility do you get? And in that build-out process, where do you come in? Are you like a late-stage purchase from a data center build-out standpoint, early you qualify?

Yes. We have 2 major segments, data center and edge. Data center today is probably 80% of our revenue. So it's growing. Has grown a lot in the last 2 or 3 years. In particular, public cloud is the subsegment inside data center that is growing the fastest. So we are already 80% of revenue, more than that in terms of exabyte. So we see this trend continuing.

Data center is growing faster than any other segment. But we also see a good demand on the edge. So today -- and the edge is where we compete with NAND. So low-capacity drive, 2 terabyte, 4 terabyte, 8 terabyte drives and more because NAND price is so high, there is a good opportunity for hard disk to increase price. And we don't have purchase order in that segment. So we can be more -- we can increase price faster. And demand is also higher. But again, it's not the majority of our business, only 20% of our revenue.

Data center, much more structured. We have those purchase order in place where the price is defined. Price is increasing, but it's a different kind of trajectory. And in the data center, as you know, we don't really compete with NAND. So the NAND price has no influence on our demand. NAND is used to run the application on the compute side, hard disk is used for storage. So very different applications.

Got you. I mean it kind of makes sense because I think hard drives is probably up to 2% of the data center CapEx, while NAND plus DRAM is probably up to 40% now with the price increase. So you're not the problem child, but they think that's very good. Is the memory price increase, i.e., DRAM/NAND price increase slowing down data center build-out?

So far, probably not much. I would say what we try to do is not to be the main problem. So we don't want to be the component that is limiting the development of the data center. So in the past, you have seen the power being the bottleneck, probably still the bottleneck. The GPUs were a bottleneck. Possibly DRAM is becoming the bottleneck. So we are always number 2, 3 or 4 in the list, is a good place to be. We still have our power in the negotiation, but we are not limiting their development. So we are not a problem for our customers.

And the other thing that I noticed is compared to last year, I think last quarter, which is obvious from your numbers and your competitors' numbers, the pricing is getting better or a little more better than historically. So historically, it was like more like mid- to high single digit. Now it looks like we'll be low double digit. Obviously, you're still not the bottleneck, so you could probably increase pricing further and not be an issue. But what is the thought process behind price increase? And how sustainable is this run rate over the next couple of years?

Well, we know it's sustainable because we already have PO in place. So as we said at our earnings release, we see not only this quarter, but we guided more precisely, but we also discussed about the next 4 quarters. And we said every quarter, you will see a revenue increase and you will see profitability increase. Of course, a good part of that improvement is coming from pricing. We are executing a very good strategy that is not being super aggressive with price, but being very, very consistent. So every quarter, you see that price up. Every quarter, you see that exabyte up, no units, but exabyte up. So revenue is growing very well and profitability is growing extremely well.

And then the other thing that is kind of interesting when you look at last year, I would say the inflection for hard drives really started when you started seeing videos -- AI videos and things like that. Today, how do you see that? Is that still a big trend? Or is it like more data retention? What is the driver today and into next year for the next leg of like hard drive demand?

Yes. I would say there are different drivers. I would say the positive part for hard disk is that on storage, on hard disk, you don't need a different mix, different kind of hard disk depending from what kind of data you want to store. So if it is a video AI or a traditional video or data in form of text, they all get stored in the same hard disk. So it's a good simplification for us in terms of what we have to produce and what we have to generate.

I would say video AI is huge. Retention started probably 2 years ago. So when companies and people were starting to use AI, the first thing we all did is stop deleting data because if you want to have a good result from running AI, you need to have a lot of data. So data retention already started 2 years ago. And then you started to have AI generating data itself as an application. And the beauty for us in terms of data generation and data storage is AI is very quick. So generate a lot of more data than human. It works 24/7 every day, no problem, and generate data that you need even as a step through the final result. So everything that AI generates gets stored and then get used again to generate something else until you arrive to the final result.

So a lot of benefit from AI, but it's not the only application. Of course, robotics is becoming more and more a data generation application. AI robotics will be another one that is just starting today, autonomous driving. You go to San Francisco, you go to Phoenix, Arizona, you see a lot of cars that are autonomous driving. They generate a lot of data every day. And all that data is stored every day for compliance and to learn how to better drive the car in the city. So when you see that expanding through all the city in U.S. and outside U.S., that is an incredible volume of data that could get stored.

So today, like you said, majority of your demand coming from hyperscalers. Is there an argument to be made that hard drive really benefits only from hyperscaler or cloud demand? Because let's just assume in the future, we go to an enterprise AI world with like inference, there's going to be more on-prem deployment. Would hard drives benefit? Or are you going to more a cloud play rather than an on-prem play?

As I said before, we see demand in both segments, so in the data center and at the edge. I would say it depends on how AI will continue to evolve. AI today use a lot of the public cloud. So it's concentrated in the public cloud. The Neocloud are compute. They don't have storage. They take their storage from a public cloud, so from an hard disk. They move into NAND into the Neocloud data center, run the application and store back in the hard disk in the public cloud. So they basically outsource or complement the compute part of a public cloud. If they want to become more independent from the public cloud and run their compute for other customers, they will have to build storage. At that point, we will sell hard disk even to the Neocloud.

At the edge, it's the same or on-prem, if you have a company that doesn't want to use the public cloud for many reasons, efficiency cost or security of information, the structure is the same. It's just smaller. But storage is hard disk. And then when they run the compute, they move the data into NAND, run it and send it back. just smaller, but same concept.

Now when we go to the edge, that will be interesting. It's probably not happening tomorrow or the day after. But when you go longer, you will have, for example, autonomous driving. We require a lot of edge storage and fast compute. So for sure, you will have a good application for the DRAM and the NAND part of the business. But then they will also start parking data for a while until they send to the public cloud. That can be hard disk. So we will see, but I think there are a lot of opportunities for hard disk even at the edge.

Would you consider your VIA China business, it kind of similar to edge? Or is it different?

VIA is a bit different. I would say there are 2 parts of VIA. So the video and image application, first of all, there are the cameras that are connected to a storage. That storage is usually an hard disk, can be -- depending how big is the security that you're adding, can be 4 terabyte, 8 terabyte, 10 terabyte. So usually is a mid-cap hard disk, not a huge. Then the companies that are providing the surveillance, they also offer storage. So if you are a building like this one, you have all the camera, you record everything and then you say, okay, or you delete or you store somewhere. And you can store with the same company that is providing you the service of surveillance.

So in that case, they buy big drives. They buy 20 terabyte, 30 terabyte, 40 terabytes because they are actually offering a simplified cloud. There are not a lot of applications. There are some applications related to the surveillance, but not a lot of application, but a lot of storage because the data is kept until they run the application and then they decide what to do with the data.

Got you. Interesting. I'll just pause to see if anyone in the audience has a question. If not, I'll chug along. So the other thing, like when you look at your business, I think historically, your cost downs are probably like high single digits, 10%. Now it's more like low double digit, mid-teens. So the cost downs are actually better than historical. Is that purely a function of moving to HAMR? Or do you think there is a lot more room for this cost decline to improve? Or do you think there's like a once-in-a-time step down because of HAMR, now we should think about more like 10% longer term?

Well, I'd say if you look at the product, so the product cost going from first-generation HAMR 30-terabyte drive to second-generation HAMR 40 terabyte drive. the unit cost is fairly similar, while you have 10 terabyte more. So you have actually a fairly huge decline in terms of cost per terabyte.

When you look at the entire company, the entire P&L, it doesn't depend only from the last product. We sell from 2-terabyte drive to 40 terabyte drive. So the cost in the period depend how this mix move up. And so it's not only the last product. But with the time, moving more and more into HAMR, especially at the beginning of the technology where you increase 33% going from first generation to second is a good improvement of our cost decline.

When you go from 40 terabyte to 50 terabyte in percentage, that is 25%, so it's still the same similar unit cost. You add the same 10 terabyte in percentage is a bit lower. So depending on how you calculate. But I'll say, of course, HAMR is advantageous for the cost. But every period is different. Every quarter is different, depending on how many PMR we are still selling and what is the segment in that specific quarter that is getting more volume. So it's not so linear. But I'll say HAMR longer term will be, of course, the way to reduce cost in the industry.

Otherwise, with the old technology, with PMR technology, you continue to add 1 disk and 2 heads, 1 disk and 2 heads until you have space in the box. But -- so the cost per unit goes up because you need to increase the bill of material. With HAMR, you keep the 10 disk, 20 heads from 30 terabyte drive, 40 terabyte drive, 50 terabyte drive. So the bill of material remain consistent. Of course, the components are not the same. So there is an increase on some of those components. So -- but the unit cost is fairly similar.

Are there any pressures on that? Because one of the things is like, for example, you get glass from Hoya or someone, but you also have increasing demand from the optics and photonics folks. So is there anything, any disproportionate pricing where other sectors are actually willing to pay a premium, causing your cost to go up because you had to probably match that to secure those components?

Well, no, supply chain, of course, is very important. And as I said before, we buy a lot of components. We buy electronics, we buy memories that, as you know, are fairly expensive today. And we buy a lot of other mechanical components. So -- of course, every year is different. There are years with higher inflation, year with lower inflation, a year where there is a shortage of one component, so you need to spend more.

So my discussion is keeping those at the same level, you have this clearly strong reduction in cost per terabyte. But as I said before, every period is different. Right now, we have memories costing more for sure. So -- but you cannot compare -- it's not because of the technology, it's because of the component. But you need to adjust your estimate based on how you see those components costs evolving, yes.

Got you. Because I think, clearly, the demand is very strong, doing these cost outs, already at like 50% plus gross margin. I think people always speculated that it's a rational duopoly, we should be at 65% gross margin. And it seems like there's a path to get there easily, maybe exceed, but I think realistically get to mid-60% gross margin. From your view, I mean, I'm not looking for guidance, but is that a fair assumption given the trajectory of demand and trajectory of your cost downs?

I would say the assumption of continuing to improve is very fair. We actually said that just a few weeks ago. We said no, but this quarter plus other 4 of improvement. And this quarter, we are guiding at the gross margin that is about 50%. So if we continue to improve, we continue to improve, so we will go. We don't have a gross margin target. So -- but it's not a number that when we achieve, we just say, okay, this is not, we stop there. We like how we run this strategy, and we have done it for 12 consecutive quarters. This is the 13. We have other 4 that we know are coming, and we will continue to do that. And then one day, we will say, okay, this is the results that we have achieved. But we don't see an end at this point yet.

Got you. Because the other interesting thing is you said earlier on, you're not adding unit capacity. You're still growing exabytes 25% to 30%. At that CAGR, your hyperscaler customers are not pressuring you to add more capacity. They are happy with the 25% to 30%.

I don't know if they're happy. I think there is pressure to do more because their unconstrained demand is, for sure, higher. But every year, there is some components that get short. Right now, DRAM probably is short and power is still short. So from that unconstrained demand, you need to go from what they can really build. And it's our assumption, it's not their assumption. Our assumption is if we increase exabytes by about 25%, we will not be the component that is gating the development of the data center. We could be wrong. But we think 25% is good enough not to become the top of the list problem.

Got you. Because I mean, the reason I'm asking is it seems like you're doing 25% or 30% exabyte growth, slowly increasing pricing more than before, but the customers are not pushing back, so they seem to be okay. In other words, if you go to like, say, 35% exabyte growth, you're not the bottleneck. So your hyperscaler customers will still be constrained. So the marginal benefit is not much for you or them for you to go to like higher exabyte growth, right?

Yes. I'll say, so far, this strategy has worked very well. We more than doubled our revenue. We more than doubled our profitability. So there is no reason to change. I think as this strategy has given us a great result and our focus is continue to execute the same strategy for the long time. And we have already done 12, 13 quarters, so it is already fairly long. We don't see this ending. We have another 4 quarters coming. And we will have more later because now we see the exabytes that our customers are demanding for year 2 and year 3 and year 4 is actually way higher than what we have in our plan.

And also, I think your own forecast is that I think probably next month or so, I think 40% of exabytes would be from HAMR. And then maybe a year from now, 70% of exabytes will be on HAMR. Is that still the plan? Or do you think that HAMR percentage should be higher than expected?

Well, we are ramping HAMR well, especially because the second-generation HAMR was qualified a little bit earlier than what we were thinking with the 2 major cloud customers. Of course, the percentage depends also from how much we produce on the old technology because to achieve those percentage, it will be very easy just not producing a lot of PMR and you have a lot of HAMR. So it depends.

We are trying to optimize exabyte and to optimize exabytes, we produce a lot of both, HAMR and PMR. So let's see. I think first priority is to achieve this 25% exabyte increase. Now as you said, in the past, we were able to do more. And that more is probably coming more from PMR than HAMR. Now we are ramping more HAMR. So no, HAMR is ramping very well. I think we will be around those percentages, but it depends every quarter how we manage the 2 technology.

Got you. I think if you do the math, you're probably going to generate close to like $3 billion of free cash flow this year, I mean, after the dividend payment, about $4 billion in debt. How do you prioritize free cash flow? Is it like to mainly pay down debt? Would you consider repurchases of the stock price? Or do you think you probably need more supply chain preparedness for the ramp that it probably makes sense to invest more in your own business?

I'll say longer term, we have -- as we have done in the past, our free cash flow is always focused on shareholder return between dividend and share buyback. In the short term, we have reduced our debt going from about $6 billion to now a little bit less than $4 billion. Now we can go lower. We still have a little bit of the convertible to buyback. We did something recently. We will do the remaining possibly next quarter and maybe reduce debt even lower. But no longer -- after we have reduced the debt, the free cash flow mainly will go to shareholders.

Got you. Got you. And is there any -- like do you see like in the past, like you mentioned that I think heads and media takes a year to bring it online. So even if you decide to add capacity today, you're probably looking at least a year before that becomes useful exabytes into the marketplace. Is this still the same? Or do you think there is like a more tightening potential in the supply chain where lead times can come shorter than today?

Well, I would say the cycle time of a wafer where we produce heads is about 9 months -- let's say, 6 to 9 months depending on which technology, which product. If you need to build capacity like a greenfield, well, that takes more than a year, I would say, possibly at least 2 years. So now you need to build the factory and then get the tools and qualify the tools. So it's long term, yes.

I mean, if you decide to add capacity, would greenfield make the most sense? Or it be more existing brownfield?

Well, right now, we have not looked into that because no, we don't think we need to add units. I would say if we arrive at that point in time or a certain decision, we will see what is the best solution. But again, in our plan, we don't have an increase of the number of factories that we have.

And one of the -- you spoke about cost reduction. When I look at your footprint in manufacturing between the heads, the wafers, the media, everything you're in like Singapore, Ireland, U.S., is there a consolidation angle in this to help get the cost reduction? Or do you think the footprint stays the way it is?

I think it will stay because we do different type of manufacturing in different locations. So Singapore, we produce the disk. We don't have any other site producing the disk. Malaysia, we produce the substrate. We don't have any other place where we produce the substrate. China, Thailand, we do assembly and final test. Again, China is used for our Chinese customers. Thailand is used for everyone else in the world. And then probably the only 2 factories that are similar are the one U.S. and Northern Ireland, where in both places, we produce heads. But because we need that level of volume, no, I don't see any reason to change that.

Got you. We're close to running out of time. I just want to see if anyone has any quick questions. If not, I'll try to squeeze one more in. Double-digit exabyte like mid-20% exabyte growth. And clearly, like pricing is pretty robust. I'm just wondering like there's an expectation that we should see double-digit growth in revenue-wise even in '27 and beyond. If the current demand scenario continues, let's just linearly extrapolate it. Is it a fair assumption that it should -- the revenue growth should be double digit because...

Yes. No, at earnings release, Dave, the CEO, said we expect revenue to grow for the next several years at, at least 20%. So he said at least, so probably will be more than that.

All right. I think, Gianluca, thank you very much for your insights. Always fun having you. Thank you.

Thank you very much.

Good morning, everyone. Thank you for being here. I'm Samik Chatterjee, and I cover the hardware and networking companies at JPMorgan. As you all get settled, I'll just introduce the next speaker here. So here for the next fireside chat is Dr. Dave Mosley, Chief Executive Officer of Seagate.

Dave, thank you for being here. Thank you to the audience as well.

Maybe Dave, I'll start you off with a question that's more broader, more longer term. You've been at Seagate for nearly 3 decades, CEO since 2017. When you look at the storage industry today versus prior cycles, what is structurally different about the demand profile? I know everyone wants to say this time it's different, but what do you think investors are still underappreciating about this current cycle that we're in?

Thanks, Samik. A couple of things, forward-looking statements, risk factors are on our website, get that out of the way. Yes, it's been a long road for me and for our teams as well. And I don't look at it as small cycles because I think everyone is different. The last downside we went through was profoundly different, but created by its own unique situations. I look at it more as a macro cycle.

I mean, from my perspective, data is always growing underneath everything. But our industry went through an enormous build-out period through the client server days. Up until peaking about 2012, for reference, we shipped 66 million drives in 1 quarter, but they were all low capacity going into notebooks and desktops. And if you think about those times back then, the drives were not full. They were not being utilized very often, just not turned on very much.

What's different now is the drives are chock-full of those heads and media, critical components that we've built out the factories for back in those days. A lot fewer drives, but they're big beasts now. And if you go into the data centers, they're working very hard to 100% utilization and full. So a very different profile for the industry as we've pivoted the supply chain for the last decade.

Okay. Okay. And maybe to get into that a bit further, you've talked about agentic AI generating much larger data sets than periodic queries, for example. Can you just go sort of one level deeper into that, decompose what you're seeing in exabyte growth in calendar '26, '27? And how much is the broadening out of the use cases when it comes to training, inferencing? How much of that is becoming a driver of exabyte growth going forward?

Yes. It's really hard to break out because the application space is growing so quickly. And so what some things are called inference 1 day, may be actually developing into something that's actually learning more to the next. I think we try to think about the -- first of all, the recovery that happened in the last 3 or 4 years was really driven by video properties first. And so it's not small text data sets. It's large unstructured data sets, which are usually video data sets.

If you think about training that happened early days, it was a lot of text for training large language models and being able to get utility out of them and they are very useful. Don't get me wrong. But now it's not about training on ages and ages worth of text that was stored somewhere. It's a lot more about understanding what's going on in the real world and being up to the minute in your cycle and ingesting a lot of the large data sets and then being able to figure out how that is changing so that you can give the proper inference, and we're seeing this all the time.

And so if I look at the workloads that are across the drives in some of those applications, they're very diverse. There's not a predictable way of using them. And depending on the day or the user or whatever, it can actually change quite frequently. And that's what I would say characterizes the AI workloads more than anything.

You've mentioned moving to a build-to-order approach with your customers. How has that impacted visibility for yourself? And is that something that you're seeing across the industry being adopted as well?

It is. I mean, we're driven by very long lead times in our wafer operations. So we have a recording head wafers that take a long time to build. It could be more than 9 months. And then the drives themselves take another quarter or something like that. So inside of that, we know what's coming out a year from now, and we've basically gone to the customers and said, "Look, if you want to plan this really, well, which should be for your data centers, we know what's coming out, you can buy this stuff up to a certain period." And so we want to keep that 4 or 5 quarters of visibility, very, very solid for what's being built.

But the demand is significantly higher than that. And we know that if we can get a little bit more out of the factories, then we're sensing the market and the market really wants the exabytes even more. So we're now giving predictability into what we're going to be building in the next year, or 2 years, 3 years based on technology transitions and the customer demand for that is still very strong. So it's not really locked in. The build-to-orders are really only locked in for 4 or 5 quarters. But beyond there, we have great visibility.

What's the financial implication of moving to this model? Like for investors that have looked at Seagate over the cycles, how do you think about what changes on the financial side?

Well, what I really wanted out of it when we started was revenue predictability. Just we want to make sure we're not building into a forecast that's not real. Now it appears that not only are the forecasts real, but the demand is significantly higher.

Fundamentally, the financials will be driven by the demand. And we're trying to answer that demand as aggressively as we can with our technology, but I think that's why the financials keep getting better. If we have predictability for 4 quarters, but then that starts to -- we start to roll through that period and we go to re-up the next one, then we're seeing that the demand is even stronger. And so therefore, the economics get better.

Okay. Okay. You mentioned this lead times are now quite long. We've seen some of the press reports as well mentioning lead times are [ over a year ], in some cases, for HDDs. However, when you take a step back, I think investors have always been concerned about the industry adding capacity right at the peak of demand. And by the time that capacity comes online, you're going through a moderation in the demand cycle. What's giving you the comfort now in terms of your capacity planning? And how are you thinking about your capacity road map?

There's a couple of different kinds of capacity. If I go back to the industry 15, 20 years ago, the head wafer of the fabs, the lead time was only a few months. So very, very fast to turn. Now it's significantly longer with the technology being as complex as it is now. And so that's really what we're gated by. So it's critical components that are long lead time. As far as the number of drives at the back end, we could be still fairly flexible on that. But because we're able to sell the capacity that's already coming out, we don't necessarily need to.

So how do I think about adding capacity? Now it's not just one machine you plug in. It has to be synchronized all the way across not only our production, but also the supply chain, but it's really long lead time now. And we're adding tools to make our capacity a little bit stronger inside of the manufacturing facilities that have those long lead times, but they tend to be the types of tools that will go through technology transitions rather than just pure capacity adds.

So maybe just to follow up on that, and this question comes up pretty often. So I'm sure you've taken this question before is what fundamentally needs to happen for you to add unit capacity or floor space capacity rather than just depending on higher capacity drives to add exabytes?

Right. There is an answer to the question, I'll get to that in a second. But what our customers are driving us for right now is more exabytes. And we believe that the way to get the most exabytes is to take our talented teams and really go through these technology transitions. We're targeting mid-20s percent growth, which is an enormous CAGR. And the only way we're going to get there is to be able to go through those technology transitions, if you will, to take a 3 terabyte per platter product to a 4 terabyte per platter to a 5 terabyte per platter year over year over year. And so that's really the way in it. If we took the teams off and started building new factories or bringing up new machines that would just take too long, you would end up more capacity, if you will, but then you'd slow the rate of growth on that technology.

So back to your question directly, the wildcard really is in unit capacity for disk drives, which we, again, could be fairly flexible with once we package those heads and media. That gets down to more customer diversification and edge and edge AI and all those use cases, which I think could come someday. So we would take the heads and media that we have planned and divert them somewhere else should those applications take hold.

Okay. Interesting. Sovereign AI, neoclouds, sort of the next tier of cloud companies below the hyperscalers, what does opportunity look like for Seagate?

Yes. It's a complex space. There's a lot of different terms being thrown around. And the way I think about it is neoclouds generally tend to be compute data centers, if you will, that are failover or being used by the major CSPs, not entirely, but can be. So there's not a whole lot of hard drives in there. And where do they get their data when they need it, they get it from major CSPs or the big installs that already exist in the world.

Sovereign is a little bit different because sovereign could be more, what I would call, traditional data centers, not just compute data centers. And in a lot of cases, people are saying, "I want my data right here next to me." So that's the term sovereign. We are seeing some of those people build out want to have aspirations themselves. Some of them are still very beholden to the architectures of the CSPs. So they're asking the CSPs what to build and they're just plugging that in. But some of them have their own architectures as well.

Is that the way I should interpret that is you're not seeing them as direct customers, but more in terms of demand coming through your hyperscalers. Is that...?

A few direct customers, but still very small. And I would say more of the sovereign behavior has been, "Hey, I know somebody is going to need this data center equipment. I'll build the shell or I'll start plugging gear into the shell that I already have and then I'll go find people that can use it at scale."

Got it. Okay. All investors today associate the Seagate investment bull case with HAMR. Maybe talk about HAMR why that's a strong differentiation for Seagate? And do you think that differentiation is defensible against competitors?

So HAMR is heat-assisted magnetic recording in case you're not familiar with it. It's really our ability to use a different material set on our disc, iron-platinum alloys. And what I would say is that the problem with making the discs out of iron-platinum is not just making them, but it's also being able to write them. So you need -- we needed a large amount of photonics integrated circuits, PICs, and lasers to be able to write spot sizes that are 20 or 30 nanometers in diameter. It's been a miraculous journey, really, frankly.

We're now at the point where we've shipped millions of HAMR drives, not 1 million or 2 million, many more than that, too. And we've got qualifications, a little bit of new news here as we've got qualifications against all planned CSPs now on Mozaic 3. We've even got the 4 terabyte qualified at 2 places, which we've already announced. So the qualifications are going very, very well. We're very confident in the technology, and we can get it to 5 as well.

This is our desire to bring more exabytes into the world. I can't really speak about competition right now because they'll probably figure out their own plans. But I would say that as we keep working really hard to keep marching up the technology curve, I think there is the best way to answer what the world ultimately needs, which is more exabytes fast.

Okay. Congratulations on the additional qualifications there. Maybe now thinking about the implications of HAMR from a financial perspective, how do you think about it leading to market share increases versus lower cost and driving more of the financial returns from a lower cost equation?

Yes. I've said this many times, we don't really think about market share. I think about how many wafers I'm going to start and whether I've sold them or not and what the economics of that is. And we're trying to get predictability not only for ourselves, but also for our customers for the next, say, 4 quarters or 5 quarters. To the extent that we go from 3 to 4 to 5, we're doing that without adding significant componentry. So the cost leverage we get is really good, actually, and we get a lot more terabytes exabytes out of that.

So that's the way I think about the advantage to the extent that you're making a 40 terabyte or then a 50 terabyte, which we talked about in our earnings call, we'd be doing by the end of next calendar year. That's a huge value proposition for power, space into a data center if you're going to be running that data center for 5 to 10 years, like that's the way our customers think about it. So I think those products are going to get very good looks from the customer, and we're going to be able to book them predictably.

There is another wild card in this and that there's still a lot of capacity, that's actually at 20 terabytes. So if you think about how you're making a 20-terabyte drive today, how many critical heads and media it takes versus at 4 terabytes per platter, 20 terabytes necessarily and 5 disks. 5 terabytes per platter, it's 4 disks. So you're able to address those price bands with significantly better cost perspective. And so that's the way we think about it, one of the reasons we're driving areal density so hard.

Got it. I have a follow-up on that, but I'll come back after this question because this will be related. Mozaic 3 took years to qualify at the first hyperscaler. Mozaic 4, you qualified at a much shorter duration, like PMR equivalent times at 2 of your largest CSPs. What specifically changed on the customer side? Are customers much more familiar with the platform? What's changing the time lines around the qualification?

Yes. I think 2 things. The first is that customers -- it's a new technology. So customers were maybe a little bit reluctant at the start. And it's -- anytime you're making big areal density jumps, people will look at it and say, "I have to make sure I check it out." And they did. And we had some early issues as well. So we've got through a huge learning curve as we went up the volume ramp.

I think the second thing is that there was a lot of noise about HAMR being different. And the reality is, once I deliver the drive to you as a customer, it's not different. It's -- but to the extent that you're not ready for a 30-terabyte drive in your architecture, the way your software runs. Now I think people are, they're ready for 30, they're ready for 40, they're ready for 50. So I think the adoption will be much quicker. And we're trying to do this all very predictably. We'll start those head wafers when we know we have demand, and we know we can get through the qualifications quickly and so on and so forth. So that's what we're in the middle of right now.

So the expectation should be, which I had for you as a follow-up is Mozaic 5 and 6 should be qualified in a much quicker time with the same group of customers.

I think so back to the traditional qualification cycle.

Okay. And going back to the last question, when you said a lot of customers still are on 20 terabytes. Now do you see them then taking the leap to 40 terabyte drives with Mozaic 4 or still opting to go through that transition with Mozaic 3 then to 4?

I think there are a lot of different use cases in the world where people really don't need that much capacity depending on what's going on, especially in small data centers. It could be on the enterprise edge, which I know needs a little bit better definition out in the world. But our traditional enterprise customers may be running small, medium businesses or even large businesses. And so they'll still be able to use low capacity drives. So not everything will move to the highest capacity.

Got it. As you think about the improvements that you're delivering on cost per terabyte with the ramp on HAMR, you've talked about a 50% exabyte crossover to HAMR in the second half of calendar '26. Once you get beyond that crossover point, what happens to the cost per terabyte curve? Like what does it look like once you get over that 50% threshold?

Again, the same number of components in the drive. If we can keep pushing areal density, the cost stays relatively the same. So the more terabytes we get out, the better. And the demand for terabytes is so aggressive that I think some of our customers are willing to work with us on the -- not only the new technology, but any new features that will get us there. So with that fundamental cost floor, if you will, I think we're set up pretty well to continue margin expansion. That all comes down to exabyte demand, though, right?

Got it. Maybe on the technology side, you've focused on HAMR and that's one approach that we've seen. There are other approaches we've seen from some competitors in terms of looking at track layout being different. How do you think about what makes sense longer term? Do different technology approaches, including exploring alternative areas around track layouts, do those have a position in your road map as well long term?

Yes. I mean things like SMR get discussed a lot. We started shipping SMR, I think, 2013, when we went from 5 terabytes to 6 terabytes. So we've shipped literally hundreds of millions of SMR drives into desktop markets and things like that. So SMR is a trick we can always pull out of our bag. There are other track layout tricks that may be useful over time. And we -- by the way, we're already doing that for major CSPs as well, just to be clear. So it's HAMR plus, if you will.

One of the things that I'm most interested about the industry right now is that, yes, you get through the transition of being able to use this new media type and getting the photonic circuits, right? There are fantastic new road maps now PICs on photonics circuits and on lasers and being able to use some of the other technologies that exist in the world and take pieces of that and apply them to areal density. So there's a lot of ideas on the table now, and I don't think our areal density road map is going to stall at all. I think it's going to increase.

And so it will be largely building on top of the HAMR platform, but with additional innovation on it.

I think we'll be using iron-platinum for quite some time. It gives a much stronger signal to be read back at small bit sizes. So bits are about roughly 30 nanometers wide by 7 nanometers down track. So they're already pretty small. You have to make them even smaller. But the HAMR media gives you a great signal out of something that's that size. So we have room to run.

Got it. So maybe just flesh out how you're thinking about density of the drive itself? Like when you think about nearline being the clear driver of growth here, you've done a quite robust road map in terms of technology innovation, when you look past the road map beyond like Mozaic 6, for example, what are the big 2 or 3 technology bets you will have to take today to enable that?

Again, back to photonics, I think we're going to be pouring information -- technology into making the world's smallest edge-emitting lasers and making them behave the way we want and as low cost as we can get them. And then there's a lot of technology to be brought to bear at integrating the photonic circuits level.

I think recording channels, still wide open, a lot of progress being made in silicon. There's a lot of progress made on things like GPUs and TPUs and learning so that we can actually apply that learning back to what's going on in each individual drive. And so that's one of the reasons I'm very excited about. We have research efforts into all these.

And when you say you will make those investments or efforts, is it all organic? Or do you see that there's a need to do acquisitions to fill the portfolio on that front?

Not afraid of doing acquisitions if we saw the right technology, but we -- some of our technology has always been so bespoke to our application that we'll be doing it. It will be organically led for sure.

Okay. Okay. So before prices of NAND went up where they are today. The popular bear thesis on hard disk drives was that you're ceding share to SSDs incrementally. How do you think about that sort of long-term road map now? And is there an overlap with SSDs when it comes to the warm tier of storage? Are you seeing that more and more becoming more prevalent? Or with the price increase in NAND, is that actually reducing in terms of how much you overlap with SSD?

Yes. I think a lot of this comes back to what happened in the PC. PC forever, the capacity points in the PC were growing and then they stopped and they're still stopped. And this still -- the PC is still the same capacity point it ever was. When that happened, it transitioned from hard drive to flash and it has been flash for a decade, right? That's the way I think about it.

So a lot of people -- it's a very natural emotional thing. You say, well, then the hard drives are going away. Not in the data tiers and the data center. As a matter of fact, I think the data tiers there are not only stuck on hard drives, but they're stuck for a long way into the future. And that's because people know how to manage them really well and the economics of replacing the amount of data that's in the data centers with flash, it doesn't make sense.

And I say this all the time, flash is a great technology. It has a lot of little niches to put product in. And there's different kinds of flash. There's fast and there's more, I'll call it, long-term flash. But it's -- it can be tricky to manage, and so on. Our customers and the data tiers and the data centers understand all this really well. They're some of the best storage architects on the planet. They've already figured out exactly how they're going to deal with hard drives and memory layers above. And I don't see that changing. If anything, now that flash has gotten considerably more expensive, they're asking the hard drives, how much more can you do at your tier and then the demand just keeps growing. So you talk about zettabytes worth of storage capacity inside of disk drives to replace that with flash would be ridiculous investment. It's not going to happen.

Makes sense. Let's talk about pricing. A big shift for the industry, particularly on a revenue per terabyte basis, I think you, your peer Western Digital is seeing the same thing. Just talk about the pricing power that you now see, particularly when you engage with hyperscalers and these pricing discussions, how has the balance of power in terms of pricing change?

Yes, I wouldn't say it's power. Fundamentally, it comes down to demand. If -- when we started out of the depths of the last downturn, we said, "Hey, we want to get paid for what's coming out of our factories, and we cannot just build it speculatively." And because of that, we would locked in certain pricing for, say, 4 quarters or 5 quarters back to the BTO discussion. As we -- as those roll forward, different customers are on different time lines, of course, you get to the end of that cycle and you say, "Okay, now there's more demand. And then so let's up." But we're still going for predictability, and that's what's driven that CSP pricing up.

The way we test this all the time is if there happen to be any swaps, anybody says I don't need product or we get a little bit more out of our factory and can take the parts that we're already making and make a different product out of them, then what happens is we go out and offer that to the markets and we see where the new pricing is. And so that's the biggest reason for things to go up. We don't see that stopping.

Okay. So maybe let me take a step back and try to address the question with a more longer-term focus is when you look pre-COVID, most of your gross margins were in that sort of 20%, 25% range for most years. Today, both you and Western Digital are progressing to 50% gross margin. You're almost there at this point. One part of it is the demand cycle, as you said. But if the demand cycle were to change in a few years, is the company structurally a better margin than what it was pre-COVID?

Yes. I think given visibility -- of course, demand is the fundamental answer to everything, right? But given visibility that we have into the demand, there's -- we actually took a lot of supply offline by virtue of what we went through after the COVID downturn. And so right now, lead times being what they are, that we can't really react on the upside to the demand. And in some respects, if demand were to take -- were to start going down, we have visibility into it, we would take necessary action, I think, again. So -- but I don't see that really right now.

What's happening in the application layer is profound. So the way I think about it, and I'm a hardware person, so you have to be careful with me, some of the software layers before that used to siphon money off the top are not there anymore. And they're -- and now what's exposed is, "Hey, I've got this really fast application, it needs data." And I think that's more of what we're seeing in the world.

Okay. So you mentioned the long lead time. And what a lot of investors ask us about is hyperscalers have been increasing their capital spending plans every time they get on an earnings call, et cetera. With the kind of long lead time that you have, the build-to-order framework that you have, what's the typical duration between a hyperscaler saying, well, we'll spend more to Seagate seeing that impact on the revenue line?

I think the hyperscalers understand what their -- what workloads are actually driving them and they are seeing these new application spaces, say the video for example, they know how much video is being created all the time at the extreme edge. They know how much video could be if it was enabled properly. They know the sources of data on the edge. So they know exactly how much people are uploading to their properties all day long. And so they get a pretty good sense of those businesses.

There's new nascent spaces like people talk about physical AI and robotics that will need mass capacity. Is that going to be serviced in the cloud or at the edge? Very, very good questions. All these applications are being forecast into the future. And and we'll have the storage components as well. So I think that's one of the reasons why they're giving us that 2-, 3-, 4-year visibility, even though we haven't locked in the BTO orders, we are getting very good visibility and a nice solid growth. That's another thing different about this demand cycle.

Got it. Okay. In the last 5 minutes, let's try and get through three questions here. A big talking point has been your 70% incremental margins that you've been delivering consistently. Talk to us about the sustainability of that. The -- as you go through the ramp on HAMR and you continue to deliver more exabytes, how should we think about sustainability of the 70% incremental margins that you've been delivering?

Yes. Again, fundamentally, demand will dictate everything. And our execution to the extent that we can, if we get through the 4 terabyte to 5 terabyte transition aggressively, while the demand is still high, I think that will serve us very well. We have all the assets in place that we needed to do that. Our team is executing just fantastically. So that's what it really comes down to is from an OpEx perspective.

Our team is aligned on the right technology, knows exactly how we got to 3 and 4. Now know what we need to do to get to 5. We have all the tools in place. We've been planning this technology transition for quite a while. So I think it is sustainable.

Okay. Does pricing and how the pricing environment give you a lever to even exceed the 70% of the incremental margins?

Pricing will always come down to demand. I do think that the difference between a 3 terabyte and a 4 terabyte and a 5 terabyte is pretty compelling if you're building a new data center or replacing the hundreds of millions of drives around in existing data centers. So from a power perspective, the rest of the equipment you need to support that, the number of exabytes you're running, I could see if the demand continues to remain strong, then we would have that kind of capability.

Okay. Operating leverage, significant expansion on operating margins, I think, 1,400 basis points year-over-year in the March quarter. You've been holding OpEx relatively flat. And as the business now scales towards a $20 billion annual run rate by fiscal '28, where does -- how do you think about the OpEx envelope that needs to support that overall revenue profile?

Yes. I mean given the downturn we went through, our first priority was giving our people raises and they deserve the bonus they've been getting the variable comp and so on. Again, we have the team, the technologists together to do what we need to do. We could spend more on OpEx over time, but I don't really see the need to right now hiring more people, great, but that it would take a long time to train them. We are just really focused right now on using the existing team that we have to go get 5 terabytes of platter going, and then we'll see what happens in the future.

Does broadening out the customer base or investing in these new technologies that you're seeing talking about photonics, does that impact the R&D space? Or is that largely sort of similar to the run rate you've been on?

So far, we've been doing that inside of our existing footprint. And we have thousands and thousands of engineers that are in tune with those technologies and markets. But again, we could if we wanted to. I still don't see the need right now. We love the footprint we have. We love the team we have. And we're trying to stay focused.

Okay. Let me end with capital allocation. Clearly, you'll be generating a lot of cash. How are you thinking about capital allocation primarily in terms of how do you invest, where do you and how much do you invest in the business, how much do you allocate to M&A or buybacks? How are you thinking over to the drivers?

Yes. No real change. I mean -- so I think I said this on the earnings call that we just had -- last year was all about patching up the supply chain, just getting working capital back. This year is about readdressing our debt and getting in fighting shape in case there is some other downturn, we don't see that, but we just want to be the best stewards that we can.

And then we'll get back to being the Seagate that we want to, which is returning 75% of the cash to shareholders. From -- and we have a strong dividend, we always did. We're going to be reassessing exactly how that weighs off versus buybacks, but we believe that this is shareholders' money, and we want to give it back to them.

So inside of that, I think we have enough flexibility to do anything we need to whether it's investing back in our own R&D or if we need to do small tuck-ins or something like that for technology access, we can do that. But we're not really focused that way. We're more focused on just getting through this year, getting that 5 terabyte per platter launched and seeing what the next year holds.

Okay. Great. I'll wrap it up there. Thank you for coming to the conference. Thank you to the audience as well. Thank you, everyone. Thank you.

Good afternoon, and welcome to the Seagate Technology Fiscal Third Quarter 2026 Conference Call. [Operator Instructions]

Please note this event is being recorded.

I would now like to turn the conference over to Shanye Hudson, Senior Vice President of Investor Relations. Please go ahead.

Thank you. Hello, everyone, and welcome to today's call. Joining me are Dave Mosley, Seagate's Chair and Chief Executive Officer; and Gianluca Romano, our Chief Financial Officer. We've posted our earnings press release and detailed supplemental information for our March quarter results on the Investors section of our website.

During today's call, we'll refer to GAAP and non-GAAP measures. Non-GAAP figures are reconciled to GAAP figures in the earnings press release posted on our website and included in our Form 8-K. We've not reconciled certain non-GAAP outlook measures because material items that may impact these measures are out of our control and/or cannot be reasonably predicted. Therefore, a reconciliation to the corresponding GAAP measures is not available without unreasonable efforts.

Before we begin, I'd like to remind you that today's call contains forward-looking statements that reflect management's current views and assumptions based on information available to us as of today and should not be relied upon as of any subsequent date. Actual results may differ materially from those contained in or implied by these forward-looking statements as they are subject to risks and uncertainties is associated with our business.

To learn more about the risks, uncertainties and other factors that affect our future business results, please refer to the press release issued today and our SEC filings, including our most recent annual report on Form 10-K and quarterly report on Form 10-Q as well as the supplemental information all of which may be found on the Investors section of our website.

Following our prepared remarks, we'll open the call up for questions. In order to provide all analysts with the opportunity to participate, we thank you in advance for asking 1 primary question and then reentering the queue.

With that, I'll hand the call over to you, Dave.

Thanks, Shanye, and hello, everyone. Seagate delivered a very strong March quarter. underscoring both the durability of demand and the leverage in our model. We grew revenue 44% year-over-year, achieved record gross margins more than doubled non-GAAP operating income and generated 1 of our highest ever levels of free cash flow at close to $1 billion.

Momentum continues to build for our Mozaic HAMR-based platforms with 2 of the world's largest CSPs now qualified on our 4 terabyte per disk product. For both of these customers, qualification time lines were in line with PMR products, underscoring the maturity of the platform, and our team's outstanding execution as we work to meet customers' accelerated demand requirements.

Our strong Q4 guidance issued today demonstrates our growing conviction in the business and future opportunities. As we look ahead, we see Seagate now entering a period of structural growth.

Our belief is rooted in 3 pillars. First is the sustainability of rising storage demand. AI-enhanced applications are accelerating data creation, expanding retention and increasing reliance on historical data sets for advanced reasoning. Extending beyond cloud data centers to the enterprise edge, these trends require storage solutions that deliver cost and energy efficiency at scale, making high capacity hard drives essential to modern data center architectures.

Second is our strategic technology road map. Anchored by the Mozaic platform and HAMR innovation, we are delivering critical technology breakthroughs at the right time to support our customers' rising demand now and into the future.

Third is our proven strategy focused on converting demand into profitable growth and value creation. Our build-to-order model enhances demand visibility and supports pricing and supply discipline. Our HAMR-based product road map enables margin expansion as we scale. And our capital allocation framework enables us to leverage our earnings growth and cash flow generation into strengthening our balance sheet and enhancing shareholder returns over the long term.

The combination of these pillars, robust market demand, a proven technology road map and disciplined operational execution is already driving performance ahead of the financial targets outlined at our analyst event a year ago. The progress we have made gives us confidence to significantly increase our annual revenue growth target from the low to mid-teens to a minimum of 20% over the next few years.

This confidence is reinforced by the strength of the current demand environment shaped by ongoing momentum from cloud investments. The March quarter marked our tenth consecutive period of revenue growth from cloud customers, who have committed hundreds of billions of dollars in infrastructure CapEx investment to support their own long-term growth in AI transformations.

Using remaining performance obligations, or RPO, as a proxy for future revenue potential, the top 3 global CSPs alone have nearly doubled their RPO to staggering $1.1 trillion, a clear indicator of sustained growth ahead.

Assurance of reliable supply is our customers' highest priority particularly for nearline products, which accounted for close to 90% of total Exabyte shipments in the March quarter. We have Exabyte scale supply agreements in place with nearly all major cloud and hyperscale customers, with nearline capacity almost fully allocated through calendar 2027.

At the same time, we are finalizing build-to-order contracts with these customers through the end of fiscal 2027, which defines specific configuration and pricing. Our value-based pricing approach enables customers to plan with confidence while contributing to sustained profit growth for Seagate. And we are actively engaged in strategic planning discussions now reaching into calendar 2028 and beyond.

Today, AI sits at the center of nearly all customer demand conversations. We are in the midst of an inference inflection where compute infrastructure is shifting from periodic training to becoming engines that continually generate mass capacity data. Leading AI chatbots now handle billions of user prompts daily, each consuming and producing multimodal outputs that fuel an unprecedented surge in data creation. Agentic AI pushes this even further, transforming sporadic engagements into autonomous workflows that continuously ingest inputs, generate reasoning and store durable outputs that are dramatically increasing data intensity and long-term storage requirements.

AI is amplifying demand across existing applications such as video, where large cloud providers are integrating AI into platforms to boost user engagement and revenue opportunities, driving new video creation and the need to store it. We believe demand will further accelerate as AI applications move beyond the data center into the physical world, powering manufacturing systems, autonomous vehicles and robotics. These physical AI deployments generate massive data streams from sensors, cameras and telemetry with a single autonomous vehicle producing up to 4 terabytes per hour. A portion of this data is reused for simulation, validation and retraining with retention requirements stretching 5 to 10 years to meet compliance standards.

These inference-based applications are creating a growing need for both cloud and local storage. We have started to see interest from sovereign and neo cloud data centers for our enterprise nearline drives and system solutions.

To manage these intensifying workloads, cloud and edge data centers deploy storage tiers that work in concert to optimize performance, cost, energy efficiency and data durability.

Hard drives are critical to these modern data center architectures, delivering scalable capacity along with energy and cost efficiencies and that form the foundation of the mass data storage tier.

Seagate's proven product portfolio makes us well positioned to address this broadening opportunity set. Our technology strategy prioritizes aerial density innovation over increasing unit volumes to address rising demand. Leveraging our technology strengths, we provide the most capital and manufacturing efficient path to scale while delivering improved cost and power efficiency per terabyte for our customers. This approach supports our goal to supply data center exabyte growth in the mid-20% range.

Our Mozaic 4+ platform is a prime example. As our second-generation HAMR-based product, Mozaic 4+, can deliver up to 44 terabytes per drive, over 30% more capacity compared to the first-generation Mozaic drives, which we achieved with the same number of disks and heads with minimal change to the bill of materials.

Mozaic 4+ also incorporates our internally designed laser and integrated photonic circuitry into the recording head. This innovation enables high-volume, extreme precision manufacturing that enhances our ability to increase drive capacity and cost efficiency.

We began revenue shipments from Mozaic 4+ in late March, and based on current ramp plans, we expect Mozaic 4+ to represent a majority of our HAMR exabyte shipments exiting calendar 2026.

We have shipped millions of HAMR-based drives highlighting our ability to engineer with atomic-level precision and then integrate that innovation into high-volume exabyte scale.

We work closely with customers to ensure our technology road map aligns with our future storage capacity and performance needs. Customer feedback consistently indicates that tiered storage architectures and software solutions meet their performance needs over the next few years. Capacity scaling remains their top priority, and we are executing the plan.

Our Mozaic 5 product development is progressing well to plan to deliver capacities at 50 terabytes with qualification shipments targeted for late calendar 2027. These drives leverage our advanced photonics expertise internally designed laser and mature tends platform to extend the aerial density capabilities. This approach offers customers a predictable path for addressing their future exabyte growth needs as well as upgrade the storage capacity of their installed base, while using the same power budget and lower space and the momentum we've seen in qualifying Mozaic products continues to validate this approach.

Today, the vast majority of HAMR supply is allocated to cloud and hyperscale customers. However, as production scales. We expect to leverage 4 and 5 terabyte per disk capabilities to produce cost-efficient, lower capacity products for enterprise data centers and edge IoT applications. This unified platform approach will simplify our product portfolio and enable manufacturing, supply chain and cost efficiencies to deliver strong economics for Seagate over the long term.

In summary, Seagate is entering a period of structural growth, powered by durable demand, increasing adoption of our Mozaic-based products and continued execution against the strategy designed to drive margin expansion, cash flow and long-term value creation.

I want to thank our global team for delivering another strong quarter and recognize our suppliers, customers and shareholders for their ongoing support.

With that, I'll turn it over to Gianluca.

Thank you, Dave. Seagate posted very strong results for the March quarter, exceeding our expectation of revenue, operating margin and earnings per share, while setting new profitability record that reflects sustained data center demand.

Additionally, we further strengthened our balance sheet by retiring $641 million in gross debt and achieved free cash flow margin of 31%.

Revenue for the March quarter was $3.1 billion, up 10% sequentially and up 44% year-over-year. We achieved non-GAAP gross margin of 47%, up 180 basis points sequentially, and we expanded non-GAAP operating margin by 560 basis points sequentially to 37.5%. Our result in non-GAAP EPS was $4.10, up 32% quarter-over-quarter and 115% year-over-year.

We shipped 199 exabytes in the quarter, up 29% year-over-year. The data center market accounted for 88% of exabyte shipments and 80% of revenue, with strong demand contribution from both global cloud and enterprise customers.

We shipped 175 exabytes into the data center market, up 6% sequentially and 47% year-on-year. Data center revenue increased even faster over the same period, up 12% sequentially and 55% year-on-year, totaling $2.5 billion.

Cloud makes up the vast majority of data center revenue and capacity shipments. We are focused on ramping Mozaic to address growing cloud customer demand. In the March quarter, we shipped Mozaic drives for revenue to 75% of the leading global cloud customers, and we remain on track to complete qualification with the remaining 2 customers in the current quarter.

In the enterprise OEM data center market, we saw a notable sequential revenue increase, reflecting growing deployment of AI application, along with renewed demand for hybrid and tier storage architectures. This proven strategy widely adopted by cloud and hyperscale customers, provide scalable and efficient infrastructure solutions across all market conditions.

Our edge IoT market made up the remaining 20% of revenue at $612 million, up 2% sequentially.

In the client and consumer markets, high supply and higher NAND cost offset the typical seasonal demand slowdown in the March quarter.

Moving on to the rest of the income statement. Non-GAAP gross profit increased to $1.5 billion, up 23% quarter-over-quarter and 87% compared with the prior year period, growing roughly twice the rate of revenue.

Non-GAAP gross margin expanded to 47% in the March quarter from 42.2% in the prior period. This improvement reflects continued execution of our long-term pricing strategy, along with improving product mix.

Together, this drove a mid-single-digit increase in year-over-year data center revenue per terabyte. We expect this trend to continue, supported by a strong demand environment.

Non-GAAP operating expenses were in line with our expectations at $296 million or 9.5% of revenue. The combination of higher revenue and expense discipline enabled us to achieve our long-term target earlier than originally planned.

As we effectively execute our strategy around advancing aerial density, supply discipline and pricing, we delivered a 30% sequential improvement in non-GAAP operating profit to $1.2 billion or 37.5% of revenue.

Other income and expense were $62 million, reflecting lower interest expense on the reduced outstanding debt balance. We expect other income expense to remain relatively flat in the June quarter.

Non-GAAP net income grew to $934 million, with corresponding non-GAAP EPS of $4.10 per share based on tax expenses of $171 million and a diluted share count of approximately 228 million shares, including the net impact of our 2028 convertible notes.

Turning now to cash flow and the balance sheet. We invested $151 million in capital expenditures for the March quarter or roughly 4% of revenue year-to-date. We expect capital expenditure for fiscal year 2026 to be inside our target range of 4% to 6% of revenue, with investments aimed at the ongoing transition and ramp of hammer based products.

Free cash flow generation expanded significantly to $953 million, up 57% from the prior quarter, representing our highest level in over take. We expect free cash flow generation to improve further through the remaining quarter in calendar '26, supported by sustained demand trends, operational efficiencies and capital discipline.

Cash and cash equivalents increased to $1.1 billion at the end of March quarter with ample liquidity of $2.4 billion, including our undrawn revolving credit facility.

During the last quarter, we recorded approximately $191 million to shareholders through dividend and share repurchases. We also retired $641 million in debt including over $600 million of exchangeable senior note due 2028 using cash on hand. Our adulting gross debt balance was approximately $3.9 billion exiting the March quarter. Year-to-date fiscal 2026, we have reduced gross debt by approximately $1.1 billion.

Net leverage ratio improved to 0.7x based on our adjusted EBITDA of $1.2 billion for the March quarter, up 28% quarter-over-quarter and more than doubled year-on-year. We expect the net leverage ratio to continue declining as profitability and cash generation increase while we plan to further reduce debt.

I'm pleased to share that Fitch recently upgraded Seagate credit to investment grade, recognizing our strengthening balance sheet and profitability expansion.

Turning now to the June quarter outlook. Despite rising geopolitical tensions, including the ongoing complete in the Middle East, we do not currently expect material impacts to the business. Our teams acted quickly to mitigate supply and logistics disruption, and we will continue to monitor this dynamic situation.

Underlying demand fundamentals have not changed. AI is reshaping data into a strategic asset, accelerating our customer need for storage capacity at scale. We see strengthening exabyte demand and continue to execute our Mozaic product qualification alongside our pricing strategy. With better context, we expect June quarter revenue to be in a range of $3.45 billion, plus or minus $100 million, which represents a 41% year-over-year improvement at the midpoint.

Non-GAAP operating expenses are expected to be approximately $295 million. Based on the midpoint of our revenue guidance, non-GAAP operating margin is expected to be in the lower 40% range.

Non-GAAP EPS is expected to be $5 plus or minus $0.20, based on a tax rate of about 16% and non-GAAP diluted share count of 231 million shares, including estimated dilution from our 2028 convertible notes of approximately 3 million shares.

Our financial performance and guidance demonstrate our focus on profitable revenue growth, alongside a product strategy designed to capture the significant opportunities ahead. Combined with the visibility gain through our customer agreements, we are confident in delivering quarterly revenue growth and margin expansion through fiscal 2027, positioning Seagate to enhance value for both customers and shareholders over the long term.

Operator, let's open the call up for questions.

[Operator Instructions] Our first question today is from Erik Woodring with Morgan Stanley.

Awesome. Congrats on the results and guide. Dave, I was wondering if you could go a bit more into the detail on specific tailwinds to HDDs from Agentic AI? And I guess, meaning like the broad tailwinds to HDD storage demand for multimodal models and physical AI is pretty clear, but it's less clear exactly what parts of the adjunctive are ripe for HDD. So I guess my question is specifically how does Agentic AI-benefit HDD demand? And does that have any impact on how you think about that mid-20% nearline exabyte CAGR you provided at Investor Day a year ago?

Thanks, Erik. Yes, we are picking up confidence because of some of these new applications. I think it's important to realize that some of the applications, while important, are fairly small data applications, some drive enormous data sets. And so when I think about Agentic AI, I think about frequently asked questions, you're -- rather than just periodically querying something you're doing as part of workflow. And when you do that, you may actually reference enormous data sets to draw your conclusion then you may actually create new data that needs to be propagated out in the world to the extent that that's unstructured data, video data, that's where it's actually hitting the storage tiers fairly hard. So not entirely related to mass capacity storage, but we're starting to see a lot of this pick up.

The next question is from Asiya Merchant with Citigroup.

Great. Great set of numbers here. Congratulations. If you could just talk a little bit about cost reductions, pretty impressive year. You guys are on the second-generation Mozaic now. How should we think about these cost reductions? And if you could just update where you think you would be for HAMR? I think you said majority of them exiting, I think, fiscal '26, if I heard that correct. But if you could just update us on where you are with the HAMR targets and the blended cost reductions we should expect as you ramp into the second generation?

Yes. Thanks. That's 1 of the reasons we try to change as little as we can, platform to platform. It just -- it derisks the product transition, but it also allows ourselves and our suppliers to leverage all the installed base as well. So we're trying not to change as many parts as we possibly can. Obviously, there are technology changes. Most of those are under our control and that heads and media. But right now, because of the momentum that we're seeing with the HAMR road map, we're seeing that we can get more aerial density for fairly small changes inside of our portfolio. Most of it affects the laser and the photonic circuitry and the material set on the media like we talked about. So all of this is not netting out to much of our bill of materials change and therefore, we're getting a lot of the cost leverage.

Gianluca, do you want to expand on that?

Yes. I'll say, if you look at our last several quarters, the cost reduction was coming from mainly 2 items. One is for sure are the mix going to higher capacity drives. And second was the full utilization of our manufacturing. When I look into the future, of course, now we are full. So that part maybe will not be so important in terms of cost reduction, but our mix change continued to be very fast. We are going faster than what we were thinking on the transition to HAMR. And now that we have second generation HAMR now, we have a very good increase in terabytes per unit. And of course, this is the driver, not adding more below material to the disk, which is the main driver for the future cost reduction.

And I think, Asiya, your second question was around where we stand in terms of HAMR. Dave had mentioned on the call that we would expect towards the end of this calendar year for Mozaic 4 to cross over with Mozaic 3, and then we remain on track for overall HAMR exabyte crossover at that time as well.

The next question is from Samik Chatterjee with JPMorgan.

Maybe on the pricing side, pretty strong price increase, both on a year-over-year and quarter-over-quarter basis here. I know you sort of expect these pricing trends to continue. But just trying to think through why shouldn't we see pricing maybe accelerate a bit as more new contracts come into play as you go through sort of end of 2026 and 2027, why shouldn't sort of -- how should we think about pricing? And why should it sort of accurate more as more new contracts come into the P&L from here on?

Yes, thanks. The first way I think about it is what is the true demand and I think the demand is rising to your point, further out in time as we roll out of 1 LTA and into the next, then the market demand dictates what the economics. We talked about this a little bit in the prepared remarks about when we set exact capacity configurations, what products are qualified with what customers and therefore, what price. As we've been rolling forward, though, we have the ability to -- just a few more drives out of manufacturing or whatever. So we can always test what that demand is and the demand keeps going up. And so we're seeing what the market price, if you will, is our goal still is to try to lock in with our customers and give them predictability so that they have a great economics plan to build their data centers out and we know what we're going to get paid for, for what we start in our factories.

Yes. We have now finalized our build-to-order for our fiscal '27. So we see how the pricing is trending, how say, there are no changes to our pricing strategy. We are continuing to execute the strategy that allowed us to increase profitability for the last 12 consecutive quarters. And based on those orders that we have now finalized in terms of mix, in terms of pricing, in terms of volume. We said that for the next 4 quarters or for the fiscal '27, we are confident in saying that we have a good opportunity to increase our profit and our revenue sequentially through the fiscal '27.

The next question is from C.J. Muse with Cantor Fitzgerald.

I guess another question on Agentic AI. And particularly as you think about the need for large-scale data lakes and overall demand for persistent memory, is this changing perhaps your product structure road map. I know you announced a partnership with NVIDIA. Curious how this is augmenting kind of your product road map? And also does this change kind of your thinking around supporting demand via only aerial density improvements?

Yes. It's an interesting question, C.J. I think architectures still are largely driven the same way they were a couple of years ago, which is -- and we said this before, our customers want more capacity per spindle and that's our highest priority. And so we're still racing on aerial density exactly to your point. There are a lot of conversations about performance tiers. Can we get a little bit more performance out of the drive. And so for example, we've talked about this in the past. We had stacked actuator designs in the past. We've shipped millions of those drives in the tens of exabytes range. for performance tiers, and we can certainly pull those designs back down off the shelf. But I would still say, while those discussions are happening, the biggest driver for us is get more capacity per drive.

Yes. On Agentic AI, you need historical data for agents to reason, and you need to store that data for compliance. So we see those huge benefit to our business.

The next question is from Wamsi Mohan with Bank of America.

So you generated almost $1 billion in free cash flow, so over 30% plus free cash flow margin in the quarter. And given your view that you're entering a new structural growth, how should we think about how you're going to deploy this cash beyond sort of the next 12 months where I think you said you're going toward higher that. And a quick clarification around pricing. When you say you have pricing locked in for fiscal how much of the capacity for fiscal '27 has pricing been locked in? And how much is sort of floating at the moment?

Yes. We said the vast majority of our nearline capacity is allocated during the next 4 quarters. So of course, it's not 100%, but it's a very high percentage.

On capital allocation, in the last few quarters, we have focused a lot on reducing our debt, especially our convertible because somehow that would have created even more dilution. We still have about $400 million of the convertible that is open, but we will probably address this quarter or next. So a little bit of a reduction in debt and then I would say the majority will probably go to share buybacks. Now we are active already today in the market, and we will probably do more in the next few quarters.

Yes, Wamsi, I would say that last year, we were focused very much on working capital and just getting the supply chain back healthy again from what we went through. Now to Gianluca's point, we have to take care of some of the debt that we have. And I think the next place that we go to exactly to your point, is back to where we were before, which is returning value to shareholders.

The next question is from Krish Sankar with TD Cowen.

Dave or Gianluca, on the mid-20% exabyte growth, are you just increasing capacity per unit or are you actually increasing the units or ahead capacity?

Yes. I would say capacity per unit is where our focus is. If you think about it, and a lot of people get this wrong in the when thinking about hard drives, it's a very complex supply chain with many different suppliers coming in. And then there's our critical components that we control, but they have very long lead times, not just for the capital to build more but also for the product itself when it's inside the machines. And so therefore, it's really needs to be well orchestrated in our supply chain. It's not like just plugging in a few more machines to get more capacity out. Our people are very much focused on increasing the aerial density, the amount that comes out of the entire fleet, that's the way we believe gets the most exhibited into the world. And if we took those people off and had them make more parts to your point, we would probably net-net fewer exabytes over the next few years. So we're very focused on with the technology innovation that we see coming in front of us continuing to drive those efficiencies. The customers benefit from those with energy efficiency and efficiency and scale as well. So this is in concert with our customers. This is the way we're driving and trying to be as aggressive as we can.

I'll say the move to the second generation HAMR now is giving us the opportunity to continue to grow and to achieve a target CAGR that we discussed about a year ago. And then after the second generation, we will have the third generation that Dave was mentioning in the prepared remarks, that is not too far in time from now. It's basically at the end of next calendar year, we will be already in call with a 50 terabyte drive. So that is our strategy and all based on technology transition and automating units.

The next question is from Mark Newman with Bernstein.

Congrats on the quarter. Just wanted to double-click on pricing. It seems like on my math, your pricing per exabyte seemed to accelerate a bit something like mid-single digits Q-on-Q. And I wanted to understand, is that more from is that more because you had a higher portion of new contracts signed this quarter versus previous quarters? Or was it just that the magnitude of the price increase on new contracts has gone up. I guess the reason for this question is we're just trying to get a sense of if this magnitude of price increase is going to continue every quarter going forward. Or was this because you had a number of new signed and so perhaps it was a bit higher than normal. I really appreciate any kind of clarity you can get on the pricing dynamic?

Yes. We are not changing our pricing strategy. So as I said before, we have executed this strategy for a long time. and we are continuing to do the same. Every quarter is different, depends from how many new contracts we have in the quarter depend a lot from the mix also know how we move customers from 1 product to the next. But in general, say, there are no changes in how we address our pricing strategy. So we have done that for many quarters. And as I said before, we have the same trend for the next 4 quarters for the entire fiscal '27 and possibly even for longer.

The next question is from Jim Schneider with Goldman Sachs.

I was wondering if you could just maybe frame for us a little bit with a little more precision as we look out, say, towards the end of fiscal '27. Given your pricing visibility, would you characterize your price per exabyte growth year-over-year in those -- for those longer-dated orders as sort of up low, mid or high single digits year-over-year?

Yes, we probably don't guide so far in time. So as I said before, every quarter, we'll be a little bit better and we expect revenue improvement. We expect profitability improvement. A big part of the profitability improvement is coming from pricing, but is also coming from the change in mix and the reduction costs that the 40 terabyte HAMR drive will give us.

Yes, Jim, that's the way I look at it is there's new products coming, higher capacity products and then we said this in the prepared remarks as well, the ability to address some of the lower price bands, if you will, with better products, fewer components in them with -- that's what aerial density provides us, and that's the way we think about it. So fundamentally, it will still come down to demand. as we play out through '27. To the extent that we can get up the ramp faster than we think on yields and get the scrap down and be able to address other people through completing the customer qualification. That product is very -- that allows us to get into those other markets very aggressively. And I think that's where we're focused. And then whatever the ultimate demand is, we don't know, but we think it's pretty high relative to our supply as well. So we'll continue to negotiate with customers to give them predictability and they'll determine what the price is.

The next question is from Amit Daryanani with Evercore.

I just have a question just on gross margins. And if I think about the Analyst Day, you folks talked about 50% incremental gross margin, seems like a while back that happened. You positing 70% plus pretty consistently. I'd Lotto understand, is this outperformance kind of driven by pricing or mix or shift to HAMR. And then importantly, is 70% incremental sort of the right framework to have as you go through the fiscal '27 model?

Yes. I think -- I'll let Gianluca talk quantitatively here, but I think that the strong demand is something that we -- even we weren't focusing on a year ago to your point. And we've executed really well against that, maybe even better than I thought we would. We're pushing aerial density really aggressively and the team has done a great job.

Yes. No, we have executed better than what we were planning a year ago from different drivers. Now I say pricing was a title it better. This was a mix transition a little bit faster. So now we can leverage more on the 40 terabyte drive. So I'll say, yes, I'm looking at what we have done in the last few quarters, and I don't see a reason why we should not do the same in the future. But of course, every quarter is different. So let's see what we can achieve.

Perfect. Congratulate [indiscernible].

The next question is from Aaron Rakers with Wells Fargo.

Maybe I'll stick with the P&L kind of similar to Amit. When we think about the model you framed out at the Analyst Day, I'm curious, as we look into fiscal '27 and given variable comp dynamics, how do we think about operating expenses? I think your prior target was to kind of maintain roughly 10% OpEx to revenue. Clearly, we're now breaking through that. So I'm curious how should we think about the OpEx trajectory going forward?

I would say, think about it as relatively flat. Obviously, if we see the need to go invest more for the technology to drive the technology even further we can. But right now, I think our team is doing very well, and we have a fairly big OpEx portfolio that we can readjust priorities inside of. So I think the way I think about it is relatively flat.

Yes. Just to be sure, flat on a dollar basis, not as a percentage of revenue, as we discussed also in prior quarters, and as Dave said, this is a good level for us. And if we need to do something, we will do it. But right now, we don't see the need.

Next question is from Timothy Arcuri with UBS.

I just wanted to clarify exactly what the message is on units. I know you and your peers stopped giving us units a few quarters back, but there was a big head supplier that did report last night. Again, it heads up 40% year-over-year, and they specifically indicated that its demand from the U.S. HDV guys. So I know maybe some element of it is that you want to prioritize internal head capacity for HAMR. But how does that fit with the idea that you're not growing units? Or are you, in fact, beginning to grow units because of some of these new demand drivers stack?

To first order, Tim, no, we're still not growing units. I mean, inside of the mix, there may be more heads inside of the drive, right? So the average number of heads per drive, which we don't talk about very much, may be increasing. It's not 20 heads, which isn't the highest capacity drive that we have yet. And there's still quite a bit of the low capacity drives that are serving customers that are very important to us. So as we look across that blend, probably more heads of media going into the average drive is the way to think about it. The total number of units is not really increasing. And I don't think it will, unless we see a resurgence at the edge. And so -- and that may be over a long period of time.

Yes. As you know, Tim, HAMR cycle takes a bit longer than PMR. So we use a little bit of PMR heads just to keep the units as they are today. Otherwise, the units will actually go down.

The next question is from Karl Ackerman with BNP Paribas.

You spoke about how the Mozaic 4 platform will command 70% of your HAMR shipments by the end of fiscal '27. But how quickly might HAMR exceed half of your total exabyte shipments? I ask because it seems to support favorable capital intensity. And as yields improve on HAMR, it seems easier and more economical for you to replace lower capacity data center edge hard drives with HAMR had in media.

Maybe let me clarify those percentages. What we said is we will achieve 70% of exabyte, nearline exabyte built on HAMR drive by the end of calendar '27, actually by fiscal '27, sorry. And by the end of this calendar '26, we said the majority of HAMR exabyte, so inside the exabyte will be 40 terabyte drives product versus the 20 terabyte in for that we were building before. So those are the percentages.

But still a quite aggressive ramp on Mozaic 4+. To your point, I think the other way to think about it is our wafer fab is relatively full. And so therefore, everything spoken for. We're making sure we do that blend just right. We're not leaning too hard into the Mozaic for because some of the other product families are still doing quite well and needed for various customers.

The next question is from Vijay Rakesh with Mizuho.

Just a quick question on the margins. Obviously, very solid margin pickup in the quarter and the guide. Just wondering if there's a way to look at it on what's the impact from HAMR mix versus utilization or price? And how does this change with the Mozaic 5, I guess?

Yes. I think as we continue to go up the curve and we can hold the line on new piece parts in the bill of materials, like we talked about before, leverage as much technology that already exists. I think that's where we get the best cost leverage. And again, the technology in the heads of media was fundamentally enables all of this. So we're -- that's why we're investing very heavily. We get more exabyte output as well, and that will help drive margins. I think when you think about a 3 terabyte per platter a 4-terabyte per platter or 5 terabyte per platter drive, that value into the data center is enormous. I mean, it's space efficiency on all the parts around it for -- on a per terabyte basis and then obviously, power and things like that. So the customers lean very hard into those things, and that's why we -- that's giving us great visibility. And as we drive that without adding too much incremental cost, I think that's why our margins are defending.

The next question is from Steven Fox with Fox Advisors.

Congrats on the great quarter. I was just curious if this latest inflection point has anything to do with what seems like a rising cost differential between HDDs per gigabyte and NAND? And if it doesn't, right now, could it in the future sort of help for a future inflection?

Thanks for your question, Steve. I say this all the time, NAND is a great technology. It has many niches that hard drives are not in. So it's -- and we need those niches to continue to grow because they serve data markets, either on the ingest side or on the consumption side as well. But in the storage tiers that we largely talk about inside the data center, I don't see the architectures changing very much. If anything, because of the economics of what's going on right now, people are coming back to hard drives and saying, what can you do? And I think that was referenced to some of the earlier questions on the performance side, is there something else that hard drives can do inside of their tier to make sure they're improving their performance. Again, we get driven very hard to just get more exabytes out. And I think the architects understand this really well. And I see these architectures pretty sticky for a long, long time into the future.

Next question is from Ananda Baruah with Loop Capital.

Dave, I wanted to ask you, just going back to your remark a little while ago about using hammer to go down to lower capacity points. Is Mozaic 4, is that sort of the model that gets either go down to 20 terabyte HAMR? And if so, like at what point of the Mozaic 4 ramp do you think that you guys might have an opportunity to do that?

Yes, we had originally talked about it, Ananda, in that context, Mozaic 4 20 terabyte, if you will, with the 5 disk. I think the demand for Mozaic 4 at the high end is so high right now that as we look forward, I don't think you'll see very many of those, but we'll see how the market plays out over the next 3, 4 quarters. Mozaic 5 obviously changes the economics again. And at some point, we're going to be able to readdress those markets in a very cost-efficient way.

Yes, I would say now in theory it is a great strategy. The problem we have is demand is so strong in the public cloud that we don't have enough volume to also implement with lower capacity base strategy based on the 4 terabyte per disk. So again, possibly, we will address a little bit later out in time.

Yes. So the economics are so attractive at the higher end that it's not worth doing yet. Got it.

The next question is from Tom O'Malley with Barclays.

Reach in DPR, there's been a lot before here on the pricing and the contract side. But just if you look at the NAND industry and potentially DRM as well, you're hearing more about potential prepayments over the course of the contract life. Are you guys seeing that in the market? And would you ever consider this as new contracts come up over the coming years and demand continues to grow just given your production footprint? Would that be something you would consider in the future?

Thank you, Tom. I would say right now, our free cash flow is very strong. So we are not looking at prepayment in particular. I think we are mainly focus on predictability of the shipments and on optimizing our pricing strategy. Now I don't exclude that in the future, we will maybe implement prepayments. But so far, we have not focused on that part.

Yes, Tom, we've really been going for demand predictability. And the customers have to drive through important architectural transitions themselves. We have to drive through the product transitions. We have to make sure all that's synced up. And so that predictability is top of mind for us, not necessarily any other economics.

This concludes our question-and-answer session. I would like to turn the conference back over to management for any closing remarks.

Thank you, Gary, and thanks to everyone who joined us on the webcast today. We're excited about the strong March quarter and the accelerating momentum building for our Mozaic technology platforms as we enter this period of structural growth. We'll keep executing with discipline to expand margins, drive cash flow and build long-term value creation. Thank you for your continued support.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

All right. Perfect. We are going to get started here. So again, welcome to day 2 of the flagship TMT Conference. My name is Erik Woodring. I lead the hardware coverage here at Morgan Stanley. I am delighted to be joined today by Gianluca Romano, CFO of Seagate Technology.

Before we get into things, let me just remind everyone to please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. Gianluca, thank you for joining us today.

Thank you, Erik. Now before we start, we'll be making forward-looking statements today, and you can learn more about the risks associated with those statements on our website. Now we are good to go.

Perfect. Good. So let's start on the demand picture. It's clear demand is quite strong. I'd love to better understand exactly how AI is becoming a tailwind for you guys, -- meaning it's clear the world will need to kind of store, retain, leverage more data in a world of multimodal models, Agentic AI, et cetera.

So the broad tailwind is clear, but can you maybe help us understand some of the emerging use cases for HDDs in an AI world, just to get some context for what's helping to drive this acceleration behind AI?

Absolutely. No, demand is very strong, as you said. AI is one of the applications that is generating a lot of data and therefore, the need for data storage. In our industry, you don't need to produce a different hard disk to store AI data.

AI is generating the same kind of data from a storage standpoint that the traditional application we're generating. So we don't have a different mix inside our products, where we can say exactly what is coming from an AI application or a non-AI application.

So it's a bit more difficult for us to perfectly quantify. But for sure, in the last 1.5 years, I would say, AI has been the center of the discussion with our customers in terms of their need to increase storage.

And more recently, probably in the last couple of quarters, in particular, Video AI has been the reason for that additional increase in demand that maybe came a little bit earlier than what we were expecting when we met in May at our Investor Day. So there is a little bit more demand than what we were expecting. But again, it's not that we were not counting on Video AI for the long term. It's just happening a little bit faster than what we were thinking.

And maybe just touching on video. The implications are significant when we think about the data requirements for a 30-, 60-second video relative to a text file. Is that -- and I'm not just trying to focus on the consumer side, but is that maybe the most exciting new application as we think about the potential to -- as we think about inferencing and as we think about the potential that AI can bring, are there any other kind of very focused items that we should be focusing on the video?

I think it's very exciting until the next up because that's the reality. 6 months ago, we were talking about something else. Today, we talk about Video AI. 6 months from now, hopefully, we talk even about something else that will be very important for people and for businesses and something that generate data and needs to be stored.

And as you know, 90% of the storage is on hard disk. So everything is important. And as I said before, for us, data is the data independently from which application it will generate the data. I think you will see a lot of increase in data also from other applications outside AI like autonomous driving or started many years ago, but has not really developed a lot.

There are many cities where you can see autonomous driving being a reality, other cities, where you don't see any. And this will continue to evolve and that needs a lot of data to work. But of course, Video AI is something that is taking a volume that is way higher than what we were expecting.

Okay. Let's talk about visibility. Last quarter at earnings, you talked about calendar '26 nearline orders effectively being covered for the year that you'd start signing purchase orders into the first half of calendar '27.

Can you maybe just give us an update on demand visibility and maybe more importantly, just beyond that, the sustainability of HDD demand kind of beyond the first half of '27?

Yes. No, what we really care is to be sure that when we start a product in our manufacturing, we already have an order for that product. So an order that covers the mix that we are producing, the price for that product and the time of the delivery.

This is why we focus mainly on the next 4, 5 quarters. But customers are very interested in volume. So of course, they want to discuss about exabyte volume when you go longer, so not only for calendar '26, but also for calendar '27 and even longer.

So we have agreements with our customers on exabyte volume for the longer term, and we have very precise orders for calendar '26. And this is why we were able to say in January at our earnings release, based on this visibility, we expect every quarter of calendar '26 to increase in revenue and in profitability. Now of course, this was more difficult in the past when the industry was working in a different environment.

Now we have this visibility, so we can predict and we can make our estimate. I also have to say that now we are beginning of March. So there are, of course, some contracts for the first part of calendar '27 that have been translated into POs.

And everything is continuing as we have done for the last 11 quarters with that pricing environment that give us opportunity to reasonably increase pricing and take benefit of the cost reduction, when we move the mix from a PMR product to the fourth-generation HAMR 30 terabyte to the second-generation HAMR of 40 terabyte...

As a follow-up on this and speaking to visibility, how are you protecting the company from the risk of overordering amidst this demand strength and this elongation of visibility? How do you make sure you ring-fence that risk to limit any future kind of cyclical drawdown?

Yes. We are very disciplined on how we deploy our CapEx. Our CapEx is not going to be utilized for increasing units. It is all focused on increasing capacity per unit. So technology transition and of course, the HAMR production is a focus. We are not going to increase the units because we think through the technology transition, we can generate about 25% CAGR in the nearline space that should be enough to cover the short-term need of our customers in terms of data centers that are really building up.

And maybe just to hit that point again because I want to be explicit. There's been -- at least you guys have talked about leveraging third parties for some head and media content. Some have associated that with greenfield unit additions. I just -- we have set the record straight for everyone. You said it once, I just want to make sure you say it again so it's clear. The view on new unit capacity is there is none coming online at the point.

Yes. I don't think there is a need for more units. No, I think us as a company have enough units to serve our customers because we are able to move those units from a certain capacity per unit to a higher capacity per unit. For example, when you go from the first-generation HAMR 30 terabyte to the second-generation HAMR 40 terabyte, in theory, if you move all your customers from 30 to 40 terabyte, you can increase your exabyte output by more than 30%.

So of course, this doesn't happen over a year, it takes a little bit longer. But there are opportunities to continue to increase the exabyte without the need to increase units that in the past was the problem that then generated oversupply instead of like undersupply and then generated that volatility and that impact on pricing. I think the industry is way more disciplined today, and this is the benefit of being in this business today.

Okay. Great. I would love for you to touch on maybe the conversations you're having with the major CSP customers. And the question is, we've heard a reference to kind of going from a transactional model to maybe a more partnership model, which allows you to have that visibility.

Just can you speak to maybe the permanence of that change? Is that just a function of the supply-demand imbalance as we sit here today? Or has something actually changed in which there's a structural importance to HDDs that perhaps didn't fully exist given the need to store and retain and produce more data?

Yes, it's probably a little bit of both. Of course, the fact that there is a little bit of shortage in storage in hard disk in particular, is helping that partnership. Now that's the reality. But I also say there is a lot of collaboration with our customers on developing the right product for their storage needs.

And the right product is the one that has more and more capacity per unit. So as you know, we entered the qualification of our 40 terabyte just a couple of quarters ago. And I'm pleased to announce today that both customers that were in full for the 40-terabyte drive has now qualified the drive. So we will start shipping some volume already this quarter and then more quarter after quarter.

So I guess you just eliminated one of my questions here. good. No, let's -- before we get into HAMR because obviously, it's critical. Just the point on pricing, right, supply-demand imbalance right now, clearly kind of insatiable demand, using aerial density to drive supply or to drive exabytes higher.

Talk to us about what's happening with pricing because you've kind of characterized it as flattish to low single-digit growth. Your competitor has been maybe a little bit more outwardly bullish in talking about mid- to high single-digit year-over-year growth. Just is there kind of upside as you see the pricing environment? Is there ability to take more price, not just as we think about calendar '26, but into calendar '27?

Absolutely. No, I think it depends what is your starting point. Usually, I tend to talk about sequential improvement. So it's more a quarter after quarter. Of course, if you go year-over-year, the increase is substantially higher than what we can do sequentially.

So I would say, generally, I guess the trend is very similar in terms of pricing. Again, the focus of our customers is more capacity per unit. So now that they can buy the 40 terabyte and those 2 big customers, of course, they will focus more and more on getting the volume.

They need to give us the time to ramp. So there is a little bit of time to ramp high volume, but it's a huge improvement for them to go from 30 to 40 terabytes. Now if you think about the cost of a slot that they have a physical slot, if they can put 30 terabytes and monetize the 30 terabyte or for basically the same cost, having a 40 terabyte, now they monetize 30% more.

So it's a huge benefit for them to increase in capacity. It's a huge benefit for us because with a similar cost per unit, we generate 10 more terabytes per unit. So the cost per terabyte now declined faster than what we have seen in the past. And again, with the pricing strategy that we have implemented and executed for more than 11 quarters and what we see for the future, we can reasonably say revenue would be higher and profit would be higher.

Okay. Good. So let's now touch on the comment that you made about HAMR. So you did start -- you kicked off qualification on Mosaic 4 products with your first CSP last July.

I think the second one started last October, if I'm correct. The comment that you're making today effectively is we have now qualified those 2 CSPs on Mosaic 4. And just to make sure -- again, to make sure we hammer the point home, you're starting to see volume shipments calendar 1Q this quarter. Is that correct?

That's correct.

Okay. Now how do we think about maybe the pace of the rest of the CSPs out there? Just maybe help us understand the interest level in Mosaic 4 from a qualification standpoint, how you think about that trajectory could look like going forward?

Yes. I would say that technology is not so important to them anymore. They have basically all qualified the first-generation HAMR. So they know it's working well in their environment. So technology is not an issue anymore. They just focus on how to get to a bigger drive because that is where they have the best return.

So I would say we need to get the time to ramp the first 2 customers that are really big customers. So they will take a lot of volume. But with the time, we will be ramping up also for other customers to get qualified and then take benefit of the bigger drive.

Okay. Awesome. Congratulations on that. So the -- you made a comment last earnings, you said the transition from Mosaic 3 to Mosaic 4, so 30 terabytes to 40 terabytes per drive you'll do that fairly aggressively. You also mentioned it will be a fairly prescriptive ramp. So obviously, Mosaic 3 took time to ramp and qualify.

What exactly does that look like when we think about that mix shift that you've talked about historically for HAMR? Does that look any different than what you talked about last May in terms of the 40% mix, the turnover or the crossover, excuse me, does that look different now that you've had these qualifications?

No, we were counting on having those qualifications. I would say we are getting those calls maybe a few months earlier, not a few quarters earlier. So it's not that we can ramp a very different number of drive for the 40 terabyte.

I would say we are still focusing on achieving those numbers that we said at the Investor Day. If we are at 70% of nearline exabyte sold with an HAMR technology by June 27, I think will be very, very good for us, especially because we start to better optimize our manufacturing.

When you start a new technology, you have a period of time where you are not optimized. You have one technology that is new that somehow compete with the old technology. So what you extract from your manufacturing is not fully optimized.

It's very good, but it's not fully optimized. There is a lot of opportunity to improve. And the more and more you ramp of the new technology, you become basically just running one kind of products, and that will help us to get the 25% exabyte or maybe a little bit more in the future.

Okay. And then just in terms of the technological volume ramp associated with Mosaic 3, but also Mosaic 4, are there any bottlenecks as you see it, whether it relates to cycle time or components, rare earths, et cetera? Just want to make sure we're kind of triaging any risks that could be associated with that ramp?

I would say the limiting factor is our own ramp. It's not, I would say, any external components that we buy is our ramp on how much we can -- as you know, the cycle time is not short.

So we need to start earlier to produce as a media and then to do the assembly and to do the final test. Even the final test gets a little bit longer, not because of the technology, but because of the capacity of the drive. So everything takes a little bit more time. That means to do a ramp of millions of units take a little bit of time. But every quarter, you will see a higher volume and a better contribution of the 40-terabyte drive, not only to the revenue, but also to the profit line.

Okay. Great. Your competitor had an Analyst Day a few weeks ago. They were talking about kind of adding more heads and platters to a single drive. You've been very focused on areal density, kind of leading the market on areal density.

Just your perspective, as you think about your technology innovation around more planners to a single drive. Is that something that you're focused on? Or is the focus kind of squarely on areal density where, again, you're kind of leading the charge there in the market?

Well, probably when you move to HAMR, the first focus is taking benefit of this increase of terabyte per disk. When you go from a 3 terabyte per disk to a 4 terabyte per disk, you had 33% of capacity, which is much more than adding 1 disk.

If you had 1 disk out of 10, you have only 10% of increase. So increasing areal density per disk give you a better return. So going to 3 terabyte per disk to 4 terabyte per disk to 5 terabyte per disk to 6 terabyte per disk is probably the best return you can get.

Now at a certain point, it will be interesting and financially having also a good return to start increasing disk. Of course, in the longer term, you want to take benefit of all the space that is inside the box, but you need to find the right time.

Right now, we get more benefit in increasing the areal density at a certain point, probably will be good for us and I think for the industry in general to take all the space inside the box. If you -- if you start increasing disk when you are at a 5 terabyte or 6 terabyte, the return on the disk and 2 add is fairly huge.

Okay. So I think maybe as the CFO, this is -- and at least as an analyst, this is one of the most exciting parts of this journey in areal density is you're not changing the form factor, you're adding more capacity to the box, but you're not adding disks, you're not adding heads. You've talked about in-sourcing the laser diodes that you've been working on innovating there.

What does that translate to when it comes to cost downs? We talked about price per terabyte earlier. You've been able to do something like 15% annual cost per terabyte declines. So now that we're mixing into more Mosaic 3 and in the second half, more Mosaic 4, how do we think about the trajectory of cost per terabyte decline?

It's actually very interesting. The bill of material going from the fourth-generation HAMR to the second-generation HAMR to the third-generation HAMR is fairly similar. As you said, it's still based on 10 disks and 20 heads.

Of course, they are not the same disk and heads, our disk with more capacity and are new heads that need to be developed. Also the components that we buy externally need to evolve to support the higher capacity, but it's the same number of components.

So the cost per unit is fairly stable, but we had a lot of terabytes per unit. So there is a huge and very important decline in terms of cost per terabyte. Now when you look at a period of time, it depends how many units you have sold for that new product. So every quarter, you will see we sell millions of units, and we sell PMR units. We sell fourth-generation HAMR.

We now sell second-generation HAMR. Every quarter has a different mix and a different capacity. But the trend is, of course, is having a good cost reduction. And the more we can ramp of the 40 terabytes, the more we can take benefit of that decline cost. So this is part of what we said before. An important reason of why we expect a better profitability is the mix moving more and more to the 40-terabyte drives.

Right. Okay. Let's take what we've learned maybe on demand and now on the cost side and translate that into financial metrics. So I believe the latest is mid-20% nearline exabyte growth in line to even stronger revenue growth when we think about the benefits that you can get from pricing that we discussed earlier.

A year ago or almost a year ago, you laid out a target of 50% plus incremental margins. Clearly, you've been well outperforming that metric. You've been doing 70% plus. Just given what we're talking here about demand, your ability to cost down, your ability to price up, that becomes a very powerful tool.

Do we expect in totality kind of growth as we now look into the second half and then margin expansion as we look into the second half? Like are those metrics that we think should accelerate just given all of the kind of goodness that is now coming through the model? I know you said we can grow sequentially. Just trying to contextualize that growth.

Yes. When we had our Investor Day, I said we expect gross margin to increase 50% incrementally starting a certain level of revenue, and we have done significantly better. The reason why we have done better is mainly because demand is a bit stronger than what we were expecting now going back to the Video AI discussion we had before.

And because with those calls that we have achieved, not only the last one that, of course, did not impact the prior P&L, but will impact the future, but the transition to the fourth-generation HMR also gave us an opportunity to reduce cost a little bit better than what we had in our plan and so to achieve a better margin.

I think we are continuing with the same trend. As I said before, the pricing situation has not changed. It's actually now extending for us to the first part of calendar '27. So that is very good. That means demand from customers is still very strong because this is a big test when you go and finally put number in a PO, you can see how real is the demand, and this is a confirmation that demand is strong and is real.

And of course, with the mix moving up and up in capacity, we expect to continue improving in gross margin. And finally, in net income because our OpEx is already at a very, very good point today.

Just very quickly touching on OpEx there. You've signaled a ton of leverage that you can drive in the model. Are there -- like maybe contextualize the added costs, if there are any associated with OpEx as we think about moving into the back half and into calendar year '27? Or if there was growth in OpEx, like where does that come from?

Yes. I think in terms of resources, we have a very good structure. So I don't expect an increase in the need for resources. So in terms of cost, it should be fairly similar to where it is today. Usually, we have the annual salary increase in the September quarter. So we could see a little bit of increase in that quarter and then, of course, in December, but kind of limited.

We've seen that this year, right?

We have seen it this year. So part is also variable compensation. So every year is a bit different on how much is a variable comp. But I would say -- if you look at calendar '26, I don't really expect a lot of difference in OpEx spending per quarter.

Okay. And then CapEx, you've been very consistent there. It doesn't seem like anything changes. But even as you ramp Mosaic 4 and beyond in testing beyond Mosaic 4, does anything change with that range that we think about CapEx as a percentage of revenue?

No, I would say 4% to 6% of revenue is a very good model, give us the opportunity to buy the tools that we need to ramp more volume of HAMR. So focusing on the technology transition and moving from lower capacity drive to higher capacity drives.

And this is what we need. Of course, there is always CapEx maintenance, there are equipment that gets old that you need to replace. But again, all the focus is on components and transition from one product to the next. Of course, we don't focus on any increase in units.

Okay. Perfect. I know your Analyst Day probably feels like it was ages ago. It hasn't even been a year impressively. Can you just remind us how to think about margins and free cash margins on the other side of the equation, right?

We're kind of dreaming the dream as we go up into the cycle. When you talk about your ability to -- I think you've guided to 40% plus gross margins, operating margins of 30% plus. Is that signaling kind of those are the floors as we think about the new model, meaning even if we go through a period of digestion at some point in the future, that's, again, notwithstanding certain quarters, but that's where we feel like the floor kind of is in our margin trajectory.

I would say we're already above those levels and the trend is to improve. So honestly, I don't see a reason why we should go in the other direction. We have all the visibility on calendar '26. I told you about the first part of calendar '27. So again, I see -- from here, I see improvement. I don't see a reason why we should go in the other direction.

Okay. And then maybe last question before we get into kind of capital allocation balance sheet. From the perspective of kind of new entrants in the market, this is a rational oligopoly right now. And I've long said rational oligopolies can be very powerful.

Is there a threat of new entrants? Do you see that? Is that something that is on the horizon? Just would love your perspective. As you look out into the market, what you see in terms of this oligopoly structure potentially changing at all?

Well, from my point of view, the difficult part of this industry is not only manufacturing, but also the technology inside the drive, especially when you move to HAMR, there are way more additional complications and more difficulties.

This is where the industry is competing is in product, in evolving the product so that we can generate more terabyte per unit for our customers. But it's no focus on volume or units or all those things. Those things are actually eventually creating a possible disruption for this industry in the future. So it's not the right focus.

The right focus is technology, giving our customers what they need in terms of the product and increase the exabyte. -- so that they can build the data center where they need and they have the storage where they need. For a new entrant will be very difficult because they need to spend a lot of money, first of all, to set up the manufacturing, second to develop the technology.

And this point is more complicated than 2 years ago or 3 years ago. Now they need to enter and having HAMR technology. If you enter and you don't have HAMR technology, you're already out of the game. So I'd say for the next few years, I don't see this happening.

Okay. Okay. Good. Let's touch on that capital structure capital allocation. First, on the structure side, you retired $600 million of your converts last month. your gross debt is, I think, right around $3.9 billion. That's exactly kind of what you've guided to or at least projected towards. Have you kind of reached the end of that deleveraging just as it relates to your gross debt?

Are you kind of done and now you're returning to buying back more stock? Or is there a desire to maybe reduce that convert a little bit more? How do we think about balancing that?

Yes. To me, reducing the convertible has a double benefit. One is reducing the debt. But second, very important is to avoid further dilution from the convertible. So it kind of could be assimilated to a little bit of a share buyback. So we have retired $1.1 billion already in the last 2 quarters. We still have $400 million outstanding. So we will take care of that part of the convertible in the next few quarters.

So in terms of debt, I'm fairly happy where we are today. Probably we will go a little bit lower. As you said, the free cash flow will be very strong. So we have the opportunity not only to do a strong return to our shareholders through dividend and share buyback, but also to reduce our debt a little bit more.

So as we think about maybe the balance of how you've utilized free cash flow, call it, over the last 2 quarters, it has been more levered to delevering.

Now it's maybe a little bit more balanced as we look forward, reduce the convert a little bit more, buy back a little bit more and kind of transition that over towards buyback as we think about a few quarters.

Yes. Every quarter will be a little bit different, depending on what kind of treasury activity we are doing in the quarter, but you will have both.

And some quarters, we'll have more share buyback and less reduction of debt and some quarters, we will maybe focus more on the convertible resources and some other debt reduction and a little bit less share buyback. But if you look over a longer period of time, excluding the debt, basically 100% of our free cash flow will return to our shareholders.

And I was going to touch on that, but my question was going to be the stated goal was kind of 75% of free cash. It sounds like given how confident you are in the outlook in your margin structure, innovation, all of that stuff, we could and/or should be expecting upside to that.

Absolutely.

Okay. That's amazing. Another good insight here. So before we end, last 2 questions. Just as CFO, just quickly addressing any risks that you think about in terms of things that you want to challenge your team to do better internally, things that you see on horizon, just making sure -- and anything is kind of ring-fenced as we all think about potential risks. And then one question after that, please.

Well, risk, unfortunately, is always there. Geopolitical risk, of course, is high in every industry, not only for us. So we control what we can. And what we can control is our manufacturing, how we address demand, how we implement the pricing strategy.

I would say where we can extract more value in the next several quarters because demand is above supply is to moving the mix more and more into the higher capacity drive, so we can extract a little bit more exabyte from our manufacturing every quarter. And of course, those exabytes has a very good return for us.

So that is, I think, the focus, focus on the quality of the product and the exabyte that we can generate quarter after quarter. And now the team is doing an incredible job in extracting as they can.

Amazing. So we've got Mosaic 4 ramping. That's new. We got more free cash flow returned to shareholders. I'm sure everyone here is happy about that.

As you ramp up in kind of the last minute here, just leave everyone with a final word. What maybe is underappreciated? What should people thinking about the messages that you want to leave as you set up stage here?

Well underappreciated could be the state of the industry. Of course, everyone is impacted by the past. I would say the past was a different business was based on client business, so on desktop, laptop, different application for the storage. year after year, this business moving into the big cloud. 80% of our business is data center.

So it's all focused on high-capacity drives. And this is a new situation, a new industry where demand is very strong, where the industry is full in terms of capacity, not adding units, but focusing a lot on technology. I think this will generate a lot of profitability.

For us and for the industry in general. And this is something that maybe you need a little bit more time for people to get used to that. In the past, it was not exactly this. So -- but now it's 11 quarters. We did this already be for 11 quarters. When we talk about the next 4 quarters. So this is not 2, 3, 4 quarter cycle. This is a huge change in the industry, in my opinion.

Amazing. That's a great place to end. Gianluca. Thank you very much for your time.

Thank you, Erik. Thank you.

Hi, everyone. Good morning. I think we'll get started. Come take a seat. I'm Mark Newman, Bernstein's U.S. IT hardware analyst. And very happy to introduce Gianluca Romano, CFO of Seagate. Thanks very much for joining us today.

Thank you, Mark. Before we start, let me remind everyone that I will be making forward-looking statements today, and you can learn more about those risks on our website.

Okay. Perfect. So I'm going to get started here and just remind everyone, we've got some prepared questions I'm going to ask, and then I'm going to open it up to the floor. There is an app you can submit a question or if you prefer, we have a microphone, and you can ask questions live as well.

So to start off, let's talk a little bit about the market dynamics in hard disk drives. So recently, one of your competitors, Western Digital, they -- or I should say, your main competitor, sorry, one of your main competitor, Western Digital forecasted mid-20s exabyte growth in nearline. And I just wondered what's your latest thoughts on growth. Do you agree with this? Or what are your latest thoughts on that?

Yes. And generally, we agree. We presented to our Investor Day May last year, a very similar forecast in terms of nearline exabyte CAGR, north mid-20s. is over a period of time of 3, 4, even 5 years. So every year is a bit different. But generally, I think we have a similar view of what will be demand and what we can increase in terms of exabyte production.

And on pricing, how I describe pricing. Recently, we've seen blended average prices on a per terabyte basis to be stable, flat to very slightly up, up something like low single digits year-on-year, something like that seems to be -- just going back to Western Digital because they just had an Analyst Day. So they recently said that they're sold out for 2026 and with some contracts going to the end of 2027 and even one contract going through to the end of 2028 and that they expect prices to remain stable, they call it, which the implication from that was flat to very slightly up. I just want -- so starting off, do you expect prices to stay stable for the next 5 years? Do you think prices could go up? Or how do you see that?

We are not changing our pricing strategy. We're starting this strategy about almost 3 years ago at this point. And we are very consistent. So every time we renegotiate a contract, there is a little bit of price increase. It's very important to push customers to adopt the latest product available because this is how we want to increase exabyte. It's not adding unit, but it's increasing exabyte from the same number of units that we can produce in our manufacturing.

So I would say we could be more aggressive on pricing for sure, but I think stability and visibility is very important. So we are very consistent. Every quarter, we do a little bit better. We don't go for unreasonable price increase, but we are very consistent. So there is always the impact of the mix from lower capacity drive to higher capacity drive. So when you look at the average, it's not really telling you exactly what we do product-by-product. But I tend to agree that based on the orders that we have for calendar '26, as we said at our earnings release, we expect pricing -- average price per terabyte to be flat to slightly up.

Just to clarify what you do -- to the benefit of the audience, I believe what you do is you -- for legacy products or products that are being renewed that are more than 1 year old, you're essentially increasing the price on like-for-like products, right?

And then that's what you're saying is you're offsetting that with new products with the higher density coming in at a higher price per drive, but a lower price per terabyte and that offsets the price increasing on the like-for-like drive. Is that a correct summary?

I would say the offset is very limited. It's not a full offset. So there is an increase of the average price per terabyte as you have seen, and you will see through the rest of the calendar year. There is maybe a little bit of incentive when a customer is adopting a much higher drive in terms of capacity because our cost per terabyte is much lower. So our profitability improved a lot despite that very small discount.

But this is valid only for the first time they buy that product. When they come back and buy again the same product, now price is going up, which is why to model every quarter is not so easy. There is a lot of mix impact and who is buying this new product for the first time, who is buying for the second time. Some have a discount; some have an increased price. So -- but when you put all together, we see stability and some increase in price per terabyte -- average price per terabyte.

Given that the market seems to be so strong right now and given that NAND flash prices have been increasing dramatically, and probably double or triple in the last quarter or so. Is there an opportunity to -- I mean, I know you touched on it, but is there a realistic opportunity for Seagate and potentially your competitors to increase price per terabyte more than the currently quite -- yes, compared to NAND and DRAM, your price increases is quite slow. So is there an opportunity to increase it more? Or is that going to be more limited because of the longer-term contracts that you have?

No, I think if we wanted to increase pricing more aggressively, we could. Demand is well above supply. So we could, but it's just as we want to manage our strategy and hopefully how the industry is viewing this is a long-term benefit for the industry to have demand a little bit above supply. And we are reasonable with the price increase, but we are very consistent. So it's stable, maybe not super aggressive, but very stable and very consistent, and this is increasing our profitability for almost 3 years consecutive every quarter. So I think if I have to decide what I want is a major increase and then possibly a decrease in profitability or is more stable increase, I always prefer the more stable increase.

Fair enough. Fair enough. Drilling down on the point versus NAND a little bit more. Given how steep NAND prices have gone up, that cost differential for your customers is much wider. So it used to be 6 to 7x, I believe, NAND being 6 to 7x more expensive per gigabyte. Now it's probably 10, 15x, maybe 20x, not that far away around the corner given how fast prices are going up.

So does that impact your potential demand? I know demand is already tight, but is there some portion of demand that can -- that's on the edge between the hard disk drive and NAND that can flip backwards and forwards -- and that potentially you can benefit from?

I would say in general, as you said, demand is above supply. So whatever we produce is going to be sold. So this additional potential demand cannot be satisfied anyway because we already sell 100% of our exabyte.

In the public cloud infrastructure, I don't see a lot of changes. It's kind of independent. The price of NAND from hard disk is independent wherever they go, very high or very low. The infrastructure has always been very similar, that is storage is on hard disk and then you move the data from an hard disk into an end when you need to run the application and then you send the data back to an hard disk.

So that infrastructure has never changed for the last 10, 15 years. It's always separating the 2 components. And I guess, their procurement, our customers' procurement, are buying those 2 components for different needs. So the overlap is very, very limited. Now when you go in other segments, low-capacity segments, there is for sure an overlap. And of course, in that case, the delta price can drive some delta volume. But because 80% of our revenue is in data center, I would say it's very limited.

So this will be some OEM business potentially?

Some OEMs, some VIA client, so video and image application, but in the client space, consumer, for sure, very sensitive to pricing, but not in the big public cloud and probably not even too much in the big enterprise OEM.

A few months ago, there was some market chatter about some hyperscalers hard disk drive demand switching to NAND because they couldn't get enough hard disk drive at that time. Did you see that? Or is that false?

As I said before, I think it's just people talking. As I said before, all what we produce is sold and is sold today for the end of the calendar year. So we sell what we produce, then you can talk and say whatever you want about the extra demand that we cannot serve. I don't think it's going anywhere because of what I discussed before. The 2 components are used differently.

Now I don't exclude that for very, very small volume, you could use some NAND to do storage, but it's not a good decision for our customers. And by the way, it's not that they are now swimming into NAND either. So I don't know how they can do it.

Switching gears to technology and supply side. You said that HAMR should enable roughly high teens areal density per year. This, though, is still a pretty large gap with long-term demand in the mid-20s. And at the moment, your shipments are matching pretty close to where demand is mid-20s. So how do we reconcile that gap between the high teens' areal density improvement and demand and shipments in the mid-20s?

Yes, it's a good question. Well, first of all, I would say it's not that demand is in the mid-20s. Demand, the TAM is what the industry can produce. So for sure, supply and demand today, if you look at what we ship, is equal to supply. It's not equal to demand. Demand is higher.

Yes, we are increasing our exabyte at -- like 25% CAGR. And that depends from the entire mix. So you need to think about not everyone today is on a 30-terabyte drive and tomorrow moving to a 40-terabyte drive. If theoretically, everyone is doing that in a year, you can increase by 33%, your exabyte CAGR.

But really, there are -- we sell drive from 1-terabyte to very soon 40-terabyte. So the increase in exabyte depends on how all this mix is moving up. It's not just the highest capacity. It's everything up. There are a lot of enterprise OEMs that are buying 18 terabytes, 20-terabyte, 24-terabyte and every year, they move up, but they don't go from 30 to 40. So when we look at that mix change and of course, our product road map and how we think customers will be qualified and what we can ramp of those products, we think about 25% exabyte CAGR is what we can generate for the next 3 to 4 years.

I see. So even though areal density is only in the high teens, you think you can still with some of that mix change.

If you think about -- if you have a customer that today is on 20-terabyte and can go quickly to 40-terabyte can double. So -- but the mix is not very variable. So when you put all together, the 25% CAGR is what we think we can do. It is fairly close to what we have done. I think recently, we have been between 25% and 30%, depending from the quarter. I think that is the range.

Is there any plans -- you probably get this question like every day. Any plans to add -- but I have to ask you, any plans to add unit capacity? Any motors board factories that could be repurposed to boost unit production?

Well, our factories are full at this point. Of course, we always try to optimize the space that we have, but we are full. We have no interest in adding units because we think with our product road map with the same number of units, we can generate that increase in exabyte that we think will be enough to satisfy the short-term need of our customers, possibly not giving them the opportunity to build a lot of inventory or any inventory, but also not so short that they cannot build the next data center because they don't have hard disk.

Today, I'm not aware of any project on a data center that is put on hold because they will not have hard disk. So we are one of the components that can be short, but we are not at the top of the list. So of course, there is a demand that our customers have and then it's what they can do based on the power they can get and other components.

There are some components in memory that looks to be short right now. Of course, GPU was another component that is probably still a bit short, some of the semiconductors seems to be short. So we are possibly a little bit high in the list, but we are not the component that is the bottleneck in building the next data center. And this is exactly what we want. We don't want to be a problem for our customers. but we don't want to produce more than what is really, really needed for the short term.

Okay. Great. Can you talk about yields in your HAMR products?

Well, yields as every new product that you have improve with the time. So of course, every quarter, we have a little bit better yield. We are now at the second generation of HAMR. So we have the fourth generation that is the platform is on a 30-terabyte and go from 30 to mid-30s. So there are different variants of that product. And we are qualifying the second generation. So the second platform that start at 40-terabyte SMR, and then we will grow with the time also into the mid-40s. So the yield will improve as always been in the past.

So at the moment, though, the yield, would you say it's not quite mature as PMR yields?

No. It's not exactly the same level. Of course, PMR, we have built that product technology for more than 20 years. We are building HAMR. We -- the first qualification was November 2024. So it's not even -- it's basically 1.5 years. So there are opportunities to improve the yield, that means lower cost for us. And of course, this is also what we include in our estimate for the 25% exabyte CAGR. So some will come also from better yield. Now the yield is good. It's not bad. It's actually very good, but it's not equal to PMR at this point.

Right. And is there a difference in yield between your first-gen Mozaic 3 and your second-gen Mozaic 4, presumably?

Second generation will be better, but we are still not really producing that. We are still in the call phase. So we are finalizing those calls and then we ramp. So we will see, but I'm sure it will be better.

But I guess my other question, though, on yields would be even though you may have a slightly lower yield for your HAMR products, is it still correct that due to the area density improvement you have, you still have a lower cost per terabyte?

Absolutely. We show that actually at our Investor Day, where I was -- a slide with the cost per terabyte of our last product on the prior technology, so on PMR, the fourth generation HAMR and then the following generations of HAMR. And you can see the fourth generation of HAMR is fairly similar to the last product on PMR in terms of cost per terabyte, a little bit lower, but fairly similar. And then you start to see the very good impact in the cost decline from the second generation on. So this 40-terabyte drive will give us a good opportunity in terms of cost reduction.

So the big benefit really is 4...?

Start at 4, was some benefit on the 3-terabyte per disk, so 30-terabyte per drive, it will be much better on 4. And then with improvement, it will continue to be a very good driver for our cost decline and of course, our profitability increase in the next several years.

So I think last quarter, I know you don't break it out exactly, but doing the math, you had HAMR was over 20% of nearline exabyte shipped. Is that -- so I assume then that's all Mozaic 3 because Mozaic 4...

It's all Mozaic 3, yes.

Yes. So not much difference between those products within Mozaic. Do you have any shingle products within?

Of course. We sell a good portion of our products are SMR, depending from the customer. If the customer prefer SMR, it's just a firm change that we do at the end of the production and we sell SMR. If they want CMR, we sell CMR. I would say the big public cloud are moving more and more into SMR. So we sell more and more of that version. But it's the same product. But when you shingle, you get a little bit more capacity. So it's good for us and it's good for them if they can use it. Unfortunately, not everyone can use SMR yet, but that is a trend. And of course, we are in favor of this change in the future.

And you share that cost benefit from SMR with the customer or you take it off?

No, a terabyte is a terabyte.

So they basically pay similar for SMR...

I would say there are no significant differences. And of course, every customer is different. But in general, today, a customer or buy SMR or buy CMR. They don't buy a mix. So you cannot really even compare. So there is a price for a certain customer based on the terabyte that they want. And then if they want SMR, they get SMR. If they want CMR, they get CMR.

So can you talk about what is your mix now, your SMR versus CMR? And is that continue to increase going forward?

Continue to increase. I don't think we have disclosed recently exactly what is the percentage, but I would say it's continuing to increase. And because customers are the same in the industry, I would say hard disk suppliers, they should have a similar percentage overall.

Right. Okay. Moving on to more the financial side. Given stable to strong pricing and your HAMR ramp, how should we think about your cost per exabyte declines in gross margins going forward?

Well, I would say we have a very good trend in the last almost 3 years, now 11 quarters of consecutive improvement in pricing, improvement in cost and therefore, in better gross margin. I think we guided another very good improvement in the current quarter. And as I said before, we are not changing our strategy. So the pricing strategy is the same. So I expect a similar result in the future or even better possibly. And they move into the mix to higher capacity drives give us the opportunity to continue to reduce the cost.

Of course, you need to ramp high volume of that product to get really the impact in the financials. So it takes a little bit of time. It's not in the fourth quarter that you get all the benefit, it takes time, time to ramp-up, but will be very beneficial.

Could you remind us what is the schedule for the Mozaic 4 ramp?

So we are qualifying 2 customers right now, 2 big cloud customers in U.S. So as soon as we qualify, we start shipping a certain volume. But of course, at the beginning is not much because before we qualify, then we ramp. So we have a little bit of volume for them available when they qualify, but then we really ramp. I would say, more in the second part of this calendar year.

Got it. Got it. Okay. I guess one of the questions I get a lot from investors is around relative profitability between Seagate and your competitor, Western Digital. Despite Seagate having a strong lead in HAMR, Western Digital still has a slightly higher gross margin. Can you talk about why that is? And will that narrow?

I don't know. I don't compare too much in details with my competitor, but I would say there are maybe a couple of reasons. One is they don't produce HAMR yet. And of course, at the beginning, when you have 2 technologies in the same factories, you don't -- you are not fully optimized. So for sure, your cost structure in manufacturing is not optimized.

No, it will be better for us when we move more and more on HAMR because at a certain point, our production will be mainly HAMR. So we will not have that little bit of disruption within the 2 technology. They will have to go through that transition at a certain point, of course, it will be good for them also to be fully on HAMR, but there is a little bit of time where you have bad disruption.

And second, I guess, they produce a little bit more exabyte probably for the same reason. And those exabytes are probably paid very well.

And presumably, the SMR shingling may have some impact on that because they have -- I think they have a higher portion of shingling, and the ultra SMR has a bit of a bigger benefit versus Seagate shingling?

I don't know. No, of course, they don't provide a lot of details exactly on the individual product. I would say customers are the same. So if a customer is buying SMR from one supplier, it's probably buying SMR from the other supplier because that means their data center can handle SMR. So I don't think there are huge differences.

And then, of course, everyone has own technology or if they can produce more terabyte from the same number of units, it's because they have a different technology. We focus on HAMR. We think HAMR is the technology of the future. And we took a little bit a different decision. We anticipate HAMR, they are doing HAMR to be later. I think it's a temporary difference. Going on with the times. The 2 companies probably will have similar technology and similar capacity per drive.

But more a bit of a transition in the meantime, though. On capital returns, so your latest guidance, I believe, is greater than 75% of free cash flow returned to shareholders. But your free cash flow is -- it's growing massively. And in my projections, expected to grow -- continue to grow significantly, rapidly shrinking debt. Shouldn't you be considering something closer to 100% of free cash flow? Percentage wise?

Well, this is what we have done in the past. We have basically returned the entirety of our free cash flow to shareholders. So I don't exclude we will do the same in the future. We have focused on reducing the debt in the last few quarters. So at a certain point, our debt was above $6 billion. Now with the last transaction that we have done recently on our convertible, we are below 4, and we will reduce that even more in the future. But except for that focus on reducing debt, I would say the other part of the free cash flow will probably go to shareholders through dividend and share buyback.

Okay. I have more questions, but I just want to give the audience an opportunity to jump in if there are any questions from the audience, please put your hand up or submit it online. I haven't seen any on the portal. No one seems to be using the portal, but feel free to put your hand out. We've got a microphone right here. No hands so far. I'm going to keep going.

So -- actually, I had a question on TCO. So HAMR versus SSD, TCO. So NAND pricing rising significantly. Has this changed the TCO gap between hard disk drives and SSDs? How are you thinking about that going forward for total cost of ownership?

Well, if you look at history, that NAND price is more volatile than hard disk. So right now, it's in a period of time where NAND price is increasing quarter after quarter. So of course, there is an increase in the gap between the 2 technology. But as I said before, this is not really impacting too much our business. 80% of our business is in data center. And in the data center, there is no overlap between NAND and hard disk. So again, demand is very strong. So we sell all what we produce. So it is kind of independent from where the NAND price is.

Now in the edge IoT part, what we call edge IoT, there is, for sure, a price sensitivity. We were discussing before saying if you go into the consumer business, for example, the fact that the NAND price is going up is probably giving us the opportunity to do 2 things. One is probably to sell a little bit more volume. Second, to have a better price. So it's helping, at least temporary is helping that part of the business, but it's not the majority of our volume, of our revenue is about 20% of what we sell in a quarter.

So the AI demand. Are you seeing some shift where sort of the customers' hyperscalers or neoclouds in AI use hard disk drives, not just for cold storage but also warm training data as well?

Well, the neocloud is a bit different structure. What I said before, in the public cloud infrastructure, they are a complete data center. So they have storage that is hard risk and they have where they run the application, and they transfer the data from the storage hard disk into NAND and they run.

The neocloud are basically running the application for the big public cloud. So we have basically the same customer. We provide the storage, they provide an additional compute opportunity. So the storage is still in the public cloud. So they import the data from our hard desk in a public cloud into their NAND in the neocloud, they run the application and send it back.

So the neocloud is not actually purchasing hard disk drives. They're utilizing the large storage network in the public cloud, which is...

They use their customer storage to provide the application for the same customer. So when they use a certain -- either on the application for a certain customer, they take that storage, they import, they run the application and send it back. When it's a different customers, it's the same, takes the data from the other customer, runs application and send it back.

Now in the future, if they evolve in a complete data center. So that means they don't -- they just don't run the application for the same customers, but they have different customers, and they want to be independent, they also need to have their storage. At that point they will buy hard disk.

Got it. During our recent Q2 earnings, you talked about nearline capacity fully allocated through calendar '26, I believe. And just in a world with booming AI demand that requires persistent access to this massive, massive data sets, are you seeing hyperscalers trying to lock in supply for longer term, 2027, 2028? I mean one of your competitors did talk about that, but Seagate has not yet talked about that. Just curious if you're seeing that...

Yes. No, I think we talk about that also. I think high level is very similar, but maybe what is a small difference on how we agree and when we agree on pricing. In terms of volume, it's very similar. So we discuss with customers entire calendar '26, calendar '27 in certain case even longer. So the volume agreement are already there.

What we have done maybe differently, I don't know, possibly different is when we discuss price. And we discussed -- we fully discussed pricing for the entire calendar '26 because when we start the product in our manufacturing, it takes about 3 quarters to come out, and we want to be sure that there is an order attached to that product. And that order has to have the customer, the time of delivery and the price.

When you go longer, we think it's probably good for us to wait a little bit before we define the price, as you discussed before, there is this demand that is very strong and a slight imbalance between supply and demand, that is, of course, helping our strategy on pricing and continue our pricing strategy as we have done for many, many quarters. So we prefer to wait it a bit longer. I would say right now, we are discussing pricing for the first part of calendar '27.

So we always say -- so we always have those 4, 5 quarters where everything is fully defined, everything. But going longer, we basically only define volume. And then we define the exact mix of the product based on what they are qualified and the exact pricing at a different time. Now for what we are discussing in the first part of '27, pricing is following exactly what has been in the past. So no changes there.

And have you seen signs that AI demand is expanding beyond and you talked about hyperscalers towards enterprise and sovereign. Have you started to see that yet?

Possibly. I'll say it's probably still at the beginning. I would say, in AI, the major recent change that we have seen is a faster adoption of video AI in China and in U.S. So there are a lot of China new video AI application like we have in U.S. That is what is boosting a little bit more demand, a little bit faster than what we were expecting.

So it's creating a little bit of a bigger gap between supply and demand in the short-term. But again, it's not that we were not expecting video AI to be a big driver of the demand. It's just happening a little bit faster than earlier than what we were expecting. That, I think, is where the exabyte, there was a big volume is currently is.

You've also identified agentic AI as a driver for persistent historical data storage. So as AI agents evolve from simple training to independent decision-making, they require constant context from vector databases stored in hard drives. And -- so does this transition move nearline drives into higher replacement cycle environment, given the faster the cycles that you may be used on that type of use case?

I would say, generally, when you look at AI, there are different phases. There is a training phase and then there is no inference. There are different phases and data is used differently. In training, you need a lot of data. So that impact retention. So if you are training for AI, you don't delete any data. You actually want all the data possible to do a better training. Then you use that trained AI to generate a data that is valuable to you.

It can be for you in manufacturing, for you in your personal life or in whatever application you're using. So at that point, AI starts to generate data. And I think we're already in the second phase. Now before we were saying, well, to us in terms of storage, the major impact from this new AI is detention. Companies are not deleting data anymore and cloud are keeping more data, more data, more data.

Now we see data generated from AI coming back into storage. And video is huge because video consume a lot of terabyte. So the more companies are keeping video for surveillance or for quality manufacturing or people to generate video for their use. This is a lot of exabytes that are consumed every quarter. So there are different applications, different cases.

Now it's not only AI. Right now, AI is a big change compared to the traditional application, but every year, you will see new things. Even application that already exists like autonomous driving are many years where we are talking about autonomous driving, but we didn't see a lot of cars going around without a driver yet. Now it's happening. Now you go to San Francisco, you go Phoenix, you go other cities, you have more and more of those autonomous driving.

This is generating a lot of data and that data gets stored. So there are a lot of applications on top of AI that will generate more and more data. And robotics will be the next one. Robotics need a lot of data, and you need to keep that data. It's also part of the training of the robotics. So it's somehow linked to AI, but it's not the pure AI. But all those new applications are all based on data and how they use the data to improve and to generate something more valuable in the future.

I want to give the audience a chance to ask a question again. Any questions? I could keep going, but I want to give you a chance. No questions? It's all clear. So actually, another question about -- we actually -- we've only got -- we don't have that much long left, but I'm going to squeeze in 1 or 2 more. So your competitor, I'm sorry to ask about your competitor, but they just had the Analyst Day. So that's why I referred to them. They recently talked about hyperscaler customers, being technology-agnostic and don't care whether they're shipping HAMR or PMR, just -- they just want exabytes, they want terabytes. Is that -- do you agree with that?

I tend to agree. I tend -- I know I'm sure a customer needs a certain number of exabyte. And they don't care what is the technology inside the box. They care about how big is the drive because this is their TCO, especially in this period of time where pricing are not going down, they're going up, your TCO comes from the size of the drive. If you can put a 40 terabyte drive into a current data center, where before you had a 20-terabyte drive for the same cost of that location, you double the storage. And they sell storage, our customers are selling terabyte to all of us and to our companies.

So for the same cost, with a bigger drive, you can strongly increase your revenue and your profitability. That's why they care about the size of the drive. Said that, if you produce hard disk, you want to produce this bigger size with a better cost. And so the number of disk, the number of ads are important to the cost. So if you can produce the same capacity with a lower below material, you come out with a better cost reduction. But I would say a customer doesn't care if a hard disk is a PMR or a HAMR. And if it has 12 disks or 10, they don't.

We care because it's a cost opportunity and also because inside the box, there is a limited space. So you can add disk until a certain limit and then there is no space anymore. So HAMR is the future because we'll continue to increase the capacity per disk, so what we call areal density. So you can store more and more data in one disk. So you avoid that problem of the physical space, and you continue to produce bigger and bigger hard disk drives. But I would say, from a customer perspective, I don't think it's a big difference.

They care about exabytes and how much they're paying per exabyte and reliability in the performance.

They care about exabyte in total, they care about terabyte per unit. So they want 1 hard disk at 40 terabyte, not 2 at 20, because of what I told you before from the same location. Now they get more storage.

So just finally, just wrapping up for me, just looking around and see if there's any questions, please put your hand up any questions. Otherwise, we're going to be wrapping up in a minute. And what do you think is most misunderstood on the Street? I mean, obviously, Seagate stock has been pretty strong. Do you think there's anything that's misunderstood? Or anything you want to point out that you think Wall Street analysts like myself or buy side are not getting quite right or missing?

What I would say, in general, I think the story is fairly well understood, the application that are running and driving the need for more storage are very well known. And is fairly clear where demand is stronger than supply. I think -- I think it's important to understand that this industry is generating an increase in exabyte year after year after year. So there is a huge increase in exabyte that maybe will not fully answer to the demand, but it is a huge increase and is a strong industry. It's a very consolidated industry.

I think there are differences between the industry today and the industry years ago.

And I think the business is very strong and the cyclicality, if any, will be very different in future compared to what has been seen in the past. I think from time to time is still hear about the NAND versus hard disk. That is a part that I don't understand. I don't understand how after 15 years of cloud infrastructure being exactly in the same way. People can still think that will change somehow. And by the way, as I always say, it's not that there's a lot of NAND available. So that is maybe some -- I don't want to say doubt, but some possible small misunderstanding.

Too much focus on that.

Yes. But overall, I'd say the story is stronger. I think is recognized from the analysts and our investors.

Okay. Well, great. Thank you very much, Gianluca, for joining us today. And thanks, everyone, for joining today.

Thank you.

Welcome to the Seagate Technology Fiscal Second Quarter 2026 Conference Call. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Shanye Hudson, Senior Vice President, Investor Relations. Please go ahead.

Thank you, and hello, everyone. Welcome to today's call. Joining me are Dave Mosley, Seagate's Chair and Chief Executive Officer; and Gianluca Romano, our Chief Financial Officer.

We've posted our earnings press release and detailed supplemental information for our December quarter results on the Investors section of our website. During today's call, we will refer to GAAP and non-GAAP measures. Non-GAAP figures are reconciled to GAAP figures in the earnings press release posted on our website and included on our Form 8-K. We've not reconciled certain non-GAAP outlook measures because material items that may impact these measures are out of our control and/or cannot be reasonably predicted.

Therefore, a reconciliation to corresponding GAAP measures is not available without unreasonable effort. Before we begin, I'd like to remind you that today's call contains forward-looking statements that reflect management's current views and assumptions based on information available to us as of today and should not be relied upon as of any subsequent date.

Actual results may differ materially from those contained in or implied by these forward-looking statements as are subject to risks and uncertainties associated with our business. To learn more about these risks, uncertainties and other factors that may affect our future business results, please refer to the press release issued today and our SEC filings, including our most recent annual report on Form 10-K and quarterly report on Form 10-Q as well as the supplemental information, all of which may be found on the Investors section of our website.

Following our prepared remarks, we'll open the call up for questions. In order to provide all analysts with the opportunity to participate, we thank you in advance for asking 1 primary question and then reentering the queue. With that, I'll turn the call over to you, Dave.

Thanks, Shanye, and hello, everyone. Seagate closed out calendar 2025, with a record-breaking quarter, driven by sequential revenue growth across nearly all end markets. December quarter financial results exceeded both top and bottom line expectations and set new company records for exabyte shipments, gross margin, operating margin and non-GAAP earnings per share.

We expanded non-GAAP gross margin above 42%, supported by the execution of our pricing strategy, along with an improving mix of our high capacity drives as HAMR shipments ramp. Looking at the entire calendar year, 2025 marked a transformational period for Seagate, both financially and operationally. Over the calendar year, we increased revenue by over 25%, improved gross margins by nearly 740 basis points and expanded operating margins by an even greater amount, demonstrating the profitability leverage in our financial model.

2025 also solidified HAMR technology as a long-term enabler of mass capacity storage. We ended the year shipping 3 terabyte per disk Mozaic-based HAMR products to our first CSP customer. And by year's end, quarterly HAMR shipments exceeded 1.5 million units and have continued to ramp. Mozaic 3 HAMR drives are now qualified with all of the major U.S. CSP customers and qualifications for our second-generation Mozaic 4 terabyte per disk products are tracking well to plan.

These developments align with our long-term areal density road map that extends to 10 terabytes per disk, which we expect to deliver early in the next decade. I want to thank our Seagate teams around the world for exceeding our performance expectations and delivering outstanding value to our global customers.

We continue to operate in an exceptionally strong demand environment, particularly within the data center end markets. In the December quarter, we saw sustained demand growth for our high capacity nearline drives across global cloud data centers as well as continued improvement from the enterprise edge. Based on our build-to-order pipeline, we anticipate these positive demand trends will continue for some time.

Our nearline capacity is fully allocated through calendar year 2026, and we expect to begin accepting orders for the first half of calendar year 2027 in the coming months. Further out, demand visibility is strengthening based on the long-term agreements in place with major cloud customers through calendar '27. Additionally, multiple cloud customers are discussing their demand growth projections for calendar '28, underscoring that supply assurance remains their highest priority.

We will continue to meet strengthening demand through our strategy to maintain supply discipline and satisfy exabyte growth through areal density advancements and without increasing unit production volume. In the December quarter, our average nearline drive capacities rose by 22% year-over-year, approaching 23 terabytes per drive, with those sold to cloud customers averaging significantly higher. This trend underscores the strong adoption of our higher capacity drives to support demand growth.

At the same time, revenue per terabyte sold has remained relatively stable, reflecting the effectiveness of our pricing strategy. Seagate is well positioned to continue benefiting from the combination of powerful secular tailwinds and supply discipline. Video applications continue to drive significant demand for hard drives with platforms like YouTube witnessing 20 million video uploads daily, up from just 2 million 3 years ago. This staggering pace of growth extends to other cloud video platforms and doesn't yet include the full surge in content generation expected from emerging AI-driven video applications.

These applications are not only fueling social media uploads, they are also transforming how organizations turn their data into tangible value, enabling personalized marketing, interactive education and advanced simulations, capable of training manufacturing, engineering, health care and other professionals. The strategic value of data is further underscored as new applications and use cases emerge across cloud and edge workloads.

Among the most promising of these is agentic AI, which relies on persistent access to large volumes of historic data to enable effective planning, reasoning and independent decision-making. Adoption is already gaining momentum with one recent survey conducted by a leading cloud service provider reporting more than half of participating customers were actively using AI agents. Early adopters are already realizing measurable returns with benefits ranging from lower cost to increase revenue opportunities.

With the deployment of AI agents at the edge, where untapped data often resides, we believe the stage is set for a sustained and meaningful increase in data generated and stored that will support inferencing, continuous training and also maintain model integrity. Modern data centers have evolved to address the complexity and scale that massive workloads bring through sophisticated data tiering architectures, ensuring that the right data is available at the right time and place.

Hard drives are essential to these architectures, anchoring the mass capacity data tier that stores the vast majority of exabytes from storing the checkpoint data sets used to train and maintain model integrity to supporting vector databases that provide the context necessary for accurate inference results and agentic AI performance. By leveraging hard drives, data center operators, whether in the cloud or on-prem, can achieve the optimal balance of performance, capacity and cost efficiency at scale.

Against this transformational backdrop, Seagate's HAMR technology road map positions us to meet growing demand and deliver ongoing TCO improvement for our customers. HAMR is a proven technology with large volumes of drives running in cloud production environments for more than 3 quarters now, and performing well across a broad spectrum of use cases. We are systematically ramping our Mozaic 3 HAMR products to qualified customers while maintaining focus on optimizing the profitability of our available supply.

As noted earlier, Mozaic 3 is now qualified with all major U.S. CSP customers and remains on track to have all global CSPs qualified within the first half of calendar 2026. Additionally, qualifications of our second-generation Mozaic 4 products are progressing well. We expect to begin the ramp of Mozaic 4 later this quarter and have multiple CSPs qualified in the coming months in line with our plans. We continue to set the pace for the industry, recently demonstrating 7 terabytes per disk capability in our labs. As one of our largest CSP customers recently aptly described, hard drives are engineering marvels, a sentiment that we obviously share.

Our deep expertise across mechanical engineering, material science, nanoscale fabrication and now advanced photonics, not only enable Seagate to deliver on the HAMR road map, but also creates a durable competitive moat for hard drive technology well into the future.

Wrapping up, 2025 was a milestone year for Seagate in every respect. Financial performance, operational execution and technology leadership. We are carrying this momentum into calendar 2026 supported by a powerful demand backdrop as new AI applications start to complement traditional workloads.

We will remain highly disciplined and focused on expanding profitability through our higher capacity product mix, underpinned by the strong economics of HAMR. Our areal density road map positions Seagate to sustain the core TCO and efficiency advantages of hard drives as data creation and storage requirements accelerate in the AI era. We believe this foundation creates a compelling long-term value proposition for the company, our customers and our shareholders.

I'll now turn the call over to Gianluca to cover our results in greater detail.

Thank you, Dave. Seagate delivered another quarter of strong year-over-year revenue growth and set new record profitability metrics in the December quarter, underscoring the durability of data center demand trends. Additionally, we strengthened our financial position by retiring $500 million in gross debt and generating over $600 million in free cash flow, marking the highest level in 8 years.

December quarter revenue came in at $2.83 billion, up 7% sequentially and up 22% year-over-year. We achieved non-GAAP gross margin of 42.2%, up 210 basis points sequentially, and we expanded non-GAAP operating margin by 290 basis points sequentially to 31.9%. Our resulting non-GAAP EPS was $3.11, up 19% quarter-over-quarter. These strong financial results demonstrate our ability to execute our strategic objectives, including leveraging our technology road map to support demand growth.

To that end, we shipped 190 exabytes in the December quarter, up 26% year-over-year, while keeping overall unit capacity relatively flat. The data center market accounted for 87% of our shipment volume, supported by ongoing demand momentum from global cloud customers and sequential growth across the enterprise OEM markets. We shipped 165 exabytes in the data center market, up 4% sequentially and 31% year-on-year.

Data center revenue grew at roughly the same pace totaling $2.2 billion for the quarter, up 5% sequentially and 28% year-on-year. Against this strong demand backdrop, both cloud and enterprise customers are transitioning to higher capacity drives. Average cloud nearline capacity increased to nearly 26 terabytes in the December quarter and will continue to grow with the ramp of HAMR-based Mozaic products. As Dave highlighted, Mozaic drives are running very well in production environment and meeting all performance, reliability and integration expectations.

In the enterprise OEM market, we are benefiting from slight improvement in traditional server units, along with increasing demand for storage servers, driven in large part by the adoption of AI applications and need to store data at an enterprise edge. The edge IoT market made up the remaining 21% of revenue at $601 million, supported by anticipated seasonal improvement for consumer products in the VIA client market. We project the broader VIA market to grow over time with the largest growth contribution coming from VIA nearline products that are captured as part of our data center end market.

Moving on to the rest of the income statement. Non-GAAP gross profit increased to $1.2 billion, [ up 14% ] quarter-over-quarter and 44% compared with the prior year period, significantly outpacing revenue growth. Non-GAAP gross margin expanded to 42.2% in the December quarter up from 40.1% in the prior period. This improvement reflects the ongoing execution of our pricing strategy and the growing adoption of our latest generation high capacity products, which collectively drove a modest sequential increase in revenue per terabyte, a trend we expect to continue into the March quarter.

Non-GAAP operating expenses were $290 million, relatively flat quarter-over-quarter and in line with our expectations. Operating expense as a percent of revenue declined to 10.3%, rapidly trending towards our long-term target of 10%. The combination of strong top line growth and significant financial leverage drove an 18% sequential improvement in non-GAAP operating profit to $901 million, almost 32% of revenue. Other income and expenses were $70 million, reflecting slightly lower interest expenses on the reduced outstanding debt balance.

We currently project other income and expenses to remain relatively flat in the March quarter. We grew non-GAAP net income to $702 million with corresponding non-GAAP EPS of $3.11 per share based on tax expenses of $129 million and a diluted share count of approximately 226 million shares, including the net impact of our 2028 convertible notes.

Turning now to cash flow and the balance sheet. We invested $116 million in capital expenditures for the December quarter or roughly 4% of revenue. We are maintaining capital discipline while we continue to transition and ramp HAMR technology. To support these objectives, we anticipate capital expenditures for fiscal year 2026 to be inside our target range of 4% to 6% of revenue.

Free cash flow generation was strong at $607 million, up 42% from the prior quarter. Looking ahead, we expect free cash flow generation to further expand in the March quarter, supported by sustained demand trends, operational efficiency and capital discipline. These factors position us well for durable long-term cash flow generation.

Cash and cash equivalents totaled just over $1 billion at the end of December quarter, with ample liquidity of $2.3 billion, including our undrawn revolving credit facility. During the December quarter, we returned $154 million to shareholders through dividends. We retired approximately $500 million of exchangeable senior notes due 2028, which serves to limit further dilutive impact from business and optimized cash deployed for future share repurchases.

Our resulting gross debt balance was approximately $4.5 billion exiting the quarter. Net leverage ratio improved to 1.1x based on adjusted EBITDA of $962 million for the December quarter, up 16% quarter-over-quarter and up 63% year-on-year. We expect the net leverage ratio will trend lower as profitability and cash generation increase while we continue to evaluate opportunities to further reduce debt.

Turning now to the March quarter outlook. The demand environment remains strong, particularly among global cloud customers. As a result, we expect data center demand will more than offset typical March quarter seasonality in the edge IoT markets. We expect March quarter revenue to be in the range of $2.9 billion, plus or minus $100 million, which represent a 34% year-over-year improvement as a midpoint.

Non-GAAP operating expenses are expected to be approximately $290 million. Based on the midpoint of our revenue guidance, non-GAAP operating margin is expected to approach the [ mid-30% range ]. Non-GAAP EPS is expected to be $3.40 plus or minus $0.20, based on a tax rate of about 16% and non-GAAP diluted share count of 230 million shares, including estimated dilution from our 2028 convertible notes of approximately 7.6 million shares.

Seagate's strong December quarter performance and March quarter guidance underscore our continued focus on driving growth, enhancing profitability and optimizing cash generation. Based on our current outlook, we expect to deliver sequential improvement to both the top and bottom line throughout calendar 2026 and remain in a strong position to enhance value for both customers and shareholders over the long term.

Operator, let's open the call up for questions.

[Operator Instructions] And the first question comes from C.J. Muse with Cantor Fitzgerald.

Given the supply-demand dynamics, you're obviously in the catbird seat. I wanted to really try to get some more detail on gross margins going forward. Your philosophy historically has been to share gains, both your customers and yourselves. But at the same time, given this tight environment, you are raising like-for-like pricing. So curious, is there a framework to think about in terms of the incremental gross margins that we should model from here?

And then, I guess, maybe bigger picture, as you think about overall average pricing per exabyte, we've gone from kind of down double digits to high single digits. And I think we just exited the quarter down 4% year-on-year. Do you see a world where pricing could flat or even move positive year-over-year?

Yes. Thanks, C.J. I'll let Gianluca chime in here as well, but the pricing will be dictated by the demand. Right now, the demand is really strong. So I think as we roll through into '27 and '28, we look at how much capacity we're having. We're bringing online by virtue of the fact that we're making all these aggressive product transitions. We'll bring more exabytes to bear and then people go out there and renegotiate for those. I think flat to slightly up is certainly possible. And that's the way we're really managing it as we talk to our customers. The value proposition of the new drives as they go up 5, 10 terabytes at a time is pretty strong.

C.J. So on the gross margin, we are executing very well, but executing a little bit better than what we discussed at our Investor Day, where we presented a model with a 50% incremental margin above $2.6 billion of revenue. We have done better every quarter, of course, is our objective to continue to optimize what we produce, what we sell and finance the profitability that we can get from the product. So the models cover over a longer period of time, now 2, 3 years, not 2 or 3 quarters, but I'm positive we are continuing to progress in the right direction.

The next question will come from Wamsi Mohan with Bank of America.

I have a similar type of question. I guess the gross margins in the guide and the incremental quarter-on-quarter gross margins on the guide are very strong. Can you maybe help bridge the drivers between mix and price? Obviously, you've got a better mix of data center revenue next quarter. But just wondering if you can dimensionalize that. And the opportunity for pricing, David, you just said sort of flat to up as possible. But as we think about the pricing that might be getting embedded within these LTAs and sort of beyond '26. Why can't that be a lot higher just given the tightness in the supply-demand environment?

Yes. I think this gets into how persistent is the demand going to be, Wamsi, we talked about 2 or 3 years from now. The one behavior change that I really like in the last year is that people are starting to say, if I can't get it now, I'll plan next year better and the following year better. So we're having great dialogues on that front. Of course, supply has risen quite a bit in the last year's supply of exabytes from the industry. The industry has reacted pretty well, but I think demand is still pretty strong.

And my perspective on this is I think demand will stay strong for quite some time. So in that kind of world, we're having great discussions with customers further out in time. And the biggest part that helps us in our planning is through these product transitions. They know that's how they get more exabytes.

Yes. And Wamsi, we are saying in the script today that for the rest of the calendar year, we expect revenue and profitability to continue to improve sequentially every quarter. So we are not implying in any way that this trend is changing. It's actually now getting better somehow.

The next question will come from Erik Woodring with Morgan Stanley.

Congrats on these results, incredible. Dave, at your Analyst Day last year, you kind of pointed to a mid-20% exabyte growth CAGR. And I'm just wondering where you think that supply growth can land this calendar year. And as you get closer to that HAMR crossover point later this year, like does that pace of exabyte growth accelerate? And I'm just asking this because demand is clearly outpacing supply. So can you maybe just help us try to better understand the shape of your exabyte supply growth because obviously, it will dictate kind of exabyte shipments for the year.

Yes. Thanks, Erik. So we are planning to transition to 4 terabytes of platter. And fairly aggressively, but I think what people have to keep in mind is that we were fairly tight all throughout manufacturing. So we have products that are in the pipeline already that are committed to customers and so on. We don't just move very quickly to 3 or 4 terabytes of platter as things come. And it's a good problem to have, actually. We're running manufacturing quite, quite tight right now.

So I think it will be a fairly prescriptive ramp, to your point. It won't be as fast as maybe we've done some ramps in the past, but it will be very profitable, and that's the way we look at it. As we go further out in time, I'm very optimistic that the 4 terabytes per platter is a very strong product. It will start to replace some of the other legacy products, I'll say, that way and because it has so much better value proposition in a lot of those markets. And then when that happens, then we see more opportunity.

The next question will come from Asiya Merchant with Citi.

Great results here. Just a couple that are related to the prior question. You guys gave some projections on HAMR, not just for fiscal year '26, but even into fiscal '27. So if you could talk about upside to achieving those targets for the HAMR rollout. And related to that, how we should think about the blended cost reductions, pretty impressive, again, margins here and guiding for improved profitability. So if you could talk to us a little bit about the cost reductions going forward, especially as you ramp HAMR here with the Mozaic 4, that would be great.

Yes, Asiya. So I would say, first of all, we are very happy with the transition to HAMR. We qualified the last big cloud service provider in the U.S. and we have qualified 6 out of 8 of the top cloud service providers. So the transition from PMR technology to HAMR technology, is progressing very well. And we are now qualifying the new product, the 4 terabyte per disk to 40 terabytes per drive. Of course, this will help with the increase in exabyte in terms of mix. We gave a good indication, I think, at our Investor Day, and we want to be aligned to that. And the cost will be favorably impacted, especially when we start ramping high volume of the 40 terabyte drive. Of course, that will drive a fairly important reduction in cost per terabyte compared to the current HAMR, and of course will be a good contributor to further increase our gross margin.

The next question will come from Karl Ackerman with BNP Paribas.

I was hoping you could clarify what portion of your LTAs for overall nearline HDD capacity has fixed or multi-quarter pricing agreements. I ask because as these LTAs roll off throughout 2026, any new agreements will be locked in at higher values reflective of not only the demand use case also widening price per terabyte gap between enterprise hardware drive HDDs.

And Karl, your -- the second part of your question was a little fuzzy. So we captured the first part, but I might ask you for clarity on that second.

Sure. Yes, I'll just repeat, if I could. As these LTAs roll off throughout 2026, I would imagine those new LTAs will be priced at perhaps a higher value or higher order value, particularly given the widening gap between hard drives and SSDs. So if you can comment on the mix of LTAs and how you think that progresses throughout '26 would be great.

Yes. Thanks, Karl. So as we roll off, say, for example, somebody might have been qualified on a 2.4 terabyte per platter product or something, and then they might be qualifying a 3.2 or even a 4-terabyte per platter as we roll forward. So we changed based on the demand that we see, we changed -- and our available supply, we changed the pricing dynamic there. I think that's one of the biggest things you're pointing out.

I'll say that '26 is fairly booked. We talked about that in the call a bit to the extent that we can out-execute our plan, it will be marginal like you saw last quarter, we get the qualifications done a little faster. We ship a few more drives. That's how we can do better than planned. But other than that, it's fairly predictable in '26, and we're looking to start '27 the same way.

Next question will come from Jim Schneider with Goldman Sachs.

I was wondering if you could maybe address -- given everything you just said about demand and about the mix effect from HAMR this year, maybe can you give us any kind of directional guidance about where you might expect exabyte shipments to end up on a calendar '26 versus calendar '25 basis relative to the sort of long-term targets you've laid out previously. It seems like you could do materially better than that, but I just wanted to confirm what your expectations were if you could give us a numerical range.

Jim, no, we are not guiding calendar '26. But we said in our financial model, we said that we expect exabyte -- nearline exabytes to grow in the mid-20%. We have done a little bit better if you look at the last few quarters, and we always -- as Dave said before, we always try to extract as many exabyte as we can from our manufacturing. So we are continuing this trend, but we don't guide for calendar '26.

But moving from 2.4 per platter to 3 per platter to 4 per platter, you can see that we're on a trajectory like you described. It's -- when it gets down to the individual customer level, obviously, we have to be very predictable because they need what they need and what they -- what we've committed to in order to build out that data center. So we'll continue to execute that plan and maybe we can do a little bit better as we transition to 4 terabyte per platter.

The next question will come from Amit Daryanani with Evercore.

Gianluca, I'm hoping you can talk a little bit about the March quarter guide because it seems to be a really sizable uptick in gross margins. I think it's up like 250 basis points or 100% plus incrementals. Could you just -- is there any you would call out in March quarter that's unique that's helping drive that kind of margin expansion? And is this really all coming from the core HDD business? Or is there a potential benefit from the old systems business helping you as well?

Amit, well, I would say, we expect it to be a very good quarter. I don't think it's different than what we have done before. It's always based on the pricing strategy and the mix, as you know. We qualified another customer on HAMR, so we will ramp a little bit more volume on HAMR. This is helping us to get better margin. But fundamentally, is not really different in how we think we are going to execute the quarter and is good. I think the incremental margin was very good.

Yes. And it's not the systems business. The systems business is doing well, but it's fairly small scale in comparison.

The next question will come from Mark Newman with Bernstein.

Congrats on great numbers today. Just want to touch again on this, the LTAs and pricing arrangements you have. Just curious, do you think there's an opportunity here for more significant price increases in NAND flash. We're hearing things like 40% to 100% up Q-on-Q for some contracts. I appreciate hard disk drives -- you have very long-term agreements. But I think there's a lot of questions I'd like to just touch on this as well. And a lot of -- as the LTAs roll off, is there an opportunity for some of those to be repriced at a more significantly higher price to change the trajectory, certainly numbers are great, you're printing.

We're just trying to figure out, could you start to see more significant price increases rather than at the moment, you're seeing kind of flattish down a little bit, up a little bit. But overall, your average prices are flat, which I understand is a mixture of like-for-like slightly up, offset by new products coming in at lower price. Just wondered if that may change. And then just a quick update on HAMR mix, if there's any update on the trajectory of the HAMR mix that you've outlined before.

Thanks, Mark. A couple of points. On the HAMR mix, we necessarily constrained ourselves on the 3 terabyte per platter because the factories were fairly full, and we knew we would be going to the 4 terabyte per platter product. So we've been leaning harder on that and making sure it gets through the development and qualification phases. As time goes on, then we'll move off and on to the 4 terabyte per platter very aggressively. So that helps you on the mix side. And the other thing about HAMR mix is it will be necessarily mixed up. I think the demand for those products will be at the high capacity points, not necessarily the lower capacity points just yet.

And then relative to pricing, I think I said before, as we -- as one long-term agreement rolls into the next year or the next year, we've satisfied our existing supply commitments, people are looking at the new products, we have constrained supply of those new products, then we look at what the demand is, and we dictate where our pricing is. And one of the very first questions I said it could be flat to up a little bit. That's the way I think about it right now, but it all depends on what the demand is.

Demand continues very strong. That's great. And again, what we're seeing is people who can't get what they need today, they're saying, okay, I need to be able to plan my data center procurement out in the future, let's get more predictable in the future, has given us better visibility, helps us run our factories for better cost and so on. So that's great.

The next question will come from Krish Sankar with TD Cowen.

I had a question, I just want to put it in 2 parts. One is how much was your HAMR as a percentage of your exabyte shipment last year? How much do you expect it to be this year? The genesis of the question is I'm just trying to figure out, obviously, a lot of questions on the very strong gross margins. If there's a way to put it in 3 buckets, like how much of the gross margin upside is driven by pricing? How much is driven by product mix? How much is driven by cost reduction by offshoring manufacturing?

Yes, there's really no offshoring manufacturing or anything like that involved. Our manufacturing operations around the world are doing quite well and quite full. So that's helping from a cost perspective. But really no change in any manufacturing strategy to speak of. Relative -- I would say a lot of what -- the benefits we're seeing is mix and mix not just because we're actually transitioning into higher and more -- better products over time, but also because the demand for those products is quite high.

You think about it, if you're building a data center with a 3 terabyte per platter versus the 4-terabyte per platter, you're going to be running that data center for a long time, you want the higher capacity point. And to the extent that we can do that as predictably as possible, that mix is what's driving the stability out in the market for us and helping us plan.

Yes, Krish, we don't give specific details on the impact of pricing, mix and cost. But they are somehow interrelated. I would say the change in mix is helping with the cost reduction and the supply-demand situation is, of course, supporting our pricing strategy. So they are all very good contributor to the increase in gross margin. And as we said before, this is going to continue through the calendar year.

How much of HAMR as a percentage of the mix?

Well, Dave gave an indication of the unit that we shipped in the last quarter. So I think you can fairly easily calculate that.

The next question will come from Steven Fox with Fox Advisors, LLC.

I guess I was just wondering on this -- on your mix question, looking at your average capacity per drive being up 22%. Like how much of that -- like obviously, the supply demand environment has tightened over the last year. And in reaction to that, are you taking steps to accelerate that mix up as the customers pushed you that way? Like I'm just curious how much you can control going forward now that we're here on even tighter supply to sort of help your customers in terms of absolute petabytes you're delivering.

Thanks, Steve. So yes, we are -- the lead time out of the wafer fab is quite long. So we have to be predictable for our customers, say, 6 months, 9 months later and so on and so forth. That's one of the reasons why we talk kind of a year at a time inside of these LTAs. So we start wafers based on what we know we're going to be able to deliver, so that we're as predictable as we can be for our customers.

As were -- if we're deploying manufacturing engineered resources we're trying to get through these product transitions. That's what gets us the most exabytes after that. And so going, mixing up is kind of our goal. So if that helps clarify what our strategy is.

It does, Dave. I'm just wondering like when you had your analyst meeting, you said that sort of a pretty well-defined time line for node transitions. Maybe just can you give yourself a report card on how you're doing on some of those time lines if we look out now versus the next year or longer term?

Yes. I think that's good. We're on the plan or slightly ahead. And again, most of that's under our control. We execute well. We've been executing well. Some of it's under our customers' control as well. And the behavioral changes we've seen in the customer I made reference to earlier, they're really pulling hard because they need more exabytes. And so that helps get the calls done quickly. It helps a road map alignment and then supply -- specific supply alignment, which helps our factories.

So next question will come from Aaron Rakers with Wells Fargo.

Congrats on the results. I want to go back to gross margin. I know you talked a lot about the pricing dynamics and the visibility you have. But the thing that stands out to me is you've been executing on like a cost per terabyte of like a mid-teens year-on-year decline in these last several quarters. As we roll out the 4 terabyte per platter Mozaic drive, how do I think about that cost down curve? Is it mid-single digits? Is it -- can you sustain at double digit and would not -- wouldn't we expect the 4 terabyte per platter HAMR drive to actually maybe accelerate the cost down, given the ability to bring that into lower end other outside of the nearline platforms. So I'm just curious how you think about that cost down curve.

Yes, we are very positive on the 4 terabyte per disk in terms of impact on the cost, as we discussed before. The unit costs tend to be fairly similar. But of course, we are adding a lot of content per unit. So that will be a good help to reducing the cost and improving profitability. So as you know, we are qualifying 2 major customers on these new products. So the time to finalize the call and their ramp probably through the end of the calendar year and for sure, well into -- the impact will be strong, I think, in the next calendar year, too.

And we plan on making a big transition to 4 terabytes per platter over the coming few years and then getting to the 5 terabytes per platter as well. We do add complexity as we make those transitions. But I'd say the first order, the things that dictate the speed of the ramp are our ability to go work scrap and yield, all through our supply chain and so on, and we're working very hard on that. I like the product. So I think it provides for a bright and stable future for us. We just need to stay focused on it.

The next question will come from Timothy Arcuri with UBS.

I want to ask about LTSAs. I think you said nearline capacity is allocated through 2026. So it sounds like both pricing and exabytes are locked in this year. But for '27, I think you said something that I took that exabyte and pricing is not locked in, but you have some sort of agreement. So I guess I had 2 questions. First of all, is it right to assume that pricing is also locked in for all of '26? And what sort of agreement are you referring to for 2027 if volume and pricing is not locked in next year?

Yes. For this calendar year, we said basically, we have PO in place for all the quarters, so volume and pricing is well defined. As Dave said before, if in a quarter, we can produce a little bit more, of course, we will sell those exabytes in the open market at a good profitability. But I would say we have -- the vast, vast majority of the volume is already allocated. Calendar '27, we will start working on that fairly soon. Of course, we have very good indication and agreement on volumes, but we have not -- we have not fixed the price yet.

Yes. And Tim, if this helps, so we haven't really started the longest lead time parts, but we will very soon for the start of '27. And we need to start having those discussions with our customers which calls are we going to get through together with -- what exactly is the plan, because a lot of them need predictability as well. So we'll have to build in our factories what -- based upon how hard they want to pull on those new products.

The next question will come from Mehdi Hosseini with Susquehanna Financial Group.

Just a quick housekeeping item. Gianluca your CapEx has been increasing on a Q-over-Q basis. How should I think about depreciation, especially since it did dip in the December quarter? Any color here would be great looking forward.

Yes. No, our CapEx is aligned to our target of 4% to 6% of revenue. We are actually at the bottom of that range. So is not increasing in terms of what we want to achieve and what we said, of course, comparing to a period where we were more into the down cycle in terms of dollars, of course, is higher. We are supporting our HAMR transition and HAMR ramp. So I would say there is nothing different than what we said.

Yes, the way I think about it as well, Mehdi is that if you go back 2 years ago, and you use that as a baseline, we were still significantly lower revenue, but also we were challenged on the supply and demand balance. Right now we're in a totally different environment, of course. So we'll probably stay well within the 4% to 6% range. But as the revenue goes up, we'll spend a little bit more and probably the first priority is maintenance tools and the things that we weren't doing a couple of years ago.

I apologize, I may have confused. I was focusing more on depreciation given these several quarters of increase in CapEx, should I expect a step-up in depreciation looking forward?

Depreciation will follow the CapEx. So you have your -- I guess you have your model on the revenue, so you can calculate the 4% to 6% of CapEx. And then depreciation for us is on a 10 years useful life. So you can probably model it that way.

It's not like -- some other fabs, it's not necessarily the huge part of the cost drivers. There's a lot of other pieces of the cost that we can go manage them.

The next question will come from Ananda Baruah with Loop Capital.

Dave, while we have you, a little bit of a technical one, are you -- what kind of activity are you seeing at sort of the so-called warm tier of storage. It's a question that comes up a bunch in our conversations. We've heard that it's obviously growing, it's growing both hard drive and flash storage is participating nicely, but would love to get your input on it. Because I think there's still -- first of all, we love to know if what we're hearing is accurate.

But secondarily, I think there's a lot of people that are assuming that that's really like it's becoming a NAND tier. Largely a NAND tier in the GenAI world. And anyway, just love to get any context there that you have?

Yes. I think you have to be a little bit careful, Ananda. So there are applications that are very memory dependent that are attached to compute and some of these applications are neat, I like them. When you get -- when you start talking about big data storage, if you will, in data centers, the tiering architecture is fairly well set and probably won't change based on economics and also architectures that are well known.

People know how to play. So if the concept is that drives aren't working hard, they're in the background, just storing data. That's not the way -- a good way to think about it. That's not the way hard drives are being used right now. They're working 24/7. A lot of times, they're optimized for performance themselves, largely streaming performance not random small block workloads. That's more of a memory thing. And so if you had an application that's random small block, it's probably memory. If you have big data, it's probably a little bit of memory on the front end and a lot of hard drive on the back end. And we think that there are applications across the entire spectrum, of course, but we think that in the future, when we start to talk about the concepts in their enormity about checkpoints and physical AI and video and things like that. It's large, large data so that the architectural tier that stores the data will probably remain constant for the next decade.

That's super helpful. I'll keep it there.

Next question will come from Vijay Rakesh with Mizuho.