Seagate Technology Holdings plc (NASDAQ:STX) Bank of America 2023 Global Technology Conference June 7, 2023 1:00 PM ET
Company Participants
Gianluca Romano – Chief Financial Officer
Conference Call Participants
Wamsi Mohan – Bank of America
Wamsi Mohan
Thank you for joining us here today at the BofA Global Technology Conference. I’m Wamsi Mohan. I cover IT Hardware here for the Bank. Delighted to welcome Seagate to our conference today. We have CFO, Gianluca Romano, and EVP as well. He has been CFO since 2021. And he has seen an interesting couple of years with COVID, with supply chain, with demand, inventory. So we are looking to get updates from him on a variety of stuff. So welcome, Gianluca. Thanks for joining us today.
Gianluca Romano
Thank you, Wamsi. Well, before we start with a question, 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.
Question-and-Answer Session
Q – Wamsi Mohan
Okay. Great. Well, with that out of the way, like, how does the demand backdrop look? And what areas are you maybe most confident about a rebound by the end of the year?
Gianluca Romano
I would say, more recently, we started to see different trends in different parts of the business, in the different market segments. I would say in China, we started to see a recovery. So we expect a sequential improvement in the revenue. Of course, it’s still very far from where it was before the down-cycle, like December 2021. But it’s starting to recover from the December level, the March level. So some good signs in that part of the world. And that is impacting a certain number of the segments. But in particular, I’ll say, video and image application. We also see some improvements in enterprise OEM outside of China, so mainly U.S. and the rest of the world. Again, still far from the peak, but at least sequential improvement.
The legacy part and the non-hard disk part is probably fairly flat at least from what we can see before entering the quarter. And as you know, as we discussed many times between earnings release and today, the part that is sequentially declining is the cloud. And we discussed what happened recently, especially in the March quarter, what is the situation with the inventory, what is the focus of our customer in the cloud space, especially in the U.S. cloud space. They are trying to increase utilization rate. They are trying to optimize what they have put in place in the last two or three years. And then we need, of course, to go through this period of time and have the inventory — their inventory to start to decrease, and then we will go through another cycle of growth.
In general, especially in that part of the business, but I would say in all the hard disk business in general, you have a period of growth where customers are focused on putting capacity in place based on their long-term demand expectations. And then when you have a macroeconomic situation that require more focus on spending, their focus change for a certain period of time into using more of what they’ve already put in place. So increased utilization, decrease inventory and try to go through that period of time. And then when the macro improves, they go back in spending more into the growth and less into the increased utilization.
So cloud is the part that right now is creating more of a decline in our revenue sequentially, maybe even a little bit more than what we were expecting at the beginning of the quarter. I would say, we still need to go through the second part of June. A lot of revenue will happen in the last three, four weeks of the quarter. But we expect maybe cloud is a bit weaker. So maybe our revenue could be more in the lower half of our guidance.
Wamsi Mohan
Okay. Okay. So maybe to — thank you for that update. So as you think about your comments first on China and the VIA market improving, are you — when you think about where your peak levels were to where you were maybe exiting December versus where you’re thinking now, like what’s the magnitude roughly, if you could share of where you’re tracking now?
Gianluca Romano
Yes, but it’s still a huge, huge difference. We said in December 2021, our business in China was about $1 billion, and 30% of our overall revenue, right now is several hundreds of millions of dollars lower even in the June quarter. But I think the good point is sequentially it’s starting to improve. And we had a fairly good revenue on — from that part of the world. But of course, U.S. cloud is predominant in terms of revenue. And therefore, the decline in U.S. cloud is more than offsetting. It’s a good news that is starting to come out from China.
Wamsi Mohan
Would you say that off the bottom, I mean, from the top, you’re still several hundred million dollars below from the bottom, are you several hundred million dollars higher?
Gianluca Romano
I would say a little bit higher, not several hundred. But no, it’s not small. It’s is a good improvement. Of course, it’s not a jump. It’s not that we go back close to where we were before. But sequentially, it’s starting to be a good improvement. Hopefully, this will continue. I would say, after the March — the month of March, we saw a good improvement in that part of the business. Hopefully, it’s continuing, of course, from China, you have different inputs, different information. Some are showing a good improvement. Some are showing a slower improvement. I just saw that on the export number that we just reported, it’s not great. But still, I think it will be a sequential and gradual improvement more than a strong jump back to where it was before. So it will take a certain number of quarters.
Wamsi Mohan
As you think about revenue coming in towards the lower end of the range, how do you think about the rest of the P&L, gross margins? Are you changing sort of underutilization impact? Is that increasing? Are you making changes to your production decisions based on how things are tracking?
Gianluca Romano
Well, I’ll say the production probably not too much. We know we already cut production. At the beginning of the quarter, production is — the cycle time is fairly long. So we look two, three quarters away. We think underutilization charges will be around $50 million to $60 million, that is what we discussed as in earnings release. I would say, no, it’s more tactical. It’s not — I don’t expect a major change. I expect a little bit of a change from the midpoint that we guided, so not huge. But tactically, some of the cloud customers could consume a little bit less than what we were discussing at the beginning of the quarter. And of course, then whatever is not consumed in this quarter will be consumed in the next two or three quarters.
Wamsi Mohan
Understood. There’s also been some of the companies like Micron have faced some recent challenges coming out of China. Do you think Seagate’s exposed to those kind of risks?
Gianluca Romano
Well, I think every company could be exposed to those kind of risks, geopolitical risk is not a small risk. I would say I don’t see Seagate to be exposed to that specific risk. But again, we have seen rules changing in the last two or three years, very often. So it can always happen. Hopefully, the world is going in the opposite direction where it looks like U.S. and China are discussing more and trying to find a different solution that is not what we have seen in the last couple of years. And that will, of course, help the business in general, not us, in particular, but in general, the technology business to go back to a more normalized way of doing business with less constraints.
Wamsi Mohan
Can you maybe provide us an update on HAMR? It’s become a fairly key part of your story as investors think through it. Where do you — what’s your confidence in sort of launch of product? How is some of the qualification — early qualifications going on that?
Gianluca Romano
Yes. HAMR is growing very well. It’s growing actually a little bit faster than what we were expecting just a few months ago. So we are starting to ship already last quarter and this quarter for qualification. Usually, a qual takes a certain number of months. So we are going through that process. And we still expect to increase our revenue sequentially quarter-after-quarter. Probably you will see a good volume from HAMR that can start to impact our P&L, I would say, in three to four quarters from now. So not very far in time. Now you will see volume growing through this quarter, next quarter. But of course, to impact P&L, you need to have a certain volume. Otherwise, it’s not big enough. I think three to four quarters, and we start to see that.
Wamsi Mohan
Where would you say you could hit like 1 million drives that you would be sharing?
Gianluca Romano
Yes, we didn’t give a specific volume, but of course, you need to have close to that volume to really start to see the impact to the P&L, to the financials.
Wamsi Mohan
Can you talk a little bit about margins on HAMR and how you expect progression on that?
Gianluca Romano
I think HAMR give us and the industry in general, because we will not be the only one selling HAMR in the future, gives the opportunity to the industry to improve the profitability mainly because you can grow in capacity per unit without increasing the bill of material. With the current technology, PMR, you can increase capacity per unit, but you always need to increase by one disk and two heads. So there is a little bit of increase of areal density, but a lot of increase coming from more bill of material. So even if you increase the bill of material, you still have a decrease of cost per terabyte. But of course, there is a negative impact of higher bill of material.
When you go to HAMR, our 32-terabyte is based on 10 disks and 20 heads. So same number of disks and head of the current 20-terabyte PMR. The following product will be a 36-terabyte and will still be based on 10 disks and 20 heads. So all the increase is coming through areal density. The following one, 40-terabyte, still the same 10 disks and 20 heads. And also the 50, we said at our earnings release, in our lab, we are already running individual disk at 5 terabytes. So we can increase capacity by a lot without increasing the cost per unit because it’s the same bill of material. So that will give the industry and Seagate in particular, because we are not the first adopter, to have a cost reduction and to maybe give some of that cost reduction to our customers, not to improve their TCO, we want them to be successful, but not all. In the past, the vast majority of the cost was passed to customers. I think HAMR will give us opportunity to still provide a good TCO to customer, but keep a good part of that cost improvement into Seagate and therefore, improve the gross margin. That is the main focus, the main strategy of the company and why we spend so much money in developing HAMR in the last more than 10 years.
Wamsi Mohan
If you think about the qualifications that are currently underway, can you talk about what all capacity points you guys are trying to qualify?
Gianluca Romano
It’s a fourth product, so it’s basically the [32].
Wamsi Mohan
Okay. And as you think about the progression of that by the time you get to mid next year, what would you say would be all the capacity points that you could have qualified?
Gianluca Romano
I’ll say with PMR, we were able to have a new product basically every 12 months, 12, 15 months. And maybe take a little bit longer, but not much longer. So I would say every — probably every 18, 20 months. So a little bit longer, but of course, increasing capacity is much bigger. And as I said before, the bill of material is the same. So I think if we can execute this plan will be very positive to Seagate in terms of financials, but also to the industry in terms of technology progression.
Wamsi Mohan
And what about down the stack, if you were to go to lower capacity points, when would those get qualified?
Gianluca Romano
That is another good advantage of HAMR. If you think 40-terabyte based on 10 disks and 20 heads, there are probably only one segment that will use 40 terabytes at their point. So it’s only the cloud. The other segments of enterprise OEM probably will be in the 30 terabytes, video and image application will be in the 20 terabytes and other legacy products will be lower. But you can see, for example, in the video and image application, they could have a 20-terabyte with 5 disks and 10 heads. So this drive will basically compete to a traditional drive with 10 disks and 20 heads. So it’s a huge difference in bill of material, a huge difference in cost. Again, it’s not that we need to give all that cost to customers. We give a little bit of that cost benefit to customers. And the rest of the benefit should stay with Seagate and, again, improving our gross margin at very different capacity points using the new technology.
Wamsi Mohan
What percentage of your high capacity drives do you think would be if you look out maybe middle of next year to the middle of the following year? Like how would you say that migration would be what percentage of your high-cap would be HAMR?
Gianluca Romano
Well, we are not developing any PMR after the 24-terabyte. So we have a 24-terabyte coming out soon, next few months, you will see it. That is the last PMR product. So I would say more capacity point above 24 PMR that is probably 28 SMR. All the capacity above that are HAMR. So again, in a couple of years, I think all the cloud guys will focus on the high capacity. They will want to have an HAMR 36 terabytes compared to a PMR 24, to me makes a lot of sense. So I think the vast majority of our high capacity business will move to HAMR. Of course, depends also from how fast the company can ramp production on HAMR. But longer term, HAMR is a solution — as you were saying before, is a solution for not only for the highest capacity, but also for mid capacity, probably not for the very low capacity also because that business is declining and in a few years, probably will not be there anymore. But again, for the rest, it’s probably based on HAMR.
Wamsi Mohan
As you think about the next 12 months and ramping your production for HAMR, how does that influence your gross margin? Because obviously, you’ve not sort of achieved the yield or scale that maybe you’re still in the process of getting there. How does that play out from a gross margin standpoint?
Gianluca Romano
Well, there is no technical reason why the yield on HAMR should be different than the yield on PMR. I said that when you have a new technology, a new product, you need to learn, especially manufacturing because we have a lot of — you spend a lot of time in R&D to develop the product and to be sure technical is working, but you cannot really try to do on very high volume. So that is what you need to learn when you really start to ramp. What we said in the past is we are not anticipating low yield, just that don’t know what we don’t know yet. So we need to go through this period of time where we increase the volume up to the level that we were discussing before and see what is the yield. If it is like the PMR, we just ramp, no problem. If it is lower, we need to slow down, understand, fix the problem and then ramp. It’s something that everyone will have to do it anyway. And we go first. So we are the first one that will have to learn and understand. But I would say so far, we have not identified anything in the technology that should create a problem in term of it. But we have not run 1 million units yet, so we need to go through that process.
Wamsi Mohan
Right. And at steady state, like when you think about your longer-term gross margin profile, would you just get back to your long-term gross margin profile historic? Or would you go — is there opportunity to sort of exceed that or move that range higher with HAMR?
Gianluca Romano
Yes. I think we need to go step-by-step. The fourth step is to go back to where we were before the down-cycle. And of course, we need to go through this period of time where we focus on a lot of cost reduction, a lot of optimization, internal optimization even in our manufacturing. So we go through that. And I think as soon as we start coming back into the upcycle, you will see that level of profitability to come even at a lower revenue than where we were before, mainly because we have reduced our cost. So we think we can give better cost structure even when volume comes back up. So that is step one. Probably HAMR will be only a part of the risk. And then when we really ramp high volume of HAMR, as we said before, we see an opportunity for further strong reduction in cost per terabyte. And as I said before, we want to give a good TCO to our customers, but we think we can keep a good part of that into Seagate and improve our gross margin further and higher than where we were before.
Wamsi Mohan
So right upfront you mentioned maybe VIA doing a little better in China and cloud customer is tracking a little bit lower. As you look out over the next few quarters, do you think that those trends kind of still persist? Or do you think that what you’re seeing from your cloud customers is maybe just a slight delay of purchasing that, that kind of recovers back in by the September quarter?
Gianluca Romano
Yes, I think we will see a similar trend. I think China will continue to improve. I think legacy reasonably will stay fairly flat. Usually that segment is the one that declined. I would say, because we should be through the end of the down-cycle probably will not decline, but probably will not even increase, but stay fairly flat. And then cloud will have to take one quarter or two quarters to deplete the inventory. And I think this is the right thing to do instead of trying to push more and more volume into customers that they don’t need right now. So we need to go through those few quarters. And then we think through the end of the calendar year, we’ll start to see some improvement also in that segment. And I think ’24 will be a much better year on all the segments, all the segments.
Wamsi Mohan
Yes. Let’s hope so for that. I do want to give the opportunity to see if there are any questions in the audience. We do have a mic. So if there are any, just raise your hand, and we’ll get the mic to you. All right. There’s one right here, please.
Unidentified Analyst
Thank you very much. Great, great presentation. Very quickly, when we look at the data center, we can see lots of the alternative solutions versus HDD against PCIe SSD. So what’s the management’s long-term strategy is how to optimize the mix ratio, HDD versus SSD, which is becoming more a PCIe-based solution? And also, when we look at the AI training area, which is based on the more powerful GPU provided by NVIDIA and others, they are really recommending the very PCIe side as well. So what is the management strategy on this or opportunities or risk?
Gianluca Romano
It’s a very good question. I think two separate questions. Maybe let’s start from AI. Everyone is talking about AI right now. I would say there are two phases of AI. One is a generative phase. So how does this work? Basically access data that is available mainly in the cloud. So there is not much to do for an hard disk, data is already there. So they access the data, and they do — they generate something new. Let’s say they generate a new video. To generate this new video you don’t need the hard disk. You need the GPU and you need the flash and you need the DRAM, you need many other components. So on this phase, hard disk is not really part of the activity. But once you have the video, what do you do? You save it and you save on the cloud. So this is where the hard disk gets the positive impact from AI, is saving all the new documents, video, songs, everything that is created gets back to the cloud.
So it’s not one or the other, are two phases. One phase is clearly more oriented to the GPU and the components we were discussing before. And one is 100% hard disk. When you have the new — the result of what we have created, you save on an hard disk. And this increase the database, because then AI has to access the new information, the new data and create something new again, it’s an evolution. It’s not that you create that new video and it’s all what you do. One week later, you access that new video and 10 other new videos that people have created and you try to create something new. And that’s something you get stored on an hard disk again. So there is a huge benefit from hard disk because you will have more data to get store. If you create a video and you delete, that’s not a huge benefit. But when you create the video, maybe you create 10 videos and then you say, oh, I like those two and I save those two. That is where you get the benefit on the hard disk. So there is a lot of benefit from hard disk.
In general, when you look at NAND versus hard disk, I would say it depends. If you go on a very low capacity, there is, of course, an overlap between the two technologies. And when I talk about the legacy, that tends to decline in terms of volume year-after-year is because the flash is overlapping and taking over on those small applications. But even on a laptop, you need to think a little bit further. So you have a laptop, and you say I can have an hard disk or an SSD. And you say, I want an SSD because it’s faster. So you say, oh, hard disk gets displaced by SSD. But once you have that — once you have the SSD in your laptop and you create your work, your analysis, where do you save it? In the cloud. So the data storage is still — even from your laptop is still favoring hard disk. It’s just physically in a different place. So when we say NAND displaces hard disk, depends. Of course, in the tool, yes. But in where the data gets stored, no. It’s still getting store into an hard disk.
In general, in the cloud, you have both. You have a lot of hard disk where you do the storage and you have the NAND where that is used to do the analytics and the compute phase. There’s no use for storage. So they access the data, they do the analytics, they store on hard disk. They access the data, which is the architecture of the cloud. And I’d say if there is something positive through this down-cycle is that NAND is so cheap that if there was a reason to replace hard disk with NAND, you should have seen it in the last year, two years. But 90% of the data is still stored on hard disk, exactly as it was before the down-cycle. So it’s just the architecture but doesn’t see the benefit of having storage into a NAND.
Wamsi Mohan
I think we have one more question back there.
Unidentified Analyst
Very quickly on the capital structure. You guys yield around 5% dividend yield, which is pretty high for this sector. We got a lot of questions about capital allocation into — there’s two or three more quarters of uncertainty, macro uncertainty, how do you think about capital allocation?
Gianluca Romano
Well, I think Seagate has always been very focused on shareholder return. We have done this through share buyback and through dividends. At the beginning of the down-cycle, we decided to stop the share buyback. Through the down-cycle, you need to manage your balance sheet and in particular, the debt. But we generate positive free cash flow. We have generated free positive cash flow every quarter. So, so far, we don’t see any reason why we should suspend the dividends.
Wamsi Mohan
Any more questions in the audience? Maybe if I wrap it up, Gianluca. Can you just maybe also talk about your recently renegotiated covenants and you’re giving yourself like a lot of room. Frankly, it could be negative EBITDA in the quarter and you could still get to these covenants. So can you help us think through — obviously, you don’t want to keep going back to the bank to renegotiate these, but is this just a case of like a lot of conservatism, so that even if the macro deteriorates, you’ve got plenty of room, plenty of leeway, plenty of liquidity. So just — can you just help us think through there?
Gianluca Romano
Yes. I think there’s a point. I think was a good discussion, we grew support from our banks. Of course, the key is always to start to improve our EBITDA. And this is what we want to do. But we created a certain level of leeway. And I think that gives us a little bit of flexibility. At the same time, we have reduced or we are reducing a little bit of our debt. At the end of September, we were $6.3 billion. At the end of June, I think will be like $5.4 billion, and we can reduce it a little bit more if we have to. So I would say it’s a good collaboration. We try to reduce the debt. They give us opportunity and support on the covenants. And so I think it was a good result.
Wamsi Mohan
Okay. Great. unfortunately, we’re just about out of time. So thank you so much, Gianluca, for being here today. Thank you for the update, really appreciate it.
Gianluca Romano
Thank you, Wamsi.
Read the full article here