Western Digital Corporation (NASDAQ:WDC ) Deutsche Bank 2022 Technology Conference August 31, 2022 11:15 AM ET

Siva Sivaram - President & Head, Technology & Strategy

Sidney Ho - Deutsche Bank

Okay. Good morning everyone. I'm Sidney Ho. I cover semiconductors semi cap equipment and IT hardware at Deutsche Bank. The next company we have is Western Digital. Western Digital is a supply of both hard disk drive and flash memory. And today we are very pleased to have Dr. Siva Sivaram here, he’s President and Head of Technology and Strategy at WD. So welcome Siva.

Thank you Sidney. Happy to be here.

Great. Well, we'll make this session as interactive as possible for those who are in attendance here. If you have any questions feel free to raise your hand and we'll run the mic to you. Before we start, I have the honor to read the safe harbor statement. So I’ll do it. So Weston Digital will be making forward-looking statements and I ask you to refer to SEC filings for the risks associated with statement. Western Digital will also be making references to non-GAAP financials and a reconciliation of GAAP and non-GAAP results can be found on our website.

So with that out of the way why don't I kick it off with some near-term question. I think inventory adjustment is one thing that everybody is trying to figure out, most of us understand the ongoing -- spending. So there seems to be more signs of inventory adjustments broadening price market and [indiscernible]. Can you give us an update on where you are seeing some of the inventory [ph] adjustments, how secure they are when compared to past cycles? And more importantly how long do you…?

Thank you Sydney. By the way you did a great job of reading safe harbor. I think we should hire you permanently for that. We announced earnings -- I'm not here to either update or reiterate guidance -- talking about trends going forward.

You're absolutely right. Things started deteriorating towards the very end of calendar and that trend has been on. As you said, the China mobile slowdown was clear even at that time and that trend has persisted even more importantly in client PCs, client SSDs as inventory correction. This truly a demand driven shortage at that time, the inventory correction was happening.

Cloud is generally okay. Cloud is not any one company. There are -- of the titans they are never in sync when one is accelerating, one is digesting, one is digesting, the next one is accelerating. They build out at different times. So they are generally okay and that trend has persisted.

Okay. Maybe I'll just jump ahead. I'm sure we have a few more questions in point. But if I look at this cycle why do you think is better for whether this cycle than it was the last down cycle?

Yeah. As you know the memory business is a cyclical business. No one can exactly predict when cycle is going to go low or high. But you know it's always going to be happening. We went into this fully planning for this. As you know we have had a complete management cycle, very, very strong division of focus and responsibility on the hard drive side of the flash. We have had a very clear focus on platforms and product lines ready to go taken on down cycle [ph], of course as a corporation, we have paid down debt -- over $1.5 billion in the last fiscal year, the time we stopped paying dividend over $2.5 billion of debt taken down in the last approximately three to four months. So we have been very, very focused on improving our performance through cycle. Cycles are going to happen. But how do we as a through cycle do it well. And that's where this focus on technology leadership, product platforms, focus on execution, making sure corporate debt is in a better place at our balance sheet front. Put them together where we are and we feel very, very confident that the company.

Great. So you mentioned something earlier on the sale side. But really I just want to drill a little more. That's the positive side of things in general that you see consistent demand from data center customers. What are some of the trends you're seeing in cloud. Especially, love to hear your thoughts between the US cloud versus China cloud. As you look in the second half of 2022 or even 2023, how confident are you that we are not going to see a broad-based inventory digestion in that particular segment?

So cloud in particular, you need to keep this in mind, it is a secular long-term growth business. Without question, the cloud is growing. Whatever the short-term phenomena, if there is pandemic coming out of it cloud grows, if there is new technologies Tiktok and volume goes out, cloud goes up. In general, given the data growth, cloud is in general going to go up. Inside that, clearly the titans are large. They are very large companies. And though they have an ability to affect demand supply in any one segment substantive. When one is expanding rather will not be in sync. They may be in qualification. The third one may be in digestion of what they develop. So this is why we say in general we are doing it.

Now in China, particularly, both the COVID-related shutdown and the geopolitical issues have had an impact. We talked about the China slowdown in our December quarter. That trend has persisted. And over time, you know they also have to bounce back. Now for us particularly, our excitement in the cloud business comes from both the HDD and the Flash side because in HDD, we have just introduced 22 and 26 terabyte. We are sort of a step ahead of everybody else. 22 is led being qualed and 26 is led sample and 22 you can go into a consumer channel today and buy a 22 terabyte drive. Nobody is even coming close to that.

On the Flash side, our NVMe drives are already qualed, our market share is going up, the big five the next-generation based NVMes are getting qualed. We feel a lot more stronger about our position in the cloud than we were in the past. So you put the two trends together, we are doing well, it's a secular long-term trend, not all cloud titans work at the same time, average it out I'm going to say generally okay.

Okay. No, no, that's helpful. The other topic that I guess I get a lot of questions on is on inventory. So with regards to the inventory on your own balance sheet, it's now sitting at about 105 days, which is not really that far off from the last piece over 110 days in 2019. How are you thinking about your own inventory days in few quarters given tend to build more inventory? And do you see the risk that at some point you build too much inventory you have to write something?

So you want to separate Flash and HDD in this. And the question we're asking I guess it's more towards Flash than HDD inventory we know in places, where as our supply constraints have eased up we are able to truly reflect and meet the end-user demand which was not the case a couple of quarters ago when we had IC shortages, shipping shortages those things are affecting. Now we are truly able to meet our customers in demand in a reasonable fashion.

In the Flash clearly customers are correcting their inventory position. We have manufacturing that we continue to do we work very, very closely with our partner to make sure we are regulating supply both from the fab and into the end market. Given that end demand has gone down and the supply is reasonably -- it's coming down. It's come down substantively from the expectation of mid-30s bit growth to probably mid-20s to high 20s bit growth. Clearly the supply has come down. But you know there is going to be some inventory buildup.

Now let me be clear. Our customers are also listening to -- the last thing I want to tell you is hey I want to buy inventory you can come in. So, we are not holding inventory. We are not trying to hold inventory in this. We will make sure that we balance this.

That's great. Maybe let's switch over to a little longer term the different business starting with hard drive. So, you had mentioned earlier that, that is that business probably more stable than normalized state than you started ramping production of 22 and 26 terabytes which to me, it puts you slightly ahead of the competition. How quickly do you expect those new drives to ramp up in volume? What does that mean for your margins of the business? And which is still below your long-term target 31% to 34%?

Clearly this is the question I was hoping you would ask, right, this is what excites me to talk about technology. We have had a very clearly articulated strategy as to how we want to do this. We are not trying to get home lens. We want to make sure we want to improve it as a whole system. You want to look at heads, media, firmware, system software, mechanicals, drive integration each one of them in parallel so that I can have a predictable road map.

The result of that is that, 20 is ramping, 22 is already shipping and is now in qualification. 26 is sampling and getting into customer sense we got where we want to be. This is the kind of cadence you want to have. In parallel, big enabling technologies OptiNAND, triple-stage actuator, UltraSMR, these ePMR these are seminal technologies that are now becoming much more mainstream.

UltraSMR in particular, our ability to get 20% more bits because of the system-wide integration out of the same device, that 20% is a big deal. It's like an x-node we are able to get done. We are able to do 22 and 26 at the same time, that gives us that leadership and predictability that I think will stay on. The roadmap is very, very strong. You can see our road map all the way to with conventional ePMR, UltraSMR and OptiNAND technologies going up to 30 terabytes. That gives us that confidence.

Got it. Well, maybe just stay on the roadmap side of things. With regards to UltraSMR, you talked about 20% increase. Previously, it's only suitable for certain use, right? And the adoption wasn't that great. But you see much more confident about the adoption this time. Is it just the 20% more? Is the possible ownership is much better, or is there other reasons why more broadly adopted?

So, I want to start with one piece of data that sort of sits this up. When we exit this year, this calendar year, over 20% of our capacity enterprise cloud shipments will be in SMR both in revenue and in bits. So that's about as strong a statement as we can make on how fast this is ramping. Cloud titans are qualifying.

You're absolutely right. When you come back and say, I have a 14-terabyte drive and with SMR I can get another 1.4 terabytes, the whole software needs to be changed, we are not that particular. But when you come back and say, I do 20% on a 22-terabyte, I can give you 26 terabytes there. Wow, that is like two nodes that I can avoid with that one call and I'm able to get that now. That requires obviously the data manipulation coming in. They are willing to do it.

We, by the way, do a lot of enabling work to make that happen, whether it is database support, utilities, application nodes. These things we actually develop and we -- by the way, for SMR, we do it for the entire industry. We do it in open source and enable our customers. That has allowed our customers and how to take that seriously. Multiple cloud titans have now qualified SMR we are on to the races. And that's why we are so confident on this technology.

That's wonderful. Just on the roadmap. You talked about earlier, you have the 22, you have the 26. How many more generations do you think CML has before you have to switch to the next technology, which I assume is EMR.

Yes. Just to quote Mark Twain that of, EMR has been exaggerated for a long time now. We've been talking about this. This is the nicest thing about our roadmap is. You want to make sure you have a good blend of breakthrough technologies and evolutionary technologies. And we have done. So, we see our path to 30-terabyte CMR with EPMR generation 1, EPMR generation 2, EPMR generations 3 and then SMR and OptiNAND on top of it.

That is -- now that does not mean we don't believe in HAMR. HAMR will come in its time. It is again a revolutionary technology. We will be there when we do it. However, we are not forcing it. You want the lowest TCO to win. And we have these two coming in parallel. And right now, the incumbent technologies and their roadmaps is very strong.

Our focus, as you would see, even during these tough times, even as we come out of it, and because the cloud is in generally good state, is to figure out as you said earlier, how we get our margins up, how do we get our share of the value into our company. These technologies provide the platform where we can get that.

Got it. So, on HAMR technology, again, I think at the Analyst Day you talked about launching the technology around 2026, just based on the roadmap. So, we shouldn't be thinking of that as a hard cutoff from CMR to HAMR, is that right?

So even now if you notice, 18 is running in volume, 20 is getting call, 22 is getting done, 26 is going. Multiple nodes are running in parallel. We are not saying everyone goes immediately gets up. Same thing is going to happen with HAMR. HAMR will be introduced, allowed its time to cook a technology, you remember when helium came, how long helium took before it became the main stream. We introduced at 8 terabyte. Really 14 was the big node where helium took over the world. We need to give these technologies time to cook in parallel you want to have a strong road map that back to them.

So if you take a step back to think about CMR versus HAMR, what are some of the trade-offs that customers should be thinking and is there a long qualification time, testing time all the stuff on people in some in the camera?

You would expect this, right? First and foremost, these are extraordinarily demanding workloads. This is not the whole client hard drive, where about 10% gets written and most of the time it sits still. These drives for next five years are going to be hit every minute of every day continuously, which means reliability is very important.

For us to have the confidence in the reliability where we build enough drives to prove that the customer developing the confidence and there are going to be quirks with any technology. The fact that the customer needs to get comfortable with that. That qualification ensuring that when you put this in a 55-60-degree centigrade ambient and you run it for five years, will I be able to guarantee its reliability as well as they do today with CMR. That's going to be the long haul in the tent.

Okay. Maybe one last question on hard drive. So there has been a lot of emphasis on long-term agreements. I think that's mostly a hard drive comment rather than flash comment. Where do you see the value of LTAs in periods of inventory adjustments? It seems like these are not including pricing mostly volume driven?

So LTAs in general, we now have gone from about 5% LTAs and hard drives to now approaching 30-plus percent in LTAs. But what they give you is a better picture of a longer-term trend what mix that we are going to be taking qualification slots, qualification times, volumes, that gives you a much better idea of when their expansion plans are, when their division plans are going to be. So we can plan better. That's the biggest advantage of LTAs in the hard drive.

Okay. That makes sense. Well maybe a policy if there are any questions in the audience, raise your hand we'll – I think we'll run the mic to you. Okay. I will keep going. So let's switch over to the flash business. A few weeks ago on your earnings call, you talked about just starting to have discussions with your JV partner cut back on capital spending. Can you give us an update on those conversations? And when should we expect to hear the outcome? When do you think those actions will help reduce the production growth?

Even though we say we're just – it's probably not a technically correct statement because we talk to our JV partner three times a day, right? It is not that we are in continuous conversation, is there an externally announceable conversation that we can have. That's probably the more justification. We have been – we monitor our overall demand together and what capacity we need to be building in an intimate fashion with our partners.

We've been together for 22 years. We have run this for a long, long time. We know each of our cadences. Now on a more normal basis, you do see this idea – that the overall industry bit growth rate has come down. It has come down from where in the last downturn happened in 2019, we went into the downturn shipping 40% 45% bit growth rate year-over-year. Clearly, we are not there. We have now dropped all the way from low 30s to now mid to high 20s is where the industry is headed. We know these trends well. So we adjust our capital accordingly.

Now we want to take extra capital actions. You know very well by the time, a tool gets ordered tool gets installed, material qualified, material goes through and consolidates a year. So we know how to manage this very well and that's how we've been thinking about, with respect to capital with our partners.

We'll be very conservative and we have always been. As you know, in general, I have been accused of this that "Hey, you guys are not spending enough capital". We are extraordinarily capital efficient with respect to our technology. That's a metric, we very carefully watched. You've seen me talk about it in various meetings. Our capital intensity, meaning amount of capital dollars needed to produce the next 1% of capacity increase, is the lowest in the industry and we will continue to focus on that.

So just to be clear, on that point, so even if we adjust CapEx right now, it is not really going to impact production for this year.

You do – we’re going to talk about it a year from now.

The best case probably in calendar -- late calendar Q1 calendar, Q2 of next year.

Okay. Perfect. So maybe just on the bit growth beyond September quarter. Obviously, September guidance implies a significant decline in volume and pricing. But beyond that, I won't ask you about the pricing. But do you think, you'll start seeing a rebound in bit shipment not production, but shipments in fiscal Q2 and the December quarter. And in terms of margin, if pricing continues to drop, are there factors that will help offset such that your margin won't drop to where it was in the past cycle.

So Sidney, you know typically flash. The second half of the year, is the second half of the calendar year, is the stronger growth time, both calendar Q3, calendar Q4 and then calendar Q1 gets to be the low, Q2 comes back. That's sort of the typical seasonality across the industry. We are going into a seasonally strong time, with this level of inventory correction. So we watch this carefully. The trends are still continuing. We still see the reduction right now.

But what happens during Christmas, what happens during Chinese New Year, are important. As those signals come to us, when we ship ahead for those events we will be watching and adjusting our inventory, incoming bit growth, outgoing bits, we will be carefully shaping it. I'm not at any position to come and tell you that there is a -- predict when this will be changing. You know the seasonality of what is expected.

Okay. That's, fair. So let's talk about Flash broadband. I think you're shipping BiCS5 right now. And BiCS6 is, I think it's -- so it's actually, you're shipping BiCS6. Can you remind us the timing of your ramp of BiCS6 and the crossover of BiCS5, talk about BiCS6 it's more capital intensive as well. How confident is that you are able to get your 15% year-over-year cost reduction for this note?

This is what the technology is to live for. There are lots of factors that help us. When you can come back and say, "Hey, Japanese yen has weakened." So you'll get the cost deduction. That's not what we are talking about. We are talking about truly technology-driven cost deduction. And we plan this over multiple vectors. You want to get on the same wafer more productivity, meaning more bids come out of the same wafer. That helps you the cost. You want to make sure that your capital intensity is low, so that that production of that incremental bit is cheaper than it was the prior time. BiCS5 was an extraordinarily very good capital intensity node, because it did not – it was – it was what you would call a tock node. BiCS4 was a tick node BiCS5 is a tock node. BiCS6 is a tick node meaning it is more capital intensive, but it gives you a lot more bits.

So then now you have the question, how much of my line will I convert to BiCS6, because if I had 1,000 lots coming, I can get the same number of bits now with fewer lots than I had before. We adjust that based on demand how fast bit growth in the industry. The pure technology is very capable of delivering that 15-plus percent cost reduction. But how much of my line will they convert to BiCS6 depends on what the market is telling me how much do they need?

We modulate that continuously. That's what decides our CapEx spending. So right now we are in that modulation phase. Normally, I'd come and say, hey, sometime in middle of calendar 2023 we'd switch over to BiCS6. I can't tell that now because I'm watching the market to make sure when I bring that up to the crossover.

That makes sense. So beyond BiCS6 you have announced – the BiCS plus and BiCS plus from my understanding is that 200-layer plus. But some of your competitors have already launched 200-plus layers. So I understand layer count is not everything that you told me in the past. But how confident are you that you can stay cost competitive with BiCS plus and beyond?

Yeah. So you're absolutely right. There are at least two of our competitors have announced that, they are going to be shipping 200-plus layers either late this year, or early next year. We don't know what percentage of their BiCS are going to be in that. But just like you said that is bad news not good news. When somebody goes to the next node a little too soon that means they are spending a lot more in capital and need to get to that for them to be competitive with what we are able to achieve with old nodes.

The biggest reason is scale and reuse of equipment. When you go to a new site and go to a new node, you have to spend a lot of capital to make that happen. Some competitors make that choice for the longer term. I'll spend right now so that over time, I can get there. But we have been in scale for a long time. Because we've been in scale we are in much more of a steady-state operation where we plan this reuse of equipment extremely carefully to minimize CapEx, and still get the cost that we need.

The technologies are already there. So we develop ahead the enabling pieces of technology whether the circuits and the array or two tier or three tier that's our bonding we develop these technologies and keep them ready. You watch the bit growth rate needed, get the 15% cost reduction at the least amount of capital. That's how we achieve it every time and we have done that over a long time.

Okay. Now that makes sense. That actually ties to my next question. So you mentioned that a little bit earlier as well about the capital intensity. But if I look at your cash CapEx is 8% to 10%, I would think for flash loan the capital intensity is quite a bit higher. Maybe you can talk about what the right numbers to think about going forward. But even if you consider that you're spending significantly less than your competitors on a capital intensity standpoint maybe just to answer your question, you're just more efficient. But can you walk us through how you do it versus how you see other competitors being higher?

Yes. So, overall gross CapEx for the company, which we come out and say is of the order of 20% to 25% for the company of revenue. Cash CapEx is different. Because of the JV structure, we do address capital in a couple of days what we paid the JV. Separately, what we spend directly in our cash as back-end CapEx and then the corporate CapEx is that go together. We have maintained that our flash between 19.5% and 20%, 25% gross CapEx as a percentage of revenue over the last several notes. Now, you never want to look at any one node, way itself you sort of on a rolling average, you'll see that that is maintained between. So you may get one note to be much more efficient. The next node will catch up. So if you average over a couple of nodes, you'll see that the gross CapEx for Flash is in that 20%, 25% of revenue.

Okay. That's helpful. I want to stop here again. Any questions from the audience that want to ask. Okay.

Thanks for taking the question. When do you expect to start seeing NAND really penetrate the nearline market?

So this question gets asked of me all the time. We have had a mental model of one replacing the other in the hard drive business for a long time. In the client space that is completely true. The client deceleration of HDD and moving into SSD is almost complete if not in the last legs of it. That's not the case in the enterprise. The enterprise runs very, very differently. Given the enterprise bit demand growth in the 35%, 40% consistently over a long time and the base numbers, meaning they want units of multi-10 terabytes. These two run in parallel for as long as I can see.

For the next 10 years, we don't see the two blending over one with the other. And the workloads are very different and each of those workloads have strong demand. So, fast data, if I'm trying to do a focus on AI ML big data B2B analysts, all going into SSD even now and that's growing very rapidly. But on storage intensive applications, video and transmission, those kinds of things, there is no substitute for HDD. Because the HDD strengths are in cost per bit and reliability and an ability to ship that over a long a continuous stream of data.

For example, if you are on YouTube, by the time the commercial place, the ad plays together, you've already gotten the data out there and you can stream it, whereas on the SSD side, you want instantaneous. If you get on a web page within three seconds, if you are not -- the webpage has not refreshed, you already moved on. I need to get that cash. That part of it is always going to be SSD. So, this idea that somehow these two are going to be a replacement for one another, we don't see it. We just -- everywhere we see our customers', qualification, demand road maps from the cloud titans. The two of them are there running in their parallel tracks. The same procurement guy buys it, the same engineering organization qualifies it, but there are two different product lines for the next 10 years.

Right. No, that's very helpful. I understand that the times like [indiscernible] storage and there are different workloads. But I also have heard that like NAND on a power efficiency and real estate efficiency is starting to just eat in on a TCO basis. Is that...

It's about 5x, 6x. So even at the best of best case when NAND is on a down cycle and the ASPs are down and we are still talking 6x 7x. We don't see it.

Any other questions? I'll stick into this one. I kind of know the answer. But if I kind of look at the strategic review with the exit of Elliott. I'm not going to ask you any of the details about it, but can you just remind us what the timing of that whole process is?

So we had the letter from Elliott now running about 10 weeks ago, just before our Investor Day. Since then, an executive committee of the Board led by Dave Goeckeler has been running a process examining all aspects of that letter. The Board is very firmly committed to maximizing shareholder value. And they are running a very thorough process. It would be premature for me to come back and give you either a time line or when they they'll do it thoroughly. And Elliott is under NDA and they are working constructively with us on this whole process. That‘s what all I can.

Yeah. That's what I expect. Thank you. Maybe just to wrap up what are some of the key messages you want investors to take away from today? What are the areas you think investors may have underappreciated with regards to WD?

Let me start from the hard drive side. In the hard drive side, the strength and leadership of our technology has been underappreciated. People talk about HAMR and all that, the ability to deliver predictably products into leadership. We have done that very well. We have this 22 and 26 as demonstration points and we'll continue to do that strong road map, I want to make sure you understand how strongly positioned we are with the cloud. The cloud in general is behaving well. There are multiple titans each in its own timeline and I think they are not all synced up.

On the flash side, our ability to get into the enterprise data center market has improved dramatically. We have qualified our NVMe Generation 1, Generation 2 is already in qual and we are clearly seeing our overall market share penetration with the titans growing. The flash road map with our stated goal of better capital intensity is very strong. It continues to be strong. The management team has very clearly streamlined the company to be an execution engine, to have well-defined business units, paying down debt, watching the balance sheet, making financially prudent decisions to make sure that we are able to take advantage of the technology leadership to improve our own gross margins and better financial performance. That I'm feeling very strongly about the management team that has been put together. That's the message that I like to give you.

End of Q&A

Great. Wonderful. I think we're just out of time. Thank you for spending the morning with us and everyone has a productive rest of the day.

Sidney, thank you. Thank you for taking the time to talk with me.