Western Digital Corporation (WDC) Management Presents at Stifel 2022 Cross Sector Insight Conference (Transcript) – Seeking Alpha

Western Digital Corporation (NASDAQ:WDC) Stifel 2022 Cross Sector Insight Conference Call June 7, 2022 11:30 AM ET
Company Participants
Gabriel Ho – Director of Investor Relations
Siva Sivaram – President of Technology and Strategy
Conference Call Participants
Patrick Ho – Stifel, Nicolaus & Company, Inc.
Patrick Ho
Thank you, everyone. Good morning. Welcome again, to the Stifel Cross Sector Insight Conference. I’m Patrick Ho, Senior Analyst here for Semiconductor Capital Equipment, Disk Drives and Lasers. I’m really happy to introduce our next company, Western Digital, one of the leaders in the data storage, disk drive and flash marketplaces. I’m going to first toss it over to Gabriel Ho, to give the Safe Harbor statement and then we’ll start with our fireside chat.
Gabriel Ho
Okay. Thanks, Patrick. And today, we will be making forward-looking statements and I ask you to refer to our SEC filings for the risks associated with these statements. We will also be making references to non-GAAP financials and reconciliation of a non-GAAP, and non-GAAP results can be found on the website. Thanks.
Question-and-Answer Session
Q – Patrick Ho
Great. Thank you, Gabriel. I’m really happy to have Dr. Siva Sivaram here with us. And this session will be a little bit different from – we don’t have to talk about gross margins or any of that kind of stuff. But I’m really happy to talk about the technology roadmaps that you guys recently outlined at your Analyst Day and dive a little deeper into that. But before we go into that, I wanted to – given your longstanding in the industry, if you could provide a big picture thoughts on the explosive growth that we’re seeing in data creation, data analytics and in turn the need for increasing data storage requirements. I think Dave Goeckeler, did a great job at his presentation of detailing the opportunities there. Siva, if you can give a little color on that, about how you see data storage growth, again, given how long you spend in the industry, particularly on the flash side of things. But again, if you could give us a little color on the growth of data?
Siva Sivaram
Thank you, Patrick. Appreciate the opportunity. Obviously, data storage is an area which is close to my heart and we can talk endlessly about it. We saw these big major transitions in our industry when we went from the PC age to the mobile age to then to the cloud. So in the last several years, you can see the large models, the AI/ML models that need to be built, the amount of data that is needed in creating these. So the explosive growth in incoming data has been massive.
We now are talking in terms of zettabyte of storage. So if you go into a normal big data center, about 75% of the space in any data center is storage, both hard drives and flash. What we are now seeing is as fast as that data center is growing, we are also starting to see because of 5G, a lot of edge and endpoint storage. Given the cost to the bandwidth to transfer data back and forth, you do end up having to do a lot of fast analytics at the edge. What that means is now I have a new source of where the data is stored. These inference engines require a lot of storage, so we are starting to see more and more and more as 5G becomes more prevalent, a large new opportunity opening up that is at the edge.
So the cloud is growing still at a place where only about 10% of the data that is produced is actually stored. There is still a need for it to – the ability to store all of this additional data from which you can monetize that information. And we are not there yet. Only about 10% or less is stored. So the cloud storage is growing and edge is also growing. There is a very famous speech by the Head of AWS, Peter DeSantis, in their latest conference, where he starts this whole presentation with saying when it comes to data storage, the king of data storage is the hard drive. And it is one of those things that need for data storage is growing so fast and hard drives are still the core where 80 plus percent of all data center storage is still in the hard drive.
Patrick Ho
Great. In fact, that’s a nice transition because I want to start with the disk drive. And I think in your presentation, you talked about the shift from PC client to the data center cloud. What I want to look from a business strategy perspective is how has this shift and focus changed WD’s focus in terms of the needs of the customers, right? So the PC client customer base is vastly different from the cloud. How has your business strategy changed to meet that, that evolution?
Siva Sivaram
Traditionally, the hard drive was seen as a shrinking business over time and it was being replaced by flash. That’s the thinking from the 2010, 2015 kind of – everybody’s mind, it is still that.
In the client space, that is completely true. It is a secular decline. Over time, the amount of hard drive both in terms of spindles, in terms of disks, it has come down substantively. What that has done is as we free up that capacity that was subsidizing the cloud growth. So as the cloud was growing, it was literally transferring capacity from the client devices, meaning the PC devices to pay for the expansion. So whenever we needed more growth in HDD for the capacity enterprise, you can go take capacity away from the client space. That’s no longer the case. That has – that transition has fully played out. It is no longer a business that’s going down over HDD. HDD is growing faster now in – both in terms of the number of spindles and in terms of number of heads and media. In all metrics, revenue TAM business is growing.
The big customers are no longer the channels. It used to be that we used to – we sell through multiple myriad channels everywhere to small mom-and-pop ordering 100, 200 drives to now the Cloud Titans. The Cloud Titans now by tens of millions of drives a quarter. So our sales motion, sales processes are all now geared more and more towards these big Cloud Titans, large qualifications. When we decide to go from the 14 terabyte to 16 to 18 to 20, 22, we end up qualifying. Once those qualifications are complete, big transitions to the next node happens. Pricing is more stable, longer term relationships. So you now how deal agreements on SLAs for longer. And we are starting to be predictable in our revenue and margins with them, whereas the consumer sentiment used to determine whatever next quarter pricing was. Now, it is truly value-based that the pricing is. So in general, the business has transitioned from being something that you sell into mom-and-pops to now being large, big customers with whom we have strategic longer term relationship, billion dollar customers that we are selling in.
Patrick Ho
Great. And maybe as a follow-up to that answer, something that hasn’t changed, whether it’s PC client or cloud is a total cost of ownership. Now it’s a different dynamic for the Cloud Titans. How do you convince them that they’re still getting value in that total cost of ownership with disk drives?
Siva Sivaram
Mechanical hard drives are an interesting concept. People have written off them forever. I mean, you can see every generation they’ll come and say some other solid state is going to take over. What we have done is predictably give our customers a reduction in total cost of ownership and that predictability of high-single digits cost reduction year-over-year. We are trying to change the dynamic from there to now being more value-based because we are starting to invest capital, but in general, people do know that the growth in demand is there and there is a technology engine that is also giving better capacity.
I mean, you can see, we just announced last week a 26 terabyte drive, a one-inch form factor, 26 terabytes in SMR drive, a 26 terabyte, nobody expected. Nobody expected for us to come in with a 10-disc 26 terabyte drive so quickly people thought we’ll run out of space. We are going to need big new transitions, HAMR has to come in place, et cetera. No, here we are using conventional technologies, pushing it even further. We are delivering [indiscernible]. So our customers now see that finite roadmap past 30, 32 terabytes. When we go into larger form factors, 50, 60 terabytes that we can deliver over a long term, they see the value of the predictability of the longer term roadmap.
Patrick Ho
Great. In fact that leads great to my next question about WD’s technology and how you differentiate from your peers. One of the things in your presentation that caught my attention was leadership in areal density. And then you talked about the different technologies of PMR, OptiNAND and UltraSMR. How do these various technologies provide WD with the areal density leadership you talked about and how do they provide WD with a competitive advantage over your peers?
Siva Sivaram
Here our technology development and product development strategy has been distinctly different from our competition. In our case, what we tend to do is we look at the hard drive holistically. The hard drive has multiple large components that go into it, the mechanical part of it, the media, the head, the recording technology, the firmware, and the overall system architecture. We look at this in a holistic sense and say, I don’t need to hit a home run on any one of them. I need to make sure this engine keeps chugging forward with predictable capabilities that I can draw everywhere. For example, I’ll use the example of the mechanical part of it. The mechanical part of it is that you are now trying to precisely position on the disk at a specific place. All right, if you need to do it, here’s our own body we do it, shoulder, elbow, wrist, get there.
When we need to large inertia big movement on the shoulder when it comes to precise, you use the wrist, triple stage actuators. These are the kind of innovations that give you precise positioning, Shingled Magnetic Recording, SMR. The read track and the right track are different. The right track is wider. Read track is narrow. Why can’t I just overlap them so that I can read closer? To do that, I need that mechanical device to read, so they both go together. So we introduce SMR. SMR is a Shingled Magnetic Recording.
Then you go back and say, hey, I have another powerful weapon called flash. Currently use DRAM small amount of DRAM to hold the metadata inside. What if, I knew all of mechanical vibrations, how much readout there is, to keep that information in flash so I can improve the precision of [indiscernible]. The control plane is now flash dominated. You add a little bit of flash not to put on the data path on controlling the drive itself, the metadata. So that’s OptiNAND, addition of NAND to make hard drive better.
You can see the recording technology with energy enhanced is getting better, using energy enhanced without trying to go to HAMR, et cetera, we have been able to get better density, but when you augment it with SMR mechanical triple stage actuator or OptiNAND now I can produce a drive that is 26 terabyte when the industry is not even – rest of them are still talking 20 terabyte. We come in with the 26 terabyte we have to sampling right now. That’s the – the beauty of the strategy of development is to do these, all these spots in parallel. You can put them together as a product when we need to.
Patrick Ho
Great. I know your focus on the Analyst Day was on the technology end and the differentiators, but I think something that I believe is important for WD is on the operational side of things. And I believe at one of the presentations, you highlighted how platform commonality, the supply chain and automation should help WD in terms of better financials in that business segment. From your perspective on the technology and can you give a little bit of color on how WD can improve operationally? Because admittedly like 16 terabytes, you guys relate to the market and at that cost us, lost share, how do you improve on the operational and given that you have the technology. So I always tell people it’s great that you have the technology, but if you can’t execute on it, it doesn’t do anyone any good, so…
Siva Sivaram
Absolutely valid point, Pat. So we did stumble on the 16 terabyte transition because we went a little aggressive on using 8 disk as opposed to go into the 9 disk we could have done easily. But what we have done now is this; Western Digital came in as a series of acquisitions, HGST, WD, Allconnect, each one had their own platforms, media, bill of materials. We have stated that for a while. We have converted them, especially on the client side, dramatically reduced the number of platforms.
Now we are one company, together we can go back and remove all these, each segment having multiple platforms, we have gone away from it, we have reduced it. Even on capacity enterprise, the large hard drives, we have now reduced it in such a way that there is – the same wafer starts that produces the heads can now be used in multiple generations of nodes. So we have streamlined it in one fashion. This is in platforms and in heads, making sure they’re coming together. We are consolidating, so that we can have ability to buy products at a higher volume, et cetera, to make the efficiencies better.
The second is the factories used to be also have different cultures. There was a WD factory, there was a HGST factory. Now over time, we have gotten them to be more like our flash factories in the sense, highly automated, very high focus on OE, operational efficiencies, making sure that we are using modern techniques that are going into these factories, inventory management, things like that, that have become a lot better given the new management focus on operations. So put them all together, consolidate platforms, ability for larger purchasing together because they are streamlined, better factories with one set of culture, fully integrated. We’ve gotten so much better at this than we were a couple of years ago.
Patrick Ho
Great. Let’s go to the flash market, which you spent many, many years going back to your days at SanDisk. Again, one of the things I was impressed with at the Analyst Day was at the highest level, you talked about the transition from a mobile to an SSD market. In fact, I think I’ve asked a few times Dave, on the call about the SSD market. He said, it’s coming, it’s coming. And then I saw on your Analyst Day, how you guys really went to a lot more detail. Given that you’re trying to capitalize on the shift, the customer focus shift to hyperscalers. Can you from a technology perspective, discuss how your differentiated technology, whether it’s the charge trap cell approach, and stuff from a technology and how that’s going to be a key differentiator especially as the marketplace evolves into an SSD market.
Siva Sivaram
Patrick, this has been a very important transition for us. An SSD consists of the flash itself, a platform on which that flash is converted to a product with a hard – with a controller and with the firmware, with an architecture that goes with it. It allows us the ability to differentiate. First and foremost, for example, we introduced the DRAM disk drives. For the value segment, we took over that marketplace because we came up with an architecture that does not use DRAM, allowed us to reduce cost, increase the capacity that we could sell. We improved our market share from high-single digits to 25 plus percent on the back of that architecture.
Second, WD Black, high performance SSDs for gamers and high-end performance. It’s still the best performing SSD in the market with multiple generations of it. WD Black as a brand has grown well. WD Black is doing very, very well with respect to revenue and margin, where we could come back and show our ability in vertical integration. The third biggest marketplace is as client HDD comes down, that market is replaced by a large external SSD marketplace. The 1, 2, 4 terabyte, external SSD, that market is ours. We now have the majority market share. We are over 40% market share in that business. Through all channels, through the retail, consumer channels, through our value added reseller channels and through our OEMs, we are starting to sell external SSDs.
This is a growing segment for us. You can see the combination on top of it, the enterprise SSDs, which is the high end of it, where we have now done two generations of NVMe SSDs. We are now being able to come back and take all the bits from the low-end bits to the high-end bits, we now have places in the SSD marketplace to sell, which are controlled by us. This is where this becomes very useful for us to develop as SSD player.
Patrick Ho
Great. Sticking on the technology side and it’s something you talked a little bit about at the Analyst Day, but webinar you hosted a few months ago, I thought it was extremely informative was how lateral scaling helps offsets some of the vertical scaling, a lot of your peers do to get more capacity per bit. If you could detail to the audience, why that lateral scaling is so important and a key differentiator for WD over the long run in terms of a technology basis?
Siva Sivaram
So if you are like me looking at scaling for the long-term, scaling benefits, when there’s a non-linear advantage, when you scale. Vertical scaling, if you add more layers, you get more bits, very linearly. If you add more layers, it costs more. To get more bits, but you’re also spending a lot more in capital. When we strategically decided that our objective is to produce the same number of bits with the least amount of capital, we had to look at this problem more carefully.
How do I get the most number of bits for the same volume, which you do by lateral scaling on each layer? How much densely can I packet? And then reduce the number of vertical layers to get the same layer number of growth. In the marketplace, I get beaten up for it. Everyone says somebody has 176 layer, somebody has 128 layer. You don’t have it. I keep telling them no. When they come and say more layers, it’s actually bad. You are being measured on the wrong thing. You want to produce the same growth with the least number of layers. It’s not that we can’t make more layers. You don’t want to make more layers
Patrick Ho
Costs more?
Siva Sivaram
It costs more. What it also allows us to do is reuse equipment, take the last generation tools and keep using it. I don’t want to be spending more on capital to produce the same number of bits. So we look at it as X, Y, Z, and logical scaling, go from two bits to three bits to four bits at the right place. Together, you get the most number of bits, together you get the maximum capital efficiency. You can go back and look at the data.
Take the last three years, look at how much bid growth every one of our competitors have done and look at how much they have spent on capital. You’ll see how dramatically better we are in making that happen. That data is there for, and I don’t have to look at it, you can get industry analyst data to go take a look at it. You’ll see this conscious strategy of reducing the number of layers to get the maximum amount of bit growth still gives me that capital efficiency, capital efficiency ultimately translates into many, many, many good things, not just capital, I mean bit cost, but you are also free cash flow and making sure that I’m not committing a lot more for my good money. And you can see the benefits of reducing CapEx to produce the same growth.
Patrick Ho
Actually, that leads great into my next question. One of the takeaways that was, you guys have the most attractive capital efficiency on the NAND flash side of things. Though, one of the arguments I get from your competitors on the NAND flash side is your joint venture with Kioxia. There are your competitors who say, hey, look, you got to work with someone else, potential technology, bumps that they may have to deal with operationally. I’m just giving you that type of feedback. Now, since Dave has joined the company, he’s been a strong proponent of that joint venture partnership. And you guys now have an agreement now for many years of going forward. Why do you believe the Kioxia partnership? Is it advantage for Western Digital versus say your peers say, I’d rather just do it myself?
Siva Sivaram
Yes. Clearly, look, if I was unconstrained, I can just go do everything. It’s all true. I’m pretty sure half the people here are married. We are a married couple, and there’s a reason why people get married. We do things in a complimentary fashion that each of us does differently, much better. We’ve been married now for almost 30 years, 1999 was when the first joint. So it’s now 2023 – so 23 years we’ve been married and we are committed to live together till at least 2034. And there are many advanced reasons why we do this over and over again. It is not that we are forced to do it. We do it by choice. We do it by choice because of two reasons.
First, we get scale. Together, we are the 33% producers of flash. So for the same amount of capital that I normally get, I get the benefit of 2.5x my capital spending. Second R&D is completely 50-50 share. Per bit, I get to spend the most R&D than anybody else because I get not just my spending, somebody else is spending on it. I am extremely efficient in R&D because of that. Between those two, in terms of scale, and in terms of R&D efficiency, we get to be the best in the business over a long time as we have shown.
I think we have done 15 generations of technologies together. It’s not like we just did started on it today. We continue to be together because we can quantifiably see our cost, our cost decline, our technology transition. Now you can always say, yes, two people have to make decisions together. It takes a longer time, et cetera. Yes. You balance that against the perspective that you get from multiple different customers to see what you do. We end up making better decisions because of it. Its tough living with somebody else, but in the end there are benefits.
Patrick Ho
Fair enough.
Siva Sivaram
And by the way, just to finish that part. To Dave Goeckeler’s credit, he recognized this immediately. He has personally invested in building that relationship further. He spends the time with Kioxia CEO, Hayasaka. So together I have been with this relationship for the last 10 years, and we have worked closely together. That relationship works only because of the personal commitment that both sides put into it. And he has done very well.
Patrick Ho
Yes. I will give Dave credit on that because I think one quarter into his tenure, he came out and said, this partnership is – one of the earnings calls, this partnership is critical to WD’s future product roadmap. So I do think that he made that, he put that flag in the ground very early on.
Siva Sivaram
And he has stayed with it.
Patrick Ho
Yes. The other aspect from your Analyst Day on the flash side of things that caught my attention as a positive was your BiCS platform, it’s been successful to date, it now continues to expand into new markets. And now you’re on the BiCS5 product in high volume. One, given everything you’ve said up to now, is that why the BiCS platform has been successful and you gave a little bit of a preview of the BiCS+ product in the future. If you could just give a little more color of some of the added features BiCS+ will provide customers?
Siva Sivaram
Yes. The BiCS platform, we started with BiCS2, which was 48 layers, and we did 64, we have done 96, we’ve done 112. We are doing 162 in high volume BiCS6, 162 goes into high volume later this year, I think early part of calendar 2023, we get into the bit crossover for that and high volume ramp happens. So this progression, we have worked very well together and the overriding principles are the same, reduce the CapEx as you go.
Now, what we also do is, we look at not just the dye, but we also look at the I/O. You can have a big pot with a little straw, that doesn’t help. You need to make sure that you have the – you can put data in and out very fast. We are the market leaders in it, add input, output operations, I/O speeds to go along with it through BiCS5 and BiCS6. We talked about the strength of the charge trap cell. What that charge trap cell us our performance, being so much better than, a good 50% better than our competition.
What it allows me to do is trade that for multiple other opportunities. I can use that for endurance, I can use that for performance, for retention, I can use that for density, many places we trade up, we can do four bits per cell out of that. What that allows me to do is in the end, have a roadmap that is predictable for another many generations. So we started with the 2-tier architecture, then we have gone to CuA in the BiCS6 generation. We have other innovations in BiCS+ coming up. You’ll see that this roadmap is strong for the next several years.
So we showed a bonding in the future, a double bonding in the future. These kind of things you add triple stacking, and then further reduction in the lateral scaling, put all this together, we have probably a roadmap that goes up to 500 layers. We are now 162 that is ramping in volume. We are now saying we have a line of site to 500 layers, that allows us to think and have a plan ready for our customers. So they know that what product at what cost deduction, is 15% a year cost reduction that we’ve been promising. We’ve been delivering it year-over-year.
Patrick Ho
Siva, I’m sure, I could ask you a lot more questions, but we are out of time. But many, many thanks for this informative session and keep plugging away and good luck going forward.
Siva Sivaram
Thank you, Patrick. Thanks again.
Patrick Ho
Thank you.

source

Share this post:

Leave a Reply