Silicon Motion Technology Corporation (SIMO) Earnings Call Transcript & Summary

Good afternoon, everybody. This is Nick Gaudois, Head of APAC Tech Research at UBS. Today, to close the sessions, we have Silicon Motion. And with me Chris Chaney, Director of IR and Strategy. Our apologies, we were supposed to have Riyadh Lai, CFO, as well, who was dialing in from Taipei. There have been some problems with connection. So I think, Chris, we're going to start the session together. And hopefully, Riyadh can join later.

Great. Thanks, Nicolas. I'm happy to be here.

Thank you, Chris. We are glad to have you. So let me start with the SSD controller market this year. You commented you expected to reach about 300 million units SSD controllers in 2020. How do you see both channels, both the PC OEM market and module houses helping you achieving this target this year?

Sure. Well, so when we look at the 2 different end market areas, we have fairly high penetration in the channel market today, fairly low penetration in the OEM market today. And I think that the biggest opportunity we have is growth within the OEM space, and that's because of our just relatively large penetration in the channel space. In the back half of the year, we expect to see both growing, in fact. In Q3, we saw client SSD revenue up about 20% sequentially from Q2 to Q3, and we expect to see further growth in Q4. And I think our best opportunity for growing in this market in the next couple of years is through the OEM space. We currently have a very good presence with one of the biggest companies out there, which is Intel. And they're our largest OEM customer. We also have others in our mix as well that we are starting to see some better penetration in. And speaking of that, when we look into next year, our new PCIe Gen 4 devices, when they begin shipping, we'll be shipping to 5 NAND customers. So these are mostly going through the OEM side. And we have a new one in the mix that is Korean. So that's very positive for us as well. So I think we're looking at a client SSD market that continues to grow next year, as the penetration rates go from around 75% or so ending this year to north of 80% to 85% next year.

Got you. Thank you for that. And, I guess, a few follow-ups on these markets. You just mentioned the new Korean customer basically ramping into next year. Could you give us some color on the linearity of its ramp. So how do you see this developing over the next 12 to 18 months?

Sure. Well, the PCIe Gen 4 devices will begin shipping in the middle of next year. And this is aligned with the Intel CPU platform called Tiger Lake. Tiger Lake is going to be introduced early next year. And we don't expect to see volume shipments from that platform until the back half of the year -- second half of the year. So I think that we'll start to see a relatively modest ramp in Q3 next year and accelerating in Q4 for the peak season, which is typically Q3 and Q4. And then the next year, so 2022, we will see the introduction of Alder Lake. And Alder Lake is expected to be the more influential platform for PCIe Gen 4 adoption. And so I think that we'll continue to see PCIe Gen 4 devices ramp through 2022, accelerating in the back half of the year and into 2023.

Great. Thank you. And again, specifically for this new Korean customer, how do you see that developing in terms of revenues over he next 12 to 18 months?

Yes. Well, this is a customer that we've had prior engagements with in the past. They're well known to us. We've been working with them to develop this PCIe Gen 4 device and they've been a significant customer for us in the past. In fact, the -- their M&A deal that they recently announced with Intel is something that was initially perceived by investors as potentially being negative for us because the general thought was that with Intel being owned by another entity that our business with Intel could potentially dwindle over time. But now I think it's more apparent that, at least with what the acquirer has said, that they don't plan to change the production technology in the Dalian fab. And with our significance with Intel, when it comes to the QLC ramp, I think it's unlikely that we'll see a lot of movement away from our controllers during the next couple of years, anyway. So they add on top of the fact that now we have this new Korean customer to sell into, which we were not delivering client controllers to in the past, this presents us with some upside. So in terms of the number or the unit volume shipments, it's still pretty early to determine what that's going to be because the shipments are still probably 9 months away from us. I think that as we get into -- closer to Q2 to Q3 of next year, we'll be able to outline a little bit more detail on that. But right now, it's still pretty early.

Understood. And as you started, you obviously, mentioned the Intel Hynix transaction, right, for NAND. So I guess, and I'm going to put words a little bit into management's mouth for Hynix. So it's not exactly what they said. But they said, as you suggested that they're going to continue, obviously, running the Dalian assets with Intel floating gate technology for late -- for longer, for another 2 to 3 generations after 144 layers, which is one they're introducing now, we don't know the advantages of floating gate for data retention versus charge trap. But ultimately, when we take control of this asset, whilst they haven't given yet a clear road map, and that's by March '25, so we see with some time. Ultimately, I think, Chris, our experience in the industry is that you're very unlikely to have 2 process technologies running side by side, right, for one device. So -- and our assumption will be that you move to charge trap, if anything, because of the scalability of charge trap and the fact that the rest of the industry has moved to charge trap already, equally you might want now. So when that happens, and this is really a long-term question for you, does this provide any risks to your position there? Or not really? Because, I guess, from an SSD solution perspective, I mean it's obviously quite clear that the Intel business itself is still a lot bigger than Hynix's, right, at this stage. But from insertion business, Intel is really the core. From a device business, longer term, Hynix is likely to be more of a core technology. So how do we think about this long term for Silicon Motion?

Yes. It's -- like you mentioned, it's still several years out. And multiple generations out as well. I think it's too early for us to comment on that. I think that depending on what the underlying NAND technology is, is less dependent on -- I think we're less dependent on that because, like, for instance, right now, we're doing -- we supply controllers for another customer's QLC technology, another U.S. customer's, QLC technology, and they're using charge trap and not floating gate. So I think that that is probably less relevant over the longer term. Our firmware can adapt and can be written to different underlying technology. So I don't think that's a major concern.

Great. And actually as you said, you went through that transition with that specific customer, right, which just went through the first generation of it. Great. So going back to the SSD control market, I guess, for the client SSD side, you -- one glitch or one issue you mentioned in the last call and prior, if I remember, is that obviously, you had an issue of availability of NAND dies, right, for the client SSD market. And that issue has eased very simply because the enterprise and the flash array market has softened, right, in the second half of the year. How do you see that into Q4 in terms of availability playing a better hand for you and beyond Q4?

Yes, you're right. In the first half of this year, the availability of NAND devices for the consumer and the client market was relatively limited. And that's one of the reasons that we saw sequential declines in Q1 and Q2 in our client SSD business. But late in Q2, that began to change as the hyperscalers started to move past that surge procurement period and more NAND was available for the client market. And then when that happened, for instance, in Q3, we saw a 20% client SSD growth in revenue. And as we look into Q4, it looks like there's still plenty of availability in -- for NAND in Q4. And into next year, I think judging by what everyone's saying about bit growth, something like 25% to 30% bit growth by most of the major players, combined with, I think, more normalized demand, we'll see declines in NAND in the probably mid-single digit range as we get into Q2 and Q3 and Q4, and that's more normal. I think we'll probably have relatively decent availability throughout the year for the client side. I think what we saw last year was a little bit of an aberration because of the surge procurement that had to happen during the work-from-home and educate-from-home period as that was ramping up. But next year, I think a lot of that equipment and storage capacity will be in place and growth from this point on will be more incremental, not so much a giant surge of that.

Fair enough. I'm glad you zoomed on that, Chris. That was actually my next question. If you -- one we're getting a lot, and I'm sure you're getting a lot as well is we had, obviously, this PC demand pull in with COVID-19, both enterprise and consumer and more so consumer. And with that, how do you assess the risk of client SSD demand being negatively impacted by this pulling fading down? That's the first question related to but the second question is, what we've seen in Q4 guidance from PC OEM and ODMs and the initial visibility, which we have, which is pretty good into Q1 is a mix shift to Chromebooks as well. Right, which is quite meaningful. And obviously that has implications on the NAND solutions you put in those portable PCs. So if you could maybe frame the 2 forces here to influencing the market into client SSD for you next year will be very useful.

Yes, yes. So this year, the PC market looks like it's going to be down 5% to 10% or so. And despite that, it looks like we're going to be growing at probably twice the rate of the market. So for instance, for the first 9 months of the year, our client SSD revenue has -- is up about 20%. The unit shipments for client SSDs for the whole market are up around 10%. So we're growing at about twice the rate of the market. But when we look at what's going to drive us next year? It's going to be, I think, the same factors that were driving growth for the past couple of years. It's not so much the PC market growth because the PC market, other than this year, has been relatively flat. Yet we've been able to grow our client SSD market year after year because of increasing penetration. And as we look into next year, whether or not the COVID surge subsides next year or there's another Windows 10 refresh to the aging laptops in the corporate side or whatever the cycles are next year. If the PC market grows by a few percent or declines by a few percent, it's probably going to be less of a factor to our growth than the actual penetration rate growth next year. So the way we look at it is, let's just assume kind of another flat year next year, with penetration rates going from around 75% last year to perhaps 85% this year. That still leaves room for our client SSD units to continue to grow next year at a probably high single digit, low double digit growth rate. So I think there's still a very healthy market.

Great. Moving to...

You wanted to mention -- sorry, I didn't answer the question about the -- about Chromebooks. And that I want to -- I wanted to address that because -- so yes, some of the growth this year is definitely due to a move to the Chromebooks, mostly for the education market. And most Chromebooks don't use a client SSD controller. SATA or PCIe, they use mostly an eMMC controller. And we do ship eMMC controllers to module makers, mostly in China that then ship out to those markets. But the ASPs for those devices are a small fraction of what we find for a client SSD controller. So it takes a lot of them to move the needle for us. So while I think there's been this -- there has definitely been a surge in the back half of the year in the Chromebook market, it benefits us, but not a whole lot. So it's a very small bump in revenues, I would say.

Right. Okay. Thanks a lot. So -- and I guess you're starting to move to eMMC a little bit. So for the eMMC+UFS controller market, I mean, you alluded to an inventory adjustment with a specific customer, basically, being ongoing up to Q3, calendar Q3. If I'm not mistaken by customer positively preannounced quoting mobile demand improvement, specifically as one of the drivers. So from there, at what pace should we see better demand coming your way?

Good question. Yes, we saw a pretty significant drop from that customer in Q3. And that led our eMMC+UFS revenue stream to be down about 50% sequentially from Q2 to Q3. Now what we've seen since then is that the shipments to that customer have begun to recover in Q4 sort of in line with what they recently told the market about their -- the revenue outlook for this quarter. So I don't think that we're going to see the revenue levels in Q4 that we saw in Q2, when it really peaked, but I think we're on track to head back in that direction next year. So we have a very -- I'd say we're very enthusiastic about where this customer is going in the UFS market next year in smartphones and they've recently begun to ramp their new uMCP product, which has a -- their mobile DRAM, their LPDDR5 DRAM and then their mobile NAND and then our controller. So to the extent that they are successful as they were in the last generation, I think that we'll do very well, and they'll do very well in the market this coming year.

Great. And actually, 3 follow-ups to that I have. The first one you just mentioned is the adoption of LPDDR5 uMCP solutions just starting. I mean, it's interesting but actually DDR5 adoption is starting with low power, right, this time ahead of higher frequency, which is the opposite of what we've seen in DDR4, DDR3, et cetera. So you have this, you have uMCP. So what are the benefits of those solutions? And how do you see the adoption going into 2021?

Sure. Well, the LPDDR5 DRAM and mobile NAND and the controller together provide a low-power solution that has wide bandwidth and a small form factor. And bundling them together in a single package is -- it meets the bill of materials requirements for the mainstream phones, which is where adoption is headed next year. So in the past couple of years, we've seen UFS adopted primarily by the high end smartphones. And the top of the line, 15% of all the smartphones is kind of where we saw UFS over the past couple of years. But in the next couple of years, we're going to be moving into the more mainstream. The new apps processors by MediaTek and Qualcomm are specced out to use UFS. And they're kind of leading the market with lower-priced apps processors for the mainstream market. And as that happens, we're going to see more UFS devices pulled into the mainstream market as well. So that's why I think we're going to see a lot of the uMCP devices in the market next year and the following year. First, our key customer is ramping the product late this year into next year. And I think that the market is going to see better adoption of the uMCPs next year.

Great. And I think a question we had from the audience was related to Micron, and I'm just quoting in their last conference call, "Over the next several quarters, we'll be introducing a slate of new SSD and mobile NAND product that leverage increased vertical integration, using our own internally developed controllers and our industry-leading second-generation replacement gate TLC and QLC technology." So in that context, could you comment on whether you think Micron would be basically releasing more UFS products with internal controllers or whether we will continue to see you working closely with them.

Sure. Well, I can't speak too much about their strategy. That's really for them. But what I can tell you is this. They had been pretty public in talking about increasing the value of their NAND product portfolio, so they have more value-added products. And that entails adding other elements aside from just pure NAND to the mix to make a solution. Where we've seen them focus their controller efforts so far has been on the high-end controllers for client SSDs. I'm sorry, not -- well, some client, but mostly the hyperscale and enterprise market. So that's kind of where I think we're going to see them focus. On the UFS side, we are doing quite well with them today on UFS. We'll be -- the new uMCP that's shipping late this year and into next year is a 3.1, UFS 3.1. And we are working with that customer on not just the current generation, but the next couple of generations as well on the UFS side as well as on the NAND production technology as well. So our road map with this customer extends beyond just the current product set and what's going to be shipping in the next couple of years actually is going to be going further. And we are doing development work with them on the next several generations of their technology, both on UFS and the NAND. So whether we're talking about higher stacking, 170-plus layer and on up as well as beyond UFS 3.1. So I can't speak to what their internal strategy is, but I can tell you that we feel pretty good about our position in UFS with this customer.

Great. Thank you, Chris. I appreciate the color. Just on next-generation UFS, actually. Could you update us on your progress just for UFS 3? I think you're mentioning earlier that you're confident starting production mid-2021. So if you could give us a bit more color on that, it would be great.

Sure. Well, we're shipping UFS 3.1 devices, controllers this quarter, and they're going to be ramping into high volume next year. So 3.1 is going to be, I think, the variety that we see in uMCPs over the next 12 months. And beyond that, there are other forms of UFS that I think will just gain adoption over time. But I think the predominant UFS next year is going to be UFS 3.1 in the UMC form factor and we're still also shipping the prior generation, generation 2. So I think we'll see a mix of both of them, but 3.1 is coming soon.

Great. Shifting gear. We have sort of question from audience on this, but let me rephrase it a little bit more simply. For -- how are sales of Open Channel solutions tracking with Alibaba? And how do you see this developing into 2021?

Sure. So starting in the third quarter of last year, we began shipping Open Channel to Alibaba. Those shipments continued through the back half of last year and then into this year. And their seasonal pattern tends to be that they start ramping in Q1, accelerate in Q2, and Q3 tends to be the peak. And then Q4, the shipments of Open Channel drop and that's because everything that Alibaba is doing up to that point is focused on September 11, which is -- I'm sorry, November 11, which is their Singles' Day. So they have to have all their equipment and storage and all their technology in place and tested by that time. So that is what causes a seasonal pattern that we see. So when we look at next year, we're going to be shipping to an additional 2 programs next year. So up till now, starting from third quarter of last year up till today, we're shipping into one of their storage platforms, one of their storage programs. In the second quarter of this year, we won an additional 2 others. So we'll be shipping to 3 programs next year versus one this year. And we haven't given guidance yet as to what that means for revenues from Alibaba next year, but I think it's reasonable to expect that 3 programs versus 1 should mean there's some growth opportunity there. And it also means that our devices are being deployed more widely through their storage network, which is a positive as well. So I think we're encouraged with what we see at Alibaba, and we think they like our product and their -- our devices have been running in their networks now for 2 Singles' Day events and have done quite well through both of them. So I think we've kind of proven that in terms of the test case.

Great. Just staying a little bit with [indiscernible] markets. How do you see the competitive environment for Shannon SSD currently? And how is data center demand in China shaping up for you?

Sure. Well, so at Shannon, we ship 2 primary products. The first one is the one we've been talking about now, and that is the Open Channel device, and that's the newer product that has a higher price and higher margin. We also ship more legacy products. These are the standard NVMe devices that Shannon has shipped in the past. And those go to a broad set of -- several customers in China. The demand for standard NVMe drives in the back half of the year in China has softened. And we communicated that we had seen softness in our third quarter conference call, and I think that's going to continue into Q4 as well. We mentioned that as well. The focus for us next year is going to be more on Open Channel. And so that's where our design efforts are, that's where our efforts with the customers are as well. And we expect that to pick up next year versus what we saw this year just due to the multiple programs that we're shipping into. But beyond that, I think that we still have these legacy drives in our back pocket. We can ship them if the demand is there. But our focus is going to be primarily on our newer products because there's just better growth prospects, better opportunity for differentiation. And I think that it's likely that we'll see competition in that space. We were first to ship Open Channel in third quarter last year. I think that over time, Open Channel is going to adopt different kinds of names, different varieties of Open Channel and Alibaba and others will likely start using more differentiated storage solutions. They can enhance their ROI by offering better performance and things like that as experience for the customer through those devices. So I think we're going to see more specialized storage devices used by the hyperscalers over the coming years. And I'm sure there will be others that want to enter the market because that's an interesting market. The interesting thing about it also is that the big players that are very focused on shipping a lot of bits to the customers don't necessarily have a lot of time to design very customized devices. And this is where Shannon excels. They're a group, a design group that designs a custom storage solution for specific customers that have very special needs. And so that's what differentiates us.

Great. And with this Chris, one last question still on [indiscernible] market. YMTC, so we see them probably having 50,000 wafers per month capacity installed at the end of this year. It looks just like Wuhan is funded to go to 100,000 at the end of next year with 128-layer kicking in on top of 64. So how are you positioned with YMTC? And where do you see yourself, I guess, really being present and developing going forward from a product perspective?

Yes. So we've been -- we started shipping silicon to YMTC last year. They started off and continue to be a fairly small customer of ours. So we're shipping silicon to them now, controllers for their client SSDs and the devices as well. And we hope they succeed because if they succeed, then we have a growing and emerging new customer. And so far, their silicon is shown to be working well, and our controllers are working well with their silicon, so we're happy about that. And we hope they continue to grow. So we're happy to add them as a new customer.

Great. And I think we just literally reached the end of our time. I -- well, Chris, I just wanted to give you a special thanks for stepping up to do a presentation for us, and the discussion was great. And hopefully, we'll get Riyadh next time. And apologies again with difficulties of a change of platform basically from yesterday to today. Thanks, everybody, for staying with us today. This is the end of the second day of UBS TMT conference. And we'll be resuming tomorrow morning. Have a great evening, everybody.

Thank you, Nicolas. Appreciate it.