Ambarella, Inc. (AMBA) Earnings Call Transcript & Summary

Seymore, semiconductor analyst here in the U.S. for Deutsche Bank. Welcome to the Deutsche Bank 2021 Tech Conference, all virtual again this year, unfortunately. But we are happy to have a great lineup for you today in the semiconductor space. And we're kicking off today with Ambarella. We have both the CFO, Casey Eichler; as well as the Head of Corporate Development and Investor Relations, Louis Gerhardy. I'm going to pass it over to the fellows here first to do a quick overview on Ambarella, and then we'll jump right into questions. So Louis and Casey, good morning. Thanks for attending and let me hand it over to you.

Appreciate it, Ross.

Thanks, Ross. I'll just give a real quick high-level introduction, and we can jump right into Q&A. Ambarella, of course, is a semiconductor company. More than half of our engineers, by the way, are involved in software and algorithm development that often gets embedded in the firmware of our chips or software on top of them. We have a globally distributed workforce. It's proven to be resilient in the pandemic operations in Europe in both Munich and Italy, U.S. of course, and then in Asia is significant workforce for us. We were founded in 2004, and the premise is that since day 1, it's video is a unique type of data in the first 15 years of the company's existence, focused on collecting that data that's coming in through a lens of a camera in a CMOS image sensor and collect that data for human viewing purposes. And 5 to 6 years ago, company made a significant pivot into using that data that's collected to enable machines to see and make decisions, either partially or fully autonomously, not just for the automotive market for the broad IoT endpoint market. When you think about AI, Ambarella focus is exclusively on the AIoT endpoint devices. We don't sell into servers at the enterprise edge or in data center. We're just trying to do as much AI as possible in the endpoint devices. And that requires a fundamentally different architecture than what's used in servers today. Just fast forward in 5 years from when we made the initial investment in CV2, most recently reported quarter, we're now showing significant momentum on the revenue line with our computer vision products. We've said there'll be at least 25% of revenue this year, given that they carry 2x ASP that translates into -- it will be about 10% or 12% of our units this year. And most importantly, the operating leverage from this incremental TV revenue is now starting to drive our operating margins, which reached 16.9% in the last quarter, and consensus has them advancing around 20% in the next quarter. So I'll stop there, Ross, and pass it off to you.

Perfect. Thanks for that kick off, Louis. So I want to get into a lot of aspects of Ambarella, but the first one, I'm going to ask every company I meet with today and host about some of the more current events side of being a semiconductor company right now. And top of mind for most every customer at least is, supply. So talk a little bit about the supply dynamic from your side. You guys had a relatively unique exposure to one of your foundries having an issue in Texas with the freeze earlier this year. But it appears that, that is now back online. So just talk about how much of your guidance for your out quarter assumes that, that Samsung fab is back online and then a general commentary about supply tightness?

I'm sorry. Go ahead, Louis.

Yes. I characterize, to summarize at a high level, then we can go into more detail. 4 issues on the supply side to talk about and one of them is getting better. The other three, we see persisting into the next year. Issue 1, long lead times. Issue #2 is substrates, shortages there. There's some things we can do about that to try to make it better, but it's still something to talk about. Third concern is, our customers having difficult times, in some cases, completing their bill of material kits. What that means is, other components that are on the same bill as materials, we are, whether it's $0.50 microcontroller or a WiFi chip or PMIC, those are the types of devices that are difficult for our customers to get. So those 3 issues, we -- hasn't really changed much from the prior quarter, and we see it persisting into the next year in constraining results. The fourth issue that you referenced is that a large amount of our video processors were produced at Samsung's wafer fab in Austin, Texas. So we use Samsung fabs in Korea, we use Samsung fabs in Austin. Well, our CV products are built in Korea. The Austin facility builds a large amount of our video processors, in other words, used for human viewing applications. And when the Texas freeze hit in February, it disrupted operations so that in Q2, which was the July ending quarter for us, we experienced the worst of the supply disruption. And with that fab back in normal operations now for the next 2, maybe 3 quarters, we see we're in a period of wafer delivery catch-up from that facility. And so this is one of the 4 issues for us. It actually is getting better. The other 3 are persisting.

Got it. Thanks for color on that. When do you think those 3 will be fully resolved? I know that's the question nobody has the answer to. But you mentioned it, lasting into next year, when do you see an equilibrium being reattained?

Yes, we're not calling an end. So from a wafer supply point of view, which was the first point I brought up, most of our wafers come from Samsung. And we just have an excellent relationship there. And we feel that we're much better off with Samsung than we'd be in other foundries that are having more difficult supply situation. So it is still an issue for us. Our lead times are long. And so not going to provide an idea on when that might end. The second issue is substrates. We're well into the path of trying to mitigate that problem. And the way we can mitigate it is to find alternative substrate suppliers. And then once we get an alternative substrate, then we have to package it up and then give the new devices to our customers to qualify, requalify. And once they requalify it, then we can ship those parts. And so substrates have been an issue for us, and we are already pretty deep into the process of asking customers to requalify some parts that have these new substrates in them. And then in terms of the third issue, shortages of other companies components, definitely, I can't answer that one.

Yes, it's hard enough to worry about your own components, let alone everybody else's.

Yes.

So why don't we pivot off to supply side? One more -- a little bit of current events type question, but with a little more Ambarella specific spin. If you go back in time, an opportunity for you and a challenge has been the concentration of your customer base in China, specifically with the 2 largest IP security camera makers there, Hikvision and Dahua. Over the last year or so, that has come down substantially. I think it's only a high percentage -- high single-digit percentage of sales right now. And I believe you're also diversifying into other vendors in China. So talk in general about the China strategy and the concentration to those 2 customers, where do you think that's going to be over time? And do you think a more fragmented Ambarella can be more stable and less volatile than it's been in the past?

Yes. So just at a high level and answer to the China-specific question, our business is becoming much more diversified. Fermi mentioned on our last earnings call that he expects this year to represent a record revenue year for us, and that would compare with our fiscal '16 when we did $316 million of revenue. In that year, our top 5 customers were 60% of revenue. Right now, the first half of this year, our top 5 customers are 26% of revenue. So our overall business is really diversifying. But to specifically answer the Hik and Dahua question, if you go back a couple of years ago, they were 25 -- combined, they were 25% of revenue. And in this most recent quarter, you're right, we said they're both just under 10% combined. And a wide variety of factors have impacted those numbers. Geopolitical forces have been in play for a couple of years, that's had a role. Shortages are now having an impact. One of those companies is now aggressively adopting our CV. The other one hasn't yet because they still have a lot of inventory to other components. So as you can see, there's many vectors in answering that question. But if I back up to a high level again for the company overall, we estimate that perhaps 15% of our total revenue is consumed in Mainland China, although more than half of our revenue shifts to China could be assembled, most of it is still being exported out of China. Hopefully, that helps.

Definitely. All right. Let's get to some of the fun stuff on the technology side of things. You guys have said before that about 25% or over 25% of your revenues this year would be CV-enabled. And I think, Louis, you said, about 10% to 12% of your units. Talk a little bit about what makes a CV product, computer vision products unique at Ambarella. And then the markets that you're currently addressing with it. And then we'll get into how that opens up new markets after that?

Yes. So CV for us represents a new product cycle in existing markets. And then it also represents an opportunity to expand into many new markets. And the simplest way to think about it would be, we're going from serving the world population of humans that see things, that's the human viewing chips we provide. And now on top of that, we continue to support it. We're now addressing the number of machines in the world that can use a camera to perceive their environment and make some partially autonomous or someday fully autonomous decisions. So that, that latter number is new. It's large and important area to watch for us. But to answer the first part of your question, CV, let me just describe how we're differentiated and why we win, and I'll just do it really quick. And then you can drill down. We're differentiated in our ability to have very unique IoT endpoint CV devices by our know-how. And what that means is, 17 years of video processing expertise, we're able to get the highest quality image through the lens of the camera, no matter how difficult the conditions are. For example, is there too much light, like the sun is shining into the camera or is there darkness, is there movement, is there a lot of vibration, is there mud on the lens. These are all things that we can work around. And by the way, the road map to further improve camera technology is extremely robust. And I think you've heard, Tesla say the same things. We are very much in agreement with that. The second way we're differentiated is, we have our unfair share of engineers who are really passionate about treating a video as a unique type of data. And the third way we're differentiated was, we take a very unique approach to designing our chips. And the high level, that approach was to acquire a leading computer vision software company, take a couple of years to reverse engineer their software stack to understand how could we create from the ground up a computer vision chip optimized to process data through the lens of a camera as efficiently as possible. So again, differentiated by know-how, unfair share of engineers and then the approach that we took. And today, we're winning because of our efficiency, which is efficiency in terms of performance per watt, performance per dollar. Second reason is that we provide an open platform. So we don't bundle the software stack with our chip. We allow our customers to capture that software value and to create differentiated products. And then we're also winning from our flexibility that we enable. For example, customers today, we just announced in the last current earnings call, a company called KeepTruckin which is a Google Ventures backed company, which is doing in-cabin -- I'm sorry, it's doing a multifunction safety system for vehicles that does Front ADAS, it does driver monitoring and it does incident recording, all in 1 chip, all simultaneously. And that's the unique capability that we bring to the table. And again, it's allowing our customers to be differentiated and creative with what they bring to market. So we're winning from efficiency, open platform and flexibility. We're differentiated by our know-how, our engineers and our approach. Hopefully, I answered everything there.

Yes. In great detail. Thank you so much for that. So let's talk about the different waves. You guys have talked about, if I recall right, Wave 1 was IP security on the professional side. Two was more the consumer side, and then 3 would be the automotive side, the three waves of CV. Talk a little bit about where you are in that right now? If you have 25% of your revenues are over that, already there, am I correct to assume that, that means the better part of approaching half of your IP security camera business is already CV? And just how do we see that penetrating from Wave 1 to 2 to 3 over the next few years?

Yes. So we gave guidance on Wave 1, 2 and 3. Wave 1 would become material in calendar year '20 last year in the professional security market. What that means is enterprise CapEx and public infrastructure-driven. That wave is very much on track and remains strong. And I think estimates like you mentioned are not unreasonable as to the contribution and proportion of our revenue. Although I highlight that from a unit perspective, it's a much lower figure. And I'd also add that, if you look at CV penetration into the overall IoT camera market, it's extremely low, less than 5% of the installed base has converted to the CV products. So just measuring the road, how far this will go and how long it will be a driver by the percent of our revenue, maybe a bit misleading because the overall installed base penetration is low, and we think CV is going to continue to ramp for the next 4 to 5 years. Wave 2 is in the smart home market. We said that would become material in calendar year '21, this year. And that is indeed happening. We have several type profile programs in mass production now. And others, we expect to be in mass production later this year. And so that one is very much on track. And then Wave 3, auto. We've said that it would be material in calendar year '22 or calendar year '23. And that business is -- it's found a quite a bit of success with revenue more than doubling this year by just hitting a lot of singles and doubles. We did announce in November last year in the automotive revenue funnel. And at that time, we articulated we had several larger wins in that funnel that would not kick in for revenue until calendar year '22 or '23. That remains the case. So those larger programs are on schedule and haven't yet gone into production. But the reason our automotive business is good now is through design wins, like on the last call, we announced Yandex, point out Solera, the old Omnitracs. We announced the KeepTruckin I mentioned. We announced the DMS win with Dongfeng. And so the auto wave, our confidence in our guidance is quite strong in that as well.

So why don't we dig a little bit into the final parts of the IP security camera market and then we'll go into the automotive side. When you talk about the CV penetration, what percentage of the cameras, and let's talk, the professional side. Do you believe, we'll actually require or desire having the CV technology. Is it just the high end? So like you said, it's 10% to 15% of your units today. The majority of that is in professional cameras. It's barely penetrated from a unit perspective, but is 100% of that market truly the SAM?

Depends on what year. So it's just like any other new technology cycle. You may -- you're probably familiar with the installed base exhibit we have, which shows the first generation of IoT cameras, mostly surveillance with analog. And that business is decreasing as a proportion of the installed base, but it's still alive. The second wave where Ambarella began to serve the market was in, let's just call it, digital camera, some people called IP cameras. And that came into the market. And it just steadily ramped and became a larger and larger percent of the installed base, so that it dominates the installed base today, even though there's still some analog left. And now you've got this third technology wave coming in, which is computer vision, which is less than 5% of the installed base. And our belief is that, that computer vision really sliver today is almost like, where the digital IP cameras were in 2014 or '15. And that every year, as customers become more familiar with the technology, and drive cost out of their systems and come up with better software, our expectation is that over time, CV will dominate the installed base, whether it's in a home or it's in a commercial application. But it doesn't happen overnight, which I think was the basis for your question.

And then the final part on the IP security camera side of things is a little bit on Wave 2. So people think of the Chinese vendors of commercial professional cameras, but the other side of it, whether it's Ring or Blink or the home security systems, people are installing themselves these days. That's another big market for you guys, obviously, as described by Wave 2. Talk about the penetration into that market, the legs of it, the competition in it. Does that tend to be lower relatively lower performance cameras that people want there? And so is there more competition? Or is CV just as valuable in that market and Ambarella therefore, can exert its leadership there as well?

Yes. So the dynamics of this market are much different than the professional or the enterprise and public infrastructure cameras. And in fact, for CV, it's really a different set of products that are ramping there, products like CV25 or CV28. And to understand what's happened in that market, the smart home camera market really emerged from Europe and North America. And that's where a majority of our revenue is. More recently, this type of product is caught on in Asia. And the products -- many of those products at the early stage are lower end, and we don't serve much of that market. So if you think about where Ambarella exist, in general, we -- in this home security market, if you look at the unit, TAM or SAM is about 1/3 of the units. But from a revenue SAM, it's about 2/3 of the revenue opportunity. So in other words, this bottom 2/3 opportunity that's emerging very, very basic cameras that work at short distance and maybe they run 1 application challenged in difficult lighting conditions, so on and so forth. We don't serve that market. So the market's is really bifurcated in this area, and we continue to serve the mid- high to high-end of that market, which is where the revenue opportunity and the margins are.

So let's get on to the automotive side. Obviously, that's the big opportunity that people look at and are very excited about. You guys have been transparent about your pipeline in that business as well as the new design wins that you've rattled off, a bunch of them even last week when you reported results. So if we're thinking about it being roughly 20%, 25% of the business today, it's growing very rapidly, probably double this year. What makes up today's business? And then what would that profile look like a year from now and maybe 2 years from now? How do you see that transitioning?

Yes. So our history in this market is providing recorders and data loggers for human viewing purposes. And as we talked about before, we've made a huge investment into computer vision, which enables machines to be partially or fully autonomous. And so our story in auto is providing both human vision in both aftermarket and pre-install. And the new opportunity, greenfield for us is enabling machines to see, whether it's a Front ADAS system, whether it's a surround view with some AI applications or an e-mirror doing blind spot detection or a more highly integrated L2+ or L4 SoC. And so the business, as it stands today, as we speak, the human viewing part of it is still significant and by far the highest growth is coming from the data logger pre-install market. That's where data logger is installed in the car at the time it's produced. It's not an aftermarket product, whereas our outlook for that human viewing data logger function, that's not going to be a key growth driver for Ambarella in the next 5 years. It's a nice business. It's paid a lot of bills. We tend to serve the high end through companies in Korea like,[indiscernible]. You can go to their product line and see what they have. So our opportunity, whether it's human viewing or machine sensing in auto is going to be driven really by our chips being installed in cars as they're manufactured. And increasingly now, and this has changed very rapidly in the last year, our automotive business is being driven by the computer vision chips, which sell into front-facing ADAS markets. It's market mobilizing, it sells into driver monitoring. It sells into e-mirrors. We've announced some wins with L2+ and L4. So really the big delta versus a year ago is that the CV in auto is now ramping. The data loggers for pre-install is very strong, and penetration into new vehicles is low there. And then the aftermarket data logger business is, it's a nice business, but it's not going to be driving much of our incremental automotive growth going forward.

So right now, like I said, you're probably 20%, 25% automotive, 3 to 5 years down the road, do you expect this to be the biggest market end market for Ambarella?

Yes. I think Fermi would be disappointed if that doesn't play out that way.

Why don't we pivot a little bit related to this to the margin side of things. We want to have Casey to have some answers in some of these questions, too. You guys have upsided the gross margins pretty consistently over the last couple of quarters. You're above the high end of your range, very healthy range. I believe it's -- the target is 59% to 62%. Talk a little bit about what's driven the recent upside? And perhaps, more importantly, as you look forward, I know CV doesn't carry higher gross margins, but it carries higher gross profit dollars because the ASP difference. Talk about the puts and takes to structural gross margins as you have these end market mix shifts to automotive, as you have CV penetrate, et cetera. How do you think that range of 59% to 62% changes, if at all?

Yes. I think as you know, a few years ago, we were under that range for a period of time as we were starting to build out the CV side of the business. And as we've been able to do that, we've been able to build a broader structure towards the high end of the range as you talk about. We've even been a little bit above the high end of the range. I still think -- for me, and I believe that, that target is the right place to be for now. And as we continue to mature in some of the new markets in CV and actually some of the new markets across some of the other areas that we're just now starting to get into, we might have a chance to revisit that and look at that again. But for now, I think that, that is the right place to be as a margin profile. And I think as you've seen in our recent, we think we're moving to be towards the high end of the range over the next couple of quarters.

As your China business diversifies or said differently, as you went from 25% to 30% of sales with 2 large customers now down to high single-digit percentage of sales, does that diversification take some of the pressure off gross margin?

Yes. I mean, certainly, now on I quote, big purchase of the margin, Dahua is still growing in the CV area. Hik is not today. And so that has helped with the margin profile there. One of the things we've also profiled that we've just started now is, we've got 05 or plus markets in China that are also starting to start new products in CV. And as that starts to become bigger and bigger in our market there, I think that will help with the profile. But there's no question over the course of time. Traditionally, those 2 markets had put some pressure on that profile.

And last gross margin question for you, Casey. As the end market mix shifts, if automotive does what we had talked about earlier and becomes the biggest market, if you penetrate the professional side, first and then you move over to the consumer/home side. Do those shifts have any gross margin implications? Or is it really leveraging core technology with a similar cost base across different markets, but no real meaningful change in margins?

I think for today, what we've been commenting on is that we think it will stay the same. Over time, as these other markets start to mature, some of the old traditional markets start to go down, I think you could see some profile change over the course of time. But for today, as we've gone through 1, 2, 3 and now other markets as we go into next year, I think we'll have to wait and see exactly what change could may or may not happen over the long term. As you know, in particular, in auto, that takes a long time to come to real maturity. And that will also change the profile as we go forward. So I think, right now, we're very pleased with the way we've gone through the targets, and we think there's good opportunities out there. But auto, in particular, it takes some time. As we've talked about in the L4, L5, we've been talking about out a while that could be -- within 10 years, that could be 1 million units. Some would say that's low. But the real opportunity a lot of these markets were 0, 2 plus in the first 4 markets, the ADAS markets that we've been talking about. Once we get out further, I think that the profile can change.

Last question for you guys, another financial one to wrap up, maybe for you, Casey. You guys have a ton of cash in your balance sheet, I think, about $12 per share, if I remember correctly. I know you have a share repurchase plan that you've authorized and extended, but you really haven't bought anything back. Just at a high level, do you view that as kind of an insurance policy more or less, give you flexibility? Or what should investors look at for use of funds for that increasingly growing fund of cash?

Yes. A couple of things there. One is, I think that as we get into automotive, in particular, in some of these other markets, they want to see people that are around for 10-plus years. And so they want to make sure you have enough on your balance sheet to be able to get through this. And so I think there's a lot of our customers and potential customers that look to see a balance sheet that give us the confidence that we're going to be able to get into some of these new markets and have that cash. The second thing I must mention is, Louis has been with us a long time and very active in looking at opportunities. And we constantly are looking at opportunities to see where we can take what less and what the rest of the team has put together as we go forward and see where there are ways we can accelerate. And so myself, Les, or I'm sorry, Louis and others go out and are constantly looking at where we can go ahead and accelerate our opportunity, not only in automotive but in some of these other markets as well.

Great. Well, guys, we are right at the end of the time that was allotted. So Casey and Louis, thank you so much for participating in our conference. Congratulations on a really strong quarter and guide last week when you reported. Probably the cleanest one I've seen in a long, long time. Very impressive. And we look forward to monitoring the progress of the company going forward. So thanks again for attending our conference.

You bet. We hope to seeing you in person, soon.

Thanks.

Take care, guys.

Thanks, Bye.

Take care. Bye-bye.