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

All right. Good morning, and good afternoon, everyone. Thanks for joining us today. My name is Adam Gonzalez, and I cover the U.S. semiconductor and electronic design automation sectors on Vivek Arya's team, here at Bank of America Securities. Today, we're really excited to have Casey Eichler, CFO of Ambarella; as well as Louis Gerhardy, Head of Corporate Development. What we'll do is we'll start with a little bit of introduction before diving into some Q&A. We'd also like to keep this session as interactive as possible. So if you do have any questions, please feel free to submit them to me in the Veracast portal, and I'd be glad to ask them if time permits. With that, let's get started. Casey, if you don't mind, would you give us a quick introduction of Ambarella, its history, the markets, customers and applications that it serves?

Sure. Sure. Ambarella is a company that was founded in 2004 and went public in 2012 based on their initial products, which were all video-based. Fermi Wang and Les Kohn are pioneers in that area. Fermi holds patents on H.264 and by reference, 265. And that is what is shown on Page 5 of the presentation if you have that. That's the video compression image signaling -- or image processor, I should say. That green boxes on that slide, that's the history and the legacy of the company, and -- is what was the foundation for the new set of chips that have just come out over the last couple of years which are our CV chips, which have not only the human vision, but also a deep neural net AI processor that provides computer vision on the same chip. So that is -- the first series of chips are shown on Page 6. We started with CV25, CV22 and CV2, kind of a mainstream, mid-range and high range CV chip. You can see some of the characteristics on Page 6. And really, what that allowed us to do is combine our superior performance per watt per dollar across a series of markets, all of those chips that can play into those markets. And that's automotive security, both consumer and professional, as well as a series of new markets in warehouse, smart manufacturing, access control, robotics, et cetera. The 2 chips that we just announced at CES this year, CV22FS and CV2FS, are ASIL-ready chips more directed at the automotive market today. And those chips were announced, as I mentioned, in January. We have since, as Fermi talked about in our call, taped those chips out, they're back. We brought those chips up, and we will be getting those out to sample with customers in the not-too-distant future. And so that's really where the company has gone today from vision-only solutions to a solution that combines human vision and computer vision. Security market is about 60% of our total revenue today, and that's both in professional and consumer. Those are the 2 first waves that we've talked about in CV. The professional started in the second half of last year and continues to ramp. We've said that the consumer will ramp in the second half of this year. And will be followed by some automotive as well as some of these other markets in CV. But today, automotive is about 20%. Legacy business, which is our old legacy consumer business, things like sport cams and a host of other markets, that's about 20%. And then these 2 markets total the balance of 60%. On Page 8, you can see the biggest players in both markets. They're all partners of ours. We're not in [ HD camera ], but we have a deep relationship, and many of them are developing today on CV as well. Page 9 walks through the different segmentation in automotive. The recorder business is where we started and continue to have most of our revenue derived in that business. But we're starting to have design wins that will be announced over the last few quarters. In all of the other markets, Forward-Facing ADAS is probably the biggest market today. Obviously, a market that Mobileye is in. But e-mirror, be it 1-mirror or 3-mirror solutions, full cabin monitoring or driver monitoring, surround. And then obviously, 10 years out, you'll start to see more material revenue in part-time and full-time. Autonomous, though, there's lots of activity and discussion in those markets today. The next page shows our human vision, which is the recorder business today as well as the computer vision partners that we have. Those partners are typically developing the software stacks that get deployed on our chip and go-to-market with Tier 1s, Tier 2s and OEMs. I'll go to our other markets today. If you look at the competitive landscape across the first markets, that's shown on Page 12. And I'm happy to address questions there. But the other revenue, which was the legacy of our business is down to, like I say, 15% to 20%, and it masks a lot of the growth that we've had. As those markets have declined, we've ramped up the security camera and automotive markets. And there's just Page 13 that kind of walks you through that. Finally, the emerging markets that I've talked about, the robotics manufacturing, smart access with players like Schneider, Honeywell, Siemens, Keyence, Cognex, a whole host of people that are in those markets. All of these markets, we've gone from small consumer markets that are hundreds of millions to multibillion-dollar markets, and several of these markets are also multibillion-dollar market opportunities that we're just now starting to get involved in. Finally, the last 2 slides talk about the announcements that we made specific to this quarter. There's a slide like this for every quarter, so you can go back and see some of the other announcements that we've made, both human vision and computer vision designs. And our current quarter, which probably most people saw yesterday, and in our guidance for Q2, we've said that we really can't provide any guidance for the second half of the year. But obviously, the headwinds are the pandemic that's currently going on as well as the continued trade tension between U.S. and China, and that's in several different ways. It's related to entity list and tariffs and a host of issues that I can walk through if people wanted to address those specifically. But you got export control, regulations and many other headwinds there. So with that, I'll take a breath and pause, but that's a fast and brief overview on the company. I'm happy -- I know that was quick. I'm happy to go back and talk about anything in particular if somebody wants to go back and address. Or certainly, Adam, if you or anybody else have questions, I'm happy to address those as well.

Sure. Thanks, Casey. Thanks for that overview. Very helpful. And I'm sure we'll dig back into some of the topics as we go through this conversation. But first, and you touched on this a little bit in your slide deck, Ambarella has really repositioned itself as an artificial intelligence company over the last several years. And I wanted to give you a chance to really emphasize how Ambarella is approaching this market different when compared to most other companies out there.

All right. I want to get Louis involved here as well, so I'll ask Louis to, kind of, sketch that out, but I'm happy to jump in as well.

Yes. And if we look at Page 5, as Casey covered, those functional blocks highlighted in green, collectively, our video processor heritage. And we're known for capturing a scene as realistically as possible and doing it in a small form factor, very low power consumption. And what we're providing now is those green blocks integrated with the blue block, which we refer to that as CVflow. What it is, is it's a deep neural net AI processor. And if you look at the AI market opportunity today, many solutions for, let's call it simple AI, which might be perfect lighting conditions at short distances where they're -- very simple AI application. That can be executed sometimes on an ARM core. What we're providing is a deep neuro net processor. Sometimes this is referred to as deep learning. And what this will enable is more accurate solutions, more AI applications can be executed simultaneously. And so we're serving the high end of this AI in the edge endpoint device applications. And so that is a key way that we're differentiated in terms of how we define products and leverage our competency in the past. Other more specific competitive advantages that we have would be, for example, in the video processing area, these green functional blocks in the diagram, we can simply see better in low light conditions. We can see better in highlight conditions. In other words, high contrast situations where HDR is needed. We can see more accurately farther away. And when you consider that this video processor front end is collecting the raw data that's been fed into an AI processor, if you don't have best-in-class video processing capabilities, it's almost like garbage in, garbage out for these highly integrated SoCs that we offer. So that would be at a high level how we're different, both in terms of how we've defined our products, the part of the market that we're trying to serve and then some of the functional blocks of technology and how we're differentiated.

Great. Thanks for that, Louis. And you guys just reported Q1 results last night. So this is a very timely opportunity to talk about that. Can you provide a recap of the results and your outlook and how the overall demand environment is shaping out for you guys? What's going well versus perhaps below expectations at the beginning of the year?

Yes. Our results for the quarter were in line or exceeded kind of the consensus that we gave at the beginning of the quarter, even with the pandemic. And so our results, I think, considering the environment, were very good. We said that we did have some visibility, and we're able to give guidance in Q2. So we guided to $50 million on the top line, plus or minus 6%. We guided for a little bit of an improvement in the margin between 59% and 60.5% in margin due to what we anticipate will be a mix change to a certain degree. And then we guided our OpEx to be $31 million to $33 million, so kind of in line with what it was in Q1 for the quarter. We ended the quarter cash flow positive. We generated close to another $8 million in cash. And so we've now got over $411 million in cash and no debt today. So we're positioned to continue to invest in the opportunity, and we're doing that with not only our tape-outs in 10-nanometer, but we also -- Fermi mentioned that we taped out a tape-out chip in 5-nanometer, a test chip. So that we can start to move to that node. And so that was all, I think, very positive. From a market perspective, earlier in our first quarter guidance, we talked about that we anticipated being able to generate roughly 10% or more of our total revenue from CV. Obviously, with a lack of visibility into the second half, we've said that we continue to have that as a goal and are driving towards that. But a part of that was going to be the second wave of CV revenue, which was our consumer security business. The first wave was professional security at the end of last year, and consumer was looking to add on into this year. And so we'll have to see how our customers are able to move forward with the products in the market today. But we're hopeful that we can continue to drive CV revenue growth in the current year. So that's kind of a brief recap of what we talked about. There was also on -- as I mentioned, on slide, I believe it's 15, we had several announcements in our earnings during the quarter, 15 customer and partner announcements. 8 of those were CV, and they were not only in automotive but also in security, both professional and home -- smart home as wearable as well as one wearable announcement with Axis for a body camera. And as you can imagine, with what's been going on in the last few days, I think body cameras will continue to be an increasingly interesting market as we move forward. So that's kind of the highlights for the quarter and our guidance, I think.

Great. Great. I think we can all appreciate that the technology and the pipeline that you're building is pretty impressive. I think on the call last night, you did walk through some puts and takes with regards to U.S.-China trade and the impact of the broader impact of the global health pandemic. Particularly, as it relates to U.S.-China trade, can you help frame that topic for investors and the impact to Ambarella, particularly, in your surveillance business, you guys sell to Hikvision, Dahua. And they -- really their option for procuring chips is HiSilicon. And they might not be able to make chips anymore. So can you help us understand what the latest impacts and the latest view you have is with regards to trade?

Sure. Well, I guess the latest impacts were a couple of announcements on the U.S. side related to their -- the entity list where they added people to the entity list and then export control issues that were talked about. And those both were kind of in May. Right now, we are able to sell all of our products or a vast majority of our products to Dahua and Hik. And we continue to feel that we're able to do that. Obviously, with that overhang, their willingness to continue to it going forward is something we're going to have to see. Because if you got key components, whether it be our components or other U.S. components in your cameras or in your products, you're going to be concerned about having that change overnight with another change in regulations or laws by the U.S. And so you've had that pressure as well as the pressure of Chinese government wanting Chinese companies to source domestically not only our products, but all products where they can. And so that puts some uncertainty onto those 2 customers and how we move forward with those 2 customers. On the other side, HiSilicon, our competitor, is a subsidiary of Huawei, and there's been some U.S. constraint on their ability potentially to get further access to TSMC and TSMC's technology. And so some people have moved away from HiSilicon related to the China content and some because of concerns related to that. And so there is probably some uplift that we get certainly outside of China with some of our customers who were possibly using HiSilicon in the past and have transitioned to ourselves, either current or new customers. So one example of a new customer would be Hanwha Techwin. Hanwha Techwin is a large provider out of Korea. They were using their own chips. And HiSilicon in the past and a few quarters ago announced that they were going to be moving all of their development to -- from HiSilicon in the future to Ambarella, and those products are now starting to come out. So there are some opportunities along with the drawback. But China is the largest market today. Hik and Dahua are the largest players in the professional security market today. So the trade considerations have certainly been a headwind and, I think, will continue to be a headwind related to our products in the security market in general.

Got it. And tying into that, in the past, you've also spoken about your expectations about $10 million of inventory was pulled in by those 2 customers in 2019, and you expect that to burn off, I believe, in the second half of this year. What's the latest on that front? What sort of things are you looking at to keep track of whether or not that is being burned off? Are there -- is there any abnormal activity in ordering patterns? Just give us an update there.

Yes. I think that those are our estimates. We don't have perfect visibility, but we estimated at the end of last year, as you mentioned in Q4, that we felt that they had built up excess inventory of about $10 million. And at the time, we said we thought that would probably be consumed in the second half of this year. We continue to believe that and restated that in our call yesterday. Again, our estimate and visibility is very, very difficult. But we think that will exist. We don't think any of that has been burned off in Q1 or will be in Q2, but there's a potential there, I guess, as well in our current quarter. So that would put some additional headwind into the second half of this year if that was indeed to happen, and that was what we talked about and, again, reconfirmed in our call yesterday.

Got it. And then on the COVID-19 front, I think that's a little bit further down the line in terms of disruption. But you did call out on the call last night that access to certain physical locations led to somewhat of a slowdown both in the professional and the consumer side. And this ties into a question that I just got in from the audience, the consumer security IP camera market, which is, I believe, 1/3 of that surveillance segment for you. But what proportion of your consumer security IP sales are online versus retail at this point?

Well, every customer is approaching the market in a different way. Some are selling purely online and some are selling purely through big box company like the Walmarts and Best Buys and all that of the world. And some are combining those. There is one customer that actually sells the consumer side door to door. And obviously, initially with everyone at home, their business, I think, had a pickup. And then when things got worse, they had to pull their sales force and so that obviously impacted them the opposite way later in the quarter. With -- as far as our supply chain, our access, our ability to continue to move things forward, there was some impact. But generally, our development as we have with this with the tape out of chips, et cetera, everything has continued on and not been terribly impacted by the COVID virus. Where I think everyone is trying to understand is really the end market demand for our customers and how that will play out as 40-plus million people in the U.S. are out of work, and we continue to struggle with going back to where we were or establishing a new normal or however you want to think about it. And so that is certainly -- visibility is way down, and that remains to be -- to play out for us and for everybody for that matter. There are certain things you can move forward in trying to close design wins or to move designs forward. And in certain times, it's helpful to sit down in the lab with our partners and our customers and help them work through issues and develop new technologies. So -- but that part, obviously, as our office has been closed, most of everyone else's office has been closed. Even as offices reopen, I think they reopen in a guarded way. So whether or not people outside the company can come into other offices, et cetera, that's still, I think, limited, certainly in the U.S. and Europe. China and Taiwan are getting a little bit back more to normal, though there still are some differences. And so hopefully, over the next few months, we continue to make progress. I think generally, everyone feels that travel and the ability to interface face-to-face is going to be limited at least for the -- in some way for the rest of the year. And so we'll just have to see how that plays out.

Got it. And notwithstanding some of the near-term turbulence caused by COVID, I would think that perhaps the longer-term opportunity in surveillance, and particularly as it pertains to computer vision, might be positively impacted by this. When you think about the context of smart cities or the increased surveillance that will be needed to, I guess, reopen safely, how do you think about that longer term? And that leads me to just talking about the security market in general. Can you talk about what you see the market opportunity there, both in the professional security side and the computer security side in the next few years?

Adam, it's Louis. I'll take a crack at that. So you were referring to security camera market, and that's correct. Today, most of what we ship to the home today and into the enterprise or smart city is for a security function application. But what's happening in that market is there's many nonsecurity applications emerging as cameras get embedded into things. And so on the question of how is the pandemic impacting kind of the megatrends for Ambarella, I'd say, in 3 regards. One, higher levels of automation, we think, will be more important. How do you achieve that? You do it with AI processing. And for us, that's on the visual AI side using an RGB camera and our AI SoC chip. The other thing that is being accelerated is a move towards contactless-type sensing. A camera, of course, you don't have to touch. You don't have to enter a code or you don't have to leave your fingerprint on a sensor. Now obviously, in this pandemic period of time here, being contactless has some pretty meaningful advantages. And so using cameras as a sensor of choice in applications like access control, that may be kind of more important for us. And then third, we're all using Zoom and other forms of video communication. We just think that video as a form of communication will just continue to become more and more important. And that obviously is good for cameras.

Got it. That's very helpful. We only have about 10 minutes. So I want to shift gears to the automotive market for you guys. And first, can you just touch on what some of the near-term trends you're seeing there are? I know it's about 20% of your business, and most of that is in video recorders, but it's really -- could be the biggest growth opportunity for Ambarella as you look out over the next 5 to 10 years. So really, how is the broad-based auto sector shutdown affecting you guys? And how is it affecting your CV design pipeline? And then can you just talk about some of the longer-term markets and opportunities and how you see that playing out with regards to CV?

Yes. You're right. Today, majority of our business is in human vision and is in the recorder business, started off in aftermarket and has moved to OEM. And that was primarily in Asia, both China, Japan and, to a certain degree, Korea. That's clearly becoming more and more popular. And that business, I think, to a certain degree, is going to get combined with other functionality. One of the things about the performance that we have in our chip and what that allows us to do is it allows us to combine functionality to where you could have a recorder, a forward-facing camera and also the same chip running cabin monitoring or driver monitoring system or a 1- or 3-mirror solution that's combined. And so some of these markets will blend a little bit. But the other markets that I think are pretty wide open as well are the e-mirror, driver and cabin monitoring, surround vision. A lot of those markets, I think, are available to us, and we've gotten several announcements and good penetration. Forward-Facing ADAS is another market that we're also -- it's the biggest market right now and that we're also very much interested in. That's the market Mobileye's in. People are interested in having options in that market because Mobileye today is really a black box solution. You have to -- if you're a Tier 1 or a Tier 2, you basically have to differentiate on the design of the form factor. But the software and hardware are bundled together, and you can't really make any changes there. What -- people like a ZF or a Conti or HELLA really want to do is have the opportunity to differentiate on a platform, an open platform, and that's what we provide. Port their neural nets and take something differentiated to their customers or to OEMs. And so providing several different sources, not being a sole source and not being a source where the hardware and software have to be bundled allows those players to differentiate and OEMs to differentiate their product. And so I think we have a good opportunity in that market, but it is the auto market, and it takes a period of time, the things we're competing for today or for 2023, '24, '25 and beyond as the auto cycle goes. Some of the other markets where we can get in a little bit quicker are really aftermarket in some of these different markets. In fleet -- in fleet management and other areas where I think you can get to market sooner with our vision-based or CV-based solutions. And a lot of that we demonstrated at CES this last year and even the year before that. And so we've had several announcements in fleet and other areas that I think can get to market a little bit sooner. But -- and then I'm not ignoring full and partial autonomous. It's just the real revenue and ramp in that market, depending on who you talk to, for a real full autonomous is 10 years before you get to 1 million units a year. And so while that's an interesting and very big and broad market and it's where I think we can play very actively and are already having discussions very actively, from a revenue perspective, it's probably even further out.

Got it. And then just framing that into context of your computer vision opportunity. You guided up 10% of sales this year. Pro security, you started ramping in the second half of last year. You've guided consumer security could start to ramp in the second half of this year. How does automotive factor into that? And what sort of milestones should we look out for in that market?

I think for CV, really, as I mentioned, it would really be some of the aftermarket and more the fleet type of opportunities into next year. Also, I think the e-mirrors, whether it be 1- or 3-mirror solution and some of the cabin monitoring are places where we can get into revenue situations with CV a little bit quicker. The areas where you're controlling the car or the actions of the car, whether that be ADAS or all the way up to full time, that just -- we just rolled out our chips to do that, and that's the FS chips. They have the ASIL requirement that you're going to need -- or the functional safety that you're going to need as a standard either in Europe or in North America. And those chips will start to sample engineering samples to customers in the summer. And by the fall, we'll be able to have more meaningful engagement there. But that revenue, again, is then going to be designed in and off further out. But some of these other areas where you don't need the FS, that's been worked on for the last year or so, and I think they can get to market real quick.

Got it. We only have a few minutes' time. So I just want to touch on robotics briefly. It's something that's come up every now and then on your earnings calls. How do you think about this market and the potential TAM opportunity moving forward? And when could we start seeing some contribution in terms of just design wins and revenue?

Adam, it's Louis. So we announced at CES a robotics platform. And what it is, it's a software development kit with some of the most common functions used in a wide variety of different type of robotic applications. This could be a surveillance robot, a warehouse robot, an autonomous-guided vehicle for delivery purposes. And so we're at the stage now where that -- SDKs in the hands of customers. We have some early adopters we're working with as they develop or port their software over to our platform. And early revenue from that market is a year or 2 away, but it's a promising category. It's extremely fragmented, by the way. So there's so many different markets that will need a wide variety of different types of automation is the simplest way to put it. And identifying the opportunities where it makes sense to engage and invest mutually with our partners in developing these new pieces of the hardware is what we have to go through. So we're in the process of sorting it out. We do have these early alpha and beta type customers. And you'll be hearing more about these in the coming quarters.

Got it. And then lastly, just wanted to touch on housekeeping items, the consumer legacy business, which is about 15%, 20% of sales. Casey, you mentioned, it's no longer a GoPro play. I think GoPro is not a customer anymore. It's much more diverse. How do you see this business winding down over the next few years? Is it a very long tail? Could there be new opportunities to perhaps inject some volatility into that? How should we think about that?

Yes. I think it does wind down, as we said over the next 5 years, give or take. To be honest with you, we were pretty conservative in our estimate there, and we've been surprised with a couple of upsides there. DJI has done well with their Mavic Mini. And they also have a sports cam, I think, did all right as well as there was a device for a video logger that did pretty well. So that's been good. As I mentioned, some of the police wearables and some of the other things might have a little bit more legs than we would have thought because, unfortunately, of things going on recently. There was a couple of other sports cams done by other people. That market generally, as GoPro has found out, has been pretty flat but -- and some of the other players have actually done pretty well in taking some of that market share. And it's been Ambarella-based. And so that's been pretty decent. But we tend to be pretty conservative on that. We'll take the upside where we can. But without focusing on vision products for those markets, over the course of time, unless certain markets develop a need for CV, where I think we could be positioned, I think that we'll continue to transition over the next few years.

Got it. And with that, we're at the end of time. I just want to take the opportunity to thank you, Casey and Louis, for joining us today. For anyone in the audience who has any follow-up questions, please feel free to reach out to me. I'd be happy to pass them along. Thanks, everyone.

Thanks, Adam.

Okay. Bye-bye.