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

Thank you to this session. I'm Vivek Arya from BFA's semiconductor team. I'm really delighted and honored to have the team from Ambarella join us this morning and really delighted to have Fermi Wang, the Chief Executive Officer of Ambarella. I'll go through some of my questions, but please feel free to raise your hand if you would like to bring something in between. But a very warm welcome to you, Fermi. Really appreciate you joining us.

Thank you, Vivek.

So Fermi, maybe let's start at the beginning, which is, for a long time, Ambarella has been known as a specialist in kind of vision processing and cameras. But you have made this interesting towards, AI. So maybe walk us through why did you decide to make this transition. And where are you in that transition right now?

Right. So we started this company 20 years ago, purely because we want to. We believe that we want to enable people to capture personal video content. That's -- 20 years ago, that's the intention of the company. So for the first 10 years, we try to optimize our video capture technology from the video processing to video compression to the storage to the local storage so that we can really help our customer build the best camera. We call it human vision camera, basically camera is designed for human to watch the videos. But -- and then luckily with that, we went IPOing 2012. Right at the IPO time become very clear that our customer wants beyond just a human vision. They start seeing that the computer vision becoming important because newer network enabled a brand-new trend of AI processing for video. And they want to enable cameras, not only in watching, but also can help us to people to understand the environment. That's exactly where we start our computer vision or AI type of video processing. It took us a few years to understand how to do an algorithm first approach to provide optimized silicon architecture for AI, for video data on home silicon. And then the product we generated -- really generated real revenues -- our second-generation CVflow engine. We start shipping CVflow engine in 2018 -- 2017, 2018 time and the CV2 engine, we have our dedicated neural network AI influence engine in there and the target at the AIoT market as well as automotive market. And throughout 2018 to 2023 last year, our AI engine based revenue is roughly 60% of total revenue. So you can see that we transition from a video, human viewing camera company to an AI video camera company today. But however, we didn't stop here. In fact, in the last few years, become very clear to us that the AI performance requirements from our customers, existing customers and future customers are going to continue to increase. And Gen AI just another take a different level, but just look at our current silicon that we are producing right now. Our CV2 first generation or second generation of AI chip, the average ASP is $20. CV5 is the top end of the second generation processor, the ASP is anywhere between $30, $40, and AI performance increased proportionately. And our third-generation AI processor we designed for the automotive, the ASP we talk about is $100, $200 to $400. So you can see that the dollar size is also related to how much AI performance you have. So my feeling is throughout the year, we start with a human viewing video, now we are doing AI for video, and we're counting on the AI performance requirements from our customer continue to increase in the future. And of course, ARM will be the next generation, and we'll talk about this for full generation AI processing engine, but that's where we are at with our current AI for video technology.

Got it. And exactly as you pointed out, I think your 60% AI exposure probably puts you next to just one large player right in the market. But if you kind of contrast for me yourself versus that large incumbent, right, there, obviously, products are designed for a very different application, right, but they also cost tens of thousands of dollars you mentioned something that's in the tens of dollars. So what is your AI strategy? How do you pick the specific markets to target? And what kind of differentiates you in those markets versus the large incumbent that's out there?

Right. So in fact, Ambarella is focusing on training, which we don't do that we focus purely on the inference side on edge device or edge server. And the other thing that we don't do is really that we don't do cloud. Data center is not our business and we purely focus on the areas in AIoT or automotive, basically edge device companies. So from a technology point of view as well as the market focus point of view, we are different. However, I really think that the reason you have so much investment on training because you want to enable more influence in the future, otherwise what they want to do training. The ideas enable different applications at the end. So influence, I think, I actually believe what we have a huge growth potential in the future. In terms of how we differentiate, it's very simple. We believe that starting from day one of the company, we believe that anything related to video or edge AI from that point of view, you need to have a specialized architecture. If you want to optimize for the power consumption, performance per watt, performance per dollar, you have to optimize for it. And our differentiation is always that we create architecture, best for the video and the best for the video AI and therefore, we can provide much lower power consumption. In fact, with a much lower, it's not at 20%, 30% compared to our key competitors, who are 2x or 3x power saving on the same performance level. And also with that, it provides a much better bond saving cost for our customers. That's basically a major differentiation for us. Of course, there are other things this is really one important thing for us.

Got it. Fermi, just a technical question that investors have started to kind of distinguish between there has been the AI that has been there for several years, and now this new large language model and just over emphasizing the word large, right? And the generative application aspect of it, right? That's a very different ball game. So when you use the word AI at Ambarella, which AI are you talking about is it both, I think just it would help us to understand the distinction.

Right. So maybe we should look at how the AI evolution in the last 10 years. When we start doing our AI for edge video device, we use this technological CN neural network, right neural network. This was the main architecture for the CV2, our second-generation architecture. Then when we come to third generation, one major difference in addition to the CN neural network, we add transformer technology support, which is another major architecture required for not only for automotive, but transformer network is really basis for a large language model recently. So the third generation really increased the support on different architecture level. And I think next generation, we need to find a way to really optimize for large language model architecture. In general purpose will help you to really go to market quickly, but for the people who really focus on power consumption or the performance provide improvement, this will become an important differentiation for us.

Do you think competitors from the mobile industry can go after the same thing because they are used to making products, right, and sort of that price envelope. And there's a lot of ecosystem that is now going after developing language models, right? I don't know whether that should come large or small or medium that are transformer-based that are going after that specific kind of device compact or do you see that as competition for you?

There is one -- if you follow the automotive software development, people start talking about end-to-end AI, which is Tesla as word. But in my mind, that's just one example of a large language model -- large model. And from that point of view, I think eventually, when you really look at in the future, when we talk about Level 2+, Level 3 car I think you're going to see larger language model in there. But you might not see 1.4 trillion parameters. You might see a fine-tuned 30 billion or 24 billion parameters running a car and provide end-to-end AI. I don't know exactly how many parameters you need because we're still developing, but I don't believe we need 1.3 trillion parameters. So for large language model, I also believe that on the data center side, people are pushing the Gen AI -- sorry, the ChatGBT 2 trillion parameters, but from an edge device point of view or some vertical applications, you're going to see smaller large language model, but fine tune for each vertical. And that's where I think automotive will be.

You described several generations of your AI inferencing technology, right? You were in the second generation, you said you added transformers to the third one. What do you plan to add to the fourth one? And do you think that will entail working on new geometries as well? Like what are the building blocks of making Ambarella move to the fourth generation of technology.

So when we support Gen 3 for transformer, we kind of define it for automotive. But if you really want to define the next-generation like transformer support for large language model, obviously, if we apply this we call algorithm first approach and understand exactly how large language model is transformer. There's tons of things you can optimize for. So I expect that our next-generation CVflow engine will be focused on optimize providing much better performance per watt, therefore, performance per dollar, architecture for ARM. That's just one direction. To reach that direction, I think we need to invest on the next-generation process node, right? And we talked about in our earnings call, we talked about we are kicking off a 2-nanometer process implementation. And we are looking -- we are finalizing our evaluation of 2-nanometer. We want to make sure that we get a commitment from our foundry partners to support us for this really advanced processing development. This is a gator all around. It's definitely, I think, difficult transition for foundries as well as for design house, and we want to make sure that we did it right.

Do you have a choice of foundries? Or is there a choice of one?

No, I think, obviously, Samsung continues to be our partners on the foundry side. We're still also talking to other foundry can support 2 to 3 nano which is TSMC only. So there's no suspicion there. There's only 2 foundry can do this at this point.

Aren't there any aspirations of taking the fourth generation and moving into the data center as well? Or do you think just by application, right? So you have been more AIoT, industrial and consumer and then you move to automotive, do you have aspirations of going into the data center as well.

There is aspiration, but also there is a reality, right? So as a small company, even we think we can build technology, whether we can penetrate the market, it's 2 different questions. So I think that we will try to take approach to that technology will be suitable for the current market that we are targeting at. Also there is a chance to upgrade to the data center. But I think although we haven't finalized this, but I really think that we need to find a way to enable our customers and partners that are willing to work with us on their current market segment as the first priority before we jump to a totally different new market and trying to compete with the giant competitors, right?

Got it. You started to talk about the different products, so kind of the 60% related to AI so CV2 and onwards. But video processor is still, I think, almost 40%. So I was hoping you could kind of double-click on these 2 and help us define how do you look at sort of the gradual kind of deceleration of one and the acceleration of the other?

Right. So first of all, last year, roughly our total revenue from our AI video chip is at 60%. The percentage will continue to go up. In fact, it's driven by 2 things. One is our CV5 family. In fact, we talk about that this is really the ASP at the north of $30 to $40 range, that will help to drive our ASP growth as well as our revenue growth. The other thing is that the point where we look at the recovery from this inventory correction, we see the strength of our AI engine chip more than on the traditional one. Our video processor, we have not taped out any chip without our CVflow engine there. So meaning we haven't really invested on the video processor only market in the last 3 years. And so that -- with that, I expect this continue to have long tails of a reduction on the video processing. We see huge deduction -- reduction last year, but we're going to see a much smaller reduction in the video process this year. I believe our growth on the computer vision will cover and outpaced our reduction on the video process moving forward.

Got it. You mentioned the third generation of AI inferencing, could you help us understand the difference between the CV3, CV7 and ones. Sometimes I get lost in these different products. So how do you kind of distinguish -- are they according to different power specs? Or are they going to different applications? What's the differentiation?

Like I said, CV3, from a technology point of view, they have 2 differentiation versus second generation. One is the transformer support the other ones, the power efficiency. Running the same workload on second generation and third generation, the power efficiency of 3x. So with the same workload, you can reduce the power consumption for 1/3. But however, from the silicon point of view, when you see a CV7 family, which is really for our AIoT. When you see CV3, is really for our auto grade Level 2, Level 2+ cars, right? So those are 2 very distinguished just like we disclose our revenue. There's AIoT revenue, there's an auto revenue. So CV7 is for this side and CV3 is for the auto side. The OI is really creative for the LLM, which we start this process just last year because when large language to pop up 17 months ago. And the OI is going to be an LLM processor that we designed for ARM, and that's definitely a focus area on to continue to invest. So we can continue to see AIoT chip, auto-grade chip in the ARM chip.

All right. Now let's kind of move to redefining your by product rather than these technologies. So right now, it's about 2/3 AIoT, 1/3 automotive. What does that target make you are working towards? I think just before our discussion, you were talking about the auto shows, if I imagine auto has got to be a much bigger part of the mix going forward.

In fact, when we started our kickoff for CV3, our intention is that automotive has to become a major portion of our revenue in the future. It has to be in the near future. In fact, with the current design wins with our current Chinese OEMs. We are talking about the end 2025 early 2026. So even with that one design win, I expect that the ratio of our AIoT versus auto or gradual to one-to-one and with more design win that will move to even bigger. So automotive will continue to -- we invest so much money on the third-generation AI engine in particular for automotive business, we need to quickly generate revenue on that. That's the #1 priority for the company right now.

Got it. How do you -- from a competitive position, kind of stack Ambarella versus, say, a Mobileye first and then against Qualcomm and NVIDIA and some other names later.

Right. So for Mobileye, I think really for people who really enjoy a black box approach, Mobileye is always their first approach. But however, for the OEMs who want to control their own software and their own differentiation like the design win we have, we talk about that's not the strength of Mobileye. In fact, in those kind of design wins, we faced NVIDIA, Qualcomm, Horizon in China, a lot more often. So I think that's a true competition. We continue to believe for the people who OEM say, okay, I need to have a product shipping in the near term. And the black box is okay, right? Mobileye usually have the edge. But however, for people really said, I have my own software, I want to differentiate then other we face other competition.

All right. Is -- how do you make the distinction for me between where does ADAS stop and where does Level 2 start, right? Because, right, Mobileye will obviously suggest they can do both, right? But today, a lot more of the focus seems to be in kind of the front right -- front side ADAS.

My definition from ADAS is just a single camera facing front with maybe 2 or 3 radars only. So that's -- we call it on 1 video, 3 radar, 1V3 or 1V5 type of solution is additional ADAS. When I go to Level 2+ and talking about multiple camera, multiple radar, sensor domain controller. That's my definition change, right? But from a function point of view, ADAS is really just emergency brake, Level 2+, there are 2 levels of auto driving in fact defined by Telsa why is really called autopilot, right? It's really highway level automatic driving with multiple camera. The next level is really FSD basically means you can do automatic driving on the local road or city roads. Those 2 require a totally different performance level. In fact, we believe that although both of the car Level 2+ and the AI performance requirement between these 2 can be 2x or multiple x difference in terms of AI performance requirement to reach those 2 levels of Level 2+. So I think Level 2 is very well defined is mobile dominant. Level 2+, there are a huge range of performance and requirement. You can go as low as 5V, 5 video, 5 camera, 5 radar all the way to 12 camera, 5 radar. This wide range can be defined by software and also a combination of hardware accelerators. So I think that's where we are focusing on really making sure that our CV3 family chip from the low-end 635 to the high-end 685. We have a family chip that can address this Level 2+ applications.

All right. One other thing I found fascinating is your participation in the domain controller market, right, with CV3? Because there, you have a different set of incumbents tied with NXP and others who have a lot of scale and incumbency in that market. So first, what made you go after that specific market? And I know you have a very important win there. When does that start to become tangible for you?

Right. So first of all, I think in any car, there are many different domains, right? The domain that we're interesting is really just a safety domain, which is controlling the car to drive. And ASP and other traditional automotive companies, they really focus on other domains for providing controllers. So for the safety domain, the biggest requirement is really video and radar perception and you do future on top of that, that requires really high level of the video processing and the AI performance. That's where we think we are better than an ASP or other traditional the controller or the microcontroller companies. So I really think that our key competitors in this safety domains NVIDIA and Qualcomm, where it will make a difference. I think -- I hope that we're going to continue to build up our momentum on the 2026 revenue from the first design into revenue, even getting radar inside of this revenue ramping up. So hopefully, that we can continue to build momentum on that for our ramp -- 2026 ramp.

Got it. Just near term, how is automotive market demand? Because that is the one market where it has not quite had the flush that we see in the industrial markets. So what are you hearing from your automotive customers?

Well, I think right now, I think a lot of our automotive customers are dealing with the inventory correction that other market has been dealing with the last few years, you see that automotive can start digesting their inventory. So that's definitely impacting the market size and momentum. But however, if you want to look at -- even though people saying even EV growth is slowing down, but I still believe that EV is going to continue to grow in terms of the market share, although it's not ramping as fast as in the past. But eventually, I think EV car is going to continue to become a midstream. And on top of that, we start seeing Level 2+ really the real autonomous driving or I should say that assistant driving to a point that is better than just an emergency break. You can help your car to navigate through the highway or local streets. I also believe that's a huge growth and penetration area. Today, Level 2+ penetration rate is probably 4%, 5% less than that. I expect that the penetration will go much higher. And that's where I think will help us to get revenue growth.

Got it. Conceptually, why should inventory matter to a next-generation technology company? Because you're still in front of customers adopting a lot. Is it that you think that they have a lot of inventory of prebuilt cars from prior generation technology? Like why should inventory matter to you if you are developing a lot of the new technology?

Not for me, but for our customers, when they start building with inventories, when their revenues started to reduce and they kind of start cutting back on the investment, right? In fact, the first thing they cut back is a Level 4 cars, which is very obvious. But however, I truly believe that the EV car and people's investment auto-driving continues. That should not impact us a lot, right? So although I based on a comment about general market situation, but I think the investment for our customers for our Ambarella to this automatic driving is going to continue to be mainstream.

Got it. Just quickly on the near term, I think you kind of described interestingly the profile in terms of there is a cyclical part of your business and then right there is a secular part of the business. When do you think Ambarella kind of transitions to something that is a lot more secular, right, than cyclical?

I think next year, we should go back to the normal business. This year we continue to deal with a little bit on inventory correction. You look at first quarter, we did 54% and now we are going to do 62% this quarter. So we're kind of still dealing with some turnover from the inventory correction, but I think that will be done by the end of the year.

All right. A few kind of financial questions on gross margins. I think you have been consistently outperforming your 59% to 62% long-term target. But do you think that as the auto mix grows as you transition to these newer products, right, that we should be expecting gross margins to come back into that long-term order? Or you think there are drivers to keep it above that long-term only.

I think that for automotive, if you look at how big our competitors are, and they are famous being cutting price to keep their market share. That's where I think that we need to be realistic that we're going to be ready to defend ourselves. So I think that we made the comment about 59% we're going to gradually go back to 59%, 60% because we need to be ready to compete in that space when our competitors, they can use this as a weapon. So I think that's just a comment for that. But however, we continue to develop new technology and that go to different markets, our ASP continue to go up, right? From the CV2 family from the $15 to $20 to CV5 $30 to $40 for CV3 to $100 to $400, when your ASP goes up, I think that's where I suggest our investor focus on. We'll continue to maintain a healthy gross margin while expanding our ASP range.

Got it. One thing Ambarella is well known for is just kind of a commitment to R&D, right, regardless of whether it's up cycle or a down cycle and that's been very impressive, and that's obviously the reason why you have all these new products coming out.

Competing with NVIDIA and Qualcomm, this is the only way we can survive, right? Without technology, we are nothing. So making sure that we are not only -- in fact, investing in 2-nanometer is like a defensive, it didn't give us any advantage because our competitor are there true. Our differentiation is from our own technology, building our own CVflow engine for AI influence within our video processing much better video processing technology. Those are the core technology that will differentiate us.

Got it. So is there a certain way, Fermi in which you manage the company and say, if, let's say, top line grows I don't know, say, 20% a year for the next 3 years. This is how much we should plan to have operating expense growth or these 2 are done independently because a lot of times you're making these investments years in advance.

Right. So first of all, for CV3 investment were done, right, because we spent in the last few years, we spent so much money to develop a brand-new architecture, brand-new market, software stream, automotive market software. This investment is really mature, and we should continue to spring our new chip and use product and given a design win without adding too much of a brand new investment into this market. The new investment is really on the Gen AI side, which from my point of view, it's not brand new because fourth generation, like I said, 2 nanometer, we don't find that before. So we can see incremental investment on the new technology. But however, how to generate a lot more revenue from our investment of third-generation AI is a key. If we can -- where we can enjoy our investment on the automotive market, I don't think that we can need another high percentage of our revenue on R&D side. We should continue to have leverage on the gross margin and operating margin.

Got it. And then China, what percentage of the mix is that? And do you think that there is any concern about whether it's a restriction or whether it is let's say, extraordinary exposure to the Chinese EV makers. How should we handicap China as a part of Ambarella's growth strategy?

Today, our total revenue selling into -- in Chinese market is 15%, right? So it's already reduced quite a lot compared to -- in past. However, I think that the geopolitical discussion is always valid, right. There are 2 directions. One direction is U.S. try to limit the high-end AI chip to the China. And also, China, I want to make sure that people use Chinese silicon. So both sides, we are facing the pressure. But however, I think the current -- the US targets to make sure that high-end AI for training, not going there, inference engine is not a target yet. If that becomes a target, that becomes really industry-wide problem. So it's hard for me to comment on that. But on the Chinese side, Chinese always want to use their own silicon, if possible. But however, if you look at that a lot of Chinese OEMs also wanted to export business. I think that, well, force them to use consider non-Chinese component for that by other business. So the balance will be there. It's not like they can continue to countdown Chinese component only for the complete automotive business.

Got it. And just finally, I know we are a little bit out of our time. But maybe just leave our audience with what are the next 2 or 3 important catalysts milestones that we should be looking forward to Arambella.

First of all, I think most important thing is the design win for automotive, right? This is something we've had so much money on the automotive side. We need to continue to find a way to create revenue on that. The second thing is really back user technology. How we go -- how we view we continue to improve our technology build up differentiation. Those are 2 things that we need to balance moving forward.

Terrific. On that optimistic note. Thank you Fermi. I appreciate your time.

Thank you.

Thanks.