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

All right. Excellent. Good afternoon, everyone. Welcome to this session. I'm Vivek Arya from Bank of America's semiconductor semi-cap equipment research team. I'm really delighted to have the team from Ambarella join us this afternoon. Fermi Wang, CEO, with me; and John Young, CFO; and Louis Gerhardy from the IR team with us as well. And I'll go the usual kind of fireside Q&A format but if you would like to bring anything up, please feel free to raise your hands.

So Fermi, warm welcome. Really glad that you could join us. And you just reported earnings, and I'm sure we'll get into kind of the nitty gritty of the quarter. But Fermi, I think it will really help the audience for you to kind of set the stage on what Ambarella used to be, what it has come to and how is your kind of strategic vision evolved in this process?

Right. Thank you, Vivek. And first of all, thank you for coming to meet with us today. Ambarella is 21 years old. And we start -- when we started, we purely focus on one idea, how to enable personal video content. And that was in 2004, there was no iPhone, there was no YouTube. So to enable personal video content is a difficult task. And we built this really proprietary video processing technology so that people can record high-quality video content in a very cheap video device. That was the foundation of the company, and we basically went to a profitability just with that technology. But very soon after iPhone come out, it become very clear that capture video has become a commodity, and we need to start adding value by go to different market. And that is the time we realized that in addition to consumer market, we need to go to security camera, drone camera or other camera like automotive camera to differentiate the product portfolio, and which we did. That's the reason we went public in 2012. And after 2012, the most important thing happened in addition to collect the money, but is to identify for us, we need to add a brand-new technology to our video processing, which is really -- at that time, we call video analytics or computer vision. Basically, you can analyze the video content in real time, or you capture the video. And based on that idea, we start producing our computer vision technology or AI for the video technology, and it took us a few years to get to the first production. And we start ramping up our second-generation computer vision technology in 2018. And in the last 6 years, our revenue go from 0 to last quarter, 75%. If you look at it, it's a 60% CAGR in 5 years in terms of revenue growth. And from there, we identify more opportunity. So in addition to just doing new neural network AI application, we identify using autonomous driving to -- basically using a video perception and radar perception to do autonomous driving. In addition to that, in the last 2 years, when Gen AI pop up, we start looking at how to apply our CVflow architecture for those more advanced AI portfolio. And so today, 70% of our revenue from IoT side, 30% from the auto side, but majority of our revenue come from our AI processors, close to [ 80% ]. So you can see that we transition from a video processing only company to AI for video. In the future, I think all of -- the only technology we're focus on going to be is how to continue to improve our AI performance for video data only at beginning, maybe moving to other data type. But just purely focused on video data type, it will give us a lot of growth on the edge AI for the endpoints like cameras or edge infrastructure, like the boxes that integrate multiple edge endpoints and put the multiple video stream into a box that we can use a more powerful chip to analyze video. So AI technology is going to be the most important technology driver for us and also revenue driver.

Got it. Now edge AI, you have a very interesting presentation on your website, right, where you lay out kind of the core and then the network edge and the application edge. What does edge AI mean for you, Fermi? And the reason I ask that question is that, yes, there is an understanding that, obviously, you can't just take products that have been designed for the core and put it on the edge, right, but then you also have a number of companies on the smartphone industry, right, in the typical conventional IoT industry who can actually participate in that edge also. So what does edge AI mean to you? And how is the competitive landscape? Like who is your true competitor in edge AI?

It's a great question because I think there's a lot of different definition of edge AI. For me, the definition of edge AI means that for each application, if you call the edge AI, majority of the AI performance or AI functions happens at edge. But there's other device, like your cellphone, you collect data, but you pass majority data to cloud, and the majority of AI happens on cloud site. Although your data is collecting on edge, I won't call it an edge device because really that's just a data collection. The AI happens in the call side. So the way I define edge AI means majority of the AI performance happened at edge device. That's why I call it edge AI.

Got it. Okay. And then competition, who do you think as competitors in edge AI?

Today, I think the edge AI, of course, that NVIDIA has some edge AI device like Tegra. Qualcomm has a lot. Then you talk about another 50 start-up companies that have been founded by the VCs. So it's a busy cloud space, but we made clear that in the last -- since 2018, we shipped more than 32 million units of AI processors to edge AI devices. And also, I think that just that one data point that put us in a unique position competing with NVIDIA and Qualcomm.

I see. A few months ago, I think we -- everyone in the investment community heard of DeepSeek, right, in a really loud way, right? I'm sure we'll debate what the pros and cons, but what does that announcement meant for Ambarella? What has that done positive or negative for you?

Right. I think there's a major impact to us positively. Before that DeepSeek, when you talk about reasoning model, everybody just assure, in fact, including myself, it has to happen in the cloud. There's no chance it will happen on the edge because of performance requirement. But what DeepSeek really showed to me is they have multiple different models, starting from 1.5 billion parameters, the smallest to [ 8 ], to then to the 400-something billion parameters. And we -- at ISC West a few weeks ago, we showed that with our CV75, which is a 2-watt chip that is capable of running DeepSeek 1.5 billion parameters model without any problem, had a really good performance. And the CV72, our 5-watt chip, 4- to 5-watt chip can run 8 billion parameter. So just these 2 things show you that at very cost effective and power-efficient solution can run reasonable model, that was not possible just 3 months ago. So I think that creates opportunity. I still don't know what's the best application I can use the DeepSeek model running at edge. We'll figure that out taking time but just that we can start showing people that such a powerful model that can only be run on the cloud now can be available at edge, I think that creates opportunity for us in the future.

Got it. Is the optimization of the -- who does the optimization of the product for these large number of large and small and medium-sized models? Is that work that Ambarella has to do? Is that the work the customer does after getting your silicon? Like whose job is it to do all that work?

Right. So for example, that's using [ puppy ] model that is generated by some big companies, right? And -- but people want to retrain it. So that retraining usually happen with our customer. Of course, we can do retraining. We are capable of doing that but however, that we try to bundle a retrain model selling to a customer, we are basically taking away the differentiation of what our customer wants. So our job -- the way we position this is our customers should do all the retraining, our job is to help them to port that retrained model onto our chip and running very efficiently. So we need to provide them compiler tools that can compile the model. It doesn't matter with CNN or any kind of Gen AI-type of models and efficiently convert that to a binary that runs our chip. That's what we should do. And after that porting, it's our job to work with the customer to optimize the model. So that, I think, is our job.

I see. One of the industry question, Fermi, I had is some of the edge AI companies that we speak to, they say, well, we can do the processor. But by the way, we also have a way to bundle the sensor, right, that is getting the information. Some will say, well, I have a great connectivity portfolio, right, because ultimately, this thing has to go back and forth, right, to some other players. How do you address that bundling argument? And how do -- is that something Ambarella will need to do, develop a connectivity portfolio or develop a sensor portfolio? Or do you think that saying best-of-breed is the right approach for you?

My gut feeling is that people want to bundle everything together. In fact, in the past because they want to integrate those functions into a single chip, but I don't think that's possible. You cannot integrate sensor or connectivity into a 5-nanometer chip anymore price efficiently. So they are really talking about they have a sensor and they have a processing chip, and they have a connectivity chip, and they bundle as a package selling at a discount price to a customer. So it's basically a business deal that -- so people -- I do agree there are people trying to bundle and selling -- using business deal to sell the whole package. But from that point of view, you won't get the best technology, right? If the people really have the better technology on all 3 of them, they will win the business no matter what. But the way we run into it is we always provide the best technology on the processing, particularly on the video processing, power efficiency for AI and also performance for AI, DRAM bandwidth, all of that is our strength. So for us, we are really just trying to compete with people trying to bundle together. But for most of the customers today at least, is they are trying to get the best technology. Cost is always an issue, but technology has to be the first priority for that. So from that point of view, although we are running to all kind of people trying to bundle solutions, but I think we still have no problem to sell our AI processors to our customers.

Okay. A few kind of near term and then we'll come to the longer-term dynamics. So you reported earnings last week, right? Q1, very good results. Q2 was good guidance, and I think you raised the guidance for fiscal '26. The stock had a little bit of a mix reaction, let's call it that, but what was your kind of impression of your earnings? How you think about the second half of the year? And if you want to make a comment on how the stock is kind of mixed?

I won't call the down 10% was a mix of responses. So first of all, our Q1 was 3% better than guidance. We guide our Q2 for 6%, then we increased our annual guidance by another 5% in the middle, right? So from the financial performance point of view, I don't see anything wrong with it. So then if I look at our script, we talk about new markets, our new chip, we continue to deliver on time. So overall, from a script point of view, I don't think we'll show any weakness. So what's the theory about -- behind this 10% drop? I heard there's a few of them, but one of them, I want to really make sure people understand, there's a thesis says because I didn't talk enough about CV3 or autonomous driving updates in my script, people think that and defocus our autonomous driving investment and focus more on the edge infrastructure. I just want to make clear that it's not the case. In fact, in the last 3 months, we continue to invest heavily on the development in customer engagement, design win. RFQ. We didn't give any update because there's -- in the last 3 months, there's no major development from the customer point of view for us to give an update. So -- but however, at the same time, the new announcement on the edge infrastructure side is not -- should not be a surprise because since edge AI come out, we talk about using N1, N1-655 to address this edge infrastructure opportunity. So that I don't think it's a brand-new thing that people should say this is what we would be focusing on. So I just want to be clear that CV3 for autonomous driving continue to be a very important direction for the company, while we should try to leverage the only investment we put into our third-generation CV4 architecture and try to identify new application, can take advantage of that technology so that we don't need to add too much more OpEx and still can enable brand new application. I think that's really the best word that we can have.

Right. And then on the second half of the year, right, I think you're using kind of your typical conservatism. Is that typical conservatism? Or is there something in the macro environment that causes you to be more conservative than usual?

I think it's a conservatism that really reflect the reality. There's a huge tariff discussion on July 2, people are going to receiving later to find out what's their tariff rating. If everything goes smoothly, that will be great. But if not, it's going to be an ugly situation immediately. So what I'm trying to say is because we don't know, we try to bake in some conservative into our second half guidance. What that means is there will be more upside than downside for us. And if things turn out to be that tariff is not an issue after July 2, I think we can get a better number than our second -- than our guidance right now. So that's the thing we try to say.

Got it. Did you observe any pull-ins in the first half at all?

Well, to tell you the truth, when you see a strong financial performance that we have, you have to suspect there are some of the pull-in, but we engage with our customers aggressively to understand their position. None of them say they have -- they are doing pull-in because they are -- most saying, "Hey, we are sitting here waiting to see what's going to happen with the tariff." So from that point of view, I suspect there are some, but that's not a major scale like what we've seen 3 years ago.

I see. One thing, Fermi, you brought up that there have not been as many updates on the automotive side. We spoke with Continental recently, and they are still very engaged, right, with the platform. So to your point, things are happening, right? It's just that end customer progress has been, right, a little bit slow. But where do you see it? When do you think you will start to -- the automotive pipeline will start to get reenergized?

Well, first of all, our investment continues. Our engagement with Conti on the current design win, for example, Aurora, [ Conti arch ] and other design win we announced already. They are all in progress, and I expect that we go to deliver revenue in 2027 as we announced in the past. So the key is, right now, we have to continue to focus on the design win and RFQ that we have. The only thing we're trying to say is look at the -- while we [ lost ] a major design in the last quarter, and we were hoping -- we have very high hope before that, after we lose it, then that become such a negative response for us. So we start thinking about how to communicate to the investors about our designing activities, setting our expectations and fail to deliver, it's just -- I don't think that's the right way to do any more, right? So we need to figure out what's the right way to communicate to investors in the future.

Got it. Okay. And then if, let's say, for the next 1 to 2 years sort of edge AI and IoT stays, what are the top 3 or 4 applications that are driving it? And then within that, if you could also give us a sense for how many of them are accretive to your average selling price right now?

So first of all, we are starting a lot of new edge AI applications that will come into our revenue pipeline. Security used to be our biggest one. Right now, we are starting -- we talk about video conferencing, portable video with wearable cameras, edge infrastructure, all of them can be meaningful revenue for us and all of them are taking advantage of third-generation AI infrastructure. And the ASP all are going up. For example, just to give you an idea, for the video conferencing, the first chip we were selling there was a video-only human-viewing processor chip at $9. Today, CV5 that people use for video conferencing, is selling between $25 to $45, depends on volume. And the ASP growth is significant, right? Because AI performance you're adding there, right? That's -- CV5 is still second-generation CVflow. When you go to third-generation CVflow, CV72 and CV75, we talk about which you're going to add advanced model that we just talk about vision language model and the reasoning model, maybe move more in the future into that platform, I expect to see more applications jump out. So the ASP is really about how much AI performance we continue to offer our customer, and I expect that our ASP continue to grow up. Today, our average ASP in the core company is $13 to $14. Our CV5 selling price is anywhere between $25 to $45 for our second-generation CVflow. CV72 is also in a similar range by third-generation architecture. CV3, we talk about $100 to $400 and N1-655 for the edge infrastructure, we're talking about low 3-digit price. So you can see that our ASP is going to continue to grow based on -- because it's really AI performance, the demand going to continue to drive up our performance requirement, therefore, our ASPs.

Got it. You still have some legacy video processor business, right? How should we model kind of the decline of that business over the next few years?

In last quarter, we have only like 25% of human viewing or video processor business. And we expect that it will have a very long tail and gradually dwindling down in that. So you should maybe assume another 3, 4 years of down. But at a certain point, majority, 99% of our revenue comes from AI-based products.

Okay. And then can you give us a sense for what your exposure is to China, both on kind of a build to and a ship-to basis, so we get kind of a true measure of what that exposure is?

So right now, 15% of our revenue is consumed domestically in China. And I also believe majority of our customers, if they don't want to consume in China, they manufacture outside China already. So our exposure is limited there. So that -- I think that's our China exposure.

You said 5?

15%.

Right. Okay. That's what I thought I heard. Okay. And any more competition in that from local suppliers?

Absolutely. But I think in the past, when you talk about competitor, you talk about the performance, the differences. But in China, Chinese government basically says encourage their system company to use domestic supply as much as they can. So that basically put the competitive edge in different type of way. But however, I still think there's a strategic reason we stay in China because I believe majority of Chinese system company doesn't matter what it's auto or others, they are so competitive in China. The price pressure in China is significant. And some of them believe now that the only way they can increase the gross margin is doing export business. With that, with geopolitical situation, they have to consider second source for other silicon that can be accepted by United States or Europe, by other regions. So from that point of view, I think for China, we work with customers that want to do export business on auto and others, and that's where we think that's the opportunity for us.

I see. One of the things I saw in your company presentation was this kind of fiscal '31, calendar '30, right, almost I think about $13 billion in the SAM that you have laid out. Is that something, Fermi, that you can do organically? Is that -- like what will it take for you to go from a few hundred million, right, to $1 billion company? What will it take?

Well, the most important thing is that we need to penetrate CV3 because if you look at the next 2 years, the growth will come from the IoT side, right? We talked about that. But if you talk about 3, 4 years, the biggest opportunity is trying to secure a major design win in CV3. And we talked about the last quarter, the design we lost is close to $1 billion opportunity for us, right? So a win like that means really solving a big problem for us. And moving forward, it's really about how to expand edge AI on the edge device, edge endpoints to the edge infrastructure. So I think that stay independent and if we want to grow to $1 billion, acquisition probably is unavoidable. But at the same time, to your point, that having a -- playing a much bigger scaled platform, it will definitely help us to get to there even quicker, right?

Right. The opportunity that you had to forego that you just referred to, what was that due to? Was it just the company's scale, was it resources, like what do you think drove it? And more importantly, is that a persistent issue? Or was that just a one-off thing that you had to [indiscernible]?

Well, one thing I want to point out that our ASP continue to grow. One of the reason is we only really focus on mainstream high end. But also, you can point to that, if we have enough resources, we should be able to even win the low end, there's no reason to leave low end. But with our scale, with our R&D investment, we believe the best way to invest for us is focusing on the gross margin generation, therefore, the operation margin generation. But from that point of view, we do work away from revenue opportunities that we could have because we just don't focus on lower margin business. And that's definitely something that can be solved with a bigger scale of the company.

I see. From a supply chain perspective, what can be the issues that can impact your cost structure if we get a different situation of where tariffs are right now? Are you sufficiently diversified?

Well, yes. Well, in fact, we are in 5-nano and 2-nano. We cannot diversify from a foundry point of view because only the largest company can have a dual source for 2 nano, 5-nanometer, we can't. So we have to pick one. But from the geopolitical situation, yes, we can protect ourselves and protect our customer by -- we are using Samsung, so we do have foundries in Korea and also foundry in Texas. So we do -- from a diversified point of view, a geopolitical situation point of view, our supply chain has been proven by a lot of our customers in terms of the robust. But the true sense of diversification is that you have dual source on any nodes of silicon, which I don't think we can do with our scale.

I see. The gross margin point you brought up is interesting because your business has been consistently above the 58% to 62% target. So is that because you're walking away from business, right? Or is it a product gap or what is helping you stay above the target? Or I guess should the target be revised given how persistent you've been?

Right. So however, through our history, we still don't focus on low end because to try to compete with the price with those companies only pay attention to that, I think it's a bad deal. However, so throughout the company, we kind of downplay the low gross margin business. But for our customer, for example, that's some of the largest security camera customer. They want from the low end to high end. And we are happy to supply to them across their platform, and we did definitely take a lower gross margin on the low-end side just because we [ want that ] business. So that we still have a combination of low end to high end. But if some of our customers just say, I don't care about mainstream, I just want to have better price on the low-end side, we tend to walk away from that.

Right. When I looked at where a lot of our peers are in terms of modeling, people always model your gross margins to get back into the -- but from what you're describing, if ASPs continue to do quite well, is there a reason why gross margins would get back to the trading range or to your target range? Or do you think they can consistently stay above that?

Well, in fact, we are in the target range of 59% to 62%, but we are kind of guiding gross margin a little lower as a trend. The reason for that is our competition, particularly on auto side, is Qualcomm and NVIDIA. I don't expect that, that will be nice to us in terms of price competition. So I think that we kind of bake in that potential competition from them. But from the IoT endpoint device point of view, I think we have a track record and also the product portfolio that can protect us.

I see. And then on the R&D intensity, Ambarella has always been a company with a very strong focus, right, on R&D. But that also means that it is very out of bounds, right, to the kind of sales growth that you're seeing. Do you think it's just a matter of time? Is it like at what point do you think Ambarella can be a company that is going earnings on a consistent basis?

So first of all, we -- if you look at the last 10 years, we invest on the CVflow for CN type new network, and then we invest on the autonomous driving for CV3. Now that we're talking about -- so that 2 generation of CVflow definitely take a huge amount of investment. Moving forward, I think in the near future, our job is leveraging that investment to focus on the applications can take advantage of those investments. So we are not looking for another market that will require a huge amount of R&D expense. Instead, I do believe our AI architecture will allow us using the current architecture and current software to tap into the new market. So from that point of view, I hope and we will continue to show more operating leverages on our bottom line.

Okay. And outside of the CV3, what are you seeing in the automotive market right now? There's a lot of concern about cyclical issues? Or what are you hearing from your automotive customers?

We saw the same thing in fact that the market has different problems and financial problems and inventory problems. So because of that, we do see people slow down their investment. Although everybody is still committed to do Level 2+, but the investment cycle and their decision cycle definitely push out. That's one thing we see. The other thing is, instead of doing really high-end advanced Level 2, Level 3 car, they focus on more on the highway level, Level 2+. So basically, changing their business model to focus on the value, more value-based engineering and try to get to market faster so they can get to profit. So I do see that the change of the -- particularly on the Western side.

I see. Does your opportunity in a car change depending on the modality like if people are using just cameras versus using cameras plus LiDAR plus other things?

Well, I don't think that changed. I think it's hard for me to believe that when you go to a higher level of autonomy, you can use camera only. So I think the domain controller that can integrate multiple sensor modality continue to be our thesis, and we believe that we can continue to benefit from there.

Okay. And then finally, Fermi, as you look over the next year, what do you think are possible kind of upside drivers to the guidance you have given? I understand macro is what it is. Is there a certain market? Is there a certain customer or application that you think can drive upside to how you think about your fiscal '26 right now?

A couple of things, right? First of all, we talk a lot of green shoot opportunity in the edge endpoint opportunities. Those opportunities, if the volume goes up, for example, the wearable camera, a lot of people -- we start seeing a lot of opportunity on wearable because it's not just policemen wearing the wearable. A lot of security guard, service people, 7-Eleven clerk put on wearables so that they can document all of the events happen when they provide services. And you can imagine that if the volume goes up, and that can be a driver, right? For us, it's all about volume. The other thing is now we're talking about edge infrastructure. We believe the revenue is going to be start second half of next year. And if we hit the market with right customers, it can be a...

Rather server-like product.

That's a server type, infrastructure type, yes, it's a server type.

You plan to sell the whole box? Or are you going to sell just the CPU on the long run?

Right. So just like a camera, we provide a complete reference design. We show a camera to customer [ without ] chip in there, but our customers will look at camera say, great, that's a good example. They will build their own box, build their own camera. So for the edge infrastructure setting, we are going to build a complete box, including the application running on top of that and give this reference design to our customer, they can find the manufacturer themselves, to manufacture the box themselves. But on top of our software, they will remove the layers of software that we provide, replace that with their own models or their own applications so that they control all of the value added. So that is this business model we are looking at.

Makes sense. With that, Fermi. Thank you so much for your time. Really appreciate it. Thank you for that discussion.

Thank you very much.