Arbe Robotics Ltd. (ARBE) Earnings Call Transcript & Summary

Great. Thanks, everyone, for joining. My name is Rajat Gupta. I'm a member of the automotive equity research team at JPMorgan. Very pleased to have with us from Arbe Robotics, Ram Machness, Chief Business Officer. We'll go straight into fireside chat. We'll start with like a quick overview of the company, and then go into more product details. But yes, Ram, thanks for joining us.

Thanks for having me.

So maybe to kick it off, just for people who might not be familiar with the story. If you could just give us a quick overview of what Arbe does, where they fit in, the whole ADAS safety systems ecosystem, just a little bit about your products, what the differences are in yours than some of the other systems out there, like who are your biggest customers, your biggest regions, et cetera.

Okay. So Arbe Robotics, we are a Tier 2 company. We are a chipset company. So we are developing chipset for high resolution and low false alarms radars. So this is the next generation of radar. It's not LiDAR, it's a radar. It's the radars that today we have in our vehicles, getting them to the next level of resolution and the next level of sensing of the environment. If I put 3 numbers that are currently in the industry that explains the differentiation and where our base taking this industry, today's radars, they have what is called 12 virtual channels, okay? The most advanced radar that is out there today in the market is 192 channels. We are talking about 10x the amount of channels. And this is Arbe that is talking about this and Mobileye. These are the 2 -- the only 2 companies that presented the ability to process this amount of channels and getting to the high resolution that is needed from the radars. These high-resolution radars enable use cases that we cannot do without those sensing radars, those imaging radars. So if it's about detecting, and I was in the Kodiak presentation just before, and he was talking about detecting obstacles, understanding what, where can I drive and where I cannot drive. Doing it based on the radar, this is something that is not feasible if you don't have this amount of channels. And we are able to solve it with the amount of channels that we have and the ability to process this amount of channels. That's what we are enabling to the -- for the industry, being able to detect, for example, an obstacle like a tire on the road, on the Autobahn, or someone changing a flat tire on the side and you don't -- and it spills over to your lane, and you want to detect it, not only by using a camera, but also using a radar. Because the camera itself is an excellent sensor, but sometimes it fails. If it's because of a direct sun, if it's because of bad weather, if it's because it's night, right? And the camera cannot see the obstacle, you need a redundant, another sensor that would be able to detect it and will be able to give the system or even the driver an alert that there is an obstacle on the road. And we wanted to go to the mass market. So it's not like for the high-end super autopilot kind of a Super Cruise, for example, of GM or something like that. This is a technology that is going to the mass market of vehicles, very high volumes. And the idea here is to enable those use cases as a safety features for the vehicles.

Understood. Generally, like in any industry, safety is always like a moving target. You want to get safer and safer. I'm curious, I'd like to get your thoughts on, just as an industry, on a scale of like 10, like where do you think safety levels are for vehicles or companies that claim full autonomy today? How would you rank that? And how does Arbe aid in like moving that further?

Yes. So if we look on safety, one thing needs to be very, very clear. The expectation of safety from autonomous vehicle is, by far, higher than the expectation of safety from a regular driver, okay? I always say we are used to. We are accepting the unfortunate effect that people are being killed by other drivers, okay, the human drivers. We are not accepting people, and we are not going to accept people being killed by autonomous driver, okay? So we have to understand that the level of safety that is expected from autonomous vehicles, from autonomous features even, not a complete vehicle needs to be autonomous, even some of the driving features, the level of safety that is expected is, by far, higher than the level that we are expecting in normal drivers. Elon Musk said back then that, initially, he didn't think that he needs more sensors because we are driving with our eyes. So camera, to be able to do that work. But that's not true because our expectation is not to drive the same as a regular driver. The expectation is much higher. You can ask my wife. She has a lot of complaints about the safety of my driving still every time she would give the kids to me to drive. That's not the case with autonomous vehicle, okay? We will not give the autonomous vehicle, the keys to drive if we don't see that it's super safe. And the ranking today is pretty low. And we believe that one of the biggest obstacles to get to the higher level of safety is the sensing, sensing of the environment. Because, again, camera is doing a great job. But even if it fail-safe for 95%, 97%, you know something, 99% of the cases, that's not enough, okay? And when someone would take those vehicles and it would be very sensitive to phantom braking, to cases where the vehicle is not paying attention to obstacles, to cases where the driving is not smooth, we are very, very sensitive to that. And we will not let our vehicles drive if we don't see that these use cases are fully covered. So we are -- I feel that we are still far from there.

Very helpful. So then in terms of the sensor suite that we're going to need to solve that full autonomous, where do you see Arbe's role in that? And then what do you think we need to do to get from where we are today to solving full-level final autonomy?

Yes. So I would separate it to level up to Level 3 and then Level 4 and 5 because these are a bit different, even their business model is different, okay? So if we talk about level -- up to Level 3, I think we'll see cameras for sure. There are always going to be. And we'll see, in most of the vehicles, imaging radar. In the high-end vehicles, we'll see also LiDARs as redundant technologies as well. So that would be -- and on the imaging radar, we see, for sure, one on the front, we might see also in the back. So for example, if you go on a highway and you want to move to a different lane, you want to see far back to see that there is no motorcyclist is coming in 200 kilometers per hour or 120 miles per hour driving there, and you got into his lane. So you need the long range and you need the high accuracy to understand where the motorcyclist is. Is it in my lane? Is it in the adjacent lane? Is it far away? And that's not a concern. So we might see more than 1 radar, imaging radar, on those vehicles. And if I refer to what -- for example, Mercedes Benz said in a public -- in a Tech AV, he said, "Imaging radar is a radar that has at least 32 x 32, at least 1,000 channels," okay? And right now, there are only 2 companies that are planning to offer this. This is Arbe that is going on production this year and Mobileye that will go to production in '25, '26. These are the only 2 companies that are offering this type of technology to the OEMs.

Got you. And so then in terms of like the content opportunity you see per vehicle, is there a difference between what you see on EV and ICE vehicle and then those larger, like the 6-wheel vehicles? Or is it all kind of the same?

Yes. So I actually referred mainly to the Level 2 and Level 3 till now. For Level 4 and 5, we see 360-coverage trucks. That's another very common use case, where we see up to 6 imaging radars over there. Again, a lot of it because of the different business model, same with the Level 4 and Level 5. They need to make sure, since the liability is on them, they need to make sure that they are fully covered. There are no exceptions, okay? So they would put more than 2 imaging radars. And probably, they will put also LiDARs on the RoboTaxi. On the commercial vehicles, commercial like trucks, heavy machinery, these type of use cases, they are not able to do -- to have a LIDAR because of the dust. It just covers the LIDAR after some time, and it's not an efficient technology for them. There are some cleaning technology coming, but that's still behind. So this will be covered also by up to 6 imaging radars as well.

That's super helpful. Could you talk a little bit about the recent win with like a truck manufacturer that you announced on your earnings conference?

Yes. So we announced recently about a partnership with well-known -- we are not allowed to say the name, I'm sorry about that, a well-known European truck manufacturer. They evaluated several solutions and several technologies, not only radar, they evaluated several technologies. And they selected, after more than a year of evaluation, our radars. And they are starting now the development of the perception stock, infusion stock, together with our radar that is based on our chipset. And the main reason for them is the fact that they need to cover all the use cases and be able to detect those obstacles, those not have the phantom braking, have a very long range for trucks because the braking distance for them is quite far away. And that's the technology that they selected.

Got it. Got it. And just going back just like product features and some of the recent wins, maybe like just on the products, could you give us a sense of what are the new products or features that you're planning to introduce over the next, say, 12 to 24 months? And how do these align with current conversations and demands from OEMs?

Yes. So if I look at right now, where the well -- our time is spent these days, so our main task right now is get our customers, the Tier 1 and the OEMs, into production, okay? That's what we are busy with these days. We are going to be -- our chipset is going to be in production next quarter. And we want to make sure that our customers, the Tier 1s, will be in production as soon as possible, okay? Some of them will be in production in 2025. And that's super important for us because that's the time to revenue for us. Before they are in production, there will be no significant revenue for us. So frankly, that's what we are busy most of the time. We are already starting the development of the next generation, which includes even a further upgrade of the system. And if I look even further out, I think that the entire sensing suite, this is something that is -- needs some integration. And having a better integration between the different sensors, including having the radars, and they come talk to each other so they can help each other in the deficiencies of each of the sensors. So each of the sensor has its own deficiencies. We want them to talk with each other, so they compensate for each other.

Got it. Got it. Just one more and then I'll hand it to [ investors ]. Could you give us a sense of how the customer perception around your portfolio might have changed over the last 6, 12 months? You hinted on the earnings call, like how customers are coming around the 32 x 32 as the minimum requirement. Just could you elaborate a little bit more on the feedback that you have been receiving?

Yes. So if we look at where the market was until recently, until we introduced this solution, the market was able to get to these 192 channels, mutual channels, to 56. That's the [indiscernible]. And they were trying to solve the problem using those type of sensors. And they spent a lot of effort around this. The OEMs, the Tier 1s, companies, other companies, third-party companies, software and AI companies we're trying to solve the problem using those sensors, everything. And I think that's a major shift that we see in the last year, that they came back with those results, that they couldn't solve the problem using those type of sensors. Plus Mobileye coming out and announcing publicly that -- and Mobileye are a perception leader in the market. And they're saying, in order to solve this problem, you need exactly the same number, by the way, 2,304 channels to solve the problem. And that's something that really helped us because the OEMs understand, and they already solved the problems that they are not being able to solve with the 192 channels, radars that exist today, and they know they need to adopt this technology. So the entire discussions with the OEMs, with the Tier 1s right now is not about is it needed, it's more about how do we integrate it, when do we integrate it, to which models we are integrating it, how we train our perception as fast as possible to take the inputs from these sensors. So it's -- it became from a strategic question to a very tactical questions and planning, I would say, with the OEMs and the Tier 1s.

Got it. So just kind of switch gears a little bit, what do the unit economics look like on your current sensors? And then what is it going to take for you guys to reach breakeven in terms of volume or cost?

Yes. So if we look at the outlook of the revenue, the first revenue or significant revenue would come from the Chinese market down much faster to get and adopt this technology. And that's where we are going to see already in 2025, probably in the second half of 2025, revenue and, for sure, significant revenue in 2026. On top of it, even before that, we might see some revenue also from the non-retail automotive. Even before that, if it's trucks, if it's agriculture, if it's infrastructure, these are different use cases for the radar. And their level of maturity is a bit lower, and then the revenue is going to be even earlier. But the revenue there is not as high as the revenue that is expected from the private vehicle market.

Got it. And then so in terms of the regulatory tailwinds of the industry, we just saw the NHTSA approved a rating change for ABS systems in '29. How have you guys seen that affect your coding activity? And do you expect other regulatory tailwinds to drive demand for your products in the coming years?

Yes. So this is something that keeps the OEMs really, really busy because, for them, they need to guess, actually, what's going to be the regulation in 2030. Because they are building now the cars of 2028, 2029, and they need to make sure that they are going to be ready for the regulation. And everybody expects the regulation to put a lot of emphasis on safety and increasing the ability of the vehicle to actively protect the environment, not only the driver, not only the vehicle itself, but the entire environment. It's not only about avoiding the accident to protect the car. It's also avoiding the accident to protect the pedestrian that is crossing the road on the red light. So that's where the regulation is. And the OEMs, they understand it, and they try to predict it in advance because they need to get ready for that now. So this is very present dilemma that we see with the OEMs. And for them, one of the solution is, let's make sure that even if we don't have the software for that, let's make sure that the hardware is there. So we'll have the sensing capability and then gradually have the algorithms or the software that will be able to utilize it. So even if, in some of the cases, they still don't have the visible line of sight to how they are getting to this level of safety, they know that, from a hardware perspective, they need to put the infrastructure there. So they would be able to develop, over time, the entire solution.

Got it. And then you kind of touched on that a lot of your revenue that you see in the future is coming from China. Given that your product is very chip-heavy, sensitive in nature, do you guys see [ chipset trade wars ] between the U.S. and China hindering that revenue growth? Or you guys see yourselves kind of insulated because you're located outside of the U.S.?

Yes. So first, we are outside of the U.S., but even regardless of that, the fact that our chipset is very tailor-made for the radar, it's not a general-purpose CPU, okay? It's not that you can take it and use it to other users. And it's very safety-focused. It has no impact on the trade wars between U.S. and China. And we don't see any influence between them. The only place that does some -- that we see some impact is actually not on our part, it's more on the partnerships that we have, okay? So for example, we do see a lot of, I would say, trend and activity around Horizon Robotics, right? There are the competitors of the NVIDIAs, Qualcomm and Mobileye in China. And we are partnering with NVIDIA, we are partnering with Horizon Robotics, we are partnering with all the players in the ecosystem to make sure we are getting those entire systems over there. But for us, there is no direct impact on the -- those trade wars at all.

Got it. And then one more for me and then we'll turn it over to the investors. But if you look at your guys' share price and then some of the other like advanced driver systems share prices in the market right now, they're very depressed, by the way, in like '22 or '21. What do you think is going to be necessary for investors to start buying the active safety story again and get the share price to rerate higher?

Yes. I think the key word that you said is story, right? I think nobody is buying stories anymore. Everybody wants to see revenue. And everybody is fed up with stories, and they want to see revenue. I really hope that with the non-retail automotive and the revenue from China, we'll be able to show not only that there is a clear path to revenue and that the technology is really being adopted by OEMs. And we said that we are planning to have 4 OEMs adopting this technology this year, okay? One of them, we already announced. One of the top 10 OEMs, we already announced, that they decided to take this technology. We hope to announce 3 more by the end of this year. We are on track for that, and more of them in 2025. But I think, right now, the market wants to see revenue or path to revenue, and then see the core margins, that they are sustainable, okay? And I'm coming -- I'm personally coming from the school of Texas Instruments, selling -- running business in Texas Instruments, and core margin is always a key over there. And we'll -- I think that once the market sees that the revenue is coming -- real revenue is coming and the core margins are good, I think that's what we'll see the effect on the stock price.

Got it. Yes, we'll turn it over to -- I don't know if there's any investor questions in the audience.

None right now. Maybe you talked about the 4 OEMs -- the 4 design wins this year, and 3 more to come before the end of the year. What -- you hinted a bit on the revenue contribution from this -- the first OEM, like, I think, starting late next year and then into '26. Could you give us a sense of how these other design wins are -- when are they going to start to contribute? How should we think about the revenue inflection?

Yes. So the 4 OEMs that we are in very advanced stages with them. Let's talk about the -- so we talked about the first one, revenue is expected in 2025. This is an accelerated program. I think we mentioned it specifically in the press release that this is under a very accelerated program. The other 3 are not in an accelerated program. That means that we'll see revenue in 2026 from those. These are vehicles that will hit the road in -- their model year of 2027 -- 2028. That means that they will hit the roads in 2027, and we'll see revenue from this -- we'll start to see revenue from this in 2026 for at least those other 3 that we are discussing.

Got it. Got it. Is there anything that's changed over the last few months or quarters around just the pace of approvals and the design wins? Could you give us some insight there on if it's getting easier, it's getting tougher, it's getting faster? Any color there would be helpful and why that's happening?

Yes. So we do see the OEMs being very -- being slower to make decisions. I think it has a global effect, one, because of the question about the EV market, where is it heading? The growth that we saw on the EV market is -- does not continue. And initially, a lot of these new technologies, the vehicle, I would say, the path to the adoption of those was linked a lot to the EV market. And I think a lot of the OEMs right now are rethinking their strategy about the EV. And this slows down their decision-making on adopting new technologies in general. So frankly, with some of these opportunities, we thought that, by now, we would be with a clear decision. By now, we have clear indications, but still no -- a written contract that one of these that we -- that I talked that I can say they made a clear decision over there. So they are slower to make decisions, and we see it everywhere. I think we saw it also on the -- some of the reports and the analyst briefings that we saw from our competitors. We see the market slower to make their decisions. The interesting thing is that they are not delaying right now their plans for the safety and the level of autonomy getting to the Level 2+, Level 3. Right now, they are not delaying those plans, okay? But things are a bit more in a flux, I would say.

Understood. That's helpful. Just on the same vein, what are the biggest challenges are you facing as an imaging radar-based company to convince customers to adapt your product over other sensor technologies who may choose not to use radar or LIDAR or more camera based?

Yes. I think that right now, we are beyond the convincing stage already. I would say, in the past, especially when we were the only one that offers this high amount of channels, it was, okay, we understand why we need it, but how come you are the only one, okay? And this was a concern with the big -- with the very big OEMs. We are beyond that. I would say, in the last year, we are way beyond that point. Right now, the challenges are more about, how do you get the safety rating there? How do you get it into production? How do you make sure who is liable to false alarms? How the OEM is getting sensors that we can start recording data? So remember, we are looking at the radars integrated into the vehicles. But before that, the OEM needs to take the vehicles, record million miles with them to train the perception stock, okay? So that's what they are looking at. They are looking at doing this beginning of 2025, recording million miles with these type of technologies. And that's what we are busy with right now, and that's what we are working on these days.

Got it. Got it. One more for me. Are there like any strategic partnerships or collaborations that you're pursuing to maybe enhance technology or just market reach across OEMs or different regions? Any color on that would be helpful.

Yes. So first of all, the -- it's not the classical, I would say, Tier 1, Tier 2 relationship usually. We are working very closely with the OEMs, with the perception teams of the OEMs, to understand what they need in order for them to really take advantage of this technology and what are the disadvantages of the other sensors that they need to compensate using our technology. So there is a very close collaboration between us and the OEMs together with -- usually with the Tier 1. So that's one. I would say, if I look at the time spent, we are spending a lot of time together with the OEMs. I did mention NVIDIA as a -- whoever goes to the NVIDIA website, you will see drivers to our radar as part of the NVIDIA driver solution. We are working with Horizon Robotics in China, working with the Tier 1s. We are using the classical model of the automotive market. We are a Tier 2 chipset providers selling to the Tier 1 that sells to the OEM. So in that sense, we are following the classical model. And there's a lot of work with the different Tier 1s to get the product where -- so it really gets to the performance that the OEMs require. So overall, I think that's where we are busy at. More than that, we do have some partnership on the IP that we are using inside the chipset. We have a very close collaboration with GlobalFoundries. They are not only our fab, they are also our production line and supply chain management. They are doing the qualification, the AEC-Q100, the automotive qualification of our chipset. So we are not doing it ourselves. We are using a partner, an external partner, to do that. So the OEMs and the Tier 1s know that this is assured by a global company and certified and qualified by GlobalFoundries. So that's -- the industry that works on the -- on those solutions, includes also some startups that are working on perception stacks that we are also collaborating with.

Got it. Got it. Great. I think that's a great way to end. So thanks so much, Ram, for joining us, and thanks, everyone else, for being here.

Thank you.