Oh, LinkedIn is much better.
Oh, actually, yeah, I think LinkedIn is something I've really dealt with when COVID, when I couldn't travel. Now I can travel and talk to people.
You can talk to people, and you can still post your videos on LinkedIn.
Right, you're right.
All right.
I do that on Twitter, actually.
Twitter. It is looking at me. OK, all right. OK, thank you, everyone, as we continue the conference. I'm very pleased to have with us Omer Keilaf, the founder and CEO of Innoviz, Innoviz's leading LiDAR player. You may be one of only a few LiDAR players.
Just a few, yeah.
A leading player in the field of LiDAR. We're going to go through a series of fireside chat questions. Anyone that has questions, you can email my colleague JR Young, J-R-Y-O-U-N-G, barclays.com. He can ask your questions. I want to start, Omer, maybe you could just talk us, zoom out, and let's look at, because Innoviz, from a product perspective, has been on a journey. Maybe you could just give us sort of an overview of where you've been the last couple of years, where you are on the product roadmap, and how everything has evolved. I like the fact that you brought a little bit of show and tell.
Yeah, yeah. We are working very close with the car manufacturers. We understand where they are trying to get: level 3 highway, level 4 robotaxis. At every point of time, you see that different OEMs have different ideas on where the LiDAR should be mounted, whether it's in the grille, on the roof, behind the windshield, the speed of the level 3 they want to support. I think you realize that there are many perspectives that you need to embed in the sensor. I'll give you an example. When an entrepreneur decides to develop a LiDAR, the last thing that he's thinking about when LiDAR is considered one of the most challenging technologies in the world is how it acts after the car bumped into a puddle of mud. People, five years ago, six years ago, were still focused on range, resolution, and cost.
You learn that eventually the car needs to drive driverless, even when it drives over a puddle of mud and the sensor is now dirty. You realize that if it does not work well in those conditions, it is all meaningless, right? It is like you have done all of this effort and developed the best technology in the world, but it only works when the car is clean. I think one of the advantages that we had, Innoviz, working with BMW, our first customer, was actually seeing those, I would say, more challenging conditions. We have baked into our second generation and our third generation all of those learnings. We show customers the way that our sensor is working, even when we throw mud at it. I always get the question of, what is the magic? Like, what are you doing here that I should know?
That's, by the way, a different topic, is the trust that OEMs have on LiDAR companies right now. This is, I think, one of the key advantages that we have. We really have our scars on our backs. And we have a product that can actually work in a real level 4, level 5, because it's a very resilient automotive grade. It meets the highest requirements of range and resolution. Now we introduced the Innoviz 3, which we just talked about in our last earnings. I hope you can agree it's a very small design. This is a compilation of many requirements we got for customers that want to embed it behind the windshield. This is where OEMs are now looking to deploy. Everything I can talk about LiDAR is actually driven from the design of the vehicle.
It's like the number one rule: the car needs to still look good. For several years, the engineers were somehow able to push away the designers and tell them, yeah, but it still needs to work. Putting a LiDAR behind the windshield adds two levels of complexities on top. As we said, LiDARs were a challenge by itself before you brought it behind the windshield. We are at a point where our new technology, our next generation, is allowing us to place it behind the windshield, where other LiDARs have failed. I believe this will be the leading, I would say, position for the front-looking LiDAR. Of course, when you talk about level 4, it does not end there. You do not have windshields around the vehicle. Even there, we have a very interesting solution, which we have not yet talked about.
Yeah, in a way, we are helping the OEMs to design their future vehicles based on the capabilities that we are able to show them. Because designers can imagine anything they want, right? They can, oh, I want the LiDAR somewhere stuck here in this size. But they can't really know if it's doable or not and whether it will meet the requirements. In a way, we are actually inviting OEMs to see our capabilities and design their future cars, concepts based on the capabilities that we are able to provide them.
This is a product that we've designed for level 3 highway or front-looking sensors. When you talk about level 4 or level 3 urban, which this is what would be next, where a person that buys a car doesn't want it to drive autonomously only on the highway. He wants it to drive autonomously also in other areas. It needs to be cheap, so affordable, and designed well, right? You cannot put nine LiDARs in a privately owned vehicle. As much as I would love to sell nine LiDARs per vehicle, it probably will not work.
Maybe in China.
Maybe in China. Yeah, if it's empty, then yes. I've seen already cars where the LiDAR was actually in an empty box. That was funny. Yeah, it became like a luxury thing to have a LiDAR on it. In any case, if you want to have a level 3 urban or level 4 in a privately owned vehicle, you cannot assume that it will be able to absorb so many sensors around it. It is actually on us to understand those challenges and design the future technologies of LiDARs that will solve it. It will not come from the car maker. It will come from us. These are the things that we are doing. Starting from LiDARs that were this big, right, to sensors that are this small and will be smaller and more affordable and better, to a point that every car would have full autonomous capabilities in an affordable manner.
OK, a lot to unpack there. Maybe we can just chip away at this. I want to just ask on the LiDAR opportunity more broadly. I think there was a narrative over the years that the LiDAR opportunity was really on this scale ADAS, L2 plus, L3, what have you, as an advanced driver assist feature. It does not seem like that has played out in the industry as broadly. Now it seems like we are pushing more toward really the LiDAR opportunity being on L4, fully autonomous, which is getting sparked interest. Maybe you could just walk us through the LiDAR opportunity. Where is the market emphasis and/or your emphasis from a sort of L2 plus scale ADAS versus L4?
Sure. I actually like to point at China as kind of a crystal ball, right? Because when you see how the industry in China has evolved to host many LiDARs on cars, I can see how that can be a good way to think how the future might look like in other areas around the world. Now, part of the reason that LiDARs were used in China was because they had premature software. They used LiDARs to overcome the challenges of using just a camera base for a level 2 system. They used the LiDAR to add safety at a shorter time with the software, which is possibly more mediocre. Now the market is shifting towards level 3 in China. I think that will also push the Western OEM. Level 4, I actually see it as I think there is a gold rush right now.
This is what I see. I'll explain how I think it's like that. Eventually, in any big city, you have a certain limit of number of cars that will be allowed to give service because of congestion. If any company would now push in hundreds of thousands of cars, nobody will make money. The traffic would be horrible. Nobody would allow it. There will be a certain cap. There will be an early mover advantage. I think that the fact that Waymo is growing their business right now, and I'm sure that you are seeing it, is pushing many players to move faster now into the level 4. They understand that it needs to be scalable, automotive grade.
We are seeing a lot of success right now because when you look on opportunities to use a LiDAR which is automotive grade in the time frame that these programs are targeted, there is no alternative. It is kind of mind-blowing. When you think about it, the only LiDAR that reached Level 3 so far in the market was either through Mercedes several years ago, but it was using a very low resolution LiDAR. Due to that, it was providing very low availability of the system. The second one is Innoviz. We are going to SOP with Volkswagen next year. We are the only LiDAR that is going to be available for SOP in the next probably two or three years. That is a big advantage that we have. I see actually quite a lot of a sense of urgency from the Level 4 players.
It's also coming from the truck market. We just announced that we were selected with a big truck company. We see it also in other areas, such as tractors and some other domains. I think that I would still say that level 2 would transition to level 3, I think probably towards 2028, when you see a bigger kind of volume. It will start at the end of 2027 and then probably would start to ramp up. I don't have a doubt in my mind that it will. I think that the fact that we are now going to enable level 3 urban would increase the value to the customer because it's not only highway. I think that people would want it. Because the fact that you are using driver people as supervisors is, I think, a very poor product for humans. Because being passively responsible, it's a very horrible and tough position to be at. I think people are really looking for the point where they don't need to do it any longer.
If we--and you said level 4 is like a gold rush right now. I think that when people were doing, let's call it like the napkin math or back of the envelope math, OK, you could see L2 plus penetration going from X to Y, ASP even at $500, what have you. You could see a very large opportunity. Once that penetration was cut, it sort of shrunk the opportunity. I think we know with level 4, the issue you're facing is the fleet is much lower, but the ASPs are much higher.
The multiples are higher, right? I have nine LiDARs per vehicle.
Sure, sure, right? Overall CPV is higher. Can you still, with a Level 4 emphasis in the near term, achieve meaningful revenue growth? Is there enough revenue opportunity on a smaller base of vehicles, but with a much higher CPV, to generate a type of meaningful revenue growth?
Yes. I'll say where it can come from. Other than robotaxis, level 4 is also trucks. Also, the count of LiDARs there is pretty high. You have buses and shuttles that are going to be—there is a huge demand. By the way, I just read in Israel they are going to bring foreign drivers because of the lack of availability of bus drivers. There is a very obvious problem of professional drivers everywhere around the world. If you think about a robotaxi task compared to an autonomous bus, I hope you agree with me that training a car to drive in a predefined route is probably easier. It's much easier to scale. I have no doubt that these shuttles, such as the ones that we are part of, are going to be a growing market.
The multiples of LiDARs around it is going to be substantial. I do not really see it just as level 4. There are many other markets that we are—agriculture, which I was referring to earlier, companies like Caterpillar and John Deere, and many tractors that are trying to improve, I would say, the efficiency. I would say lowering the human intervention in several tasks is really growing. Of course, automotive. We are seeing today a huge demand of LiDARs. To be honest, Innoviz just started going into the non-automotive, I think, five months ago when we announced the InnovizSmart.
I am overwhelmed, really overwhelmed, with the value that we are going to add there. I will give you an example that I think will—I was surprised. I used any opportunity because I am so surprised. I will give you a very simple task: security, perimeter security. You know, in Israel, there is a lot of discussion about the fake feel of security when you put so many smart systems and eventually you don't trust them. I've learned that just the best-in-class solution yet today is far from being safe.
It's all vision-based.
I'm talking about a solution that is used all around the world. We were in a conference in GS6. It's like a big conference for security. It's the same security solution that is offered by everyone. We were positioned next to a solution that was just recently picked in an RFQ for perimeter security villages. The team there brought a talent, I would say, educated team that knows how to penetrate fences. With the other solution, they were able to penetrate four out of 10 times. They were able to get to the fence and cut it. With our system, it was zero. They were shocked. The funny thing is that if I tell you how easy it is to do it, you'll be shocked. Because all you need to do—it's not like a James Bond. You need to go and put music and flip-flops.
You just need to go slow enough or go between the trees. I saw the videos of how the team has managed to penetrate those systems. I was like laughing. Like, seriously, this is still the best-in-class solution that you're using for perimeter security? When you're using a LiDAR, you cannot avoid light. You cannot. You're just seeing. To me, it was really surprising. I opened it. I expect that all of those systems that were expected to be installed with that unit will be replaced with ours.
This is what we're heading. This is just one example. I see the same in ITS. I see the same really in tolls. I'm quite surprised to see how technology hasn't really evolved in so many areas. When I come and show our LiDAR in these events, I'm looked at as if I landed from the moon, really. I show them the LiDAR and the point cloud and the 3D. People think that I just landed from some other planet. I think that there is a big opportunity there. The ASPs are significantly higher than automotive.
Maybe just to peel that one back, can you just outline for us, remind us where ASP or CPV, however you want to define it, between what was a scale ADAS vehicle, which I'm guessing was probably in the $500- $1,000 versus a full AV, so let's call it, for instance, one of the ID. Buzzes that we've seen that you're specked on versus a non-automotive solution or perhaps versus a tractor, how do the ASP or CPVs differ?
Yeah. We always look on the cost of the product, the total product, understanding the sensitivity of the pricing. When you talk about a tractor that costs $350,000, you can imagine that they are probably less sensitive on the piece price of the LiDAR if it gives their customer an edge, right? Therefore, we price differently. If you talk about a low-cost car or a premium vehicle or a vehicle that the total cost of the platform is high, we are trying to make, I would say, our value in the right position. Of course, the sensitivity of cost is very high to everyone. We are also working towards new generations to improve. Generally, the piece price, if you talk about non-automotive, it can be up to $10,000 per device, right? When you talk about high-volume automotive, it can be $500.
It also depends on time. I mean, the first year of SOP, where the ramp is still occurring, the volume is lower, the prices are higher. It's not $500. If the car has nine LiDARs, like we have with the ID. Buzz or the platforms that we are working with Mobileye or the truck company that we were just awarded, we also have multiple sensors there, then obviously, you can multiply it by nine. You get the general.
Nine times $500 or nine times $1,000?
More than $500. For automotive, we price between $500-$850. It depends on volume and time. Yeah, it's a value product.
We're talking about $6,000-$7,000 of content in a vehicle.
Yeah.
OK. What about, let's call it, the commercial justification? Because one of the early problems that ADAS faced, and it's not just LiDAR, and we saw this with I can think of other companies that have been out there, that they would get these robotaxi bids, and they would spend money on a program. There's a lot of upfront validation cost to meet the specs of each program. Then it's on a very small number of vehicles. That's a recipe for losing a lot of money. That was always the whole justification for why scale ADAS. If you're on a program with 100,000-200,000 vehicles, all of a sudden, you have an upfront spend, but you can advertise over a wide set of vehicles. With a presumably smaller set of volumes, how do you justify, from a commercial standpoint, expanding to these other areas where the scale may not be there? Or is there not that type of validation expense?
OK. So you're asking about the commercial motivation of car makers to work on those programs? This is the question?
No, I'm asking.
Or our commercial.
How do you justify it from a P&L standpoint if the problem is that each program you take on has a lot of upfront spend, but the volume is high?
No, no. First, from our point of view, OK, there are many overlaps. I mean, basically, today, we are selling the Innoviz to all of our customers. It's the same product, right? There are many overlaps. They are all coming from the same production line, from the same testing facilities. There are some, I would say, freedom of design that we are allowing at the outer shelf, window slope, the connector, the interface, etc . Therefore, our ability to benefit from additional programs in terms of offsetting our spending is really high.
We are not allowing a lot of creativity on the customer side. By the way, it's also for their benefit because they get a more mature product, a shorter time to the market. Yeah, we understand that eventually that. The OEMs understand that they are not paying for the entire cost of development. They are participating in it. Therefore, they do not get ownership of the technology because they are not paying for the entire program. It is offsetting our spending.
OK. So there's not necessarily that risk of sort of for every program, even for small volume programs, the validation expense.
No, no. I mean, look, we announced just last earnings that we have over $110 million of NREs that we've booked that are going to be paid between this year and next year, primarily, and also in 2027. That's a really high number that substantially offsets our spending. This comes on top of everything other that we are selling as products and, of course, new programs that we expect to be awarded to in the future. There are several RFQs that we are participating in now.
OK. Maybe you can talk about the competitive dynamics because I think you've mentioned the LiDAR market. It is consolidating. You see a winner takes more scenario. Most. I know we've seen some pressure that some of the other players. What is the competitive set right now? How is it looking? When you're going to RFQs, how many other serious players are you going up against?
Not many. Maybe I want to say one, but I'm not really even sure who. Depends on the need, OK? What we've learned over the last couple of years is that it depends on the application, level 3 and level 4. I would say that there are segments that we don't have competitions. Because, look, the segment of LiDARs is split between long range and short range. On the short range, we were super successful recently because, actually, there's no one offering short range. It's one of the key advantages of our technology, its ability to be very flexible, to allow us to provide different configurations. Innoviz 2 was a platform that allowed us to generate multiple variants of it, one that was a fit to Audi, and one a fit for Volkswagen, and one a fit for Mobileye. It's all modifications that were done very easily.
On top of it, we've provided a short range that was set for Volkswagen, and then another one that was for the truck company. They are all based on the same technology base. The optical backbone is identical. Nothing changed. The only difference that we allowed is related to the scanning pattern or the outer shell, etc . We are trying to always be very conservative. This is something that our competitors are incapable of doing. You have a company that is only focused on automotive, so irrelevant. You have a company that is focused on very niche highway truck, very narrow field of view, which is irrelevant for, I would say, 99% of the market. That's it. Surprising, right? It sounds like, how could it be? That is the situation. Of course, there are many Chinese LiDAR companies with good products. They have their own issues, right? That is where we are.
OK. On that, right, OK, I think we've known there is this bifurcation between China and the West. We know that China has had much more LiDAR uptake. China, in general, on everything, has a cost lead, OK? If we take a player like Hesai, and we do not need to talk about anyone specifically. The opportunity for Chinese LiDAR players to sell into the West, is that possible?
Look, eventually, it's a political discussion, right? I'm not a politician. What I can tell you is that LiDARs are security-sensitive sensors. From a national security point of view, a LiDAR is a mapping instrument. When you think about what you are capable of doing by just collecting data, enormous data, when traveling in a car, you can imagine why people are not really happy about the idea that you will use a technology that possibly is somehow involved with government support. You see that DOD is intervening. There are many Congress people that are intervening. I would say that US OEMs will take a huge, huge reputation risk if they will work with Chinese-based technology. That's something that I think will not happen. When you think about it, the US OEMs, they are selling many cars to the government, to the army.
You think that the army will not want an autonomous car if they are developing? Of course, they will. I mean, I would assume that a US company that develops an autonomous vehicle, one of their key segments would probably go to defense and army. Do you think that they will use a technology that might be forbidden by the DOD? Probably not, right? They will not develop something specifically for the army and for the consumer environment. I think from that perspective, that is, I would say, how I see it. Again, I am not a politician. I do not set those rules. I follow the news. I think just yesterday, there were some new adjustments for two new LiDAR companies that are now from China that are going to be included in a kind of list of LiDARs not to work with.
I would say the dynamics are in the direction that I would say probably not a risk. And not only automotive, also non-automotive. I can tell you that we are talking with different municipals that don't want to have a sensor that was developed in areas that might create a problem for them. I think that even from that perspective, when you look on the non-automotive market, it's Ouster and us, right? I mean, it's kind of like, you know, five years ago, you would say, oh, there are so many LiDAR companies, right? It's not the case any longer.
OK. If we add all that up in your commercial path specifically the next few years, can you just walk us through? You started on BMW, on I7.
I7, yeah.
You've won a program with Volkswagen. Can you just sort of add it all up for us over the next few years? What is driving sort of the path to commercialization and scale for you and just sort of the specific programs you have? This new program you have on trucking, sort of the timing and any parameters you can provide?
Yeah. OK, sure. BMW was our first customer. Following that, we were awarded by Volkswagen for a Level 3 program for the long range LiDAR. Following that, they decided to use our long range LiDAR for the ID. Buzz for the Level 4.
With Mobileye.
With Mobileye. In a way, they asked us to develop a short range LiDAR. They had to replace, displace the short range LiDARs they were using for reasons that some of that we've talked just now. This was actually an unplanned tour to develop a short range LiDAR, realizing that we will eventually be the only LiDAR company in the West offering a short range LiDAR.
It became a huge success. Once we were on that program, Mobileye decided to stop their internal development for LiDAR and use our LiDAR sets, long range and short range, to all of their other customers. In terms of the rollout of programs, we have the ID. Buzz Level 4. Following that, there are several Mobileye platform-based customers, such as Verne and Holon, that are going to come out following that. There is a program with Audi Level 3, which are using Innoviz 2, that will follow those.
That's Mobileye Chauffeur.
That's the Mobileye Chauffeur. There is the truck program that we were just awarded that will be in a similar time frame. We didn't say at the time, and since we didn't yet name the customer, we hope to do so, by the way, very soon. I just came back from, you know, they are U.S. They are going to operate in the U.S. Those are the customers that I think we've already shared. As I kind of hinted, there are several other Level 4 activities that I see now. Many of them are changing their sensor suite because many of them have developed their sensor suite based on products that were available and for geopolitical reasons or just the fact that LiDAR companies they've used are going out of business, etc . There are many opportunities now at our table that I hope during 2026 we probably will share more. They have similar target timelines. I would say 2027, 2028, that is the automotive space for you.
I guess questions.
There is the top five OEM that we talked about, level 3, yeah.
Just maybe on that issue, I appreciate your point about the politics and security. Just outside of that, how is the cost structure of the product that you've developed? Can you talk about your cost curve for the next three, four, five years, whatever time frame you're comfortable with? Then how you do compete in some regions, at least, where you will be competing against the Chinese? How can you either better technology quality or even close the cost gap?
Yeah, sure. We are now offering Innoviz 2, our second generation. The cost saving between the first generation and the second generation was, I think, 70% more. We had a huge cost reduction. In terms of performance improvement, it was 50 times better, OK? Innoviz 2 was a revolution. Now we are introducing Innoviz 3. We're going to show it working at CES for the first time. It's 60% lower in volume. It will also have an impact, obviously, on cost saving of using less material, less electronics, less size of optics, etc . We talked about prices of $500-$800, so you can understand where this is heading. Eventually, LiDAR technology doesn't—there's nothing fundamentally expensive about it. When you think about it, there is a laser diode, very much like you have in any lead environment.
You have a silicon detector that is actually smaller than a camera in terms of the footprint of the die. You have a lot of electronics around it that today have not gone through yet the vertical integration to a chipset that you put three chips and a flex. These are all industrialization steps that will be done over volume, right? Because now, obviously, our BOM is reducing by technology. We have not done the vertical integration. Like in iPhone, you had multiple chips that eventually were diode stacked of dice, right? If you open the box, you will still see several boards inside. It will not be like that in the future. Eventually, it will be one board with a flex, and that is it. The assembly will be easier. These are things that we are not doing yet because the technology or the market has not stabilized yet.
You make those investments when you know that you will not need to change the technology any longer. I will not invest $40 million to make the BOM so significantly cheaper if I know that in two years I need to do a new design because this customer wants this, and that customer wants this color. This is a very standard process where technologies go through these cycles. We are now in our third generation. I think that this technology is obviously going to produce a significant cost reduction. I hinted also about Level 4 technologies. I think there are still cycles that we are developing. Eventually, there is nothing fundamentally that LiDARs would not be as cheap as cameras. I know it's kind of like people can't really grasp it. There is no reason it shouldn't.
OK. We'll leave it there. Thank you so much.
Thank you.