Good morning, everyone, and welcome to day two of Rosenblatt Securities' fifth annual Age of AI Technology Summit. My name is Kevin Garrigan, and I am one of the semiconductor analysts here at Rosenblatt. We are pleased to have with us Innoviz's CFO, Eldar Cegla, for this fireside chat. We have a buy rating on Innoviz with a $4 price target, and we are bullish on the company because of their leadership in the automotive LIDAR space, design, traction, and inflection coming in 2026. Throughout the fireside chat, we're going to ask for any questions from the audience. To ask a question, you can click on the quote bubble in the graphic on the top right-hand of your corner. I'll then read the questions to Eldar. With that, thank you, Eldar, and great to see you again.
Thank you for having me. A pleasure always.
For anyone listening that may not know the Innoviz story, I thought we'd just start out with a brief overview of the company and why does Innoviz win in the market.
Yeah, sure. I think when we started, we started back in 2016. Our main thought was that there is this new emerging technology, autonomous driving, and it's missing a very important part. In order to enable autonomous driving, in order to have the driver really disengage from the driving, the system needs more sensing capability. It had the camera, it had the radar, but still, in order to disengage the driver and have full redundancy for the system, you needed a third type of sensor, and it was the LIDAR. LIDAR is basically a 3D sensor or a 3D camera, and this is how we set ourselves to this mission.
On top of it, and very importantly, was the AI software, which basically takes the data and turns it into something meaningful for the rest of the software stack in order to have these features available, in order to have driving decisions. I would say that these kinds of AI technology that were emerging were actually the enabler on top of the hardware that enabled this technology and these features to come into market, not only on the sensor side, but also on the software, the driving decision platform side. Once we started to develop these technology, both on the software and the hardware side, we gained the first traction with the first customer, BMW.
We were a very young company, a very young startup, so BMW requested to have another partner, a Tier One partner, to some extent buffer between BMW, the OEM, the car maker, to the technology provider, and we partnered with Magna back then. During this process of bringing this technology into the market and the BMW vehicle rolled out last year, you can actually buy this platform or this car, i7 cars, as of last year. During the process, the development process, we gained a lot of knowledge, a lot of experience of how to run such an automotive sort of technology project. We understood that it is very important for the OEM, this direct connection with the technology provider because of the different flexibilities that we can provide, the different tasks that you take upon yourself in qualifying the technology.
We decided following the BMW project, as I said, is already SOP, it's already in production, we decided to turn and become a Tier One. Our second client was Volkswagen, and when we competed on the Volkswagen programs, we approached them and offered them to be a direct supplier, give all the benefits of a Tier One in terms of flexibility and the services that we can provide based on the experience we gained with BMW. On the other hand, offer a better cost structure because basically you are eliminating the middle Tier One, which adds to the margin stack.
It's something that Volkswagen, after a very rigorous audit that they did to Innoviz, that took more than a year, basically, in which they audited every aspect of the company, not only on the technology side, but how we manage the company, risk management, quality, production, and so on and so forth, they have qualified us as Tier One. Since then, the industry has accepted Innoviz as a Tier One. Today, when we are approaching project, basically we are approaching it as a Tier One. I think maybe the last step, the latest step that we took is in two aspects. One is we started, when we started the company, we were more focused on level 3 application, consumer applications, because we thought it's the biggest market. Around 90 million vehicles are being sold every year to consumers. We thought this is the biggest opportunity.
Over time, and together with projects, specifically the Volkswagen project, it turned out that level four, not consumer, but commercial applications, like robotaxis, shuttles, buses, and so on, are gaining traction because the technology is maturing. Mobileye, NVIDIA have put in place a level four platform for commercial applications, which are maturing. On one hand, our technology matured as well with the capability of maturing our production capacity. All this came together. We are today offering both level three applications, mainly for consumer, and level four for commercial application, which basically doubled our business opportunity. Today we are looking at two main goals. A, win additional programs with our software and hardware. B is to ramp our production line capacity, which basically we are using contract manufacturer, Fabrinet.
It's ramping up as we speak and will be ready before the SOPs, the main SOPs that are planned already for next year for the Innoviz team. This is, I think, a short introduction to Innoviz.
Wow. I appreciate that overview. Sticking with the L4 and robotaxis, they've become a big buzzword recently, and we're seeing more and more investment. How do you kind of see that market playing out? I know you talked about it's becoming more mature. Do you see it kind of becoming full scale before kind of L3 does?
First of all, I think there are already Level 3 consumer applications that are rolling into the market, mainly on the high-end vehicles like BMW, Volkswagen, Mercedes, and maybe soon to come other OEMs as well. Having said that, we learned that, and we're seeing it happening as we speak, that the Level 4 commercial applications are getting a lot of traction. The reason is, I think, from two directions. A, the technology matured, but the business model makes a lot of sense because basically you have here a platform that eliminates the professional driver, like in a taxi or a bus or a shuttle.
It is very easy to justify putting a lot of sensors and technologies and softwares into such a vehicle and having it ride all along, in addition to the fact that it is getting ever more hard to get professional drivers for these kinds of applications, including trucks, by the way. This makes a lot of sense. Technology maturing and a good business model. On top of it, there is some, I think, boost from the government. I think the new administration has allowed to deploy a greater number of vehicles per year. It was restricted until this government, the Trump administration, came into place. I think the restriction caused it to, to some extent, slow down a little bit the penetration. Once these restrictions were lifted, we feel there is a certain land grab war between the different platforms.
You hear announcements from Uber, you hear announcements from Lyft, obviously Waymo, which is very far ahead as well at this. We see a lot of acceleration. In fact, for Innoviz, the next SOP, next platforms that will be deployed are the Level 4. The Level 4 applications are planned for next year, the SOP start of production. There will be already vehicles tested this year, this year and the beginning of next year. As of the second half of next year, they will be deployed, which means for Innoviz, a sort of acceleration and additional volumes that will be expected already in 2026. Hopefully this will even grow faster in 2027. On the Level 3 front for consumer application, we see it happening or the next deployments will happen as of 2027 and beyond.
Actually the Level 4 we are expecting as the next big step for Innoviz, the next big volumes to be provided to the market.
That makes sense. Can you just kind of frame the opportunity for Innoviz between L3 and L4, ASPs, volumes on cars?
Sure. Sure. The Level 3 applications that we see in front of us, as I said, the consumer opportunities are deploying one LIDAR per vehicle. The Level 4 commercial applications are deploying nine LIDAR per vehicle, three long range and six short range LIDARs, all provided by Innoviz. At ASPs, roughly speaking, for these kinds of applications at the earlier year, let's say between $600-$800, then for the commercial applications, it means for Innoviz, on average, nine LIDARs times the ASPs, let's say a few thousands of dollars, $6,000, $7,000 per vehicle. It is very meaningful. Even if the expected volumes of vehicles is lower than the Level 3 applications, the consumer application, still because of the multiplication of the number of LIDARs, the content per vehicle is very meaningful. Level 3 is one LIDAR, a few hundreds of dollars.
Level 4 commercial application is nine LIDARs, a few thousand dollars. Very meaningful, even if the volumes are lower.
That makes sense. Yeah. Good content for Innoviz. Relating it back to AI, how have you guys seen automotive companies trying to incorporate AI or machine learning into their platforms, into their cars? I mean, has it kind of helped expedite any processes at all or not really?
I think without AI and without this, let's say these platforms, these algorithms, the new algorithms, you don't have autonomous driving. It's simply impossible. This is a significant breakthrough. It goes in every level. When Innoviz started, it was very clear and it's one of our differentiators that we are not only providing the hardware, the LIDAR, the 3D, the missing component of a 3D sensor that is required in order to reach autonomous feature, you have to put on top of it some AI deep learning algorithms that take the data and turn it into something meaningful that the driving decision software or platform can use in an effective way in order to have driving decision. This was very meaningful.
As I said, it started from the sensor level, but then you have additional components, software components along the way that are also AI driven, let's say, like the sensor fusion. You can do sensor fusion, low levels fusion, high levels fusion, depending on the architecture. Even when you go to the driving decision, there is a lot of AI, both, let's say, traditional, classical AI, which starts very low at the very raw data, then maybe the perception software, which is taking the output and turns it into something meaningful, object detection, classification, tracking, drivable areas, can drive above, under, and other additional important features, and goes up to the stack. Without, and going back to the opening statement, without AI, you don't have autonomous driving. It's fundamental, and it's fundamental in every level in the system.
Very, very important for moving forward, I will say.
Yeah.
Okay. Perfect. I'll stop here real quick and see if we have any questions from the audience. Again, if you want to ask a question, click on the quote bubble on the graphic on the top right hand of your screen, and then I'll read out to Eldar. Okay. I guess one other question that I kind of had, your recent wins and their importance, especially with the platform players, so NVIDIA, Mobileye, and even Qualcomm, can you just talk about how important those are for helping you guys out?
Yeah. I think it's helpful for us in more than one aspect. It starts with that, with the fact that I think these platforms are an important key in order to have these solutions in an effective way, because putting a platform in place is a costly, very complex task, and not every OEM is up to the task. I would say that maybe two, three years ago, the industry thought, okay, everybody will have a project, everybody had some kind of autonomous driving project, and they tried to get to the solution themselves with partners, without the partners. The problem is it's a very difficult problem to crack. You need a lot of time, cover a lot of edge cases, a lot of engineers, a lot of costs. In order to reach for a safe, qualified system, it takes a lot.
I guess in the past, many of the projects failed. Today, basically, you have a few very strong players like Mobileye, like NVIDIA, that took this task upon themselves and are basically offering a solution that saves time and saves risk and cost to the OEM. Basically, it is an important maturity step for the industry today when an OEM needs to go, wants to go into these kinds, and they are seeking to offer these kinds of features. They have already the main building blocks ready for them. They have the sensor suite ready, like a LIDAR, Innoviz LIDAR, which is ready and mature and already deployed, the software in place, the platform themselves.
Being integrated into the Mobileye platform, both on the Level 3, the Chauffeur platform, or the Drive platform, being integrated and being a sort of, I would even dare to say, a sort of a reference design on the NVIDIA level helps Innoviz to win business because we can say, okay, we are already working with these platforms or integration compatible to these. The OEM saves time and saves resources. He does not have to start all over, and he can make a decision and have his solution ready on time and at a reasonable cost. You see that the whole technology, all the components are maturing and are ready. I think today we are at, if a few years ago, we were in a certain hype phase.
Today we are in a certain maturity phase and production phase, which is much better for everybody, all stakeholders in the industry.
That makes sense. Your partnership with Mobileye, you guys are the sole provider that if they get a design win and the customer says, hey, I want to use a LIDAR, they would go specifically to you guys compared to NVIDIA, who has a couple of different partnerships, but you may be trying to do the same thing with them. Am I thinking about that right?
Yeah. So I would say the following. NVIDIA, by the way, has more than one platform. They have the Orin and they have the drive, the Hyperion drive platform. And it's more to how this industry is working. Usually, Mobileye is offering a closed box solution. It doesn't mean that some OEM cannot decide that for his own reason, he wants to use different sensors, not only Innoviz, different camera, different radar, but the cost of doing this and the risk of doing this and the time it takes in order to qualify a new platform with new components.
Also on the NVIDIA side, even though it's, let's say it's more open, eventually at the end of the day, the de facto logic is we go with what is already available, what was already qualified, the technology that is already deployed and has gained some experience in the market because we have always to think what this technology does. This technology drives cars without the attention of the driver. We cannot have that this car will do an accident, run over people, and so on. This is a mission critical, functional, safe component or feature. Once something was qualified, usually the OEMs tend to stick to what's working and not try to spend more money, more time, and a lot of risk in qualifying different technologies.
That makes sense. It looks like you guys are making a little bit of a bigger push into the non-automotive market. How do you view that kind of market developing? I mean, do you guys feel like you can be a winner in that market as well?
Yeah. So I think it has to do something in the fundamentals that the way Innoviz thinks. Innoviz always, I would say, took the longer, more strategic route to reach its goals simply for the fact because there are no easy shortcuts. We decided when we started strategically to focus on the automotive market because we thought this is the biggest opportunity, but it came with the price. It's much more difficult to provide for the automotive space. You need to meet very stringent qualification and requirements. You also need to meet a very aggressive price point, ASP. We decided to focus there. Once we reach maturity in automotive and now we are ramping up our capacity, we are starting to look at other opportunities as well because we can provide for them.
There are already applications that over time, other applications like in security, like in smart cities, smart junctions, smart infrastructure, industrial that are already using LIDARs and they are looking for better and cheaper or more cost-effective solutions. Innoviz can today provide them because we have an automotive-grade LIDAR, which is very good, top performance. On the other side, it has the cost-effectiveness of an automotive solution. We are looking to tap into applications that already understood the value of LIDARs and already use LIDARs. It's relatively, basically they approach us. It's not something that the whole company is now set to pursue, but rather have very few partners like big integrators that have already projects that are requiring LIDARs and stream or offer our solution into these ready-made applications or projects.
How does the LIDAR sensor differ in the non-automotive market compared to automotive? I mean, can you use the same hardware for both markets? You just change up the software? Or how does that kind of work?
At least our application is a very simple answer. Yes. We are in many cases providing more than is required by the industry and in a much more effective price point. It is very easy. The LIDAR, the performance is very high in many aspects. We learned also that most applications are requiring exactly the sort of LIDARs that we are providing. We do not see any, I would say, barrier for us to enter into these kinds of applications.
Yeah. Yeah. I know yesterday you guys had announced the Innoviz Smart. I think smart cities are a very interesting use case for non-automotive LIDARs. You're seeing a lot of governments these days just talk about AI and incorporating them into anything and everywhere possible. I guess on smart cities specifically, what are the big kind of use cases you guys are seeing that you think LIDAR could be good for?
Yeah. Since this, basically you can think of the LIDAR as a 3D camera. Basically, wherever you have a camera, you might think you could use a LIDAR. A LIDAR has a few benefits. A, it is an active sensor, so it sees at night and it's less prone to the sunlight, getting direct sunlight, so it doesn't get blinded and so on. In addition to that, it provides instead of 2D data, it provides 3D data, which is much more easy to process. In any scenario that, for instance, smart junctions to monitor and control traffic, it's very clear to understand how the traffic flows because you get the 3D data or restricted areas. It's very easy to monitor areas for security or other reasons that you want to restrict because it's a 3D sensor.
Definitely in smart monitoring of crowds like in airports or in any other such application or security smart fences like the U.S., there is always the talk about the fence, the smart fence between the U.S. and Mexico. All these kinds of applications can benefit from a 3D sensor that can provide much more data points than a 2D sensor and can be more resilient to different environmental conditions.
Okay. Yeah. That makes a ton of sense. In your prepared remarks, you talked about Fabrinet and how important them being a contract manufacturer for you guys is. Can you just talk about the importance of them to your kind of growth trajectory?
Yeah, of course. First of all, we are a Tier One, so we are responsible for the production. In essence, we plan the production line, the process. We are responsible for the testers and some of the machines that assemble a LIDAR, the optics, and so on. Using a contract manufacturer gives us all the benefits of using a contract manufacturer in terms of relatively staying CapEx light, having a partner that has the capacity or the experience in automotive production and meeting automotive standards. Of course, having facilities not only in one place gives us the flexibility based on our customer needs. Fabrinet is deployed in many places in the world, mainly in the U.S. and Asia. This gives you full flexibility to give the required or answer the required needs of our customers.
The importance, obviously, is once this production line is up and running, we have the capacity to provide for our clients, mainly the automotive clients, but also outside the automotive space. Another important aspect of that is that Fabrinet is a partner that can take some of the burden of the working capital needs as we ramp over time. Having a strong partner that can take some of the burden of that is also important. It answers many needs of the company.
Not having to spend on CapEx for something like that definitely helps the cause. Relating everything back to AI, I mean, how are you guys kind of incorporating AI into Innoviz, Innoviz the business? On kind of the perception software side, again, you talked about it a little bit before, but how are you guys using AI and machine learning for your perception software?
Sure. Innoviz is providing a software stack which has many elements in it, starting with very low-level, I would say, sort of image processing, which is 3D image processing. The output of a LIDAR is what we call a point cloud. This is something that needs also processing in order to get a very good point cloud.
From that, we have additional software elements that are using different kinds of AI algorithms up to the, let's say, the top of the stack, which is the perception software, which is using deep learning algorithms, which basically do the object detection, classification, tracking of objects, and other important features for the car, understand the way it can drive and where not, where is the sideway, if I'm driving in front of a truck, how close can I get if it has some kinds of beams outside of it, so a better understanding of the environment. Can I drive under something? Can I drive above something if there is some debris on the road? There are many aspects that we take care of. When we are approaching our clients, they can choose which part of the software they would like to use for their needs.
Obviously, we price it differently. I believe it's a very important differentiator because it's something very specific to a LIDAR. Today, LIDAR is still not generic. Each LIDAR is different. Because we are very intimate with the hardware side, we can provide much better software, which is very well tuned for the hardware and eventually giving a very good software stack to our clients that at the end of the day gives them a very good understanding of what the LIDAR sees. This is very important. Again, as I said, we offer many tools. It's not just one element, but it's a few elements that we have in the stack.
I know there are multiple layers of the software stack in a vehicle, but is it easy to work with the ADAS platforms who are also developing, I know, a higher level of software, but is it easy to just kind of combine software stacks with them?
Yeah. First of all, the answer is I wouldn't call it easy. In terms of the collaboration, it's easy. In terms of the technicalities of working, you need to do the integration. Obviously, when you are working with a platform like Mobileye, Mobileye is an expert in perception software. Still, they will require from us some elements of our software. Some of them are AI. Maybe some of them are more traditional AI software systems or elements, I would say. Yes, they would like us to provide the best output that we can provide them. It doesn't end by just the raw data of the LIDAR.
They are happy to get from us some additional output from some of our software layer, even not maybe the perception they can do themselves, but definitely some lower-level AI features that we offer, definitely they are using on their system. It is a nice collaboration, and there is a very nice synergies between the hardware, the software that we are providing, and the platform itself, like NVIDIA or Mobileye.
Okay. Yeah. No, that makes sense. Okay. I know we only have a couple of minutes left. I guess just the last question that I kind of had, what are some of the most significant trends that you foresee impacting the automotive sector in the next five to 10 years? Also, how do you view the current state of the LIDAR market today versus how it may be in five to 10 years?
Yeah. Usually what I'm quoting here is the market research that we have in hand. Usually, I'm quoting the IHS number, the S&P market research that is around the consumer application level three, level four. At least to our understanding, this predicts that in 2030, there will be something that looks like a 10% market penetration, approximately 10% market penetration, which means for us, if the industry is selling 90 million vehicles per year, it means something like 8-9 million LIDARs being sold in 2030. For us, it's a very good number. Reaching a few millions of LIDARs a year is very meaningful, very significant for us.
We think the industry is on that path because at least on our side of the pipeline, when we are looking at the programs, the different programs that we are competing on, basically every name, every brand name that you can think of, major brand name has some kind of a program underway at different stages. The industry definitely is going towards adopting autonomous features both because of safety consideration, but also because of differentiation and margins. These kinds of new features provide the margins that the industry is looking for, nice margins. The industry is definitely motivated, and the technology has matured. We are in a much more mature stage. On top of it, definitely, and as I mentioned before, we see that commercial applications have reached maturity as well. As I said before, they are looking at it from a different angle.
It's very clear for them the business model that makes a lot of sense, that has a very definite, very clear ROI, return on investment. Since this has matured quicker than we thought initially, we see it as a very meaningful opportunity that basically doubles our business opportunity that we see in front of us up to 2030. We feel very strongly that the market is maturing both on the consumer and the commercial application, and hopefully, Innoviz can be a very meaningful player in these markets.
Yeah. Yeah. As we said before, starting today and even I think 2026, we see an inflection. We are looking forward to kind of seeing how everything plays out. But Eldar, it looks like we are just about out of time. I appreciate you joining us for the conference today.
Sure. Thank you very much for inviting me. Always a pleasure. Hope to see you soon.
Yeah. All right. Have a great day, everyone.