Luminar Technologies, Inc. (LAZRQ) Earnings Call Transcript & Summary

Thanks for joining us at the Barclays Automotive and Mobility Tech Conference. I'm Blayne Curtis, semiconductor analyst at Barclays, but today we're talking lidar. Very happy to have with us from Luminar, Tom Fennimore. He's the CFO. Welcome, Tom.

Thanks, Blayne. Great to be here.

Very interesting space. A lot of companies newly public. I thought maybe a good way to start off is, out of these 100-plus companies that are chasing this lidar opportunity, which is really -- maybe you could spend some time just walking us through how do you differentiate? Everybody's got their matrix. They show all the advantages. If you believe everybody, everybody is going to win. Maybe you just talk for a few minutes just why is Luminar differentiated for the competitors? And why are you going to win?

Sure. Great question, Blayne. Look, I agree with you. There's a lot of lidar companies out there. It feels like there's a different one trying to go public each weak. But I'm going to talk a little bit more about how our technology is different, which is really what is critical to drive future commercial success and determine who the ultimate winners or winners are going to be. But my guidance to investors is look who's getting the OEM wins because at the end of the day, the view of the technology that matters the most is the customers. And are they buying your products in your technology, not to put on a bench somewhere and test it, but actually buying that technology, going through the very expensive and time-consuming process of developing it and vetting it to put on vehicles that they're going to sell to the end consumer because what's going to make the winners in this industry is getting your product on vehicles in scale. That's how you're going to get real revenue, that's how you're going to get scale, that's how you're going to get access to continuous data to continue to evolve and develop your products. And so from Luminar's perspective, we set the bar very high on what we count as a win. For us, it needs to be a conscious selection by a customer to incorporate our technology into a series production program or equivalent. And I say equivalent because Airbus is one of the largest aerospace companies out there. They don't series produced vehicles, but in what we're doing with them, we're confident it's going to be a major commercial opportunity. And so we've announced deals with Volvo, Daimler Truck, Mobileye, the NVIDIA one this week, Polestar, Shanghai Auto and a few others. And so these are customers that are taking our lidar and in some cases, our software and they're going to put it on vehicles that they're going to start selling to consumers as soon as the end of next year. So then why are they choosing our technology? Austin Russell, our CEO and Co-Founder, did a great job back in 2012 when he started Luminar as a [indiscernible] major, really developing our lidar from the ground up. We operate at a wavelength 1550, which just allows you to do a lot more with the lidar because you're not limited by eye safety restrictions in terms of how much power you can put out through the laser. Now the issue with 1550 historically has been it's not cheap. And you need to use a certain type of receiver called InGaAs, which is very expensive, right? A wafer of InGaAs can cost up to $10,000. What we've been able to figure out is to just take a tiny gap of that InGaAs, put it on a receiver and take the cost of that receiver down to a few dollars per unit. And so that allows us to make a very robust lidar at a price point where it makes it commercially viable for automakers to put on their vehicles. And so we cracked the code for that. Anybody can put together a great slide with PowerPoint or say that they can see an object on the road, 250 meters in front of you in the middle of the day, if it's a bright white object. But when you have your lidar on vehicles that are on the road, with real people in them that are going to drive in all conditions, day and night, bad weather, good weather, you got to be able to see everything at all the time. And that's what our lidar enables as OEMs as well as great technology companies like NVIDIA and Mobileye, work with all the variety of lidar companies out there, and they make that conscious decision on which technology they want to put the vehicles almost every time they're selecting ours. And from an investor perspective, follow the wind, follow where the automakers and the other key technology players are choosing -- what lidar company they're choosing to incorporate that lidar in their actual products they're going to be selling in scale for revenue.

Obviously, yes, that's the feedback is, hey, it's just brute force, but it's also expensive. What have you done? You talked about this approach of using just a tiny bit of InGaAs. What have you done to kind of protect your approach in your supply chain so the others can't copy that?

Sure. So once again, for the critical components, we've custom designed them. Easiest way to build a lidar is you operate at a different wavelength, the [indiscernible]. There's a lot of off-the-shelf lasers, receivers, ASICs that you can buy, assemble them together, make a lidar. I've already talked about the underlying limits on the functionality of that. We had to go in and custom design a lot of these components from start. And we've, over the years, identified what I would say, critical suppliers for us about it's very smart engineers that are very good at using this InGaAs technology or lasers, et cetera. Historically, we've entered into exclusive supply relationships with them, making sure that they're only working with us. And then ultimately, what we've done for both our the firm that we're using to design our ASIC chips, which was Black Forest Engineering, as well as our receiver, which was OptoGration, is we decided to purchase them. And the reason that we did that is for a couple of reasons. Number one, these are great companies working them forever, bringing them in-house to the Luminar team, something we're very excited about. Number 2 is exclusive supply relationships are great. But now that we own the technology, it kind of ensures that only we can use it. But more importantly, now as we're on the verge of this series production ramp-up that we're going through, taking these smaller companies, which BFE 20 [indiscernible], OptoGration, a little less than 10, getting them in a position where they can ramp up, do automotive great scale and automative great quality it helps derisk our execution as well. And so this is a way for us to expand our competitive moat, make sure that these very unique components that we designed stays in-house at Luminar. And it also, we believe, helps with our execution by allowing us to ramp up and meet all the different automotive-grade qualifications a lot easier.

I do want to ask you, you said follow the wins, and you have a win with Volvo that's production when there's lots of wins, right? There's development wins, which may or may not turn into something that there's production, which is tied to particular vehicles. So I want to ask you about one. And then you mentioned NVIDIA, which is an interesting announcement, what's -- this is a development platform, right? So maybe you could just walk us through what does that mean? And how will that translate to production rates?

Sure. Let me start with Volvo. Volvo was our first customer that has assigned us a series production win. We announced it in the spring of last year. We are designed into their SPA 2 platform, which is not only one of the most, if not the most important platforms at Volvo, but Volvo also shares that SPA 2 platform with other members of the [indiscernible] family, including Polestar. And then they launched certain vehicle programs of that SPA 2 platform. So the initial vehicle program back in the spring of last year that we were going to go into Volvo was as an option on the XC90. And that was a great win because it's our first the XC90, the successor to that program is Volvo's flagship vehicle. But as an option, the consumer actually needs to make a conscious selection to put that technology on their vehicle now when you look at kind of the historical take rates, you can get comfortable that there's going to be a fair amount of value there. But at the end of the day, you're kind of making guesses on what take rates are going to be for a new technology. And there's a fair amount of uncertainty there on what the ultimate volumes are going to be. This summer, Volvo came and said this is very important technology to us. They've said publicly that their ultimate goal is to put this on every vehicle they make. And the next step from taking it as an option in the XC90 or the successor program is to make it standard on every electric version of the vehicle that they produce. And so that's a great vote of confidence of Volvo and us. If we weren't executing, Volvo wouldn't be giving us more business. But more importantly, it gives us a lot more volume certainty, particularly in the near term of what we're going to do or what we're going to be able to produce. And there's a lot of benefits from that in terms of scale, getting better deals from our suppliers, et cetera. And so that is a great double win for us because you go from an option to a standard. And look, I'm pretty confident that over time, our relationship with Volvo is going to continue to grow. Because you got to remember, getting your foot in the door with these initial autonomous programs with automakers is very important because they are designing this autonomous system around your lidar. And they're spending a lot of time, a lot of money on, in my opinion, the most important technology they're introducing on their vehicles in a very long time, right? This is enabling the vehicle to effectively drive itself under such conditions. And so once you develop this platform and then you go from the XC90 to the XC60, which is their mid-sized [indiscernible], are you really going to design another autonomous platform from scratch? Or if that one is working and the supplier team on that is executing and you're getting reasonable pricing, you're just going to incorporate that on every other vehicle program in your portfolio. And so that is the importance of winning these initial programs getting designed in initially because as long as you continue to execute, the rest of the business at another OEM is yours to lose. Same thing goes with NVIDIA. So NVIDIA, amazing technology company, a company we've been working out with years, dominant chip supplier kind of in that high-end automotive autonomous space. And NVIDIA has worked with a lot of other lidar companies, we've kind of found ourselves in the same room, same OEMs working together on the same autonomous programs. And so what NVIDIA is doing with their Hyperion platform is they're going to OEMs and selling them a complete solution. Here's our sensor suite. Here's the software to the back debt -- You OEM should take this reference platform. And if you want to make vehicles with it, this is what you should use. And that's very powerful. And on that reference platform, there's one lidar company, which is ours. And so what that means is that for every win in the automotive space that NVIDIA gets they're likely going to pull our lidar alongside with them. I guess it's theoretically possible that an OEM can say, hey, I'll take your reference platform, but I want to switch out a lidar company, but then you're going to have to almost start from scratch again in terms of validating that system, making sure that all the sensors are talking together, making sure that you're getting the lidar data you would get with another centers instead of [indiscernible] . It's just you're starting from scratch again and so the benefits of scale that you get from NVIDIA you're going to lose. And so my personal opinion, the odds of an OEM switching things out if they buy the Hyperion platform or switching a lighter out are very low. Now there may be circumstances where they just buy the NVIDIA chip, and that's different than the Hyperion reference platform. But this is an amazing opportunity for us because it puts us alongside NVIDIA in terms of selling autonomous systems to OEMs. And it's also another validation by a very smart technology player their view in terms of which lidar out there is the best. And remember, they worked with a lot of lidar companies. They're out there working with various OEMs, so they get the feedback from the OEMs perspective, and they selected us.

I want to ask you just kind of maybe we jump right to the wins, but just you were talking about the market in general, there's a lot of momentum to get some of the safety aspects through, I think there's some forecast to say that maybe 2/3 of cars in -- by 2024 will have some form of ADAS or autonomous functionality. Clearly, if you're going to do a full autonomous car, you're going to need lidar, you're going to need a lot of sensors. But where the volume may be would be kind of more kind of assisted driving. So I'm just kind of curious how you think about the market. You have the Volvo win, but you maybe just kind of think about where lidar fits in. And clearly, RADAR are trying to improve their performance, and there's a big gap in cost. Maybe you can address kind of both things. Where can lidar costs get down to? And is there room in that kind of mainstream volume where it's not a full autonomous car?

Sure. So this is a little bit of a weird thing to say as a lidar company. But if you kind of take the existing ADAS systems on the vehicle, and if you add a short-range lidar that doesn't have the necessarily a functionality of ours. Our opinion is that the improvements you're going to get to existing ADAS systems are modest, and those modest improvements are not worth the incremental cost of the lidar. And with camera and radar and the other existing centers out there, you can do Level 2 just fine. The Level 2 is imperfect. I think we've all kind of seen some of the recent public cases that highlight some of the issues that it's facing. But ultimately, you're going to need a lidar to go to Level 3 and above. And I'm always amused because you have Level 2 and then you have Level 2 plus, and then now there's Level 2+ plus. And to me, that just signals that everybody is like trying to get as close as they can to Level 3 but you're out there. Our lidar enables 2 levels of functionality. One is what we call proactive safety, which really takes existing ADAS systems to the next level. And this is really just more Level 0 functionality. And it's AEB and other type of systems that we can just do a lot better. And the reason for that is with cameras -- and you're more or less, it's a 2D image. And so you can take a lot of those 2D images, you can do a lot of computational work to get pretty good guesses on where the objects are around you. But at the end of the day, they're just guesses, and the more time you drive, the more errors you're going to have. And you can't set up a system that has too many false positives to compensate for that because then you're just going to be driving around in the car is going to be stopping for no reason, and it's going to create a miserable consumer experience. The benefit of our lidar is it's 3D data. It gives you very precise location 10 times a second, where every object is 250 meters in front of you in a 30-degree vertical field of view and 120-degree horizonal field view. And so that enables us to do what we call this proactive safety product. We've been releasing demos that we've done initially in August, then at IAA in Germany. We're going to be doing more in CES. But this is really taking existing ADAS systems to the next level. It's going to substantially improve the safety of that vehicle. And that substantial improvement is worth and then some the cost of lidar. And in fact, we're looking at ways to try to monetize those safety improvements, be it insurance savings to try to capture that. That is what drove standardization at Volvo. Volvo is probably the automotive brand, most anonymous with safety. And so highway autonomy is great. That's the sizzle that kind of is cool to talk about. That's kind of what gets you in the door. That's what gets you in as an option to make our technology standard on every vehicle. You really need to have a substantial improvement in the quality of these ADAS systems to justify the cost of the lidar. Now the good thing is, and this is really the business model that we're working hand-in-hand with our automotive partners to deploy is the same lidar that goes on the vehicle that drives proactive safety is the [indiscernible] that is in the vehicle to drive highway automate. The only difference is the underlying software to drive that different functionality. So if you have a lidar standard on every vehicle, you can then upsell the consumer to a highway autonomy package, either at the time that they purchase the car or you can do it via over-the-air of that by unlocking that functionality and making more money from the customer over time. You don't have to add any additional hardware to the vehicle. And so if you look at kind of some of the pricing that Tesla is getting for what they call full self-driving capabilities, for the highway autonomy functionality that Volvo is planning to introduce, that's going to be a lot better than the existing Tesla system there. And Tesla has been getting, on average, a 30% tick rate higher for the SNAX, lower for the [indiscernible] given those price points. They've been giving $10,000, historically $3,000, then $7,000, $8,000, then 10,000. And so if Volvo charge is somewhere near that and gets a 30% acceptance rate, that's a lot of money that drops to the bottom line, but you have to have that lidar on the vehicle to enable that.

Lots of impact there. And actually, I was going to ask you later about the insurance angle, but since you brought it up, when you look at -- 2 ways to really kind of push these systems in lidar with it is 1 regulation, it just takes a long time. I mean I have -- cars have some very basic emergency braking. And you see a collision, and you say, wow, I don't know if I would have caught that, right? So I mean -- but you brought in the insurance angle, which is an interesting one because, obviously, you don't need to wait for countries and to implement these kind of mandates like they did for backup cameras and such. So maybe can you talk on both efforts, but I'm really intrigued by the insurance angle. What are you doing to drive that? I think when you buy a car, I think the how often you ask insurance, I think, is what you told me before. What are you doing to turn that around to make kind of the insurance companies drive this effort as well?

Sure. And look, I really got to give Tesla kudos because they're the automotive company that I think is the most forward thinking in terms of that. Despite all the public stuff out there that the safety of Tesla's autonomous vehicles, when you actually look at their vehicles, the average vehicle in the U.S. gets into it has about a 6% chance of getting into an accident. Per year at Tesla, it's about 4%. And so Tesla actually does design safe vehicles. I think their issue is more they're not where they need to be. We don't think they're going to be able to get there in terms of the autonomous package that they have. And so what Tesla kind of realized is insurance companies were not giving them any credit further increased vehicle safety. Part of that is the way that insurance companies price their data is they have years of actuarial data and they want to see years of results before they adjust the pricing. So it's an extremely lagging indicator about the safety of the vehicle. Our technology is going to be a significant improvement of the safety of the vehicle. And if we don't do anything, we're going to make the insurance companies a lot of money because they're going to capture those safety employments and they're not going to adjust their pricing as rapidly as they should. So -- and we'll talk a lot more about this probably in the near future, but the person who set up Tesla's insurance business is now a Luminar employee. We have a partnership with Swiss Re, who's the largest global automotive reinsurer. And we're working to put in place an insurance program where we can partner with our automakers to subsidize the cost of our technology, both to the consumer as well as to the automaker. And so this is the way for us to monetize the safety improvements that we're going to be bringing to the vehicle to subsidize the cost of our technology to the consumer into the automotive OEM. And at the same time, hopefully, make some good profit for the home team here. Regulatory, absolutely, we're all over that. Our Head of Government Affairs used to be the former General Counsel for NITSA. We're going to be doing more to gather the data that demonstrates the safety improvements that we're going to be bringing to the vehicle. We're going to educate the necessary regulators on the benefits, but that also is going to take time. We need to show real results to the regulators about the safety improvements that we're going to bring. We need to show that the cost that we can do with that is reasonable. And over time, that is something that we're going to do to drive adoption of our vehicles. We think our technology can save a lot of lives, and that is an argument we can make to the consumer, to the automakers and to the regulators to make sure that our technology can get on as many vehicles as possible to help reduce the too many deaths that occur each year on vehicle [indiscernible].

Exactly. Super interesting. I wanted to back up, we just jumped in that because you brought it up, but I want to ask you on the software angle. I know I think being a semi guy, a lot of times, companies have a hard time monetizing software. I think all the lidar companies are talking about, hey, we're going to layer this perception software on top. Kind of just curious where you think you can monetize versus slim your customers are going to have a very big development effort themselves. So first part of the question is, what is the software opportunity for you? How far of the stack would you actually go? And then a second part of the question I'm actually curious on, I've covered Mobileye and NVIDIA, and there is this first-mover advantage where you don't -- it's hard to kind of get the software right if you just don't have the data, right? So maybe you can just address the data portion, what access to data you have? As you start to ramp production wins, who owns it? And kind of what's the kind of the critical mass you need to kind of get that software angle right?

Great question, a lot to unpack there. Right now, we have just as many software engineers as hardware engineers at Luminar. We have multiple series production wins for our lidar. We have multiple series production wins for our software. And to be clear, not every person who's using our lidar is using our software, but there are several that are using our software. And we really have an a la carte menu that we offered to our customers. If you just want our lidar, no problem, here it is. If you want to do the software yourself, both the perception and the full stack, that's great. We're ourself toward teams here if you need us if you stumble along the way. When you look at the software that we can provide, there are several layers, but I'll divide it into 2 main categories. The first is the perception. The outside of our lidar is what we call a Point Club. It's basically just a bunch of dot of lights. And so you have a bunch of them once again in a 3D image. You then need the perception software to say, "Hey, this group of dots is a vehicle 100 meters in front of me. This group of dots is tire on the road 225 meters in front of me. This group of dots is a pedestrian 50 meters off to the side of the road." And then because we're measuring at 10x a second or so, you can actually do object tracking, velocity, et cetera. That's our perception in software. Will it make sense for every OEM to do that themselves? Some are going to try and some are probably going to be able to do it. But ultimately, the more data you get from that, the better your object -- your perception software is going to be. And so that gives us an inherent advantage because we're going to have access to a lot more data than an individual or a [indiscernible]. So we're seeing a lot of interest in our perception software for that reason. The next layer on top of that is the full stack software. So this is then taking the perception software output from our lidar. It's taking the output from the camera, the radar, whatever other centers are on the vehicle, it's doing the sensor fusion, and then you do the decision-making in path planning. For Volvo, we're actually working with Zenseact, which is Volvo's those autonomous software arm to develop this. And for -- particularly for highway autonomy. The benefit of this is once that is done and ready and working to the point where Volvo is comfortable deploying it on their vehicles. We then have the right to take that full stack software and deploy it to other OEMs. Now why would Volvo and Zenseact allow us to do that? A couple of reasons. Number one, they're spending a lot of money to develop this. Clearly, they're going to deploy it on the vehicles, but getting more volume helps to kind of amortize that fixed cost advancement over more vehicle volume. The challenge they're going to face is when they go to market and go to work with other OEMs, are other OEMs going to really work with Volvo and Zenseact to do this? And so we're going to be the go-to-market. We have the ability to take that software. Look, each OEM is going to want to make their various modifications to it, and we have the software team to do that. But I think we're going to be in a much better position to monetize that to other OEMs. Now look, there's going to be a revenue-sharing agreement, all those not going to allow us to do this without sharing some of the revenue with them. But that allows us to offer a full stack solution to other customers as well. Now a lot of OEMs, particularly the larger ones have already built big autonomous software teams, they're going to try to do it themselves. But for a small or medium-sized OEM that hasn't built that out and need to get this technology there. We have a one-stop solution where we can give them the lidar, we can give them the reception software, and we can give them the full stack. And so that's our strategy. It's to be flexible. It's being able to service our customers and bring autonomy to their vehicles. And was it to hardware-only, hardware-plus perception, hardware-plus full-stack software. I think the final part of your question, and Blayne, let me know if I missed the point on it, was [indiscernible]. You're absolutely right. If you look at some of the robo-taxi autonomous companies out there, they're burning through a lot of money. And one of the reasons for that is if you want to build a test vehicle to drive around and collect data, you're going to spend up to $1 million per vehicle by the time you buy the vehicle, equip it with all the sensors pay somebody to drive it around, pay for the maintenance, gathering all the data, it's very expensive. And a lot of these robo-taxi companies, whether gathering a lot of data, they're really doing it in geo-fenced areas or very restrictive geographic locations. We're going to have our products on multiple customers on many vehicles starting around the end of the next year. We're getting access to the data in most cases, from those vehicles to allow continuous improvements to our products. And so -- and once again, going back to like the standardization volumes that we've talked about in the past is winning these initial Series production program is not only important to get designed into those autonomous reference platforms, get in those initial vehicles, get the economies of scales. But you're also getting access to the data in a real-world environment, right? These are cars being driven on the road every day. right? So you're going to be getting a lot of data in real-world conditions on a lot of roads. That is going to allow us to continue to evolve not only our hardware, but our software and expand our competitive moat.

[indiscernible] topic and time's has flown out. Maybe just one more question, just in terms of -- obviously, of working units, you've gotten production wins, but a much different story to produce tens of thousands, hundreds of thousands, a million units. So kind of where are you in this process in terms of scaling out your manufacturing?

Yes. Look, Blayne, our top focus by far is to execute on the wins that we have because if we fail on that, everything else is irrelevant. And in fact, when we talked about all the things you need to do to make a commercially viable lidar, one of those things is to create a manufacturing process and a design that is scalable and that you can produce in scale at automotive great quality. And this isn't easy, right? You're making very complicated opto-electrical devices in scale at auto grade standards. Our team has been working around the clock to develop this blueprint. We made a conscious decision to kind of act as a Tier 1 as opposed to using a Tier 1 partner. And we've done a lot of questions both from automakers as well as investors about if that was the right decision. And, One of the reasons we made that decision is we've looked for the right Tier 1 partner that has that expertise in manufacturing those complicated optoelectronic devices. And if there's somebody out there that can do that, we were unable to find them and when we ask our OEM partners, they're unable to find them, too. And so we have our own advanced manufacturing line. I'm in our Orlando office today, where we kind of -- where we've been producing our units, figuring out the right way to make these work. And then that blueprint that we've developed, we're handing it off to our contract manufacturing partner in Celestica, who's now ramping up to make these things in scale. And so we're now -- we said on the earnings call yesterday that we're now entering the sea phase. So in the automotive industry, you have A-samples, B-sample, C-samples and then you guess at the D-samples and then you kind of go to SOP, that B2C sample transition is the most difficult because you got to be C-sample, you have your production going in place, which we do. You have to have your design frozen the going series production in which we do. You have to work through the manufacturing issues to build these things in scale, which we're now through. And so the next stage for us is to really get to the C-sample, start making those units down in Celestica and go from really start ramping it up to the point where you can make these things in serious production scale. It's tough. The team's been working around the clock. They've done a great job figuring all this out. We're on track to get that SOP at the end of next year. We still got some more work to do. But I'm very confident in terms of the progress we've made so far and the quality of the team that we built that we're going to get to where we need to get to by the end of next year.

Well, with that, yes, time's flown by, but thank you very much for your time, Tom.

All right, Blayne. Take care. Thanks for having us.