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

All right. Good afternoon, everybody. Thank you so much for joining us for this session with Luminar as part of Deutsche Bank's Global Automotive Conference. My name is Emmanuel Rosner, and I'm the lead U.S. Autos and Auto Technology Analyst here at Deutsche Bank. We're extremely pleased to be joined by Tom Fennimore, who's the CFO of Luminar. As many of you are probably familiar Luminar is a lidar technology company founded in 2012. It has partnerships with many of the top global OEMs and multiple production contracts with major series program awards with Volvo, Mercedes, Nissan and more. So here to talk to us about all this, hopefully, latest development, all fresh and new information. We have Tom. The format for today will be a fireside chat on some of my prepared questions. I'd also encourage you to ask questions as well. So we'll probably around about 20, 25 minutes some of the topics, and then we'll hand it over to you if you have any questions.

Great. Good. We're ready to go.

Awesome. All right. So just to set the stage first, before we dive into milestone and operational type of questions, can you provide us with a state of the industry update in terms of lidar penetration, safety standards, like what do you seeing in terms of where we now, automakers, they're finally adopting it for ADAS purposes, for Level 2, Level 3 purposes? Where are we in the adoption curve?

Yes. So we're still in the very early innings as we sit here today in the middle of 2022. But as you know, Emmanuel, in the automotive industry, it typically takes 3 years for an OEM to start developing a program to when they're actually able to launch it and start selling it to consumers. And so we, Luminar, are on the verge of our first phases of launching with our initial customers. And we really have [ 3.5 ], which are ready to go sometime around the end of this year, and that's Shanghai Auto, who is the largest automaker in China; Volvo, who was our initial customer; Polestar, who's going to follow quickly after Volvo, I'm counting them as my half; and then Mobileye, which is putting us on their mobility-as-a-service vehicles, which they're planning to launch in 2 cities later this year, Munich and Tel Aviv. And so that's going to be our first wave of launches. We then have a second wave that is coming with Mercedes Benz, Nissan, a few others, sometime around the middle of this decade. And then we are in active discussion with several other automakers that would probably be launching in that 2025, 2026 time frame. So the level of activity has increased substantially over the last couple of years and not only increasing with our existing customers that are looking to put our technology on more vehicles that they're producing, but with also new automakers as well. Now going back to the functionality. I know that everybody kind of talks about all these different levels from L0 to L5. And then you're starting to see a little bit of inflation, that's now L2+ and then L2++. And we -- at Luminar, we look at it in a more simple point of view. When you actually look at the technology that you're putting on a vehicle, it's either assisting the driver or it's driving the vehicle, ensure autonomy. There's really no in between. And so we think of it more as a matrix, assisted and autonomy. And then the other one is the actual domain. And so a lot of times when people think of level 4 or level 5 autonomy, they're thinking of autonomy everywhere. Robo-taxies are kind of the example of that. We think that, that is a very, very difficult problem to solve. And from our perspective, that's probably end of this decade at the earliest when you're going to see a large number of full robo-taxi vehicles on the road. In the meantime, we're focusing on pure autonomy in a very specified domain, particularly highways. And that's what we're working with Volvo and Polestar as our initial launch customers on it. Now I'm going to dramatically oversimplify the problem for highway autonomy, but it kind of [indiscernible] down to 2 rules: Stay in your lane, don't hit anything in front of you. Now on a highway, there's hopefully going to be a lot less obstacles on the road, right? You shouldn't have children, hopefully, be running on the middle of I-95, or dogs, et cetera. And because you have -- and highways around the world generally look the same. And so highway autonomy, it's staying in the lane, don't hit the object in front of you. The challenge is you're operating at very high speeds, 70, maybe if you're like me, 80 miles an hour. And so you need a sensor that can see very far with a high degree of resolution, typically 250 meters. Why 250 meters? Because that is the distance that you need to detect an object and bring the vehicle to a full stop, gives you about 7 seconds reaction time. What our lidar does, we operate at a very specific wavelength called 1550. A lot of lidar companies operate at a different wavelength of 905. At 905, you're very limited in terms of the laser power you can put through a 905 wavelength. 905 is just outside the visible light spectrum. It's actually regulated by the FDA because you get too much power through a 905 laser, it can blind you. So you can't have a bunch of vehicles driving around putting through lasers that blind everyone. You don't have that problem at 1550. We need a different type of architecture, which we have. You need very specialized components to bring the cost down to where it needs, which we have. But effectively operating at 1550, we can put 17x more photons in the air than at 905. And so that allows you to see further with a high degree of resolution and basically see all objects in front of you at a high degree of resolution in all conditions, 250 meters out, 120-degree horizontal field of view, 30-degree vertical. And so that enables highway autonomy. And so what we're focusing on today is that highway autonomy, some call it Level 4, but it's really eyes off, hands off the road driver completely out of the loop. And then we're also working on what we're really calling the next-generation ADAS systems for what we're calling proactive safety. I would encourage you to go out and look at some of the videos we put online that demonstrates just how much better this system is because today's ADAS systems are primarily using camera or 2D technology. The problem with 2D is you don't know how far everything is away. You have to do -- take a bunch of pictures and do a bunch of computation to try to estimate it. But if you're -- once again, operating in that 250-meter, 120-degree, 30-degree cone and you're doing real-time 3D laser mapping, where precisely every object is in front of you, and you can create an ADAS system that stops when it needs to stop and doesn't have any false positives. Those are the 2 areas we're focusing on. We're seeing a lot of interest from our customers. That technology is going to be ready to go at the end of this year. And you fast forward 3 or so years from now. We're going to have several hundred thousand vehicles on the road with our technology, driving around real highways by real people and getting that technology on the road.

That's a great overview. And in terms of geography, any notable differences in terms of who is moving faster, I think? Yes, it is part of some of our virtual conferences. My colleague, Edison, was having fireside chat with some of the Chinese automakers, it feels like half of them are coming up with lidar for the rest of this year. Is China moving faster? Is it somewhere else?

So I'm going to -- the 2 fastest movers, one is going to be geography, China, we'll talk a little bit more about that in a second. And then I would say on the Western side where you're really seeing the technology -- the OEMs that are moving the fastest, it's less geography-based, it's more the luxury and the premium makers. Makes sense. They're the high-end vehicle with the highest price points, that's typically where new technology is launched. Let's go back to China because you're right. So China, as probably everybody knows, is the world's largest automotive market. China, however, they're making good progress on EV and some battery technology with CATL. But historically, they have not been a technology leader in the automotive industry. Historically, they've been buying a lot of the cutting-edge technology from some Western suppliers. China -- and I spent 3 years living in Beijing, China, in the 2010, so I have some first-hand knowledge of the experience and stay very close to a lot of the OEMs as well as some of the Chinese regulators there. China views this upcoming autonomy wave as an opportunity to become a technology globally. The Chinese government is heavily encouraging automotive companies to adopt autonomous technology to put lidar on as many vehicles as they can. And then the overall Chinese government, if you look at both the central government as well as the local government, they themselves, the government in aggregate, buy millions of vehicles a year. So they can be a very big consumer of this technology as well. What I would say, one of our first initial launch customers is Shanghai Auto. And what I would say is Chinese OEMs have a different risk tolerance than some of the Western brands in terms of how much of bedding and testing they need to do of the technology before they put it on the road. And so when I kind of look at which of our customers we think is going to get there first in terms of putting our technology on vehicles they sell to the consumer, my personal money is on Shanghai Auto because I think that they're -- once again, just have a different risk tolerance. We've talked about Pony.ai as one of our initial customers in addition to Shanghai Auto. When you kind of look at that robo-taxi or car sharing model, the number of vehicles on the road in the U.S. per licensed driver is about 1.1. Effectively, everybody who has a license in the U.S. has a car. There's practically very little obstacles in car ownership in the U.S., if you want to own it. In China, if you look at the major Tier 1 cities, whether it's Shanghai or Beijing, there are limitations on how many vehicles can be sold in those cities each month. The government will either limit it via lotteries or the number of license plates that they issues. And the reason for that is there's just so much room on the road for vehicles on the road. So if you actually look at the car ownership rate in Shanghai and Beijing, it's about 20% to 25% of the population compared to about 70% to 80% in the Western world. There's just no more room to put the cars in these major cities. So when you actually looked at the number of cars on the road in these major cities in China to license driver, it's about 3 license drivers for every car. So there's actually going to be big demand in China for car sharing and robo-taxies in some of these mobility services. And so the Chinese government is really pushing the local automotive industry to become a leader in autonomy and thus lidar as a result. I talked about some of our wins with Shanghai Auto and Pony.ai. A couple of weeks ago, we signed a deal with ECARX, which is the technology arm of Geely Automotive. Geely Automotive owns Volvo and Polestar, which are 2 of our initial customers as well. What that agreement does is we're basically going to work together with Geely's technology arms to take the technology, not only our lidar, but the software that we're working with, with Volvo and Polestar too, and we're going to localize that together for the China market and eventually deploy that to the Geely family of vehicles in China, which are Lotus, Lynk & Co, Geely, a handful of others. There are going to be certain things we can't do in China as a foreign company, such as data storage, mapping, and local software work. And then there's also going to be what I would say, certain engineering support, where if we have the right local engineering team in China that can move a lot quicker and interact culturally better with Chinese OEMs. So China is a big market. They're moving very fast. I think they're going to be a leader in autonomy. We already have a very good leadership position in China there. And I think we already have some partnerships, which we're probably going to need to be successful in there and probably more to come there.

Right. Let's shift gears to some of the milestone operational progress that we could expect out of the company. I think last quarter, you commented that Luminar remains on track to achieving serious production, readiness, as well as for the software -- for both the hardware and the software this year in 2022. Can you please update us on this process? And then as part of the -- I guess it's part of this, but -- this -- 2022 is also the year where you're starting to put it on a serious production, right? The Volvo XC90 will have your lidar starting this year?

Yes. So -- as I mentioned, we got 3 initial customers, Volvo/Polestar, Shanghai Auto and Mobileye. All of them are expected to get to SOP around the end of the year. Polestar is probably going to be a little bit later than Volvo just given the production timing. To kind of summarize our goals, by the end of this year, we want to be in a position to make our hardware that is going to be sold to our OEM customers and they're going to put it on vehicles to sell to consumers. That's what we're going to do. We're calling it SOP readiness. By the end of this year, we want somebody in the world or somebodies in the world to be able to buy a car with Luminar technology on it. Based upon what we're sitting here today, part of that is going to be within our control to make sure that we're ready to go. And that's within our control, and I'm confident that the team is going to get there. But we also need our OEM customers ready to go as well. And look, we live in an uncertain world, you have chip shortages, global conflict, et cetera. We can't control if our customers naturally delay for reasons unrelated to Luminar. But I'm confident we got 3 shots here. At least one of them is going to get there by the end of the year. And then if they don't get there by the end of the year, they'll do shortly thereafter. So by the end of this year, somewhere in the world, you'll be able to buy a vehicle our technology on it. And that's ultimately what we're going to do. The hardware has to be ready to go day 1, it's going on a vehicle. You can't do over-the-year updates on hardware. The software you need to have initial capabilities ready to go at the end of the year, but software -- and Tesla does this very well. It's something that's going to constantly involved and there's going to be increased functionality to add. Hardware has got to be ready to go. With the software, you need an initial batch with continuous improvements thereafter.

Can you talk about your initial capacity and then expansion plan?

So 6 months ago, our plan was to do our series production at a shared facility of Celestica down in Monterrey. Celestica is your traditional contract manufacturer. We were going to be one of many customers that they had in this Monterrey facility, and we're on track to kind of ramp up that initial production line within Celestica's shared facility. What since has happened is -- as we've kind of worked with our customers in terms of the initial volume that they want us to be prepared to produce that in 2023, it was coming in much higher than expected. And so we needed to do 2 things as a result. One is we needed to accelerate the purchase of automation equipment to ramp up at much higher levels. And so this year, we're going to be spending close to $30 million on automation equipment, which is expected to be delivered and installed around this year. In addition, we -- Celestica ran out of space for us at their shared facility in Monterrey, Mexico. So our initial products that we're making this year and maybe the early part of 2023 will be made at that shared facility, and we're in the process of getting that ramped up and ready to go. But more importantly, we've identified a brownfield site down the road from Celestica in Monterrey. We're going to be building out that facility as a dedicated Luminar facility that is going to be operated by Celestica. So when the automation of equipment arrives at the end of this year, it's going to be still in that brownfield facility. That facility is going to have 6-figure initial annual capacity, the ability to ramp up there. And the goal will be during the early part of 2023 to have that facility up and running and ready to operate at the run rates that our customers want us to be prepared to produce that.

So one of the goals for this year is to grow your forward-looking order book by 40%. Who are these conversations with, the existing customers that would expand it to other platforms? Is it new customers that you're speaking to? And how are these conversations like?

We're going to see growth this year in the order book, very confident in that 40% number, and it's at least 40% is the guidance that we've given. And it's going to come from both existing customers that are giving us new business as well as from new customers. So far this year, we have announced the Mercedes-Benz transaction or contract. And then Nissan at the end of April. And Nissan is a little different than all our other customers because they're the first mass market brand that is really adopting our technology. Nissan is developing their next-generation safety system around our lidar. We've been working with them for over a year on this. The plan is to get that system ready to go when you say I'm going to start deploying it on their vehicles around the middle part of this decade and Nissan has said publicly is our goal is to deploy this next-generation safety system on every car that they produce by the end of 2030. And Nissan makes about 4 million vehicles a year.

So how would you go about adding this to the order book? Would flow out of your work...

Yes. So like it's -- we're very, very conservative in terms of what we put in our order book. We only put things in there where we have a written contract with our customer, an agreement on what vehicle programs are going to be in and what -- and pricing. And so I'm not going to -- I'd be very surprised if we put all 4 million units for Nissan in the order book. We only -- we update our order book at the end of every year. We're working with Nissan on what that vehicle deployment schedule is going to be and the specific commercial terms for each deployment schedule. And once we have agreement on the specific schedule in the specific commercial terms. That is what we include in there. And so those discussions with Nissan, there's some agreement today and hopefully more of them by the end of the year. And as we come to that formal agreement, we'll include it in there. We're not going to have all 4 million vehicles in there day 1. It will only be when the specific terms, schedules, timing are agreed to.

But to be clear, should we also expect additional agreements before the end of the year, meaning with additional...

So we've announced effectively that -- we call them major commercial wins when we kind of have an agreement with our customer that they're going to use our technology to put on vehicles they sell to consumers. That's our definition of a major commercial win. It's not they're going to buy our lidar to put on some lab test [indiscernible] it's -- we're actually going to work together. We're going to take this technology. We're actually going to put it on vehicles we sell, and there's some agreement to do that as well as identified vehicle programs in commercial terms. So one of our milestones this year is to get at least 4 major merchant wins. We've announced 2 so far. Momentum remains intact for us to get there. So I don't think we're done winning yet this year. And as we win, that will also help [ further work ].

Right. When you look at your existing contracts as well the conversation with -- towards additional ones, how important is software part of it? Like are you selling this separately yet? Or is it sort of like just embedded inside the hardware that you're selling?

So it works both ways. So you need software in order to enable our lidar. Our lidar is just a very expensive box that produces some nice pretty point-clouds. You need the underlying software to actually do the object identification and tracking that is produced by our point-cloud, we call that perception. And then the output of that perception software needs to be fused with the other sensors and then do decision-making and path planning for what the vehicle does. We call that full stack. Not only do we have multiple wins for our hardware, we also have multiple software wins. What we're seeing with some of our customers is as we get in there and they try to do some of the software themselves, they realize that developing perception software for 3D point-cloud is a lot different than developing perception software for cameras or radars. And so if we have a solution that is ready to go, it may be easier to kind of pivot to use ours as opposed to use it themselves. So software is an area that we see great opportunity. As I mentioned, we already have multiple wins. We're working jointly with Volvo to develop it together. And then a couple of weeks ago, we hired CJ Moore to run our software business. CJ built the ADAS software business at Tesla. We tried to hire him a year ago when he left Tesla. He decided instead to go to Apple to work on their car development program and we were able to pull him out of Apple after a year to come run our software business. And say what you want about Tesla and what they've been able to achieve in autonomy, but Tesla has probably done a better job than any other automaker in terms of taking software and industrializing it and putting it on vehicles and figuring out a commercial plan to sell it to consumers.

I agree with that. And then pricing-wise, on the software side, is it -- that part of the question, is it embedded the way you...

It works differently with each customer. One of our customers, the initial ones, we have an all-in price with them. And for that to get the hardware and the perception, we didn't bifurcate it. But in that case, it's substantially higher, right, that ASP than what we're selling to lidar only customers. For some other customers, they kind of want a onetime fee for perception. A lot of historical OEMs, they view the pricing as transactional. I get my money when I sell the car to the consumer, I want to pay my suppliers onetime. Other OEMs are trying to move the way that, once again, Tesla is going, which is working on subscription models and we're going to follow them. And if they're going to charge the consumer for a subscription pricing for the software, we'll take subscription. So how we do our pricing is going to match how the OEM charges the consumer.

Understood. Let's talk about your bill of materials that the initial target or assumption was -- by the time you get into production, you would get it towards the $500 per BOM? And then longer term, maybe towards $100. How are you tracking towards that?

Good. So a little bit of background there. So as I mentioned earlier, we operate out of wavelength of 1550. Easiest way to build a lidar is you go buy components at the 905 wavelength. The telecom industry grew up around that wavelength. So if you wanted to go buy an ASIC, a chip and a laser, you go by 905. We're at 1550. The military kind of grew up around 1550. Military doesn't really do stuff in scale and military is known for doing stuff cheap. So we've custom designed a lot of our components. We've actually vertically integrated by buying those suppliers. There's only so many engineers in the world that have optoelectronics expertise at that 1550 wavelength. And most of the good ones now work for Luminar, so that helps expand our competitive moat. But because we've custom-designed a lot of these components, where our BOM is going to be heavily dependent upon volumes. Now the good news is that getting volume certainty from having things like standardization on the next-generation Volvo SUV gives you a very visible path to get in to those high volumes. So what we said publicly is our first full year of series production, right? So we get in series production in scale next year. It's probably going to be more in the 2024 time frame. We want to get to that $500 BOM. The team has done a great job over the last year of getting us there. We're not 100% of the way there, but we're getting very close. And then when you think about what our variable costs are going to be, the 2 biggest components are going to be that BOM, which we want to get to that $500 target. And then once again, we're using Celestica as our contract manufacturer, and it's going to be whatever Celestica is going to charge us to assemble each of those. And our target, once again, kind of when we get to that first full year series production scale is to get them into that $100 plus or minus range. So that $500 plus $100 is going to be our biggest variable cost. Now look, there's going to be some additional fixed costs on top of that, but that fixed cost is going to significantly ramp down as a percentage of revenue as we ramp up as well. Longer term, Iris or Model I, as we call it, is our current generation -- our initial generation of what's going on in series production vehicles. Our next-generation product, which we're creatively calling Model J, after Model I, will -- we're probably going to be talking more about what that's going to look like at some point in the near future here. One of the biggest changes in the architecture is going to be on the laser. 1550 laser and Iris is going to be a fiber leaser. Part of what we got the technology with Freedom Photonics, which is the acquisition we did a couple of months ago, they're able to do [indiscernible] laser, which is going to allow us to take a significant cost out of the laser, which is one of the most expensive components in our BOM. So Model J is going to have a significantly lower BOM. There's going to be some improvements in the specs, but we're kind of already doing what needs to be done to enable autonomy, so we don't need any step-function improvements. It's still going to be better. And then it's also going to be a much more compact form in that. So cheaper, better performance, easier to put in the vehicle is going to be the foundation of Model J.

I said -- can you speak a little bit about pricing. So we spoke about the costs here. What is your pricing model? Are there differences between mass market and vehicles? Is it feature defined? Like, how does it work?

Yes. So it's -- for the underlying hardware, we've said publicly for Iris that it's going to be in that $1,000 ASP range. In smaller volumes, it's going to be -- it's volume-based pricing. So a smaller volume is a little higher larger, volume is a little bit lower. Where I think you're going to see the step function down change in pricing is when we get to Model J. The lower BOM is going to enable us to price it lower than that $1,000 plus or minus, which will allow us to penetrate better the mass market brands.

Understood. And maybe just finally for me. Can you provide us with an update on your collaboration with the NVIDIA? This is a partnership that we announced, which seems that it could provide a pretty good platform for multiple automakers. How is that going?

So we're working real time with NVIDIA on Mercedes-Benz. So we're the lidar provider in Mercedes-Benz. NVIDIA is doing the full stack software. NVIDIA is designing their autonomous system. Mercedes is going to be their initial launch customer, we're in there with them. So they're really designing everything around our lidar. Lidar isn't like camera. You can't plug and play a lidar maker there. Take a look at everybody's point-clouds, everybody's point-cloud is dramatically different. And so NVIDIA is going to be building their initial platform and the reference platform around our lidar and that's going to give us a significant advantage of getting designed into their system with very high switching costs. And so as NVIDIA goes and wins more and more OEMs, and they go to them with a system that was initially designed to work for Mercedes-Benz we're going to be in the lead position to piggyback on that business as well. So that's our -- where we are with NVIDIA. And we have a very good working level relationships where the teams are working together on almost a daily basis on the Mercedes-Benz side.

Right. Let me open it up for your questions. Any questions in the room? I asked all the questions.

Good job.

That's basically it. So -- in that case, really, thank you so much.

Thank you, Emmanuel. Thank you for having us.