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

So hello, everyone. Thank you for joining us for the session with Luminar Technologies as part of Deutsche Bank's Global Auto Conference. My name is Winnie Dong. I'm a senior associate here at DB's U.S. Auto Research. Luminar is a LiDAR company that was founded in 2012 and has partnerships with many of the top global OEMs, a major series holder awards with Volvo, Mercedes, Nissan and more. The company has recently started production for Volvo recently and also showcased their next gen LiDAR product, the Halo, at your Investor Day. So I'm very pleased today to be joined by Tom Fennimore for a fireside chat and Q&A. The format for this session will be a fireside chat around some of my prepared questions. But also, if you have any questions from the audience, please feel to raise your hand, and we'll bring a mic over to you. So with that, Tom, thanks for being with us.

Thanks for having us, Winnie, and good to see you again.

Any sort of beginning remarks that you want to make before we get started?

I'll be brief, and we'll get right to the Q&A because I know that's the part that most people find helpful. For Luminar, we've been public for a little over 3 years at this point. We've met recently one of our primary milestones as a company, which is to enter into high-volume series production with Volvo on their flagship SUV, the EX90. And so our LiDAR is going to be standard, i.e., on every EX90 that Volvo starts selling this summer. So we're very excited about that. It was the end of the beginning in terms of our first stage of industrialization. It took us a little bit longer, a little more money, a little bit more blood, sweat and tears to get there, but we're there. And now we're focused on ramping that up, at the same time, ramping our costs down. We've taken some initial steps with that. And then really over the next 3 plus or minus years, we have an additional 20 vehicle lines that we're going to be launching with Volvo, Polestar and Mercedes. And so this is really in the ramp-up stage here for Luminar, and we're excited about it.

Perfect. Thank you. So I guess before we dive into milestones and aberrational type of questions, can you provide us with maybe a state of the industry with regard to the appetite for LiDARs from both a safety standpoint but also in the context of autonomous driving? I know we sat here exactly a year ago in the same spot. So what has changed?

Yes. And what I would say is the amount of interesting dialogue that we're having with automakers continues to increase. And I would say there's two primary functions of the LiDAR. One is really enabling that next-generation safety. That's what Volvo is initially using our LiDAR for. Nissan, who is generating their -- or developing their next-generation safety system, is incorporating our LiDAR into their development system as well. One of the things we released a few weeks ago is we partnered with Swiss Re, one of the leading global insurance companies. What they did is they took a vehicle equipped with our LiDAR and ran it through their few dozen scenarios that they've built over the years and with billions of miles of real-life data and says these are the scenarios which typically cause accidents, which of those can LiDAR avoid? Relative to the best vehicle out there, a vehicle equipped with our LiDAR, relative to the best vehicle, not the average vehicle, but the best vehicle; can avoid nearly 30% of the accidents. And for the accidents that you can avoid will reduce severity or the amount of energy in that crash by nearly 40%. So that means instead of hitting something at like 30 miles an hour, you're hitting it at a much lower speed. And so the amount of safety improvements here can be immense. And then the other thing that LiDAR can enable is what I would call that L3 or higher autonomy, whether that's highway autonomy or lower speed autonomy. And having that LiDAR in the 3D sensing technology, I would say almost every automaker would say that you really need LiDAR to do L3 or higher levels of autonomy.

Okay. So I think just drilling into that a little bit more, there's been ongoing debate in terms of the requirement for LiDAR, depending on the self-driving approach. Teslas and [indiscernible] saying, they don't need a LiDAR. So I guess, based on your conversations with your OEM partners, what are you hearing in terms of the approach that they may be biased towards? And then for some of the OEMs who might be on defense for it, what are some of the pushbacks like here?

Sure. Let's start with the safety system. You can design a good L0 or all the way up to a good L2+ system without LiDAR today. There are systems at work. They can be camera-only or camera and radar. What LiDAR enables for those system is really a step-function improvement and change, and I'll refer you to the preliminary results of the Swiss Re really takes those systems to the next level, but you can buy good enough system today. Now NHTSA, a few weeks ago, introduced some rules for AEB braking and other safety functionality that they expect to be standard or mandated on vehicles in the U.S. starting in 2029. Our system with our LiDAR will allow you to meet those rules, and we believe you're not going to be able to meet those rules without a LiDAR. And so from a safety perspective, while you can buy a good enough system today, that system won't necessarily meet the new mandate in 2029, and you get significantly better results. Now going to L3+ autonomy, Tesla, particularly Elon, he has been very public in his remarks saying that he can make an L3 system or can do autonomy with camera only. His argument is we humans drive only with our eyes. We don't have laser shooting out of our heads. And so -- thus, you should be able to develop an autonomous system with camera only because we're able to work for the human. I'm not going to get into a science debate, I would say almost every other automaker has a different point of view. And what I would say, if there's any technology company that should has the skill set and should be motivated to make a camera-only autonomous system work, it's Mobileye. They're one of the best, if not the best in the industry, developing these software-based systems. And what they've even said publicly, based upon everything they know today, you can build a good up to L2 system with camera only. That's their SuperVision system. But if you want Chauffeur, which is L3 or above, you need to use a LiDAR. And as today, they're designing our LiDAR into that system. And so Mobileye, they're pretty smart people. And if -- they should be motivated to make a camera-only system work, but their designs have LiDAR incorporated into their system for anything L3 or above. And most automakers feel that way as well.

Yes. I think in a market like China, for instance, there's increasingly this notion that you have to equip your vehicles with advance features, safety features to even compete in the market. And so there's this narrative going around where is LiDAR being commoditized over the long term. So do you feel like this is the case? And is there a potential that OEMs become even more cost conscious towards this system over time?

In our experience, the automakers are always going to be cost conscious, and they're going to be always asking us for lower pricing, cheaper pricing, et cetera. And that's just the nature of the automotive industry. The way that we prevent that is by continuously to innovate. We introduced our next-generation product, which is Halo, which is about half the cost of our initial version Iris. Not only is it cheaper, it's a lot smaller, integrates into the vehicle, et cetera. So we're taking proactive approaches to really bringing the cost of our system down. However, what we're also trying to do is instead of just having the traditional OEM supplier relationship where it's just transactional, bring us down to the price, let's talk about the value being created in the ecosystem. So our LiDAR with those safety improvements, if you look at a 30% reduction in accident frequency and then a 40% reduction in accident severity, that can reduce your insurance cost by up to half. And so if you're spending $2,500 a year on insurance for a Volvo SUV and you're able to reduce the insurance cost by a significant amount, well, that's going to be multiple thousands of dollars over the life of the platform relative to our cost of the LIDAR, that's a good investment. If you're able to upgrade the consumer to L3 or other autonomous functionality, that can be a lot of value that also resides to the automaker as well from selling to consumer, either at the time he or she purchases the vehicle or any time over the cost of ownership, upselling them to autonomy. And so instead of looking at it as just the cost of a LiDAR, which can be expensive where it is today, but going down over time, looking at the safety improvements, the insurance savings, upselling the consumer autonomy, then it becomes, well, a costly -- relatively costly addition. It actually becomes a very good investment, given all the different ways that you can kind of monetize the ecosystem that LiDAR can enable.

Yes. So what is the competitive landscape right now? I guess, like when you bid for SKUs, who do you run into?

Yes. And I would say that, that the competitive landscape is windowing. I mean, there's been -- I've lost track, 10 or so LiDAR companies that have gone public over the last 3 or 4 years. We're now at the stage where if you haven't industrialized your product in that multiple wins, that raises some question marks. And I think there's only a handful of companies that have been able to do that. You have to have the balance sheet and the technology road map to not only be around for the next several years, but to really bring the technology and your cost point to where it needs to be. And so when you look at all that, there are only, I would say, a small group of companies that meet that bill. I mean I think out of China, you've had Hesai and RoboSense that have done that. Valeo, we're seeing more out there because I would say they've done a good job of industrializing their technology. It's more at a 905 wavelength. And I think Valeo is a company that's going to be around for a while as well. The other, I would say, new factor that has been introduced here probably over the last few months has been the geopolitical concerns, right? There's been some public headlines about U.S. banning China LiDAR. If that happens, how are -- how is China going to react? And so I think everybody is looking at their supply chain and making sure that if these geopolitical winds continue to blow in that direction, how can I continue to produce vehicles that will have LiDAR that doesn't violate any of the laws out there. So I think that's also a new factor that's going in where you take potential bans into account, how does that impact everything. And so I would say it's becoming more narrow. We're now at the point where automakers don't want to look at PowerPoint presentations. They want to look at real-life products, results and then look at everybody's technology road map and make sure that it's going to continue to evolve with their ongoing needs.

Yes. Let's dive a little bit more into the operational progress that you guys have made. And as I mentioned in the opening remark that you recently commenced the start of production for EX90. Can you tell us more about the ramp as it currently stands? What kind of visibility do you have in terms of ramping up volume for the rest of this year? Your Mexican facility is being brought online recently. Should we expect some sort of full utilization of that in the next, let's call it, 12 to 18 months or so?

Yes, I think that's the right time frame here. I would say we're in the early stages of this ramp up. Particularly with new technology, the automakers tend to -- want to keep volumes low initially, work out the kinks. And I think that's what you're seeing with Volvo. The initial SUVs, I think my understanding of what Volvo has said publicly are expected to hit the showroom floors at the dealers later this summer. When you look at IHS, I think you're familiar with IHS, but for those that aren't, it's a third-party forecaster, which I would say a lot of people use in the industry. Right now, I would say Q2, there -- what IHS is forecasting, it's not a material amount. And then it kind of ramps up for Q3 and Q4 of this year around in that 15,000 unit plus or minus range and then really ramping up even higher there towards the middle part of next year. The other thing for us is right now, Volvo's initially producing these things at the Charleston plant. The second plant that they're making the EX90 is over in China, which is expected to come online later this year. And then shortly after that, will be the Polestar 3, which we're also on as an option, not a standard for the time being. So I think that, that is in the process of ramping up. I think a little bit Q2, more Q3 and Q4 and then higher there next year as Polestar 3 and their China plant starts to ramp up from there.

Yes. Let's talk about your order book for a second. You have -- I think you last updated $3.8 billion. You're also guiding to softly growing this order book this year. So I guess, as we stand midway through the year, any sort of color you can provide us in terms of the growth trajectory for that this year?

Yes. It's still going to continue to grow. We haven't given any formal guidance yet. The reason for it is we've seen some headwinds recently in terms of automakers making decisions on what their -- the next wave of LiDARs are -- vehicles that they're going to put LiDARs on. My personal opinion is there's two drivers for that. One is with the slowdown in EVs, a lot of automakers are revisiting their future product planning cycle. How many EVs do they want to make relative to hybrids or internal combustion engines? We are powertrain agnostic. We work on any vehicle type, whether it's EV, hybrid or internal combustion engines. But we need the automakers to make formal decisions on what that product cadence is going to look like. They need to make that decision, then they'll decide what vehicles they're going to put LiDAR on and ultimately decide the RFQs. And so that's been a little bit of a slowdown in the decision-making. And that's kind of why we've been not giving any formal guidance on what the order book growth is going to look like. The other thing here, and we've seen this with Volvo, but we've seen this with almost everybody we're working with; is getting the software ready to fully unlock what the LiDAR is capable of. It's taking not only the automakers, but some of the technology providers longer to do that. Now, once they do that, then it's very easy to replicate that autonomous system to other vehicles. But given that, that's taking longer, that's also been a slowdown as well. So we will have order book growth this year. But I think what we want to do is see some of the delays that those two things have been happening -- causing, we want to see that a little bit more visibility on the timing of those decisions before we start forecasting order book growth. But we are confident this year that we will have order book growth, both from existing customers as well as new customers as well.

Thanks for that. So just maybe go into that a little bit more. So higher autonomy is often talked in tandem with electric vehicles. But actually, as you mentioned, right, ADAS and autonomy can go either EV or ICE. So do you think it's for the industry as a whole that eventually higher ADAS adoption might really have to go hand in hand with like EV adoption? And then, do you see that sort of reflect in your existing order book?

Yes. I would say our order book today is probably a little overweighted to electric vehicles. And the reason for that is unless you need EV to enable autonomy or to fully unlock the LiDAR, my impression of why that is, is as the automakers were allocating their new technology dollars, they were allocating them more to the EV budget, particularly at the higher end because you need it to compete against Tesla. And one of the ways that you wanted to do that was by putting as much new tech on those vehicles as you can. And so what we've seen is, I would say, an overweighting of LiDAR to those EV vehicles as a result. Now, some of our order book will be on every variant of a vehicle, whether that's an EV, HEV or internal combustion engine. So as more ICE engines are sold relative to EV, the order book will naturally weight itself there. Going forward, what we're seeing now in our conversation with customers is, as I mentioned before, the powertrain agnostic. They want to build the flexibility to sell vehicles in the near term, which I think are expected to be more hybrid and internal combustion engine weighted, but still at the same time, continuing that longer term investment in EVs because I think that's how they kind of see the world continuing to move. And so as we're working with our automotive customers, it's going to make sure that we can be on every variant that we have. But luckily, as I mentioned before, we're powertrain agnostic. And what we need is once you have the LiDAR and the software system that works, we can go on any vehicle that an automaker makes.

Now that Volvo is in start of production, what's sort of next?

Right now, what we're working on is, as I said, ramping up with Volvo and ramping our costs down. We, I would say, spent almost all of our engineering resources up until recently making sure that we're ready for Volvo SOP. We needed to make sure that we were ready to go, and we were able to make sensors at that volume and that quality, and that's something that we've proven that we can do. What we're focused on now is bringing our costs down. Now that we've kind of reached SOP, we have a lot more credibility with our suppliers. So we're meeting with them and I would say, looking at entering into longer term strategic relationships, bringing the cost down, we're going from really a start-up company to, I wouldn't necessarily say a mature company yet, but a maturing company. We are in that high volume start of production now we've reached that. And that gives us the credibility to go in and get better deals from our suppliers. And so that's what we're focused on now. We're also looking at what are called in the industry VAVEs, value-added, value-engineered changes, so how can we kind of take cost out of our product through designing changes which don't materially impact the performance of the product? And so what we're doing now and I would say, over the next few months is really focusing on bringing our unit economics down to where they need to be to reach profitability on a unit economics level. In addition, we're focusing on Volvo's 1 of 20 coming over the next 3 plus or minus years. Now your first one is the toughest. And the industrializing their first product gives you a lot of experience to do the next 19. They won't be as difficult as the first one, but there's still a fair amount of work that we need to do there. And as part of that, it's really bringing our next-generation product, Halo, to the market as soon as possible. And we've really moved up the timeline with that and really having that kind of go -- ready to go with our initial customer in the second half of '26.

Yes. So, yes, you recently unveiled the Luminar Halo. I guess as we think about the composition of the order book, how fluid it is? Like in terms of evaluating your -- one of the customers like in your current order book, and I'm looking at this product that's much less cost, would I want to wait for that?

I think we're now at the point now and part of the reason we brought that up, typically, anytime you win an initial customer, it's going to take you 3 years from when you win it to when it's ready to go. So we're sitting here in the middle of '24. Any new piece of business that we win today with a new customer is likely going to be ready to go sometime in, call it, '27 on average, maybe a little bit sooner for certain OEMs, maybe a little bit later, but that general magnitude. Now that Halo is kind of ready to go in that time frame, there are some new customers we've been talking about, which were going to use Iris or a variant Iris for a year or 2 before transitioning. It's now moving to the point of how can I transition right to that one. We didn't bring it here, but we have from one-on-ones. But when you kind of see the size of that relative to Iris, when you look at the cost, the ability to more easily integrate it in the vehicle, it's very clear what the new investors or new customers want. And it's kind of aligning naturally with that as well. Now the question is for the existing customers, how do we get them to Halo as fast as possible. And that's something we're working with each of our customers, I would say, on a real-time, right? It's not as easy as just saying, "Okay, we're going to switch over now." There's -- I'll give you an example with Volvo. When you do switch over, one of the things that are important to them is make sure that the point cloud output is as equivalent as possible to Iris because that allows them to put it in and minimizes the retraining, if any of that needs to be done with the software. And so while we can make the point cloud better, they're kind of like, no, like if you make it better, it does everything we want to today. But we want to minimize the retraining of our software because that was very difficult to kind of build out initially. So we don't want -- necessarily want to go back and retrain that. In addition, you got -- Iris is kind of designed to fit into the roof line. If you go with something smaller, you got to change the roof line around that, et cetera. So you have to work with each automaker on a vehicle-by-vehicle basis. The natural point is when you have the platform turnover where one ends and the new one begins, that's the easiest point. Next easiest point is a mid-cycle refresh, which is another good entry point. And then the question is, is there more creative stuff you can do in the near term? Because the best way to get the cost reduction that everybody wants from us is to less beat us up on pricing for Iris, but to convert to the next generation as soon as possible.

Yes. And you mentioned cost, right? How should we think about, I guess, gross margin for the rest of this year? Because now that you've successfully launched the EX90, on one hand, your ASP will be lower because you normally into series production type of pricing versus prototype before. On the other hand, you're executing on cost and then ramping up higher volume in the second half. So sort of like how should we sort of view the balance there? And then I guess, what are the differences in pricing as you move into...

Yes. And so I think that this is one of the things, and we talked about it on our last earnings calls, I would say over the next 2 to 3 quarters, what you're going to see is some gross margin headwinds because we are going from prototype pricing with Volvo to series production pricing. And so that is a lower ASP than what it's been historically. At the same time, while our costs are coming down, they're not coming down at the same pace. And so it's probably going to take us to the middle of next year to start getting to some of our cost targets. And I think in the meantime that, that's going to produce some gross margin headwinds. We've said historically that it's going to take us kind of that first full year of series production to get to scale, work out the kinks and get to our supply chain to where it needs to be in order to kind of see what those unit -- some of those unit economic targets that we've talked about.

So you've killed the least, you've indicated that Tesla was one of your top customers for the quarter. Anything to say to that regarding that relationship?

Nothing really that I can add. We needed to disclose that because they were north of a 10% customer. I've said on the call that their sales have been lumpy, right? They haven't recurred every quarter. And that wasn't the first time that they [ bottomed than ] that. And look, they haven't shared with us specifically what they're doing. It's nothing related to series production. It's not going to be recurring on a quarterly basis, and I can only speculate what they're using it for. So I really can't say anything in addition to what I said on the quarter when it comes to that.

Yes. Okay. So from a industrialization perspective, it seems like you're opting for a different approach going forward, right? Maybe can you talk about that strategy there in relationship with TPK? You have an existing relationship with them already. So how is that...

Yes. We've had a relationship with TPK for over a year now. There are -- the initial relationship is they're our contract manufacturer over in Asia. One of the things is as we spent the last 4 years industrializing our product, we realized -- we stepped back and we looked ourselves and did a honest assessment. What are the things that we're good at and we do better than our partners? And what are some of the things that quite frankly, we could have done a better job at and some of our partners can do a better job than us in that? And so one of the things we did once we formally reached start of production was take the first step in outsourcing more of the industrialization functions to some of our partners, of which TPK is one. And that could be everything from -- I'll give you an example, we did prototype manufacturing in Orlando manufacturing facility. I think we're going to be transitioning most of that to TPK. A lot of the supply chain or quality management of our suppliers, we did a lot of that ourselves. I think we're going to be outsourcing more of that to TPK, a lot of the design validation testing, a lot of that we did ourselves, we're going to be outsourcing more of that. And so a lot of those industrialization activities, we're going to be outsourcing it. I would say the core engineering, the core product development, the core R&D, that's something that we're going to continue to do and we'll do in the foreseeable future. But this will allow us, we believe, to move faster, more efficiently and more cost effectively. And part of the reason why we were able to bring the Halo product development timeline in further and accelerated is because we are, a, we learned a lot from industrializing Iris and thus, we think we can do it a lot faster the second time. But because we're relying more on our partners, we think that that's going to give us the ability to move faster as well.

Yes. What is the current status of partnership with Mobileye, NVIDIA, Qualcomm?

Yes. I mean, look, we're all moving forward with them on a variety of projects, right? The Polestar win we had with -- our Polestar 4 win we had, that is actually of -- it's a little bit different than Polestar 3. Polestar 3 was off of Volvo platform, the same one that they're using the EX90 for. Posters 4 is off of this ECARX platform where they're actually using -- Mobileye did a lot of the software. So we're working real-time on that. NVIDIA, we're continuing to work real-time on the Mercedes as a joint customer together. So right now, I would say the relationship with those continues to move forward, and we're neck-deep working together on execution to jointly serve our customers.

Yes. Maybe talk about the role of AI in Luminar LiDARs? I think you have touched this during your Investor Day. You have a partnership with Applied Intuition, for instance. What is the goal there? And then how are you looking to leverage AI?

Yes. So what we're doing in-house with AI is really on our software side, trying to take what we call our proactive safety software to the next level. And so some of you will call it geometric-based or rules-based perception software is -- and probably if you just looked outside in the street of Manhattan, that's probably a good example. The way that a lot of the rules work today, particularly with our LiDAR, is our LiDAR is kind of measuring the objects in front of you. And ultimately, what you want to do to build your collision or blend-in software is if my vehicle is on a collision path with the path of an object that I don't want to hit, then I need to either stop the car or change its course, right? That is ultimately the problem that you're trying to solve. Now if you're on a highway and you're having a car that's kind of moving straight and you're looking at a vehicle, the path of that vehicle tends to be predictable. If it's driving on a highway, it's probably going to stay between the two white lines and it's going to move forward at the velocity that it has. Where things get a little bit more tricky is when you have a more complicated environment with more unpredictable objects, right? And probably the streets of Manhattan are the perfect case study of that, where you have -- particularly when you start introducing people that are like crossing the streets or walking Manhattan, those then become unpredictable and a lot of the rule-based predictive algorithms go out of the way. And so one of the things that we're doing is kind of building AI into it to do a better job of predicting the path of objects to make your collision avoidance systems, I would say, better, right, as opposed -- instead of assuming things are going to continue to move in a linear fashion, adding more, I would say, predictability in using AI-based algorithms. So that's one example. I would say the more advanced example, and this is part of what we're doing with Applied is one of the areas that I would say a lot of the software providing is with sensor fusion. A lot of ADAS systems that are built today are build off camera. And what you're doing is you're kind of appending the LiDAR on to that, and you're taking the output of that sensor, and then you have to fuse it with the other sensors. If all the sensors agree, then you really don't have an issue. Where you have the issue when you need to figure out is when the sensors don't agree, which one do you believe? And one of the way you do it is you just drive the vehicles around, identify the scenarios where the sensors don't agree and then kind of tell the system which sensor to believe in certain scenarios. And that is a very, I would say, burdensome and time-consuming way to build your sensor algorithm. What we're doing with Applied and the other way to approach it is instead of appending it on there and kind of driving around and doing -- figuring out the tiebreaker for each scenario, is you drive it around and then you have an AI algorithm that kind of does the center fusion naturally. And there, you're kind of starting from scratch as opposed to appending it to a preexisting code. And so I would say that, that's another example of where we're using AI to do the software in a different approach and partnering with some, I would say, newer companies like an Applied to do that together.

Yes. So you have a couple of minutes left. I want to see if there's any questions from the audience.

Thanks for the presentation. Just a quick comment. I think if you think about safety, Volvo, it doesn't get too much more of a brand than that to be tied up with. So I think that's great. I had a quick question for clarification and then one question just for you to opine on. One, when you said that you were running the simulations that reduce the accidents by 30%? Or was it 30% of the accidents, if you could just repeat?

Yes. So I think the exact number was 28% -- 27%, sorry. And so that's not relative to the average car what they did so and I'm happy to go through. Swiss Re took the 15 -- looked at the 15 global OEMs and took the safest car of each of them. And then what they did is they ran through a series of over 35 tests, which based on their experience, caused the most accidents. And then they did it with the cars that are on the road today, and they took the vehicle equipped with our LiDAR. And relative to the best car that they tested off the 15, there were 27% fewer accidents in those scenarios where they tested, and then the severity was reduced by close to 40%. Okay. And so -- and they published the blog with all this, we can send you the exact stats and everything. But yes, so the way to think about it relative to the best car, nearly 30% fewer accidents and then approximately 40% for the accidents you can't avoid, fewer severity because you're just hitting it at a lower speed.

That's really powerful. Do I have time for a quick question? Just -- you mentioned a little bit the kind of fear around China and maybe like a little tit-for-tat trade war there. I think you guys are building out a second facility in China, has really great efficiencies, but you're also working in a plant in Mexico. Can you just maybe take us through kind of how you're hedging against that risk, against a potential trade war? And kind of what the exposure is for the company?

Yes. So one of the things we're doing is exactly, as you said, which is we want to have a plant in China and a plant in Mexico. Mexico will -- plant in China to serve the China, maybe other select Asian markets. The plant in Mexico to serve North America and the rest of the world. So that's one way we can do it. Where it starts to get a little trickier for us is if they start banning less LiDARs made in China, but anything American related to it in there. I think that, that's something that we're looking at other clever strategic things we can do, whether they're joint ventures or something like that to mitigate it. It's tough to really think about what that is until you see what the final rules are. The other thing that we're doing is really looking at our supply chain as well and really taking this to the next level, which is, it doesn't just stop at LiDAR, but there are other sensitive components. And so we're also looking at our supply base and making sure that for each of our suppliers, we have a China and a non-China solution, in case the world continues to move, at least in my opinion, in the wrong direction. One of the questions we get is now that we did expand our relationship with TPK to rely more on the industrialization side, does that introduce more risk to us? And we've purposely structured that deal so that, if, God forbid, we need to limit our ability to operate in China, TPK is a global company, they do have operations outside of China, and we have the ability to kind of pivot to, I would say, jurisdictions that we don't believe are going to be impacted by China bans or other bans that we need to see. So it's something we're monitoring. Has it had a direct impact on our business today? Not necessarily. But what I would say is it has the focus of our automotive customers, which means it has the focus of us. And we're just trying to build the flexibility into our business model. We don't know which way the world is going to go, we're engineers, we're not diplomats or statesmen, but we just need to be prepared.

Okay. It's always a pleasure. Thank you so much.

Thank you, Winnie. Good to see you again.