2023-08-11 11:57:12 ET
Luminar Technologies, Inc. (LAZR)
JPMorgan Auto Conference Call
August 10, 2023, 09:30 ET
Company Participants
Thomas Fennimore - CFO
Conference Call Participants
Samik Chatterjee - JPMorgan Chase & Co.
Presentation
Samik Chatterjee
Good morning, everyone. I'm Samik Chatterjee. Pleasure of hosting Tom Fennimore, CFO, of Luminar for the next session here. Tom, thanks for coming to the autos conference at JPMorgan. Thanks for taking the time to do that. We really appreciate your time here.
Question-and-Answer Session
Q - Samik Chatterjee
Maybe I'll just dive into the fireside chat questions here directly. And if you have any prepared remarks, please feel free to sort of append that before the first question. But overall industrial production, you've highlighted that multiple times, you're on track for industrial production later this year. We're just curious what part of the process is remaining to be executed on from Luminar standpoint to deliver on these production units for the vehicles that go into production.
Thomas Fennimore
Sure. And first, thank you for having us, and it's good to see you again. So when you look at our industrialization progress, we reached SOP in low volume last year for our initial vehicle platform, which is actually in China with Shanghai Auto. And so we've proven that we can make these things, put them on vehicles at the right quality, but in low volume, right? That was never going to be a high-volume program. This was -- it was a good starter program for us to kind of prove out the concept of making these things at the quality that can go on consumer vehicles. So I think we crossed an important threshold last year.
Now what we're focused on is building these sensors in very high scale. The first thing we needed to do was to get our manufacturing plant up and running. And so originally, we were doing the low-volume production at Celestica's existing facility in Monterrey. That's where we made the units for Shanghai Auto, and we were making the units for our other development customers.
There wasn't enough space at that facility for the volume that Volvo wanted us to produce that. So we went 20 minutes down the street in Monterrey, found a cold shell of the building. We spent the last year plus building that out, installing a clean room and starting the automation equipment. We turned on that switch at the facility at the end of Q1.
And now what we're doing is we are producing parts at that plant, but those parts are going through the testing validation that we need to get ready for the high-volume production with Volvo next year. And so that goes through an increasingly higher production volume, looking at the units, testing them as they're coming off the line, working out the kinks of the manufacturing process. That's not cheap, right? Because I got to buy all the material for that. I got to run them through the plant. You don't do it perfectly the first time, so you got to work through the kinks.
Unfortunately, once you have assessed that, that kind of goes through the testing validation. You can't sell it, and you have to scrap it. So that results in a lot of high costs that we're currently going through. And I tried to highlight some of those during our recent earnings call. So the goal is at the end of the year to have all those kinks worked out and to be able to produce. It doesn't mean we are going to be producing, but we've kind of gone through the testing and proved that we can at that volume rate and at the quality that Volvo wants us to for their initial launch. So that's what we're working through. So I would say we crawled, we walked, now we're working on sprinting and going through that evolution.
Samik Chatterjee
Okay. So just to summarize then, in terms of your readiness for Volvo, you'll be ready by the end of the year...
Thomas Fennimore
Yes. That is our target.
Samik Chatterjee
Then customer conversation.
Thomas Fennimore
I would say to borrow the baseball analogy, we're probably getting close to the eighth inning heading into the ninth inning. There are still issues that we need to work through, but they're the type of issues like how much torque do I apply on this crew, what's the right level of a proxy. And so I would say for all the known issues, we have a plan to address that we're confident the risk as you get to the ninth inning and it's those unknown-unknown issues, and you have a band ninth inning, but we're well on track to be where we need to be at the end of the year.
Samik Chatterjee
In relation to then the customer readiness on behalf of Volvo, I think there have been multiple press reports speculating that Volvo is finding it difficult to integrate the LIDAR. Can you talk to some of the challenges that the OEM might be facing and how independent they are of what Luminar's deliverables are to the customer?
Thomas Fennimore
Yes. So look, I'll let Volvo speak on behalf of Volvo and the issues that they're facing. I know that there was some noise in the market a couple of weeks ago. I think there was something around the LiDAR integration. Volvo did put out a clarifying statement. That said, the issues they're facing are not related to LiDAR integration. It's related to the software and getting that ready. We're trying to control what we can at Luminar, which is to be ready to go at the end of the year.
I think Volvo said that their SOP is going to start the middle of next year. So we have a little bit of cushion. We didn't change the time line for our team when there was an SOP delayed. We're doing this for the first time. And so we're sticking to that original end of December, and we want to make sure that we're ready to go there. And God forbid, if there's any unforeseen issues, we have a little bit of buffer built in there. But the delay is not related to LiDAR integration or our readiness. It's related to getting the software ready on their side. That's what they've said publicly.
Samik Chatterjee
Okay. One of the investor questions I've got on this topic, and I think we were discussing this offline as well, like the risk that investors want to sort of get a better handle of is, is there a point at which Volvo says I have to delay the launch by a longer period because of the software issues. Or given some of the press reports around LiDAR being the integration, which you clarified is not the case but one of the questions that's been coming up on account of that is, do they get to a point where they say, I don't want to put the LiDAR in this generation of the vehicle. I'll just postpone it to the next one. Any thoughts around those risks?
Thomas Fennimore
I think the latter risk is very low. Volvo has been front and center both in the past, present and future of marketing their EX90 vehicle with our LiDAR. You guys can go and look at some of the videos. I've personally been to launch events in Sweden, China. There's other stuff that we're planning to do. And so I think Volvo is very committed to LiDAR. We are standard. And so their roof and production process that they setup for the roof and the system has our LiDAR being integrated into that.
The software doesn't need to be perfect at the time of launch. Software, you can update over time via over-the-air updates. And in fact, I think that that's what Volvo plans to do. And so it needs to be good enough. And the issues that I think Volvo is facing on the software side, they're all imminently solvable. And so I think the likely outcome there is they kind of get the software in a good enough shape, start production and then continually update it over the time with software updates.
And so this is Volvo's flagship vehicle. They are very committed to it. I'm sure it's a big contributor to the revenue or -- and profit. And so they're going to do everything they can to launch this vehicle.
Samik Chatterjee
Okay. Fair. Update on the $1 billion order book. I know you've said you're not going to give continuous sort of updates in terms of quantifying it. But maybe give us some -- anything to flesh out more in terms of where you are related to that $1 billion order target. How lumpy also should we be thinking it is, like is it going to be many sort of small wins contributing this year or are you expecting it to be lumpy and sort of 1 or 2 OEMs contributing to that one?
Thomas Fennimore
Yes. The answer to that is both. We've already made progress to filling that $1 billion. We calculate it at the end of the year. We don't want to get into -- we don't want to kind of disclose the order book increase upon each win because our customers don't like that. So there's a reason why we do it at the end of the year. We've had wins so far from both new -- from existing customers as well as new customers, some of which like Plus have used a legacy LiDAR and are upgrading their next-gen system to our LiDAR. And so there -- look, it can be lumpy when you get a big win from a big customer that adds a lot to the order book. The timing of when that happens, though, is difficult to predict.
The OEMs move at their own speed. A lot of the business we win is one-off, right? It's conversations that we have, particularly with existing customers. New customers tend to run a broader RFQ process. And we've been enough in those RFQ process to know that they typically don't happen at the original time line and each of the OEMs move at their own pace. And so that's why we don't predict like the timing of RFQs because they're largely out of our hands. I know other companies are talking about we have this many RFQ process, we're the finalist. What I would say for the RFQs this year that we've targeted as must wins, we are active in this RFQ process and even finalist in some, and we haven't lost anything this year that we've wanted to win.
Samik Chatterjee
Okay. On the earnings call, there was one comment I think Austin talked -- mentioned that definitely stood out, which is you mentioned there are certain features on a vehicle that only Luminar can enable and competitors cannot. And this was in response to, I think, one of the competitive landscape questions that was brought up. I mean one of the things -- can you shed maybe firstly start by shedding some light on what those features are, what do you see as that differentiation where competitors are not being able to enable features that you can?
Thomas Fennimore
Let me give you 2 real-life examples. Look at what Mobileye does, and they have a wide array of products. Mobileye has a supervision system, which gets you up to L2+ that works pretty good. You can meet today's NCAP standards. You can deploy good L2+ system. And look, if you want to build an ADAS system or an L2+ system that meets today's standards, you don't need our LiDAR. In fact, you probably really don't need any LiDAR.
If you want to take your ADAS system to the next level, and we're starting to see some, I would say, regulatory guidance to do that. And if you want to build an L3+ system at all relevant speeds, you need -- or and if you want to build an L3 system, you need a LiDAR. If you look at Mobileye's Chauffeur system, they have a LiDAR on it, our LiDAR. Their Mobility-as-a-Service system, it has our LiDAR. I know that there's one OEM in particular who thinks that they can do it all camera. My point of view, if there's any company that's motivated to do it all camera, it's Mobileye, and they've effectively admitted that they need a LiDAR.
So if you want to go L3 and above, you need a LiDAR. If you want to take your ADAS system to the next level, we think you need a LiDAR, particularly our LiDAR as well. Look at what Nissan is doing with their next-gen system. They wanted to significantly improve their safety of the cars, take it to the next level. They ran a process a couple of years ago. They selected our LiDAR. We're in the development stage to build that next-generation safety system with our LiDAR. And Nissan has said publicly that they want to start deploying it on vehicles around the middle of this decade and have it on every vehicle -- virtually every vehicle that they make by the end of this decade. They've started to release commercials showing the benefits of that, and we're making great progress with that.
There's a lot of things you can do on the ADAS system with our LiDAR. Cameras are ultimately guesstimating. If you take a picture of an object, you don't know how far it is away. There's a lot of calculation and guesstimating that goes in by taking a bunch of pictures, looking how the size of the object changes. If I gave you a laser pointer and you know exactly how far away each object is, and that's effectively what our LiDAR does. So that's an example of where you need a LiDAR, develop that next-generation ADAS system go 3 -- L3 and above.
Now why our LiDAR, particularly 1550? And I'll go 1550 versus 905. You put too much power through a 905 laser, you're going to blind the human eye. And there's actually regulations that limit how much power you can go through that. There's no such limitation on a 1550. And so we're able to put 15x more photon out in the environment. And so because we're putting more photons, we can see further and better. And we also have the ability to pump up the power in the laser in tough environments, fogs, bad weather, really want to get a sense of that. So we can put -- we don't do this all the time, but we have the ability to put 1 million times more power in the laser when it's needed. And so that allows us to see the environment in a much richer way as well as farther.
And so look at what Mercedes did. Mercedes current generation system uses a 905 LiDAR. And that's able to deploy L3 at speeds of up to about 35 to 40 miles an hour. The Germans like to drive faster than that on the Autobahn. So for their next-generation system, one of the biggest reasons why they selected us is because 15x more photons, a million times more when you need it, you see farther, and so we're able to go at speeds that are 75 miles an hour or plus because you need to see very far when you're traveling at those speeds, so you get that vehicle enough time to react.
And I'm not talking seeing a bright white object in the middle of the day. It's seeing everything, right? And so that's what we're able to do. Mercedes upgraded for that reason. We're seeing other companies start to upgrade for that reason. And so when we get an RFQ from an OEM, we look at the vehicle functionality that they're looking to deploy. If it's just an L3 Traffic Jam Assist at low speeds or an ADAS system with only modest improvements, we typically deprioritize that because, quite frankly, our technology is not differentiated.
And we know eventually that OEM will want to upgrade that functionality, and we'll have another bite at the apple. And so we tend to focus our initial customers because we can't win everybody at once at our early stage. We got to execute flawlessly. We focused on the vehicle programs where the OEMs wanted to work to unlock the full functionality of our LiDAR. And when you do that, it's more or less only our technology that can deliver that.
Samik Chatterjee
Yes. Interesting. Very helpful. Maybe changing gears here a bit, can you outline a bit more about your partnership with NVIDIA? When you work with partners like those, how are you thinking about when in bringing a full stack solution to the market, what portions can Luminar contribute? What are you looking for your partners to bring to the table in that sort of full stack?
Thomas Fennimore
So we're working with NVIDIA on a daily basis on -- at Mercedes. NVIDIA is doing the full stack software with Mercedes. Our LiDAR is incorporated around that. And so what's going on now, we just started a B sample delivery earlier this year of our Iris Plus product, of which Mercedes is going to be the lead customer. And so what you do once you have B samples, that product is mature enough for you to put it on development vehicles.
So Mercedes and NVIDIA have however many development vehicles driving around, I'm not going to say the world but driving in real-time environment, and what they're doing is they're gathering data and they're training their software algorithm. And so in NVIDIA's Hyperion software system, that drives the autonomy that Mercedes wants. NVIDIA is developing that software system around our LiDAR. And so we're the reference LiDAR in their system.
Now we won Mercedes by winning it on ourselves. We went directly to Mercedes. Mercedes ultimately chose their LiDAR supplier. And I think most OEMs are going to do the same thing, right? The OEMs are -- they tend to want to select each of the different components. However, as NVIDIA sells that Hyperion system that they developed with Mercedes to other OEMs, the quickest path to market, the most cost-effective path to market and the least risk path to market is to keep our LiDAR on it.
If you switch out LiDARs, each point cloud is significantly different between the LiDAR companies, they need to then go back and redevelop most of their software system again. And so that's more time, more risk, more money. And so the OEMs are going to want to ultimately select the LiDAR supplier, but we're going to have an embedded advantage as a result. Not only do we have that relationship with NVIDIA, we have that relationship with Mobileye, and it creates stickiness for us, not only with our OEM customers, but some of the leading system integrators as well.
Samik Chatterjee
Okay. Let's talk about another feature that you've talked about, which is with the acquisition of Civil Maps, the mapping capabilities that you've brought in-house, any update in terms of road map to commercializing it, monetizing it with the OEMs? Where are you on that front?
Thomas Fennimore
Sure. And just for a little bit more background, bigger picture on the -- what I would say, the M&A front, it's a tough environment out there for a lot of tech companies, particularly companies in the mobility tech space. And so there were a lot of -- there's been a lot of companies who are running out of cash and struggling to raise their next funding round. Civil Maps was one of those companies last year. We kind of did an acquihire, paid a modest fee. I think it was like $6 million to the -- to some of the shareholders to get the deal done. And they actually had a decent 3D mapping code.
And most of the mapping systems today are camera-based, right? They're built with cameras and are 2-dimensional maps. You kind of see these HD maps, which brings better clarity, more detail on it. Our LiDAR data, and remember, over the next few years, we're going to have several hundred thousand vehicles on the road, driving around, collecting real-time data. And so one of the ways that we can work with our OEM partners to kind of monetize the environment created or the ecosystem enabled by our LiDAR is how do we kind of collect that data and then convert it into a 3D mapping tool and then kind of work with the legacy mapping companies to utilize that data, right?
Because the way that a lot of the mapping companies build their map today is a lot like software teams. They build cars, drive them around and build their mapping that way. And so the interest from the OEMs has been, I would say, a lot faster than I was expecting because you kind of go to them and say, "Look, we got this thing, you're putting our LiDAR on the vehicle anyway, let's work together to do this. And so one of the things we announced on our earnings call, and we'll go into more detail on this probably later this year is we did sign our first development contract with a major OEM to figure out how you can utilize this mapping software, mapping engine that we have.
And so I think at some point in the near future, we'll go into more detail with our OEM partner in terms of how we're doing it. And I'd say how we're doing it is going to be a little different than what we were expecting. And so when we did that acquisition, we thought this would be a good fit for us. It was a $6 million acquisition. So it wasn't like we were making a big bet, but it was a bet that if it worked out would be you get a pretty good return on it. And I would say we're starting to see early signs of the fruit bearing from that acquisition.
Samik Chatterjee
And maybe just following up on that, I mean, once you have -- once you go through the development phase, is the end goal to get to something what Mobileye has with REM in terms of being able to charge OEMs on a sort of per vehicle basis for utilizing that data? Is that sort of how investors should think about...
Thomas Fennimore
Yes, that's one path, but there's also a path, I would say, to more collaboratively work with your OEMs in terms of gathering that data and how you monetize it.
Samik Chatterjee
Okay. Okay. That does bring me to maybe an interesting question that we've got from investors, but -- which is related to the China market. But maybe before we get into the ability to sort of collect that data on the mapping front in China. That's where we get a lot of questions.
Thomas Fennimore
Yes. So China, so let me just clarify here. This would exclude China. China has -- China is very protective of its maps even to the point where they kind of put constraints on what Tesla's -- the data that Tesla is able to capture. LiDAR data in China is classified as mapping data. Mapping data can only be held by Chinese citizens that either have a Class A or Class B license. Class A allows you to operate anywhere in the country. Class B allows you to operate like in certain regions. So think about it as having A license to operate in the United States or a B license would allow you to operate like in the State of New York. And those license are very few and limited.
In China, we -- there are a lot of rules that we need to follow as a foreign company. So for example, we have 2 server rooms in our China office. There's a server room that is in a locked office that I don't have access to. There's one person in our office who's a Chinese citizen that can get in. It's owned by a registered firm in China. All the data we collect goes to that server. That Chinese firm then filters it and then kind of we then have the second server room, which we have access to. And so we're very, very limited as a foreign company in terms of what we can. And so we don't have any plans of kind of being successful in the mapping business in China just because we don't see a feasible path to do that.
Samik Chatterjee
Okay. Okay. Maybe just last one.
Thomas Fennimore
But the rest of the world is still a huge opportunity we can do.
Samik Chatterjee
And maybe just last one before I open it up to the audience. So outside of mapping, what are you seeing in the China market? And I see a lot of other LiDAR companies not willing to participate yet in the China market in terms of going after wins. You've obviously been successful. But what are you seeing in terms of just appetite from the OEMs? Particularly with some of the local competition also coming up, what are you seeing in terms of competitive intensity there?
Thomas Fennimore
Well, I think the other -- I don't if the other companies aren't willing to be in China. I just don't think that they've been successful. We have to be in China. Our -- if you look at any major Western OEM, 20%, 30%, 40% of their business is in China. And you got to be able to serve them in China. And so yes, it's really not a strategic option for us not to be in China because you need to be there to serve your Western OEMs. They're half the market.
And so part of the reason why we announced a partnership with TPK earlier this year to build out a manufacturing presence in China because our Western OEM makers wanted it to be there. And in fact, the Western OEMs are starting to realize that they're losing market share in China, particularly to these new EV companies that are really being aggressive on introducing new technology on their vehicles. And so they're seeing their market share decline. They realize they need to react faster, particularly with new technology. And one of those things, it's actually what's happening in the China market is creating, I would say, a demand pool from our Western partners because they see themselves losing market share and need to move faster.
So you need to be in China if you want to be a global LiDAR company. Now there are a lot of challenges of being successful in China. I talked about some of the limitations that we have on the data side. I spent 3 years living in Beijing a few years ago. And so I've had first experience of what it takes to be successful in China. So we're doing a few things. One, I already talked about how we have that local China manufacturing presence. I think that gives us a lot more credibility. Two, I think this is a true statement, but I'm not 100% certain. I think we're the only Western OEM that has won a LiDAR business with a local China brand with Shanghai Auto. So I think that, that gives us, for lack of a better word, some street cred in China, that's been helpful.
Three, we partner -- we need to partner with people. We did it with TPK on the manufacturing side. We have a strategic partnership with ECARX, which is the technology arm of Geely. And then four, you need to be very smart about the local customers that you work with. There are Chinese brands that want to put a cheap LiDAR on their vehicle. That's not a business that we're going to win. If it's a game of cost, we're going to lose. If it's a game of picking the best technology that can deliver the best value, we're going to win that in a level playing field.
And so we're very smart in terms of who we target as customers. But look, we're -- we got to be very nimble in our strategy in China because I think the geopolitical landscape is continuing to evolve, and we need to be very smart about it, including how we protect our IP. But we want to be a major player in China. We realized that our market share there is probably not going to be the same in the Western world. And we need -- we're probably going to need to partner with the right players to be successful there as well.
Samik Chatterjee
Let me open it up to the audience if anyone has a question. Any questions?
Thomas Fennimore
Yes, you want to...
Samik Chatterjee
So the question is LiDAR versus radar and the competitive dynamics between the two.
Thomas Fennimore
Yes. And so pardon me. I think with radar, you're probably talking more about 4D radar, which is kind of like the next generation. And what that does, it actually gives you a better resolution than the kind of traditional radar. The way that it was described to me is it still is a very, very blurry image, and so it limits your object identification. And so I think 4D radar can help supplement an L2+ system. It can probably help replace the 905 LiDAR. But it doesn't -- it's not going to get anywhere close to that long-range resolution that we have.
So we're not worried about 4D radar. But what it can do is it can provide, I would say, modest improvements to an L2+ system. And I do think that there is a case, particularly for 905 radar where a 4D radar is kind of good enough. One of the things going back to the China market, I think that this is one of the things that a lot of Chinese OEMs are actively debating. What happened earlier this year is they're now in a price war. And so they're becoming a lot more cost effective. And so one of the things that they're debating is if you wanted to buy a cheap LiDAR today in China, it's probably like a $500 LiDAR. I think a 4D radar is kind of like less than half that price.
And so for an OEM that wants to kind of continue to differentiate themselves from a technology perspective, you can put a 4D radar on the car. It sounds cool. You're probably going to get just as much functionality or pretty close to it as you would at putting like a $500 LiDAR on it. So that's the kind of progress that we're seeing with 4D radar. The people that put our LiDAR on the car are the ones that kind of want to introduce that significant improvement in ADAS systems or that high-speed autonomy, and 4D radar just can't provide that.
Unidentified Analyst
On the -- kind of similar but just thinking about price downs over time as LiDAR potentially becomes a bit more ubiquitous. Can you give us a sense of what the price of, I guess, like other Luminar's LiDAR or high-quality LiDAR, however you want to define it, like what that is today and what the pace of cost/price declines might be over the next few years?
Thomas Fennimore
Yes. So for our current generation product, Iris and Iris Plus, what we said publicly, it's around $1,000 per unit. And -- but most of our large-scale customers, we have volume-based pricing. So in low volumes, it's a little bit higher; in high volumes, it's a little bit lower than that. And so the OEM culture is to go to their suppliers and ask for 2% annual price downs.
We've been successful so far combating that by saying, like, look, this is new technology, will give you volume band pricing. And if your -- as your volumes roll out and you introduce more platforms, you're actually going to get better pricing and more price downs from that higher volume than you would by the 2%, 3% annual price downs. So we've been successful doing that.
Our next-generation product is probably going to cost us roughly half to produce what our current version is. So that gives us a little bit more flexible -- it gives us flexibility to effectively lower the price without, I would say, sacrificing margin. The other thing that we're doing and working with our customers on is building the tools to monetize the ecosystem, the value in the ecosystem that our LiDAR creates. I've talked about 3D mapping, working with our customers to do that. We -- let's talk about the upgrade package. Our LiDAR enables both the significant improvement in ADAS, which delivers safety improvements to the vehicles, but it also can deliver highway autonomy.
So part of our pitch to OEMs and part of the reason why Volvo standardize our technology on their EX90 not only does it significantly improve the safety of the vehicle, but then you can upsell the consumer to highway autonomy. And the consumer has demonstrated an ability and willingness to pay for that, particularly on the high-end vehicles. When you look at Tesla FSD, which is an L2+ product, for the Model X and the Model S, their high-end product, $100,000-plus, that $15,000 trim package is getting a take rate based upon the data we see of about 40% in the Western world.
For the 3 and the Y, which is a lower-priced vehicle, the take rates are kind of in that 5% to 10% range. If you just play with the math and if you're able to charge a $10,000 trim package on a high-end vehicle for highway autonomy and you get a 30% take rate, that's $3,000 of incremental profit per vehicle that accrues to the OEMs. And if you standardize our LiDAR, you can -- we haven't figured a way yet to do kind of do an over-the-air update for installation of our LiDAR, but you can actually do over-the-air updates to either deploy or unlock the software on the vehicle.
So -- and then I would say, so 3D mapping, you kind of need our LiDAR to do it. The upgrade option on the consumer, you kind of need our LiDAR to do it. And then the other thing, it's like we kind of looking to get into the insurance business, we don't want to be an insurance company. But the reason we're doing that is we want to build a tool to monetize the safety improvements in -- that our technology brings. We're in the best position with our OEM partners to underwrite that. It would be great if the legacy insurance companies gave a discount. But in our conversations with them, they wanted to see years tables. So we're willing to bet on ourselves with our OEM partners.
And so what we're trying to do is change this less into a cost game, which is -- eventually hardware is always -- is going to be commoditized. I think that's going to be some point in the future. But what we want to do is build these tools and turn it into more of a value equation with our OEM partners. Here is the value that our LiDAR enables. And we can unlock it with mapping. We can unlock it with this highway autonomy up-selling the consumer. And one of the things that we're kind of having discussions with on -- with our customers is what we call dynamic pricing, right?
We'll take a lower cost on the LiDAR upfront, standardize it, but then we'll take a cut of the revenue on the back end when the consumer upgrades to autonomy, right, bet on ourselves. And then also on the insurance, like partner with our OEMs to do that to kind of capture those savings.
And so yes, whenever you have discussions within automakers, that upfront cost, just culturally, that's what they want to have, but we also want to have a more strategic discussion to say, look at the value and the ecosystem that's created with our technology. Let's build tools where we can capture this together. And when you look at that value and if you're able to capture it, it's a very, very big return on capital and investment.
Unidentified Analyst
Sorry, I'm new to the name. I'm just curious to know, have you provided a cash flow profile like looking years out this is what we expect to produce. And then part of the same question, what are the milestones required to that achievement? In other words, we need to have this much in volume, and we need to be at our next-gen system that brings our cost down to a certain level. We're going to spend at this rate. What are some of the key things that you want us to watch as investors?
Thomas Fennimore
Yes. So I would encourage you to -- and I'm happy to set up a call afterwards. At Luminar Day, we kind of gave some near-term, medium-term and longer-term guidance around it. I'll focus on the near term because that's the most near term and based upon, I think we have a lot to catch up. Okay. Well, I'll focus on the near term, then I'll give you the longer term.
So near term, first -- over the last 3 quarters, I've been burning in the mid-70s. And the reason for that is I'm industrializing my product. I'm putting in a lot of cost. I'm launching this new facility. And I got to talk to you about a lot of those costs. As we're successful in industrializing our product just by taking those costs out of the system, our target is to cut that mid-70s burn rate roughly in half by the end of the year. That will lead me with about $300 million of cash. You kind of do the back of the envelope math on that. That gives me at least 2 years of runway.
My cash flow should then improve again, right, the net stuff -- so first step function is a lot of those upfront investment costs start to rapidly decline as we're successful. Step function 2 is getting to that Volvo SOP. And so that will then take our cash burn levels down again. Our goal is to get to company-level breakeven by the end of '25. That's when we should have enough volume to get to that. And then when you get to 2030, so we'll kind of go to that level, our -- what we kind of said publicly as our targets is to be producing about 5 million units. Call it, that will be roughly $5 billion of revenue.
And we think, particularly with our next-generation product, we're going to have some pretty healthy margins. You look at high-tech auto suppliers such as Mobileye, or even like a Gentex mid-30-ish range as well. And so that, I would say, is the road map that we have. Our next-generation product, we have the design in place. We'll be talking about that more in the future. There's one component in particular that I would say is the most expensive. We kind of figured out what the right technological solution is. And so that is something that will cut the price of what it costs us to produce a sensor roughly in half.
Samik Chatterjee
Thanks for that. I'll wrap it up there given we are out of time. But thank you, everyone, for coming. Thanks for the time.
Thomas Fennimore
Thank you.
For further details see:
Luminar Technologies, Inc. (LAZR) JPMorgan Auto Conference Call Transcript