Mobileye Global Inc. (MBLY) Earnings Call Transcript & Summary

Okay. Thank you, everyone, as we continue day 2 of the Barclays Global Automotive and Mobility Tech Conference. I'm Dan Levy, I lead U.S. autos research coverage, and I'm very pleased to have with us Mobileye, one of the leaders in the active safety autonomous driving space. And we have Dan Galves. Dan used to do what I'm doing.

I did. Yes. And not as good as you're doing it. Yes, we share some previous employers.

That is -- that's correct.

No, it's good to see you back at Barclays. Before we start, I've been asked to like start reading some disclaimers before my fireside chat, so I'm going to go through this pretty quickly. Some of my statements today may be forward looking and are subject to meaningful risks, uncertainties and assumptions, including those set out in our cautionary statement on forward-looking statements contained in our third quarter release and 10-Q. I may use non-GAAP financial information in answering some questions. Reconciliation of those statements can be found on our website under Investor Relations, including in our third quarter release. We provided guidance for 2023, most recently on October 26, 2023, and we will not provide guidance for 2024 until early in 2024. The 2023 guidance was only effective as of the date given, and nothing in this presentation should be viewed as me in any way confirming or updating this guidance or providing any view on our potential results for 2024.

Got it. Done. Okay.

It's easier when there's a slide.

Okay. Let's actually just start. I think the big event that you have coming up is CES. So whet our appetite.

I think that CES is kind of our main promotional marketing, kind of visibility event of the year, and it will continue to be this year. Last year, we provided some information on kind of future revenue pipeline, design win activity for the year that had just ended, and I think we'll continue to do that. But I think the most important thing about our CES program most years is it -- Amnon is very -- our CEO, I think he likes to give a real kind of state-of-the-industry, state-of-the-union type of speech that's not necessarily 100% about Mobileye, but about kind of trends he's seeing in the industry. I think one thing that we'll focus on quite a bit this year is how we've been able to find ways to really kind of allow automakers to turn our core technologies into products of their own. We talked about a concept called tunable parameters in association with the Porsche announcement, but we'll go into a lot more detail about what that means. And kind of the sneak preview is essentially your -- our system can -- it comes with essentially tuning knobs that can kind of tweak, fine-tune hundreds of different parameters about how the vehicle actually acts. And that's, I think, where the OEM's focus is most, is creating a driving experience that's aligned with their brand that doesn't feel the same as every other competitor, and we think that that's an important way to collaborate with our customers. We'll also have our CTO, for the first time, give a speech on the Wednesday of CES that's going to kind of drill down on some of the technology items that Amnon speaks about the day before.

Great. In general, I know last year, you provided some backlog or order book color. Is that the type of thing that -- not -- maybe not necessarily for next year, but something you want to get in the habit of doing or maybe was that more one-off?

No, I think that, that's something that we want to get into the habit of doing. Like one of the benefits of our business is we have long-term visibility based on kind of the contracts that we've won that typically start 2, 2.5 years after you begin the contract and then last for 8, 9, 10 years. So we're currently signing agreements that run, in some cases, through 2035. So it gives you a good -- kind of a good sense. Nothing is for sure, because there's no take or pay in the auto industry. So it's ultimately all going to depend on how many cars are produced that are targeted within that contract. But because we have this good visibility, we think we -- it's kind of a helpful framework for investors to use to value us.

Great. Let's pivot just broadly to what we're seeing in the industry. And I think the theme of the year has really been this pullback on EV euphoria. Another theme that's emerged more lately is yet another trough of disillusionment on fully driverless AV. We know that the push to a more advanced ADAS, more advanced driver-assist type features, that was also a separate theme, somewhat of a conflation between that and EV and AV. What are you seeing on overall the bidding, quoting activity, the interest by customers in really adopting this technology? Is it in line? Is it accelerated? I'm guessing, based on the commentary you provided on the third quarter call, it's accelerated.

Yes, I mean it's a complicated question. I think that if you go back 5 or 6 years, kind of the interest in terms of fully autonomous vehicles was more about fleets of autonomous vehicles that were providing kind of robotaxi services or goods delivery services kind of within a city, deployed in networks. And I think that you had some OEMs in the industry have the sense that we could wake up one day and nobody is buying cars anymore. So there was a lot of investment in that side. But Mobileye always believed that the world had two parallel futures. And one would potentially be -- or the world could -- the AV world could develop in two different ways. One would be this kind of robotaxi, charged by the mile, owned fleets of cars deployed in networks, the cars would operate 60%, 70% of the day and everything would be more efficient. But we also believed in driverless technology or what we now call eyes-off, hands-free. So in a wide variety of conditions, you can do other things on your commute and let the car take over the driving for you, deployed in vehicles that you or me buy. We saw that we wanted to create a technology approach that could win, no matter which direction the industry followed. I would say that there's been a huge pivot over the last couple of years towards developing technology that could be deployed in consumer-owned vehicles that would kind of evolve, progressively take over more and more of the driving. And that's -- so that's the direction, and our -- the product portfolio we built while we were inside Intel serves that perfectly because our approach is targeted at a complete cost to the OEM of a self-driving system of $5,000 or $6,000. That is relevant for a consumer-owned vehicle. And building blocks of that approach can be deployed in systems that are not eyes-off, that are eyes-on. So our SuperVision system is supported by 11 cameras around the vehicle that complete -- that provide a complete 360-degree view of the environment, high-definition mapping, decision-making software. The failure rate is not going to be better than a human, so you have to keep -- you still keep your eyes on the road, but you can provide value. And that system is already on the road with several different customers generating revenue for us. So I think our -- one of our principles has always been, you should look at an approach that will get to fully autonomous vehicles when it gets there, but pieces of that approach should generate revenue today or over time. And that's kind of -- so I think it's really the way that the industry has evolved is kind of playing perfectly with our approach. I think the idea of all of the cost pressures that are in the industry today, whether it's from union negotiations or EV margins or kind of high levels of investment to support EVs is also a positive for us, because I think it makes OEMs more likely to look for kind of a high probability, delivered solution from a supplier like us at a reasonable cost point with proven performance and good time-to-market rather than internally developed systems that have -- carry higher investment costs and have, I would say, lower probability of success versus something that's already on the road.

So this point here on why customers want to engage with you. The data point that you gave on the last earnings call was that you're engaged or you're in advanced talks with 10 automakers on either SuperVision or Chauffeur. It's across regions. There's no EV pure plays. Is this -- what is it that's driving this level of engagement? Is this exactly what you're saying, that this is the easiest and most cost-effective path of scaling on this technology? And then I would just ask as a follow-up, we have these awards, and I think the metric you gave is like 50%-plus likelihood of converting into something. So how do we take this from a discussion to a hard award?

Yes. No, it's another good question. So I think that what we've seen, because if you go back to the beginning of the year, that probably -- that number -- those metrics would have been maybe 3 OEMs representing 9%, 10% of the industry. And now it's 10 OEMs representing 34% of the industry. I think what's really happened is that -- I think that the industry has seen a major -- kind of a very kind of fast-moving development of hands-free systems in China. We have companies like XPeng and Li Auto and NIO all deployed systems that, in certain environments, navigate the car for you. Again, you have to keep your eyes on the road, very similar to a client with a Tesla FSD type of technology. And consumer demand for these systems and interest by the media in China is like off the charts. And so there's huge demand to put these systems on the road very quickly. We deployed a system with an automaker called ZEEKR that has similar type of functionality. We think better, but that's subjective. So that kind of fast level of development in China has, in combination with many announcements of Chinese automakers looking to export vehicles into Europe and potentially North America at some point, has led the legacy global automakers to focus on more pragmatic concerns like time-to-market, cost and performance. And I should say like the alternative for the legacy automakers has been a desire to develop internal software to support this type of functionality and deploy it on a processor from maybe NVIDIA or Qualcomm. So the alternative to Mobileye, kind of an outsourced solution for Mobileye has been internal development. And these programs seem to not be kind of resulting -- they're either being descoped or the cost is too high. So -- but I think that the kind of the aggression by the Chinese automakers in this space has led to a focus on more pragmatic considerations by the OEMs. And then the final piece to kind of this flywheel was we successfully delivered the SuperVision software into ZEEKR vehicles in August, more than 100,000 vehicles. And the system has been performing, I think, better than we expected. It's getting great consumer feedback. So I think in some ways, until you actually see software in consumers' hands and out of the kind of the demo world, you don't really know for sure. And so I think that, that was a huge proof point to the industry that what we've been talking about is actually a real system that can be deployed quickly. So I think that that's how we think of kind of why this traction has really been kind of increasing over the course of this year.

And the path to -- like what are the gating factors? Is this just timing, companies have to go through their processes to grant the awards?

Yes. I mean I think that these are strategic decisions. Like the typical kind of our core products, driving assist products, like you have a normal sort of RFQ process where you kind of submit your bid, it's evaluated by purchasing people, the engineers kind of all know your capability, and it's a very kind of clear kind of cadence of timing. With these kind of more strategic decisions, there's a lot more factors. Like if you have a large internal team, like what do you do with them? What are their roles and responsibilities? Because I think the automakers, they want to have a role in these programs and they should because there's a lot of consumer-facing aspects to it. How does the vehicle drive? Do I want to follow the common speed or the legal speed? There's a lot of decisions that the OEM has to make. So you need to kind of work out who's going to -- what are the roles and responsibilities. There's more aspects of terms and conditions and probably a little bit more liability discussion. So it just takes a little bit longer.

Can you address -- actually, before we go to 2024 SuperVision, maybe you can just give one more on the scope of opportunities here, right? Because we know that there's SuperVision, there's Chauffeur. You have hardware, hardware/software. Maybe give us a sense of how we think of the split on hardware/software, SuperVision, Chauffeur. And then how many of these are for standard fit vehicles?

So I think that we're hardware/software integrated. So we -- the lifeblood of our company is software, but we design our own chip and the software comes deployed on a chip. Now the way that we're selling SuperVision, the way it works today is that for all ZEEKR 001s, ZEEKR 009, there's several other models, the hardware comes on each car that's produced, and it provides like a basic kind of Level 2+ capability like adaptive cruise control, lane centering, things like that. But if the consumer wants to have the full kind of navigation capability of SuperVision, ZEEKR calls it NZP, so kind of navigate-on-pilot type of technology where the car understands where you're going, knows where to position the car to get off the exit, makes automatic lane changes, the consumer has to pay extra for that. And we would get kind of a back-end software licensing fee for that. But the bulk of the revenue per unit is coming from the initial delivery of the hardware. Does that answer your question?

For customers, right, like I think there was -- it's either a software-only solution or it's software plus the domain controller. So to what extent are the customers interested in, okay, we just want software, which is like $750 ASP, or we want that plus the hardware that -- that's what gets you to the full $1,500.

Yes, you're talking about more are we the Tier 1 or are we the Tier 2.

Yes. Sorry.

No, that's fine. As the Tier 2, which is our position in kind of the core driving assist business, we would just deliver the chips and the software. With -- and then if we're the Tier 1, we're delivering those chips and software on a domain controller. Right now, all of the kind of the booked business is Tier 1, so it's the domain controller. If we look at the pipeline, it's a mix of Tier 1 and Tier 2, and that's actually best for us because I think that being the Tier 1 requires more kind of on-the-ground development and validation work directly with the OEM, which requires us to build teams. So there would -- we want to limit execution risk as much as possible. So if we can have a mix and we can have partnerships with kind of current Tier 1 partners to kind of help validate and develop some of these programs, assuming we win a lot of them, then that would be best for us.

Great. One of the key things for the stock, obviously, on a more near-term basis has been the volume of SuperVision. This year, you've guided to -- it'll be in the low 100,000 range. I don't know if there's any updates you can provide for fourth quarter. But as we think about 2024, which I think the indication from the last earnings call was something in the low 200,000 range, maybe help us bridge from where we are today to next year. I think there's a variety of models. 001 is still the key model, but Volvo EM90, Polestar 4, Smart, how are we bridging from today to next year on that low 200,000 number?

Yes. I think we -- what we said was that that's where consensus was, and it seemed reasonable, right? That's what we said on the Q3 call. And I think that what we're seeing or -- so we started this year with only one car in market, ZEEKR 001. And then the ZEEKR 009 launched in -- kind of sometime in Q1. And that's really been it until last quarter, the Smart #1 launched, Polestar 4 just started shipping to customers in China or delivering to customers in China a couple of weeks ago. So you really have kind of a full year effect of some of these launches, as well as the ZEEKR 001 is going to be sold in Europe. They're starting to deliver to customers in Europe now, so that should drive incremental volume. Polestar 4 currently is delivering only in China, but it launches in Europe, I believe, in Q1, and in North America in the middle of next year. And then FAW launches towards the end of the year. So essentially, I guess the takeaway is we don't require any sort of incremental -- we don't require any growth in terms of vehicles that are being shipped today that are sort of at run rate, like we don't need that run rate to go up next year.

So it's just about expansion of new models.

It's about expansion on the new models and then getting some volume from Europe and the U.S., which we're really excited about. Because again, like we want -- we think that these proof points in front of our other customers are really important. So if we have a successful launch of SuperVision, not just in China but in Europe and the U.S., like I think that's a big benefit to us.

And as we think about in 2026, if we think more about just what the timing is for winning a program and getting that to be in 2026, how far in advance do you have to win that program?

So with a Chinese automaker, so for example, like we signed the agreement with FAW in September, and that launch is in late 2024. So a little bit over a year from design win to launch. And we think that that's pretty typical of Chinese automakers. So if you win business today or in the next couple of months, it's going to launch in like early '25. For non-Chinese automakers, it's about a year longer. So I would say like the designs that we're trying to win today with non-Chinese automakers are more like first half, middle 2026-type of launches.

Okay. Great. Competition, and I think you -- I think we can look at competition in two ways. One, I think you addressed internally why it makes more sense to use a Mobileye versus an internal solution to scale and whatnot. But maybe you can talk about the external competitors, be it a Qualcomm Arriver or an OEM that is doing things internally using NVIDIA, other Tier 1s. I mean what are you seeing really as -- when you're getting RFQs, what are the names that you're seeing? And then is there anything you can comment on share or win rates?

Yes. So I think that it's a little bit confusing, because even if you hear about a kind of a relationship with Qualcomm or NVIDIA, in our space, it's the OEM that is responsible for delivering the software, right? So they're really -- it's a combination of internal efforts on software from the automaker that is deployed on kind of a high-powered processor from Qualcomm or NVIDIA, maybe Ambarella. And these are typically the names -- not so much Ambarella, but these are typically the names that we hear about from the customers that we're pitching SuperVision or Chauffeur to. The 10 OEMs that we've talked about that were in advanced discussions have essentially told us that they're no longer evaluating those -- that kind of direction. And now it's more about, do they want to pursue this? What's the timing? Kind of roles and responsibilities, like I was talking about before. So I think that there was this view that for an automaker, it makes sense. If you can control the IP in a high-value space like this and you have the ability to completely customize the system, you could differentiate from your -- the other automakers and you could end up with a high-value position if you did it better than others. I think that the -- and this has been going on for 5 or 6 years, I think that the success stories are few and far between. And it's also very expensive. And so it -- so it goes back to the original comment of this flywheel effect where you see sort of the fast-to-market development of the Chinese. Whether it's using us or whether it's kind of internal development by some of the start-ups, you see the consumer uptake and it's -- you see the potential for exports outside of China and you start to focus on we need a higher probability of success. We need to get to market quickly. And Mobileye, we think that we have significant advantages in those areas. Like we think we have -- our system is kind of very low compute versus the Chinese startups that are doing this. They have like 1,000 TOPS of processing power in their vehicles. Our system has about 50 TOPS in the ZEEKR, so you've got 5% of the compute. You're not using LiDAR. We have a crowdsourced map that's scalable to kind of large geographies. We have 90% coverage of our mapping in the U.S. and Europe. So this all leads to advantages in cost and performance. Time-to-market is clear because you're already on the road with the product, so there's really no uncertainty there. So I think that this is where we're winning right now. There's a lot of traction. So I think that the competitive landscape is more about internal development of software combined with a processor. Maybe that processor comes with toolkits and libraries and software development kits that are very helpful. But at the end of the day, someone needs to come up with the software, and we feel like we have significant advantages there.

Can you talk about the competitive dynamics in China?

Yes. So I think in China, you have -- I mean in Tesla, I think has had a lot of success in this space, right? And we have a lot of respect for what -- for the system that they've developed, like Tesla FSD. It's a very good system. We still think ours is better, but there's been -- it's impressive, what they've been able to do. And I think part of it is because they've built their company around a culture of software engineering. Now you have startups in China that kind of modeled themselves in that way, right, NIO and Li Auto, XPeng. NIO hired a guy from Tesla in 2016. He built a team. They've been able to put a pretty good system on the road. So I think that this combination of internal OEM development and processors from -- whether it's Horizon Robotics or NVIDIA or Huawei, it's shown some progress in China. But what we're seeing is that the legacy automakers don't have that type of culture. Because there's legacy automakers in China, too, companies like FAW or Great Wall or BYD. So we feel like because the market is developing so quickly in China, it's very important to win there. And the fact that we have delivered what is being called by third parties kind of the best performing system on the road despite not having LiDAR, despite having lower compute and as a result, significantly lower cost, we -- this is kind of leading to much more substantial discussions with the rest of the Chinese industry as well, which I think we're kind of waiting for the system to get on the road and see how it worked.

Great. Let's pivot beyond SuperVision. And I want to start actually with AMaaS. Just given everything we've seen with this slowdown in fully-driverless AV, and you provided some commentary earlier this year that you were pulling back on some elements of AMaaS, autonomous mobility as a service, that was maybe not as scalable. So where do your efforts stand on AMaaS? And to what extent could you further pull back efforts to just focus more on the more scalable SuperVision, Chauffeur opportunities?

Well, the great thing is we don't have to pull back, because the technology that we're developing for Chauffeur, which is the consumer autonomous vehicle technology, is really the same technology that we're targeting for the mobility-as-a-service business. So it's a complete set of cameras that's capable of driving the car on its own. It's a complete set of radar and LiDAR that's capable of driving the car on its own. It's the high-definition mapping. It's the decision-making software, and then it's our EyeQ6 High that can be deployed in 2, 3, 4 chips. So we don't have to pull back. I think what we've pivoted to -- a few years ago, we did have an ambition to own our own fleet of cars in Israel. Maybe it was supposed to be more of a proof of concept than something scalable. But we've really kind of pulled back from that. And we said like so we need to focus on what we're good at, which is software and hardware and then find -- you also need a car, right? So we have relationships with 3 different platform builders that are building sort of autonomous shuttle type of vehicles for us. I shouldn't say just shuttle vehicles, because I think using like just a regular SUV to try to do the robotaxi business doesn't make sense. You want to be able to carry 6, 7, 8 people at a time. So we're -- we have these kind of vehicle platforms being developed. They launch in 2026. And then we look at the demand side in terms of like where is the relevant use case for this technology? And one that we found is in Germany, like a company like Deutsche Bahn, which is the public transit operator in Germany, they have these like 50-person buses running around, 10%, 20% utilized with a very expensive driver that they can't get enough of. So if they were able to remove that driver and replace that bus with kind of a 10-person autonomous shuttle and allow their customers to schedule rides and you could create routes and essentially make that business much more efficient. There's still a lot of like positive kind of bullishness in that type of space where you have -- I think recently, we -- the Minister of Transportation did an event in -- the German Minister of Transportation did an event in Hamburg and talked about 20,000 autonomous shuttles in Hamburg by 2030. Now will that come true? I don't know. But the technology, we believe, is progressing very quickly, but I think we -- it's just as important to set up the right business model where you have kind of the right customer that understands how to maintain vehicles, that understands how to charge vehicles, that has a customer base. And then you team up with a vehicle builder that understands how to create a vehicle that is preengineered to integrate with your system. And then we're kind of in the center, providing the self-driving system. So we're still very positive on that business, but the great thing is we don't need it to develop next year, the year after or the year after. We're already generating a lot of cash flow. We have a very kind of significant potential for semi-autonomous systems to be inside vehicles, which will create more growth for us and kind of continue to fund the development of fully autonomous vehicles.

What percent of your operating expenses is autonomous mobility as a service?

I think around the time of the IPO, we talked about that 50% of the R&D was related to kind of eyes-off technology. So Mobileye Chauffeur and Mobileye Drive, and then maybe another 20% to 30% was related to SuperVision. So 70% to 80% of the R&D is related to products that are not generating anywhere close to the majority of our revenue yet.

Questions in the room?

A follow-up on the pricing. You mentioned $5,000?

So this would be total -- I'd like to talk about total cost to the OEM, right? So a typical kind of collision avoidance system that's in most cars today is maybe anywhere from $200 to $350 total cost, and we get $50 out of that. That's our average price of the single chip system. SuperVision is more like a $2,000 total cost to the OEM. It's -- today, it's $1,300, $1,400 to us for the domain controller, but the OEM has to buy the cameras, there's extra electronics, so it's about $2,000. We're targeting total cost of OEM of an eyes-off, hands-free kind of consumer level system of $5,000 to $6,000. Of -- go ahead.

Is that apples-to-apples with FSD? $15,000 back in the day, lowered to like $1,000, it varies depending on promotion. In some places, it's like $200.

$200 a month? Yes. Yes.

So we've got great details on the take rate.

Yes. So I think we -- okay. So Tesla FSD is an eyes-on system. Maybe there's ambition that someday, you'll be able to like go to sleep or rent out your car. But today, it's an eyes-on system that they're charging, I think, I don't know if it's $10,000 or $12,000 for the consumer. And the take rate from what I read is pretty low. I think what we're talking about by a $5,000 to $6,000 system cost to the OEM, that would be eyes-off, right? So you could, at least on the highway, do other things, right? And it would be engineered to do that from the beginning with multiple levels of redundancy, the two different perception systems. So I think it's a different product. Like I said, like maybe Tesla FSD will get to eyes-off, but right now, it's not.

Just a quick follow-up, pure redundancy radar, LIDAR type of camera, do you just view that as like [indiscernible] eyes-off?

We think that -- we think you have to prove that you have a failure rate better than a human in order to satisfy regulators in order to get the consumer to actually activate the system and read a book and to also protect yourself and your customers from liability, right? So our view is that a camera-only system, at least in the foreseeable future, can't get to that type of failure rate, but it can be pretty good. And we think that if you have two separate perception systems, each of which has low failure rates, but nowhere close to a human level, if you have both of those systems on the same car, then the risk of both of them failing at the same time goes down exponentially. That's our approach.

Maybe just a quick one to close, and touching on the Chauffeur opportunity, higher ASP. To what extent is your development of radar and LiDAR a gating factor? Or is that something that you could work with external -- it sounds like radar, the imaging radar you have is unique. LiDAR is a bit more of a -- you can work with external players?

Yes, I think we're definitely going to be working with external LiDAR suppliers for at least this kind of first generation of launches. We do have -- we do believe in our LiDAR product, but that's not as strategic as the imaging radar. The key to the imaging radar is, we believe you can get -- the development of our technology will result in output that's very similar to LiDAR. So you could limit the number of LiDARs that you need around the car, right? And this is a key to getting to this kind of $5,000 to $6,000 price point, is having a technology that can -- you can put 4 or 5 on a car that gives you similar output to LiDAR, but still at like 20% of the cost of a LiDAR unit. So it's an enabler to get to that price point that we think is important because we don't think people are going to be paying $15,000, $20,000 per car for an autonomous system.

Okay. We'll leave it there.

All right. Thanks, Dan.

Thank you so much. See you at CES.

Okay, thanks.