管理層發言
Greetings. Welcome to Mobileye's Second Quarter 2026 Earnings Call. As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Dan Galves. Mr. Galves, you may begin.
Thanks, Sherry. Hello, everyone, and welcome to Mobileye's Second Quarter 2026 Earnings Conference Call for the period ending June 27, 2026. Please note that today's discussion contains forward-looking statements based on the business environment as we currently see it, including regarding our future financial outlook. Such statements involve risks and uncertainties. Please refer to the accompanying press release, which includes additional information on the specific factors that could cause actual results to differ materially. Additionally, on this call, we will refer to both GAAP and non-GAAP figures. A reconciliation of GAAP to non-GAAP financial measures is provided in our posted earnings release. Joining us on the call today are Professor Amnon Shashua, Mobileye's CEO and President; Moran Rojansky, Mobileye's CFO; and Nimrod Nehushtan, Mobileye's EVP of Business Development and Strategy. Thanks. And now I'll turn the call over to Amnon.
Thank you for joining, everyone. Mobileye's core business continues to perform very well in 2026, providing a strong foundation as we build towards upcoming advanced product launches. In the second quarter, EyeQ volume was up 3%, outperforming the volume of our top 10 customers by more than 8 percentage points. While ASP came in modestly below our expectations due to a higher contribution from China OEM export volume, revenue was essentially flat and still outperformed production at our top customers by roughly 5 percentage points. Profitability benefited from recognition of the new R&D credit law, which came into effect shortly after the end of Q1 and is retroactive to the beginning of 2026. Adjusted operating profit was up 46% year-over-year and adjusted operating margin expanded by 10 points to 31%. Moran will get into detail on the credit law itself, but we view the benefit as sustainable. This policy was primarily designed to offset the higher Israeli corporate tax rate and to act as a retention tool to encourage continued R&D activity in Israel. Considering the R&D credits booked in Q2 also included in the Q1 impact, we think that the first half provides a more representative view of Mobileye's underlying performance. On that basis, results were very strong with first half revenue up 13% year-over-year compared to our core customers' production volume decline of 3%. First half adjusted operating margin was 23%, up 6 points on a year-over-year basis. Cash flow remains robust with $210 million of operating cash flow generated in the first half of the year. Overall, our current business is a robust, extremely profitable foundation. We see multiple secular drivers, including growth opportunities in India, China OEM exports into emerging markets and new customers, all of which are supporting continued volume growth above the broader market with Surround ADAS as expected to drive ASP growth starting in 2028. Moreover, we continue to secure new design wins across virtually all of the high-volume mass market applications our core customers need. When we move beyond high-volume ADAS into more advanced technology, such as point-to-point eyes-on, hands-off and Level 3 eyes-off, the pace of additional traction will depend in part on demonstrating real-world performance through upcoming SuperVision, Chauffeur and Drive launches. We have been consistent on this point the last few quarters. Until OEMs see a vertically integrated supplier system performing at scale on the road, some are choosing to experiment with multi-supplier architectures that they integrate themselves, typically on lower volume programs where the commercial and operational risk is more contained. The recent Stellantis awards are a good example of this dynamic. Mobileye won the high-volume 2027 program with Cloud-Enhanced ADAS, supporting highway hands-free driving and a cost-efficient package for the OEM, while a lower volume, later timing and higher risk program was awarded to other suppliers. We view that outcome as consistent with how OEMs are allocating risk today. Mobileye continues to win where scale, reliability and production readiness matters most, while some OEMs continue to test alternative architectures on platforms that carry less risk to their business. On Cloud-Enhanced ADAS specifically, this product provides an attractive economic bridge between base ADAS applications and more advanced autonomy, with gross profit per unit roughly equivalent to Surround ADAS and well more than double that of a base ADAS program. On the robotaxi front, we remain encouraged by the progress with Volkswagen Group company, MOIA, both in terms of the performance testing of our self-driving system as well as the build-out of MOIA's infrastructure. The program achieved a very recent milestone when MOIA began public rider testing with safety drivers in Hamburg, Germany, in vehicles equipped with our self-driving system. We're looking forward to additional milestones throughout '26 and '27. As the confidence in our technology has grown, it has led to our decision to establish a fully vertically integrated robotaxi offering where Mobileye will control all aspects of the value chain. This initiative, which will proceed in parallel and leverage the same self-driving system technology we have been developing for the last several years, targets launch in 2027 in at least one U.S. city. We believe this is a natural evolution to expand our potential share of this very large TAM. In addition to technology confidence, data from early-stage commercial services has clarified many question marks. Consumer demand has been strong, revenue per vehicle is robust and will be profitable under our cost structure and the regulatory environment is supportive. We have also noted slower-than-expected scaling by the perceived leaders in the space, which we see as providing ample time for us to build our ecosystem. This combination of factors made this a very clear decision for us. We have already established a cross-functional team to build this business and are deeply engaged with subcontractors for the vehicle platform, self-driving system installation and vehicle uplift and local logistics and infrastructure support. Our Moovit division is fully engaged and is a high-value asset for fleet supply demand optimization, trip planning and rider engagement. We are forming joint Moovit-Mobileye teams to leverage Moovit's past proof-of-concept work with multiple mobility partners and accelerate development. Moovit will restructure its resources, moving away and reducing headcount from the B2B side of its business in order to later focus on this new strategy. Finally, we will add the Mobileye logo to Moovit's app in the U.S. across hundreds of thousands of consumer users to enhance consumer recognition of the Mobileye brand. We see value in building the capability to participate across the entire robotaxi value chain. But the most important action is responsibility for the vehicle as it comes out of the uplift facility as a fully validated driverless car. This opens up many go-to-market options for us, including operating the vehicles within our own service, deploying them on a third-party platform or selling the vehicles to robotaxi service operators with recurring revenues as the vehicle generates rider fares. We also expect operating outside of an OEM to lead to more rapid validation and software update cycles. Before turning it over to Moran, I'll say a few words on my decision to step down as CEO once we appoint a successor. Mobileye is my brainchild. It started as an idea 27 years ago that machine learning can transform a monocular camera empowered by an appropriate system-on-chip into a system that can warn and mitigate imminent accidents at scale. Since then, we have delivered more than 250 million units of product, generated more than $13 billion of revenue, created thousands of jobs and saved scores of lives by preventing and mitigating accidents. We helped create an industry and changed how the world thinks about road safety. Eventually, fully autonomous driving became the overarching goal. Today, with SuperVision, Chauffeur and Drive moving towards commercialization, Mobileye is entering a new phase. Our business that runs through automakers remains central to Mobileye. At the same time, robotaxi and humanoid robotics are major long-term opportunities built on the same physical AI foundation and require new operational models and new go-to-market strategies. Our next decade may be even more ambitious than the prior 27 years. That is why I believe this is the right time to begin a search for a new CEO. The next leader should bring an operating profile to scale these opportunities and the mandate to lead the company into its next stage. This will be a singular opportunity to lead one of the most important physical AI companies in the world. As for me, following the nomination of my successor, my goal is to contribute to the technology strategy, innovation and long-term opportunities that can shape Mobileye's future. I'm proud of what we built, and I'm even more excited about what comes next. I will turn the call over to Moran.
Thank you, Amnon. Before I begin, please be aware that all my comments on profitability will refer to non-GAAP measurements. The exclusions in Mobileye non-GAAP numbers are typically amortization of intangible assets, which is mainly related to Intel's acquisition of Mobileye in 2017, and stock-based compensation, including the partial offsetting impact of the new R&D incentive law. This year, we also exclude the goodwill impairment loss that occurred in Q1 and transaction costs associated with the Mentee acquisition, which closed in early February. Second quarter revenue of $508 million was relatively flat compared to last year's Q2, which was our highest revenue quarter of 2025. Volume of 10 million was again above our expectations, driven by higher share within certain OEMs, higher ADAS fitment rates in emerging markets and upside to China OEM export volume. These three positive trends were consistent throughout the first half and pushed our volume above the approximate 9 million quarterly average we had experienced during 2025. On a sequential basis, we were down almost 1 million units from Q1 to Q2. It's important to remember that we believe Q1 included about 1 million of safety stock increase at customers. We believe that safety stock did not materially change during Q2. So underlying demand was basically flat from quarter-to-quarter. SuperVision deliveries of around 20,000 units were also above our expectations with a little over 40,000 delivered in the first half compared to end market demand for the vehicles of around 30,000. We do believe there is some intentional inventory building to protect against component shortages. This will be consumed during the second half. As such, we are incorporating a reduction in shipment volume in the second half of the year compared to the first, and maintaining our outlook of slightly below 60,000 units. Adjusted operating income was $155 million, up 46% year-over-year. Adjusted operating margin was 31%, up about 10 percentage points versus Q2 2025. The income and margin growth was more than accounted for by recognition of $93 million contra R&D expense related to an R&D incentive enacted by the Israel government during our Q2, as Amnon mentioned. Approximately 50% of the benefit was related to Q2 and 50% related to the Q1 impact recognized in Q2. Regarding the R&D incentive, this new regime became law during our Q2, but is effective as of the beginning of 2026. After a deep analysis and consultations with a variety of groups, we arrived at a specific recognition for the first half and have enough clarity to incorporate further benefits in the back half of the year that are expected to be similar in magnitude to the first half. A couple of points of detail around this new item. It's obviously a large positive impact to our P&L, which we expect to continue in future years. There is no end date to this regulation, although it's, of course, subject to future changes in the law. There is potential for some volatility in the quarterly recognition of the incentive. It can be impacted by meeting recognition thresholds, measurements of qualifying R&D expenditure incurred in Israel and exchange rate differences. It would also be impacted if Intel were no longer a controlling shareholder of Mobileye. This is a cash benefit, but the timing of cash inflows is quite delayed compared to the accounting recognition. For example, we expect the cash impact of the benefit recognized in 2026 to occur two years later gradually from the beginning of 2028. There is also expected to be some level of offset on the tax line, although not relative to Street expectations that Mobileye mid-term tax rate will be in the 20% range. Currently, in 2026, our effective tax rate is expected to be in the 8% to 10% range, which is generally aligned with the cash tax rate. As a result of Israel's implementation of the OECD Pillar Two global minimum tax rules, we would expect both P&L and cash tax rates to increase up to 15%, potentially starting in 2027. Again, we believe that it already incorporates into Street estimations. Turning to full year guidance. We are increasing the full year revenue outlook to $1.995 billion at midpoint and tightening the range, implying 4% to 7% revenue growth across the range. We see this as strong growth in the core business, given that the production volume of our top 10 customer is expected to be down about 4.5%. Our outlook midpoint is underpinned by a bit above 39 million EyeQ units, which is up almost 1 million units compared to our prior outlook. Partially offsetting the volume upside coming mainly from China OEM export volume is modestly lower expectations in the aftermarket and Moovit business and some push out of advanced product samples, which carry very high per unit prices into 2027. Additionally, the approximately $10 million of higher-than-expected SuperVision revenue in Q2 is related to timing and results in a lowering of our SuperVision revenue expectation in the back half, as mentioned above. We are increasing our outlook for adjusted operating income to $395 million at the midpoint, up from $210 million in the prior outlook. The guidance range for adjusted operating income actually widened slightly as the R&D incentive introduced an extra layer of potential volatility. We incorporate $180 million to $200 million in our full year outlook for this item. That positive impact, as well as contribution from higher revenue, is partially offset by some increase in expenses to support initial activities for our robotaxi expansion and a very modest increase to operating expenses related primarily to foreign exchange. Our assumption for full year non-GAAP operating expenses is approximately $910 million at midpoint, including the expected $190 million R&D credit. Turning to third quarter. We are assuming between 9.3 million to 9.5 million EyeQ units and for revenue to decrease approximately 5% to 6% on a year-over-year basis. We would expect gross margin to be slightly below Q2 levels based on the mix of orders we are seeing currently and for operating expenses, excluding the R&D incentive, to be slightly up from Q2 based on typical seasonality of higher operating expenses in Q3. The R&D incentive will go down significantly in Q3 from Q2, since Q2 also includes the retroactive impact of Q1. To conclude, I'm very pleased with the higher foundational margin of the business as well as our ability to largely keep operating expenses consistent with our initial outlook, offsetting some fairly severe FX headwinds that we were mostly able to offset with our hedging program and operational efficiencies. At this point, we are approximately 85% hedged for the second half, which should continue to reduce volatility. Also, operating cash flow was over $200 million in the first half, given that over $90 million of our adjusted operating profit was related to R&D incentive recognition in Q2, which won't be paid in cash for some time. This reflects very strong cash flow from the core business. Finally, I was pleased with our initial execution of the share buyback program. We were able to deploy approximately $24 million at an average purchase price of $9.37. Subject to market conditions and share price, we would expect to grow or maintain that pace over the course of the year. Thank you, and we will now take your questions.
分析師問答
Our first question is from Joshua Buchalter with TD Cowen.
This is Lannie on for Josh. Can you hear me okay?
Yes, we can hear you.
Okay. Great. My first question is on your robotaxi deployment. So I understand that Mobileye is moving into a vertically integrated strategy. But I do remember Mobileye talking about not wanting to own the fleet itself a few years ago. Can you address any details on the strategy shift and how your partners are kind of reacting to the change in strategy? And I have a follow-up.
Yes. And thank you for the question. I think what has changed in the last few years is more clarity about the business and some barriers that made vertical integration less attractive five years ago have changed. For example, Level 4-ready base vehicles are becoming available. In the past, each player had to solve this alone at enormous cost. Fleet management and customer-facing software can be efficiently built with modern AI and Moovit assets. Compute and sensor stacks have matured to a point where the driving platform is essentially settled. And when we look at the entire business case, there's more clarity about demand and about projected revenue for a robotaxi. So the numbers show, looking at our competitors, somewhere between $100,000 to $200,000 per robotaxi per year. Our calculations show $125,000, which is quite conservative. The cost of a vehicle with our sensors and our compute, given a very lean cost structure, would be below $100,000. And the operating expenses per year per car is going to be a few tens of thousands. So overall, this is a very profitable and very strong ROI. And there are many ways to fund the CapEx even when we go to the tens of thousands of vehicles. First, our cash reserves, the $1.2 billion to $1.3 billion, the expected operating cash of about $350 million per year going forward. There are many opportunities for external funding that we will start investigating when the time comes. So we are less concerned about the commitment, the financial commitment of owning vehicles because we see a very strong ROI to this business. And overall, what we're looking for is flexibility in the go-to-market. Now in certain geographies, we would go full vertical, including the demand generation through a customer-facing application, the fleet management, all in-house. In some territories, we would integrate to a TNC network. In other territories, we may license the vehicle. So it provides us flexibility in go-to-market.
Thank you for all that detail. I really appreciate that. My follow-up is on pricing. So I understand that typically the automotive industry has annual contracts included, but we've heard a lot about input costs going up across the semiconductor supply chain. And I was wondering if you would be able to address anything you're doing regarding managing margins and if you're able to pass or share any input costs with your customers?
Yes. So I think we can divide into two. First, in our EyeQ business, our exposure to cost, at least on the memory side, is indirect and we provide only the chip. And we haven't seen P&L impact of memory or some cost increases this year. And on the SuperVision side, there is a memory component that we buy. And we have observed some price increases this year that we fully passed through to our customers. It means obviously some headwind in SuperVision gross margin, but still it's a relatively small portion of the activity. So it doesn't impact significantly the total gross margin.
Our next question is from Joe Spak with UBS.
Amnon, congratulations, it's been great having a relationship over the years. I guess my first question is also related to your own robotaxi unit initiative. Is there any initial reaction from some of your existing customers in terms of how they now view their relationship with you that at least it could be viewed in a part of their potential future business you are a competitor as well as a potential partner?
I think overall, the response is positive because it means that we are doubling down on development of robotaxi and the self-driving system is a core element in it. And this, I think, overweighs any other consideration there is.
Okay. And then maybe just another one on the Stellantis win. My understanding is that was sort of upgraded from an existing program you had with them. So it seems like that's, I guess, to use non-engineering terms, a relatively less painful upgrade. And I know you mentioned that they have some other what you termed higher risk programs. I guess the question is as you sort of look at your overall customer landscape and programs, do you see a lot more potential for the sort of upgrade from current functionality to either Cloud ADAS or even Surround? Like is Surround also a less 'painful' or less risky upgrade for these customers?
So I think for the Stellantis program it is indeed an upgrade for an existing project that is already in production. Basically, the decision was that the next start of production milestone for that program would introduce REM through Cloud-Enhanced ADAS, which is a relatively straightforward implementation compared to designing a completely new architecture, a new ECU with a completely new sensor set and so on. For the OEM, it's a very modest investment, and there is significant upside. And for us, it's mostly about software. So it's again an easy execution. And it does provide some tailwind in terms of the ASP for Stellantis. Their intent is to adopt this pretty much across the board for Stellantis vehicles effective 2027 gradually, of course, but ultimately getting to a standard fit integration of REM in the Stellantis fleet. I think we've seen a couple more examples like this from OEMs that see this as kind of a low-hanging fruit. I think we are on pace to really expand our REM ecosystem as a consequence. Surround ADAS is a little bit different because it does require a different design. So it's a new chip; it's EyeQ6 High. In that case, there's already an EyeQ6 Lite program with Stellantis. So EyeQ6 High requires designing a new system and then integrating this, validating it and then launching it. So I think it's different profiles, but we are encouraged by this win. Of course, it's an important milestone for us.
Maybe I could sneak one more in. Just to be clear on your tax and the R&D credit comment. What you're suggesting is that this is not even really sort of earnings neutral. It's actually sort of like slightly earnings positive, the net of the two changes?
Yes. I mean what it means is that the tax or benefit regime has changed now in Israel. Israel used to give a reduced tax rate to technology companies, and having this multinational minimum tax rate has changed the way the Israeli government splits the benefit. Yet, it's a positive for Mobileye.
Yes. And I'll just follow up. This is Dan. It's more than slightly earnings positive because especially compared to Street estimates, which are essentially that we'll have a tax rate of around 20% in future years. So this incentive changes our R&D expense sustainably. It does not change the way we'll pay taxes in terms of relative to Street estimates.
Okay. So it's effectively a $200 million benefit then, is what you're saying?
Correct.
Our next question is from Chris McNally with Evercore ISI.
First, Amnon, I just wanted to send my best in your decision because I would say learning from you from the outside over the last decade, it's truly been, I think, one of the highlights of my career. So I wanted to say thank you first before the question. But maybe to that idea of the decision, I think what a lot of people are curious about because I imagine it's bittersweet is how much the team has accomplished to this point, but we're sitting at the dawn of full autonomous as you've kind of alluded to. Could you maybe give us your opinion on things like the perception in the RSS software stack where you've been so heavily involved? Like how much is left to 'be solved' or is left to be implemented by whoever your successor may be?
Well, thank you, Chris. We built a very strong foundation. We are at the point of really a big phase transition in the company. From the point of view of solving problems, we believe that we've solved the major problems. We have a Gen 1 system, being replaced by a Gen 2 software stack in about a month or so, and being replaced by a Gen 3 system, which is incredibly sophisticated and advanced by the end of the year. The KPIs for robotaxi look very good. We believe that we can start commercial deployment in Orlando by the end of the year, as we said in the past. I think the big problems are solved and are very close to final execution. Humanoid robotics is the next big thing, and we are still bullish on 2028 to be able to ship robots for the first use case, which we believe is going to be B2C and not B2B. So everything is ready from a technology point of view. But now this phase transition is more about operations. It's more about go-to-market. We need to explore new go-to-market approaches, especially in the robotaxi and in humanoids. Operations infrastructure should expand in order to support robotaxi. At the same time, AI is moving very fast. I'm also contributing a lot of science into Mobileye and I foresee that in the future it could be a stretch to continue to manage daily operations and be strongly hands-on on this fast-moving train called AI. So it is the right time to find someone who will take the growth of the company and the operations, and let me focus more on long-horizon thinking. It's never a perfect time, but among available time frames, I think now is the best time.
Sort of from the software level to the physical AI ops level is the way I'm thinking of paraphrasing that. And just a real question on the duration of this R&D benefit that Dan talked about. Maybe a follow-on to Joe's question. I think you mentioned that this is the level under Intel's ownership. Could you just talk about what that level would be if Intel was at a lower percentage or if the stake was ever owned by someone else?
Yes. So first, this benefit is definitely sustainable. Over the past ten years, Mobileye enjoyed as a preferred technology company a 6% tax in Israel, which was very sustainable. Now when there's a new tax regime, this benefit is no longer relevant, so the way to give the incentive to tech companies is through this R&D credit, which is sustainable and supersedes the lower tax rate. Regarding Intel, Mobileye being part of the Intel Group, because of the revenue threshold, there's a revenue threshold of ILS 10 billion, which entitles us to the highest level of benefit, which is a grant rate of 25% to 30%. But if there is a change in the controlling structure or Intel is no longer a controlling shareholder, the benefit would be based on the location of the R&D activity. Since Mobileye is very much centralized in Israel and specifically in Jerusalem, which is a preferred area that's also entitled to a high level of benefit, it would be entitled, bottom line, to about half of the grant we currently estimate. So approximately $100 million instead of $200 million, which is still high compared to most Israeli companies.
Makes sense. So it's a consolidated revenue and R&D look specifically for the benefit; whether it was Intel or someone else, it'd be whatever that R&D revenue share was applicable to the threshold.
Our next question is from Aaron Rakers with Wells Fargo.
I know you're not giving longer-term guidance beyond calendar 2026. But when we look at Street estimates, there clearly is an acceleration of growth into 2027 and 2028. And obviously, you guys have been building design wins. Could you give a quick overview of some of the key programs you see as really kicking into 2027 and driving that assumed accelerated growth that we see in Street estimates?
For 2027, we see the Porsche SuperVision program starting to ramp up. But 2027, we don't anticipate a big volume; it only starts to ramp up. That's the main growth driver in 2027. There is more: the robotaxi launches in 2027 with the Volkswagen Group and our own launches.
Okay. On the memory side, you talked about shipments above end-market demand and your memory attributes. As you think about that dynamic and given that memory may not loosen up in the near term, why would we expect your customers to compress their inventory? I'm trying to understand your thought process around the component supply chain and expectations into the back half of the year, particularly as it relates to memory.
We need to separate between our base ADAS business and the SuperVision one. For base ADAS, the overall system cost to the OEM is relatively modest and the memory component is small. So far, we haven't seen indications of changes in volumes or order intake for the upcoming few quarters. For SuperVision, we sell an ECU that includes memory components, so we do purchase the memory and incur the added cost, which we've transferred to our customers. As Moran said, so far that has been the case. Overall, the volumes for SuperVision in 2026 comprise a smaller portion of overall revenue, so these fluctuations do not have a meaningful effect on average gross margin or profit.
I mentioned it in my remarks: that's the reason we didn't increase the yearly forecast for SuperVision. We shipped between 60% and 70% of the volume in the first half, and we didn't increase the yearly forecast because our customer may have wanted to protect themselves from component issues, and there might be some headwind in the second half. But in terms of revenue, we didn't increase the guidance for SuperVision.
Our next question is from Mark Delaney with Goldman Sachs.
Let me add my thanks, Amnon. I appreciate all of the time you spent sharing your insights on the industry and taking our questions over the years and wishing you the best going forward. I had a follow-up on the Mobileye Drive business. You spoke a little bit on the progress that VW MOIA is making. Could you expand on what you're seeing there? I think there was a plan to have a driver out and start commercializing around the end of this year in L.A. Is that still on track? And as you think about the time frame to roll out the 100,000 overall order with VW, has there been any shifting in your expectation for that?
The launch in Orlando is on track. We believe we will get to the KPIs by the end of the year. Last week, there was an event showing end-to-end, including the teleoperation and the customer-facing and the entire end-to-end driving experience, and it was very successful. KPIs are on track. LA is scheduled for 2027. So in terms of commercial driverless deployment, testing will start early 2027, but the commercial driverless deployment is scheduled toward the second half of the year and additional cities. As for the volume of 100,000, it's difficult to give precise numbers now. If the technology performs as we expect, given the cost structure and Volkswagen production, this number could eventually be greater than initial estimates, but it's difficult to say right now.
My other was on Mentee and the humanoid market opportunity. You said earlier in the call that you're still expecting 2028 to be an important year for commercialization with a B2C focus. Can you speak a bit more on the path to get there? I think the company has been looking for proof-of-concept deployments in 2026 and some of the R&D you're working on with V3 and V4. It would be helpful to better understand that path.
On the hardware side, we are on the robot called V3.2, which was assembled a few weeks ago with additional capabilities. V3.5 will be ready in about a month and the model that is going to be produced in volumes is called V4, which should be ready Q1 2027. On the software side, there's accelerated development in reinforcement learning across thousands of different settings and objects and use cases focused on home use. We believe that the first deployment should be a B2C deployment to focus on a precise product definition and not start customizing to each business requirement. Then later, we will move into B2B. We believe that in 2028 we'll build around 500 units to start.
Our next question is from George Gianarikas with Canaccord Genuity.
I also want to say thank you to Amnon and best of luck. So my first question, you mentioned on the call that there's been slower scaling in the robotaxi space by the leaders. I'm just curious as to what your thoughts are as to why that's happening.
It's difficult to speculate, but I think it's around the cost structure. That could be one reason for slow scaling. But you should ask those companies. We believe that we have the right cost structure. We'll obviously not be the first in the market, but we believe that we can scale the fastest in this market. Therefore, the timing right now is really optimal.
Maybe as a follow-up, there's so much written around Volkswagen and what their intentions are in the restructuring. I'm just curious if you can update us on what your relationship is like with the company.
Our relationship with Volkswagen is very robust. We have multiple programs, not only the program with MOIA, we have the Chauffeur, we have the SuperVision. They are on track to commercialization. The first one is the SuperVision led by Porsche. With respect to robotaxi, Volkswagen is undergoing changes. We don't know how this would affect the different branches or activities of Volkswagen. But so far, our information is not different from what you have.
Our next question is from Dan Levy with Barclays.
Amnon, congratulations and best of luck on the next phase of your career. Wanted to start with a question on the go-it-alone portion of robotaxi. I think we've seen from others in the past that scaling can be extremely capital intensive. You previously had an approach with Mobileye Drive, where you effectively left the scaling to others and your approach was much more capital light. What's the strategy here to do this in a way that is not as capital intensive? And maybe you can give a sense of what the additional spend requirements are now that you're going alone, both in the near term and the midterm?
A few years back, we were contemplating whether to go full vertical or just to be a technology supplier, and we opted for the technology supplier route and have two major customers, the Volkswagen Group with MOIA and the Holon Group. Today, we believe that the conditions are ripe to expand beyond being a technology supplier and to go full vertical to diversify opportunities. From a revenue perspective, going full vertical increases revenue per car materially—about five times per car in our model. The CapEx investment is not prohibitive because there is a very strong ROI on this business, and there are multiple ways to fund it. For the first 10,000 to 20,000 vehicles, we can fund it using our own cash reserves and operating cash for the next five years that we generate. There are attractive ways to create external funding, which we'll investigate as the time comes. We're not deterred by the CapEx element because the ROI is strong, and we have conviction in demand and average revenue per car per year to build the business case. This is why we decided it's time to expand from a self-driving system supplier to a full vertical operator.
Great. As a follow-up, could you give an update on the business in China? I assume that's provided some upside on EyeQ volumes. Also, there are media articles that talk about one of your large customers increasing collaboration with some other competitors like Horizon. What's the confidence that some of these Chinese competitors can't expand to other parts of the world and that competitively you're still tied with your customers outside of China?
Regarding China volume growth, we've worked over the years with leading Chinese OEMs, most notably Geely and Chery. In the past couple of quarters, they have been increasing their export volumes significantly, and the vast majority of these volumes is with the Mobileye system and our EyeQ. We've benefited from their successful export into new markets, and they have nominated us for future programs as well. We see this as a vote of confidence in the Mobileye system for export markets, whether penetrating Europe or selling into emerging markets. This is the second year of steady growth in that regard, with a more accelerated pace this year. Regarding domestic Chinese competitors, they are talking about exporting, but so far we haven't seen a successful launch of a Chinese solution in major foreign markets like Europe, U.S., Japan or Korea. Outside of small examples in India or much smaller markets, they haven't successfully launched abroad. So far, we feel fairly confident in maintaining our position in those major markets.
One other point: regarding OEM export, the China OEM export business into emerging markets is creating a compounding effect where bringing vehicles with good ADAS into emerging markets is pushing legacy automakers to also increase ADAS penetration rates in those markets. It's a big reason our growth over market is strong this year.
Our next question is from Tom Narayan with RBC Capital Markets.
Amnon, hopefully we'll see you at CES. Your talk is one of the highlights of that event. On that topic, could you talk about CEO succession planning as it relates to Chris' question? I guess how does this inform how CEO succession planning might happen? And then I have a follow-up.
The Board has assembled a search committee. I'll help in the search. We are casting a wide net and not specifying a narrow profile. We want the best CEO for the coming ten years of Mobileye's growth. As for humanoids, this is exactly the area I want to focus more on. I believe autonomous vehicle technology is largely solved for our programs; I don't see open scientific problems there. With humanoids, I see a lot of potential and opportunities to innovate. This is an area I want to spend more of my time on. The new CEO will take care of company growth, especially areas where the technology is mature, and I can focus on long-horizon thinking and work with the Mentee team to ensure a successful launch in 2028.
Understood. On the owned-and-operated robotaxi with Moovit, the target is 17,000 fleet in five years. Some third-party forecasters have 300,000 to 500,000 fleet for robotaxis in the U.S. alone. Is there optionality to expand this approach further? You talked about the economics and why it's better to have an owned and operated effort. Could this expand if you wanted to, or is the strategy to keep it at this level so you can also be a supplier as well?
We are open to expand. We'll know better in 2027. In 2027, we plan to start with a fleet of 100 to 200 vehicles in a single city, which we will disclose when the time comes. The success of that deployment will determine how big of a fleet we want in the next five years. Some of it can be funded internally and some through external funding. We'll know better in 2027. Regarding the larger market forecasts, the 17,000 we talk about is a meaningful market share in the next five years.
If I could just clarify, the first Volkswagen Group vehicle that you may have SuperVision on would actually be Porsche, not Audi? Is that correct?
Porsche is leading the program in terms of the car models; it includes Audi car models for SuperVision as well.
Our next question is from Colin Rusch with Oppenheimer & Co.
Can you talk about the maturity of your humanoid simulation platform and how quickly you're iterating with that, and the cadence of impacting the hardware design from those learnings?
We have a simulation infrastructure and foundation models focused on about 4,000 different objects that the robot needs to identify and plan a grasping trajectory toward. We are on track with that effort. We are building a stack of both imitation learning and reinforcement learning through simulation for the home use case, and that's on track. With the V4 robot in Q1 2027, we could have it all integrated. The compute will be running on an NVIDIA Thor, the latest chip NVIDIA has in this area, and it will be strong enough to support our needs.
Excellent. One practical question: as you think about introducing humanoids into homes, how mature are your conversations with insurance providers and other services to ensure consumers are comfortable having these robots in their homes?
This is a conversation that will start in 2027. It's too early to have these discussions now. It's best to first demonstrate the product and have beta sites, then have those conversations. I believe there will be demand; our competitors are also focused on home use as a major market.
We have reached the end of our question-and-answer session. I would like to turn the conference back over to Dan for closing remarks.
Thanks, Sherry, and thanks, everyone, for joining our Q2 earnings call. We'll talk to you again in three months. Thank you.
Thank you. This will conclude today's conference. You may disconnect your lines at this time, and thank you for your participation.