Why Mobileye isn't worried about a recent loss of VW Group business

Volkswagen Group’s announcement in January it would switch from Mobileye’s to Qualcomm’s assisted-driving chip/software for its next-generation electric platform signaled the loss of one of the Intel-owned company’s biggest customers. Since then, however, VW Group changed management at its Cariad software arm and Mobileye subsequently announced potentially its most significant supply win yet. Its new SuperVision driver-assistance system will go into next-generation models for VW Group’s flagship brand, Porsche. Automotive News Europe Correspondent Nick Gibbs recently spoke to Mobileye founder and CEO Amnon Shashua about the company's turnaround.

Cariad said in January it had partnered with Qualcomm for Volkswagen Group’s next-generation software/vehicle platform, replacing Mobileye. Now Porsche has announced it will take SuperVision. What is the significance of this?

This opens the door to much broader volumes. It’s not just Porsche. Success here really means, I believe, a very, very long-term relationship because this system is so complex and the performance is so high that once it's in the car, you don't go and replace it with something from some other supplier. It's not just the brand, it is an entry to the entire group of what we think is a redefinition of premium.

Will SuperVision be integrated into the PPE electric platform that Porsche and Audi will share?

They call it e-architecture 1.2 [the new VW Group premium software platform launching with the electric Porsche Macan, the first PPE-based model].

Qualcomm looks to be going into software 2.0, potentially launching with the delayed Unified/SSP platform that’s designed to replace the MEB and PPE platforms at VW. Can you win that back?

At the end of the day, we are a supplier. Architecture 2.0 has aspirations around Cariad, Robert Bosch and Qualcomm, but that's 2028-29. I don't know where architecture 2.0 stands in terms of specifics. Architecture 1.2 is very, very specific. What matters to us is volume production. Give me programs where there is a start of production. As for [VW Group’s] deal with Qualcomm, let's see what happens.

You call SuperVision a redefinition of premium. What does this mean?

Sit in a car with SuperVision and you will know. You cannot have something more premium because it allows you to do hands-free driving everywhere. It’s already in 100,000 Zeekr vehicles. You can drive two hours in city traffic and not have a single incident, not a single intervention. It's mind blowing. And you can take that to an eyes-off system by adding imaging radars and a front-facing lidar. With today’s ADAS [advanced driver assistance systems], you have a front-facing camera, sometimes a radar, which by itself is complicated. Now you have a system with 11 cameras, five radars, high-performance computing, over-the-air updates. It's huge. If this is a success, you won't replace it. So, this is why we don't care about the [VW’s 2.0 operating system].

You are talking about Level 2 plus – hands-off, eyes on, right?

The driver is responsible. It's powered by high-definition maps that are covering the entire globe. This is one of Mobileye’s distinctive capabilities, the REM [Road Experience Management]. We have all of Europe, the U.S., Japan and China mapped. So, you can drive handsfree on every road. Punch in an address and you will get to that address. It depends on local regulation, but the capability of the system is that it drives everywhere. You can't have more premium than that.

Mercedes-Benz is already there with Level 3 in Germany and California. Does that put an pressure on you?

This is why we stopped saying L2, L3, but eyes-on, eyes-off. What matters now is the operational design domain, the ODD. You can start with a very, very simple ODD that provides no lane-changes, a 60 kph speed limit and only works with a lead vehicle, which is what Mercedes did. When we are talking about eyes-off in 2026, we mean full highway driving at speeds of about 130 kph, with the ability to keep functioning when you switch from one highway to another. So, for example, five hours of driving between San Francisco to Los Angeles could be eyes-off autonomous.

Is this with REM high-definition mapping? Other players say you need lidar for that.

This is the beauty of it. The conventional way of creating a high-definition map with lidar and cameras on a specialized vehicle costs about $10 million per city. Multiply that by the number of cities you have in a territory and this is huge. And then you need to update them. There is also a lot of information about how humans drive that you don't find in high-definition maps. Our map has all the accuracy of high-definition map, but with additional information from human drivers.

VW calls it swarm data. Is that your data?

That is us.

The former Cariad CEO Dirk Hilgenberg said in January that Cariad was created to take control of this data because they didn’t have access to it.

That was really untrue. Volkswagen can do whatever they like with this data. They can build their business, but we get a copy. There is a lot of IP on creating the data. It's not the raw data that you send to the cloud. It's processed data, about 10 kilobytes per kilometer. So, we have deals today with nine automakers that are sending us data, and we have millions of cars mapping the entire world every day at a very low cost.

How widespread is use of your EyeQ camera/chip?

We are all over. The first step is basic ADAS. If you buy a car with an EuroNCAP five-star rating, there is probably a Mobileye chip in there. Last year we shipped 34 million chips with the front-facing camera.

Are all of those 34 million cars harvesting REM data?

Not all of them can send data. Some of them are the older EyeQ3, or EyeQ3 chips [EyeQ5 is the latest system].

What are the benefits of REM for cars with normal ADAS chips?

It’s the cloud-enhanced data. If you add REM to a basic ADAS then the car can do lane-keeping assist even if the lanes are not visible or it can warn against running a red light. VWs Travel Assist 2.5 launched a year ago uses REM. It's really an amazing product. It’s just a front-facing camera, but with REM you feel like you have a system with many more sensors. Just the combined business of two leading customers -- Volkswagen is one -- is worth than $1 billion until the end of the decade, just from REM.

How many automakers have signed contracts to use SuperVision so far?

By 2026, there will be six automakers and nine brands with SuperVision. We have already announced Polestar and Zeekr. There are two additional Geely-owned brands that have not been announced yet. We also now have Porsche.

Are automakers making SuperVision standard or will they offer it via a subscription or as part of an options package?

With most of them, the system comes standard. One example is the Zeekr 001. I don't know how Porsche will decide to market it.

Are you worried more automakers will insource their assisted-driving solutions, following the same route as Tesla?

These systems are complicated. It's not just the technology, it's also the validation. Mobileye now has 400 petabytes of data to validate its systems. It doesn't make sense for an automaker to replicate what Mobileye is doing. We work with about 50 automakers and hundreds of car models. On SuperVision, where we have been working for a decade to perfect it, it doesn't make sense [to try to do something like that in-house]. That’s not to say that the automaker should do nothing. There is a lot to do. But you need to source what makes sense and develop what makes sense. For the stuff that Mobileye does, it doesn’t make sense to insource.

What areas should an automaker focus on instead?

Augmented reality and the driving experience. There are lots of things to do that that create the distinctiveness of the product for the automaker.

You are not competing in the infotainment sector. Is that something you are looking at?

If we do look at it, it will be from a point of view of synergy. For example, a hands-off system needs driver monitoring. A camera watching the driver. A stand-alone system doesn't know what the car sees, but we do. Therefore, we can create a much better package by combining that information. It’s the same for infotainment. If it's just doing the same thing that infotainment companies are doing, then we will not get into it. But if by combining the assets that we have, we can create a higher level of infotainment. We will look into it.

Do you work with any of the pure play autonomous car companies? Where do you see them going?

You have to map autonomous into two groups. One group is the robotaxi group -- Cruise and Waymo -- targeting geofenced areas. The cost of the system is very, very high. The second is consumer cars. There is always a human in the car, so we don't need a teleoperation, but you can expand it until it is completely end to end. The problem with robotaxis today is not the technology, it’s how you make a business out of them. Is that going to take a decade? Is it going to take three decades? Nobody knows. Mobileye is active in both camps. We are working with the likes of Schaeffler and Holon in mobility as a service, and with Volkswagen on the autonomous ID Buzz for 2025.