For the intelligent driving industry, 2024 is the best year and the worst year.
The good news is that the terminal market is rapidly embracing intelligence, whether it is a high-end market of more than 300,000 yuan, or a mass range of 10-200,000 yuan, and even traditional fuel vehicles have begun to roll up intelligent driving functions.
But the competition has also become fierce. Taking the "urban NOA" that major manufacturers are very concerned about as an example, in order to realize the navigation assistance between two points in the urban area, compared with traditional models, it is necessary to install cameras, lidars and other perception hardware in the hardware, and also cooperate with different auxiliary tools such as end-to-end large models, rule algorithms, and high-precision maps to escort intelligent travel.
Therefore, in addition to achieving technological breakthroughs and mass production of products as soon as possible, another important proposition for all autonomous driving companies or OEMs is:
How to achieve intelligent cost reduction.
At present, the two basic technical routes that have reached consensus are the removal of high-precision maps, and the switch from the traditional rule-based technical route to the AI-driven end-to-end route. However, there is still a disagreement on the technical route, that is, whether it is necessary to add more expensive lidar to L2 level intelligent driving.
Mobileye has already gained a certain share in this wave of intelligence, with its SuperVision product line shipping more than 100,000 units in 2023, and has transformed from Tier2 to Tier1 direct-to-OEM. But at the same time, in the field of high-end intelligent driving, they are facing fierce competition from NVIDIA, Huawei, and the self-developed teams of OEMs - and high-end intelligent driving and even higher-level autonomous driving is just a huge share of the cake, which can bring tens or even hundreds of times the growth of basic ADAS products at the level of single vehicle revenue.
During the 2024 Beijing Auto Show, two executives from Geek Park and Mobileye had an in-depth communication. In the process of communication, Mobileye firmly embraced the "pure visual route" (L2 level) and made an anti-consensus judgment that "more than 2 million vehicles in China have urban NOA" capabilities.
They also talked about their perception of the Chinese market as an international company, and they believe that "younger and more tech-loving Chinese customers" are becoming an important driving force for technology companies and car companies to invest more in smart driving.
The following is an interview with Elie Luskin, Vice President and Managing Director of Mobileye China, and Johann Jungwirth, Executive Vice President of Mobileye's Autonomous Driving Business, compiled by Geek Park and other media.
Elie(左)和Johann(右)接受媒体采访 | Mobileye
From low-level to high-end, the revenue of smart driving companies can increase by up to 800 times
Q: Where does Mobileye see the future growth of autonomous driving technology? What is the plan for autonomous driving?
Johann Jungwirth:
We believe that the entire ADAS (Advanced Driver-Assistance System) and AV (Autonomous Vehicles) market is in a period of rapid growth, and many new manufacturers are emerging and entering this market. So the potential of this market itself is huge.
For Mobileye, we currently have three product lines: the first product line is SuperVision, which mainly relies on cameras and radars; The second product line is the Chauffeur series, which has reached the level of L3 and has cameras, lidar and ultrasonic radar; The third product line is the Mobileye Drive series, which is mainly aimed at Robotaxi.
For SuperVision products, we were originally a Tier2 supplier, but now we are Tier1 and sell directly to OEMs. For ADAS products, each car can bring in about $60-100 in revenue, but in the AV field, each car can bring in $1,000-50,000 in revenue.
Reporter: Now that the cost of lidar is getting lower and lower, how does Mobileye view the competition between the pure vision route and the lidar route?
Elie Luskin:
In fact, we believe that for ADAS-level intelligent driving models, there is no need to add lidar or radar, and only the camera system has Mobileye EyeQ chips. The situation is different in the high-end AV market, and we know that Chinese automakers targeting the high-end market have used very expensive sensing equipment and integrated chip systems from the beginning, but we think this is not sustainable. They're also thinking about reducing costs, and the first thing they want to get rid of is the sensing device. Especially for some "hands-off but can't-eyes-off" solutions, the (redundant) sensors may be the first thing to be removed.
Johann Jungwirth:
LiDAR is an option for our SuperVision solution. We believe that at the L2++ level, LiDAR is only to provide more safety and comfort to the vehicle. When it comes to the L3 and L4 levels, LiDAR is a must-have. At the L3 level, the front-end system requires lidar; At the L4 level, not only the front, but also three long-range lidars around the perimeter are required to assist.
The 2024 all-new ZEEKR 001 equipped with Mobileye's intelligent driving solution is equipped with LiDAR as standard Extreme Krypton
"There are 2 million vehicles in China with urban NOA functions"
Reporter: At present, NOA (Intelligent Pilot Assisted Driving) is the focus of attention of Chinese car companies and users. However, the cost of some high-speed NOA products is very high, and urban NOA will face many problems. In Mobileye's view, what changes need to be made to make consumers go from being able to use to loving to use?
Johann Jungwirth:
Regarding NOA, the biggest complaint from customers is the problem of coverage, because "NOA can only be used in certain cities and certain environments", so the core is to expand the coverage. Mobileye has REM intelligent driving map technology, which is a crowdsourcing solution, as long as the car with our system is installed, it can give us feedback data, and we no longer need to send collection vehicles on the road to collect road information, so we can quickly improve the coverage and scale up.
Elie Luskin:
We analyzed the data of the Chinese market in 2023 and believe that there are 2 million cars running on Chinese roads, and their on-board systems are technically fully capable of supporting urban NOA. Therefore, we believe that NOA will be able to take root well in China. Seeing this, Mobileye will also provide full-stack technology solutions according to the needs of end users and Chinese car companies, and we believe that Mobileye's solutions will be very marketable in China, considering that there is a lot of pressure on the profitability of the self-developed intelligent driving systems of car companies.
Mobileye副总裁及中国区董事总经理 Elie Luskin | Mobileye
Reporter: "2 million cars can support urban NOA", this statement seems to be inconsistent with the number of cars that have actually been pushed to the urban NOA function. How did Mobileye make this judgment? What are your criteria for judging whether a car can support an urban NOA?
Elie Luskin:
This 2 million units, I mean that they have reached NOA's standards in terms of hardware, including Tesla.
Johann Jungwirth:
The key to judgment is the sensor's perception and computing power. The sensors must be surrounded, maybe Tesla is 8, Mobileye's SuperVision solution is 11, but the number is not critical, the key is the information collection of the surround. The key is that there must be a camera or sensor in the two rear corners of the car, otherwise the car will not be able to change lanes safely and smoothly. In addition to sensors, there are also chips, whether it is using Mobileye, NVIDIA, or Huawei, the computing power must be able to cover the processing needs of the sensing side.
Chinese consumers are pushing traditional car companies to upgrade their technology
Reporter: As a global company, Mobileye has a presence in different markets such as China, Europe, and the United States. For other markets, what do you think is the biggest differentiation for Chinese customers? Is the Chinese market the hardest thing to do?
Elie Luskin:
Chinese consumers and Chinese automakers are more technological, which is why our SuperVision debuted in China. In fact, after the launch of SuperVision in China, there were a lot of followers around the world, who came to us and said, "We also need you to give the same solution to Chinese car companies", such as India's Mahindra and Europe's Volkswagen Group. I think the Chinese market is leading the global trend in this regard.
Johann Jungwirth:
In the Chinese market, Chinese customers actually have higher expectations. It may be because Chinese customers are on average younger and more tech-loving, like Xiaomi's users are very tech-loving, so we think that Chinese consumers are now forcing technology companies and car companies to invest more in these high-end technology systems, in fact, we also know that some traditional car companies may really need some of this kind of push in this regard. So I think from a technical point of view, the possibilities are limitless, and there's a lot of potential. For consumers, what they might expect is this: I get in the car, I type in an address, and the car will be able to take me there, and the car will probably be driverless most of the time, and the middleman will take over. That's basically what consumers think.
Elie Luskin:
Chinese customers were early adopters of advanced technologies.
Johann Jungwirth:
It is also the most demanding, users often say "hurry up OTA", want "new features", and they want the highest technology in terms of advanced technology. China is leading in this market.
Mobileye副总裁及中国区董事总经理 Elie Luskin | Mobileye
用低阶智驾为Robotaxi输血
Reporter: Tesla's FSD system has recently dropped in price in the United States, which actually reflects the relationship between cost and market competition. How do you understand this balance between cost and security? Especially in the domestic special volume market?
Johann Jungwirth:
In the long run, it is possible to balance cost and safety. Because from Mobileye's point of view, even if we can't say absolutely that it's a pure vision solution right now, at least it's still camera-first. We believe that cameras are still the most natural, the most natural form of sensor like the human eye. Now we have a front-facing camera of 8 million pixels, which is higher than other manufacturers. As for the system between Tesla and us, 8 cameras or 11 cameras, the difference is not big. Over time, these costs should be able to come down gradually. But there can be no compromise on security.
Reporter: Because I just mentioned the intelligent driving taxi, this business is indeed very popular in China in the past few years, and in the past two years, it has begun to move towards the L2 business. Recently, Tesla mentioned that it wants to be a robotaxi. I'd like to ask Mobileye how you see the robotaxi market? What are our current products or solutions in this area?
Johann Jungwirth:
In fact, any technology product has to go through such a growth cycle until it matures. When it comes to driverless taxis, because the front-end is more expensive, and the current L4 and L5 systems are more expensive, only commercial, such as your own commercial fleet types, can afford to use them. So I must start from this and expand it when the cost comes down. So, if you think back to the past, you used to use horse-drawn carts and ox carts to pull goods. We also look at it from the perspective of development, and in the future, trucks will be driverless trucks. It's cyclical.
Elie Luskin:
I would add a very important point, because Mobileye does both, but in fact it is losing money doing driverless taxis right now. But because Mobileye is making money in consumer-grade ADAS, it can feed back to Robotaxi.
Robotaxi on Mobile 在德国亮相 | Mobileye
Johann Jungwirth:
Because a few years ago, in the middle of growth, when the market was at its hottest, many large companies and some small companies started to do it. But now going further down, when it reaches the bottom, it has no money, so the big companies go back to do L2 again, and some of the ones that specialize in driverless taxis simply disappear. They can't find it, but Mobileye can still do it because we insist on both consumer and commercial grades.
Elie Luskin:
And it is very important that as our ADAS technology becomes more and more advanced, it can also be used for higher-level unmanned driving. So as I just mentioned, in 2026-2027, our driverless taxi technology will mature, especially in the commercial field.