When will companies be profitable on the autonomous driving track? It's always been a secretive question, and no one seems to be able to say the answer clearly.
A few days ago, Wang Rui, CEO of Shanghai Youdao Zhitu Technology Co., Ltd., gave the answer in the process of communicating with the Auto Observation interview team: In 2026, Youdao Zhitu is confident to achieve profitability. "The business scale will be about 1 billion yuan in 2026 and about 200 million yuan this year." Wang Rui said firmly.
As a member of SAIC Group, especially in June 2024, Youdao Zhitu became the only technology company in the field of trucks in mainland China to be selected into the first batch of intelligent networked vehicles in the country to access and pilot the road traffic.
The core issues that the outside world is concerned about, such as how does Youdao Zhitu achieve a monthly revenue of more than 10 million? What is the main business model? The logic of cutting into the industry from containers, etc.? Wang Rui responded frankly one by one, and expressed his understanding and thinking about the autonomous driving industry.
The following video is from Commercial Vehicle Watch
Loading...
(The following is the main content of the interview)
01
Relying on the capacity service model, the annual revenue will be 1 billion in 2026
Auto Observation: Youdao Zhitu is responsible for the intelligent transformation of SAIC Group, from the actual situation, Youdao Zhitu focuses on commercial vehicles and opens resources to the outside world, has our role and positioning changed?
Wang Rui: There is no change, we use intelligent technology to commercialize autonomous driving. SAIC is a big team, business covers vehicle manufacturing, service trade, parts, including cutting-edge technology research and development, etc., we belong to the SAIC group of a large team.
Youdao Zhitu belongs to SAIC Group's commercial vehicles, focusing on the field of autonomous driving of commercial vehicles. Commercial vehicles are the fastest scenario to achieve the commercialization of autonomous driving, and they can also achieve commercial closed-loop and large-scale application and promotion.
Auto Watch: As a member of SAIC, will Youdao Zhitu give priority to SAIC Group's subsidiaries when cooperating with vehicle companies?
Wang Rui: Autonomous driving technology needs carriers to realize, and our carriers are all types of vehicles, including Hongyan heavy trucks, Yuejin light trucks, Maxus light buses and some special vehicles.
At present, we are advancing according to the maturity of commercialization, so we will give priority to the application of vehicles developed by SAIC, because the autonomous driving technology has reached the L4 level, and the requirements for reliability are more stringent than the algorithm. We are familiar with the vehicles developed by SAIC, especially the vehicles we developed together, which will have better support for our autonomous driving technology and be more adaptable. Of course, we have no restrictions on the follow-up, and we will cooperate with the entire industry.
Auto Observation: At present, Youdao Zhitu's monthly revenue of more than 10 million yuan is mainly the profit income of transportation capacity service, do you think this service-oriented profit model is phased, or will it become the mainstream of the autonomous driving industry in the future?
Wang Rui: A lot of things are spiraling, and the same is true for autonomous driving, at the beginning, the company will have to have a few cars as a pilot, at this time, the company will buy the car, and the customer will buy the service, but at a certain stage, when it can be applied in batches, the customer is also willing to hold the vehicle.
Autonomous driving is a groundbreaking and disruptive new technology, is it reliable? How easy is long-term maintenance? How much does it cost? Including how difficult it is to maintain, it is not well judged, and at this stage, the application party will let the autonomous driving company hold the vehicle. After a period of time, with the iteration of technology, verification of operation and maintenance reliability, etc., the customer will consider owning the vehicle.
Auto Watch: From the perspective of Youdao, is it better to sell cars or to provide capacity services?
Wang Rui: I personally prefer to do service, I don't want to sell things and leave on the key, I hope that one point can be expanded to the surface as soon as possible, to the entire industry.
The service can have in-depth cooperation with various departments, and can promote the construction of the entire customer system and improve the autonomous driving technology, which is conducive to the large-scale application of our products, so I am willing to do the service. After a certain stage, it will inevitably face a problem, that is, holding assets is relatively heavy. It's okay to have a few hundred cars, but when you have a few thousand or tens of thousands of cars, the assets are heavy, so I hope that when we have a scale of hundreds of cars, we can make the service better, the process more lean, and the efficiency to the extreme, so that no matter who holds the assets in the future, everyone has the enthusiasm because it is profitable.
Auto Watch: Do you mean that the capacity service is a profitable model or a phased model?
Wang Rui: Yes, after the business model matures, other companies may need to participate, such as SAIC Anji Logistics, and these professional companies will be more professional in holding assets. We are a technology company, focusing on algorithm research and development, so I think this thing is a spiral development, when it reaches a certain scale and a certain stage, there will be a different division of labor, and maintenance, asset holding, scheduling, etc. will be completed by different enterprises.
Auto Watch: According to the capacity service model, an autonomous driving company has profitability, how big is the business scale of this? How long will it take for Youdao Zhitu to be realized?
Wang Rui: About 1 billion yuan, for Youdao Zhitu, we hope that the business scale will be about 1 billion yuan in 2026, and close to 200 million yuan this year, so there is still a lot of work to be done. The first is to make the existing projects bigger, and the second is to expand a number of new projects.
This goal is a planned thing, and it will be slow to go from zero to one at the beginning, and it will be very fast to implement and promote after doing a good sample. At present, we have more than 20 projects, but there are not many hundreds of millions of yuan, if there are more projects of hundreds of millions of yuan, and a few small projects, the goal can be achieved.
02
Container terminals are the easiest to promote, with a market size of about 10 billion in the country
Auto Watch: The Shanghai Port project and the Hainan project must be built into benchmark projects, what are the characteristics of benchmark projects?
Wang Rui: The Shanghai Port project actually represents the second generation of automated terminals in the world. The first generation of automated terminals used ATV technology, which operated automatically along a given route, and the terminal was unmanned. In addition to automation technology, there are also some modest vehicle-road collaboration technologies, including cloud scheduling, etc., which make vehicles more flexible. In this case, the construction of the terminal does not need to adapt to the vehicles, the investment cost will be reduced, and the efficiency will be greatly improved, and from this point of view, these two projects can become industry benchmarks.
Auto Watch: How to understand that the second generation of automated terminals makes the overall investment of the terminal smaller? For ports or terminal companies, what is the impact on the use of autonomous driving technology?
Wang Rui: Mainly due to the impact of the construction of the container area, the construction of the second-generation automated terminal box area will be simpler, and the terminal equipment will be reduced accordingly, such as the number of bridge cranes and rail cranes.
The earliest docks were horizontally laid out, checkerboard-shaped, and inlaid with internal wheel cranes, so that it could move around the various container areas. The first code head is a little more complicated, because the automated dock needs to be paved with ATV porcelain nails, in order to adapt to the magnetic nails, the dock layout is made vertical, a column layout, one end of each column is ATV, one end is the outer truck, and the two are connected by rail cranes. In addition, each rail crane can not take into account the next box area, which requires two rail cranes in each column, one for ATV and one for external trucks, so the wharf investment is much more than that of the traditional wharf.
The container area of the second-generation automated terminal has returned to the original checkerboard shape, reducing the equipment investment of the first-generation automated terminal. In addition, the energy consumption is also reduced, because the checkerboard layout has many interactive nodes, and the vehicle and the crane workshop travel distance is short, which not only has less energy consumption, but also has high operation efficiency.
Auto Watch: How much TEU is a container per year for SIPG? What percentage of autonomous driving is possible?
Wang Rui: SIPG Group was about 50 million TEU last year, I don't know the development strategy of SIPG, but in the long run, it will achieve 60%-70% automation.
Nationally, I think maybe half of them will be automated in a few years. The scale is very large, and the overall freight is about 10 billion yuan per year, and half of it has a demand of more than 5 billion yuan.
Auto Watch: At present, it is mainly container transportation services, but the terminal also includes bulk cargo transportation, oil tanker transportation, ro-ro transportation, etc., can only containers achieve autonomous driving?
Wang Rui: One of our projects in Hebei Port is bulk cargo transportation, mainly iron ore and coal. Bulk cargo terminals can do automatic driving, bulk cargo terminals and container terminals have some particularities, the dust is relatively large, which has an impact on sensors, which requires a lot of adaptive development, but automatic driving is a very widely used technology.
First, the container terminal is relatively structured, with a good foundation, a high degree of informatization, and complete 5G construction, so the business can be carried out quickly in this scenario. Second, the enthusiasm of the port is relatively high.
Auto Watch: Will the adaptation development of various terminals increase the R&D investment of autonomous driving companies?
Wang Rui: In all of our current scenarios, nearly 20 projects across the country use a set of algorithms. We have made a building block flexible architecture, more than 90% of the code is the same, but we will make some building block combinations according to different scenarios and different sensors, which is also a major goal of our heavy investment in research and development in the past two years.
03
Closed scenarios are the focus, and open roads are cost-sensitive
Auto Watch: In the future, we mainly develop mines, is it because the port is currently saturated with autonomous driving, or is it a strategic consideration of the company?
Wang Rui: This is related to the maturity of the industry and the attitude of practitioners towards autonomous driving. Ports were the most active before, so we focused on ports.
In the past two years, for various reasons, the demand for autonomous driving in mines has begun to flourish, so we will focus on promoting it in the mining field, and just like I just talked about, there are also many similarities in the algorithm between ports and mines, so we give priority to the promotion of the two tracks of ports and mines.
Auto Watch: Is there a difference between the operation of the mine and the port? Or is it that all scenario operations are also as common as technology?
Wang Rui: From the perspective of algorithms and intelligent driving, they are all from point A to point B, but different scenarios have their particularities, for example, the particularity of the port is that there are thousands of vehicles in the interior frequently interacting, which has high requirements for vehicle safety and flexibility. In addition, the requirements for port machinery inside the container terminal reach 3 to 5 cm, otherwise the container cannot be caught, so the requirements for vehicles to stop are very high.
The mine is characterized by high dust and road conditions that are not as good as those of container terminals. In addition, whether it is ore transportation or earth-dumping operations, the loading and unloading points are flexible and changeable, and the requirements for high-precision maps are very high.
Auto Watch: Is there a difference in service model or profit model between ports and mines?
Wang Rui: At present, the mine is to be a virtual driver, the partner holds the vehicle, we do the intelligent driving service, we earn a fee similar to the driver, this is a relatively new model, and there are some differences with the port service model, but whether it is a capacity service or a virtual driver, it is actually a transportation service income.
Auto Watch: In addition to the mine, what other scenarios are there to be developed next?
Wang Rui: There are also some steel mills, large logistics parks, etc. We focus on promoting some closed scenarios, which do not need special regulations to support, the technology is in place, the partners are withdrawn, and our projects can be implemented.
Our goal is to remove most of the labor, because autonomous driving is very expensive in all aspects, and if there is no way to reduce labor, it will inevitably lose money, so what we are focusing on now is to reduce the area where we can reduce labor. At present, the places that can reduce labor are mainly closed scenes, of course, we are also laying out social roads, such as the Donghai Bridge, which will be quickly replicated and landed in the future in Inner Mongolia, Gansu and other western coal transportation lines.
04
The commercialization model is becoming clearer and clearer, and all parties involved are profitable
Automotive Watch: When will the cost of autonomous driving come down? How big is the space?
Wang Rui: Different tracks have different requirements for cost, and the goal of passenger cars is to drive together with man and machine, liberate people's energy to the greatest extent, reduce the difficulty of driving, and then reduce safety accidents, so intelligent driving equipment cannot be reduced at will, but whether it is Tesla's visual solution or the map solution of other car companies, the essence is the cost problem. If the cost of intelligent driving is a few thousand yuan, most customers can accept it, if it is tens of thousands of yuan, only a few customers accept it, which goes against the original intention of the industry promotion, so the cost is a roadblock to the current large-scale promotion of passenger cars.
For commercial vehicles, why do we need to push L4 full automatic driving now, especially in some closed scenarios, because in these scenarios, automatic driving can replace manual work, and the scale of these scenarios is relatively large, 24-hour operation, can replace two or three shifts of labor, so that the intelligent driving system increases the cost of 100,000 to 200,000 yuan, and it is easy to save by reducing labor, so the cost of intelligent driving in these scenarios is not sensitive. Of course, in the future, commercial vehicles will be promoted to social roads, such as trunk logistics and other scenarios, and the investment in intelligent driving costs will also be sensitive, although not as sensitive as passenger cars.
As for how much room for compression, I think there is a lot of pressure on passenger cars to go down, especially for devices such as sensors, where there is room to go down. New things have to follow Moore's Law, such as domain controllers, lidar, etc., and I expect that in two or three years there should be a significant reduction in costs. In addition, with the improvement of computing power, the requirements for sensors will also be reduced, so that R&D investment will also be reduced, and the overall cost of intelligent driving will be further reduced.
Auto Watch: In trunk logistics, how much does it take to achieve the current operating cost of autonomous driving before it can be promoted on a large scale?
Wang Rui: There are two kinds of trunk logistics: one is that in the case of non-liberalization of regulations, like passenger cars, the cost that the market can accept may be thousands of yuan, or even less. I think it's hard to achieve this with the current generation of drivers, who are hard-working and willing to work more intensively and spend less money. The next generation of drivers may be different, they can accept a cost of several thousand yuan, but the labor intensity is less. Second, if the government can release unmanned vehicles in specific locations, and support the attrition of at least one vehicle queue, the market will be less sensitive to the price of autonomous driving. An increase in the cost of a truck of about 100,000 may be accepted. For car owners, a fleet of 5 cars, the last 4 are unmanned, and the increased cost of 100,000 yuan per car is easy to be offset in the short term.
Autowatch: What are your expectations for the future of autonomous driving?
Wang Rui: Many industries have begun to accept autonomous driving, for example, many terminals in Shanghai Port are all unmanned vehicles, and autonomous driving has matured in many links and can replace manual labor. With the launch of some benchmark projects, it will bring greater confidence to various industries, especially in the field of commercial vehicles, in all links of the industrial chain, everyone is profitable, the investment is profitable, and the business model will become clearer and clearer.
![Painter Wong Sui-fu, vice president of the Hong Kong Artists Association[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)