The wave of intelligence is sweeping the world and profoundly reshaping the transportation field. Chip, lidar, AI algorithm, vehicle-road coordination and other related technologies are rapidly iterated, and scenes such as unmanned taxis and unmanned distribution are gradually landing, pushing the commercialization of intelligent assisted driving into a critical period.
In 2025, the intelligent assisted driving track is no longer just a contest of parameters, but also a race and a dark battle-when the market value list of car companies forms a fault stratification, when the full-stack self-research group and the open cooperation camp are arguing endlessly, when the price of laser radar is halved, when domestic chips are accelerated to replace, everyone wants to "squeeze on the train", who can stay?
The Beijing News reporter initially sorted out that there are at least 300 listed companies involved in intelligent assisted driving business in China, covering the A-share, Hong Kong-share and US-share markets. According to comprehensive data and research reports, there are thousands of companies in the field of intelligent assisted driving in China.
Based on this, according to the public data and corporate information, the Beijing News reporter made an inventory of the functional level of the assisted driving industry, the market value of listed companies, and the number of cooperative customers. Among the A-share listed companies involved in intelligent assisted driving, BYD is the trillion-dollar leader with the highest market value and the highest sales volume of new energy vehicles, Huawei, BYD, Tucki, etc. rank first in the intelligent assisted driving function, and the number of customers of cooperative car enterprises is ahead by Mom en ta.
Behind the competition among market value, ecology and industrial chain is a three-dimensional picture of China’s spiraling intelligent assisted driving industry. Beijing News reporter Lin Zi
Market value competition trillion bibcock and chaser
According to the ranking of the total market value as of May 14, 2025, the top five companies involved in the concept of intelligent assisted driving and the total market value in A shares are BYD, Cyrus, SAIC, Great Wall Motor and Zhongji Xuchuang.
BYD ranks first among A-share intelligent assisted driving companies with a market value of 1.11 trillion yuan. The competition of the second echelon is even more intense. The market value gap between Celestial (213.4 billion yuan), SAIC (192.2 billion yuan) and Great Wall Motor (172.9 billion yuan) is only in the order of billions. In the US stock market, the market value performance of Tucki, Ideals and other new forces is also neck and neck.
Some enterprises related to the industrial chain told the Beijing News reporter that now the automobile factory directly faces the terminal market, which has gained higher attention and higher voice and pricing power. Although technology suppliers are more professional on the technology track, their right to speak is relatively low. As far as the capital market is concerned, it will also favor the automobile factory with larger volume, more brand effect and more "storytelling". This is reflected in the market value. In the Hong Kong stock market, the market value of supply chain enterprises such as Horizon and Black Sesame Intelligence is less than HK$ 100 billion.
Sales volume comparison
Technology Pratt & Whitney and Brand Premium
Further according to the sales comparison, BYD’s sales of new energy vehicles are riding a dust. In 2024, BYD’s global sales of new energy vehicles reached 4.272 million, up 41.26% year-on-year, becoming the global sales champion of new energy vehicles, and its huge sales provided a data foundation for its assisted driving development.
BYD’s competitors are not only domestic car companies, but also compete with Tesla as representatives of China car companies.
While BYD’s sales in 2024 surpassed Tesla’s, the competitive logic behind the two is completely different. Byd relies on the large-scale strategy of "smart driving and equal rights" to sink the intelligent assisted driving technology to the 70,000-yuan market, while Tesla relies on the FSD technology premium to maintain the mid-to high-end positioning. Behind this market value competition are two routes: people-friendly and brand premium.
Battle of ecology
Full-stack self-research and open cooperation
At present, the intelligent assisted driving technology has entered a period of rapid development, and various enterprises have launched differentiated solutions through self-research or cooperation to promote the popularization of functions and upgrade the experience.
Among car companies, BYD and Tucki are both representatives of self-research of car companies, and they also belong to the first echelon. For example, BYD’s self-developed high-order intelligent assistant system is called "Eye of the God". Tucki, Ideal and other car companies are also constantly exerting their efforts in assisting driving.
Some companies do not fully agree with the self-research model of car companies. Wu Yongqiao, president of Bosch Intelligent Driving Control System in China, has publicly stated that in the future, intermediate intelligent assisted driving will become the standard, and the main engine factory will invest hundreds of millions or even billions of dollars in full-stack self-research every year, which is costly and not cost-effective, and should be handed over to the supply chain for processing.
Huawei has always been listed as the first echelon of intelligent assisted driving system. Jin Yuzhi, CEO of BU, Huawei’s smart car solution, said in 2024 that 10 brands of 7 car companies will adopt Gankun solutions, including Chang ‘an Deep Blue, Celestial, etc. It is estimated that there will be more than 500,000 cars driven by Huawei.
In contrast to self-research, Huawei’s "friend circle" model shows another possibility. At present, Huawei has three modes for automobile business, namely, parts supply mode, H I(Huawei Inside) mode and intelligent car selection mode.
Providing spare parts to BMW, BYD and other car companies, referred to as "spare parts supply model". Relying on Huawei’s intelligent solution cooperation, such as Beiqi Extreme Fox and Aouita, referred to as HI mode. Car companies have deep cooperation with Huawei, and Huawei participates in product design, marketing and sales. These models will appear in the Huawei Experience Center, and the intellectual and intellectual circles belong to this kind of cooperation, referred to as the smart car selection model. Among the above three modes, the one with the closest cooperation depth is the smart car selection mode.
SAIC has publicly stated that "the soul should be in its own hands", and now it also chooses to cooperate with Huawei. While Huawei’s partners get the benefits brought by the sales growth of cooperative models, they also need to face the question of "whether the car companies themselves have mastered the core technology", such as whether Cyrus and Jianghuai can independently build high-quality models of their own brands.
More noteworthy is the rise of the middle route. Technology suppliers such as self-driving unicorn M omenta (Beijing Chusu Technology Co., Ltd.) and Baidu Apol?lo are exploring their own unique routes.
Mass production and large-scale application are the key words of M omenta. On May 12th this year, Xiangdao Travel signed a strategic cooperation agreement with Momenta, and the two sides will jointly build the world’s first robotaxi (self-driving taxi) fleet based on the pre-installed mass production scheme, and take the lead in starting operation in Shanghai, which will become a robotaxi fleet that is "born for large-scale application in cities".
Baidu radish run is aimed at overseas. On March 28 this year, Baidu Radish Run announced the signing of a strategic cooperation agreement with Dubai Road Transportation Bureau, and plans to deploy more than 1,000 fully driverless cars in Dubai. This is also the first time that Radish Run has carried out unmanned large-scale testing and service outside China.
The transformation and reconstruction of industrial chain
When the intelligent assisted driving function changed from "high-end label" to "basic configuration", a reshuffle sweeping the entire industrial chain has already started.
The Beijing News reporter combed and found that the competition in the number of customers of assisted driving supply chain enterprises is characterized by wide coverage, large number of cooperation and diversified customer types. All enterprises are striving to expand their customer base in order to gain an advantage in the fierce market competition.
Among them, Momenta is the company with the widest coverage of cooperative car companies, which covers 7 of the top 10 car companies in the world and has more than 130 cooperative mass production models.
In the field of lidar, the two "leading players" are deadlocked and the competition is extremely fierce. Sagitar Juchuang cooperates with 92 models of 28 automakers and first-class suppliers. Hesai Technology cooperated with 120 models of 22 car companies, and won big orders simultaneously under the price reduction strategy.
In an interview with the media in November, 2024, Li Yifan, the founder of Hesai Technology, publicly stated that the company planned to reduce the price of its main lidar products by half to US$ 200 in 2025, which means that electric vehicles with a price less than 150,000 yuan can also use lidar. On February 11, 2025, Hesai Technology announced that various models of BYD will be equipped with Hesai’s lidar products, and it is planned to achieve mass production delivery in 2025.
These news not only show the "involution" of lidar track, but also reflect the bargaining power of car companies. Shen Ziyu, co-founder and chairman of Yikatong Technology, publicly stated in March 2024 that "there will be no laser radar companies and millimeter wave radar companies in the future, only incremental component companies for smart electric vehicles."
The competition on the chip track is even more cruel. Horizon used 9 years to load more than 4 million Journey series chips, which replaced M obileye in L2 market. Under the reality that NVIDIA O rin chips still dominate the high-end market, domestic chip companies still have a long way to go.
In addition to hardware iteration, the battle for software route is equally fierce. At present, the "map-free" route led by Tesla is in fierce confrontation with the "high-precision map" faction adhered to by NavInfo.
Behind the multi-faceted competition, enterprises are also facing common challenges. In the field of intelligent assisted driving, L3 and above autonomous driving technology has not been allowed to go on the road, and no more detailed regulations and regulations concerning the definition of responsibilities and road regulations have been issued at the national level. The commercialization of intelligent driving technology still faces uncertainty.
In addition, whether it is the full-stack self-research of automobile enterprises or the research and development of industrial chain enterprises, from chips, algorithms to system integration, every link needs a lot of financial support. In the future, the investment of tens of billions will continue to raise the threshold of intelligent assisted driving.
Despite this, the panorama of China’s spiraling intelligent assisted driving industry has also begun. BYD’s "intellectual drive and equal rights" have brought assisted driving into thousands of households, and Huawei’s "multi-party cooperation" has continuously broadened the boundaries. The rise of Hesai Technology and Horizon proves that there are no eternal giants in the trillion-dollar market.
How to balance the technical innovation and safety bottom line of assisted driving of the circle round table school?
Since April this year, the Ministry of Industry and Information Technology has carried out strong supervision on the promotion of intelligent networked vehicles, requiring automobile companies to fully carry out the combined driving assistance test and verification, clarify the functional boundaries and safety response measures of the system, and refrain from exaggerating and false propaganda. China Automobile Industry Association and China Automotive Engineering Society also issued a proposal to standardize the promotion and application of driver assistance. For a time, car companies have renamed the sayings of automatic driving, autonomous driving, intelligent driving and high-order intelligent driving as assisted driving.
With stricter supervision, how to balance technology popularization and safety in assisted driving? How to establish a database of assisted driving accidents for cross-vehicle enterprises? The Beijing News reporter invited nine guests, including enterprises and experts, to participate in the "Circle Roundtable School" on topics related to assisted driving development. Beijing News reporter Wang Linlin Linzi Zhangbing
Beijing News Cartography/Xu Wei
Viewpoint 1
Popularizing assisted driving requires safety as the premise.
Beijing News: How to balance the popularization and safety of assisted driving technology?
Lu Fang: First of all, the popularization of assisted driving technology should follow the gradual development path, and each stage should be fully verified to ensure reliability, so as to avoid premature pursuit of "completely unmanned" and ignoring potential risks. At the same time, key systems (such as perception, decision-making and execution) should be designed with multi-sensor fusion, algorithm redundancy and hardware backup, and data should be accumulated through massive virtual simulation tests (covering extreme scenes) and real road tests to optimize the algorithm’s ability to cope with complex scenes. Only when the technology is mature can security be guaranteed.
In addition, when promoting assisted driving technology, we must talk about equal rights and take safety as the basis, so that users can clearly know what the capability boundary of each car technology is. The ultimate goal of assisted driving is to improve users’ travel safety. Enterprises should take "safety first" rather than "speed first" as the principle, and the public should rationally look at the potential and limitations of technology.
Wu Huixiao: Great Wall Motor has always stressed that assisted driving is still in the primary development stage at this stage, and there is still a long way to go before real automatic driving. The core value of assisted driving is to improve travel safety and efficiency, not marketing gimmicks. In addition, we suggest establishing a national public test database for intelligent driving, sharing desensitization accident data, and accelerating the establishment and iteration of safety standards for intelligent driving.
In product communication, enterprises should face up to the development stage of technology, respect its development law, and truly inform users of the availability and applicability of technology, so as to avoid the risk of misuse caused by exaggerated publicity and even affect the overall cognition of China automobile industry.
Wang Haowei: The smarter the algorithm, the greater the security challenge; On the contrary, the safest algorithm is the simplest logic. Therefore, the core issue of the popularization of assisted driving is to find a balance between intelligence and safety. Absolute security can’t be achieved. What’s more important is to let the end users know the capability boundary of the system, instead of relying too much on the system capability. Safety is a comprehensive project, and there is no once-and-for-all approach. Efforts should be made from system design, algorithm, security mechanism, user training, regulations and new technologies.
Zheng Zhuangcheng: The popularity and safety of assisted driving are not antagonistic, but "two sides of one body" that need to be promoted together. Balancing the two needs to be realized from several dimensions: first, technical iteration is carried out with ensuring safety as the lowest level of assisted driving requirements; The second is to establish an industry safety baseline; The third is user education, which pays equal attention to popularization and transparency; The fourth is ecological synergy, and the responsibility of the whole industry chain is shared.
Dai Yifan: The essence of assisted driving application is to improve safety. It is meaningless to talk about popularizing assisted driving system without safety guarantee. Therefore, there is a "priority weight" relationship between the two, and it must be on the premise of ensuring safety and improving the existing driving safety before promoting the assisted driving system.
Viewpoint 2
Need to break the "absolute trust" in assisted driving
Beijing News: In view of the public’s "absolute trust" in assisted driving, how to rebuild a transparent communication mechanism?
Wang Haowei: Absolute trust is a dangerous term, because there is no absolutely safe system. The relative trust of the public in assisted driving is what the industry needs to pursue. Relative trust means with boundaries and conditions, and it also means that we must pursue the best under certain conditions, not the global optimal solution.
Zheng Zhuangcheng: Transparency is a technical safety belt in the era of assisted driving. NavInfo rebuilds public trust based on three core measures: technical transparency, real-time update and "data security is the cornerstone of building smart driving security".
Dai Yifan: At present, assisted driving requires drivers to stay in the loop and stay awake, especially in some abnormal road scenes. Assisted driving requires popular science rather than hype. The so-called "assistance" means helping drivers, and there must be some guidance in propaganda.
Zhuang Jinggan: Many people think that the "smart driving" system is safe and reliable enough, but this is not the case. This misunderstanding largely comes from the early propaganda strategy of car companies, which makes consumers mistakenly think that technology is mature enough to replace human driving, ignoring the capability boundary and potential risks of the system.
Viewpoint 3
The database of assisted driving accidents suggests that the government should take the lead.
Beijing News: It is suggested that a database of assisted driving accidents across car companies can be established to provide a technical basis for the determination of responsibility. What do you think?
Wang Haowei: To establish a database of assisted driving accidents of cross-vehicle enterprises, it is necessary to introduce relevant data uploading standards at the national level, and it is best to enforce them.
At the same time, the balance between the bottom line of safety and the development of the industry should be taken into account in the determination of responsibility. If all the responsibilities are put on car companies, it will lead to the lack of awe of the public in using the assisted driving function, which will eventually inhibit the motivation of car companies to develop assisted driving.
Zhuang Jinggan: "Government-led, joint car companies to build a unified national database" is helpful to standardize industry practice. At present, there are great differences in technical routes and data formats among car companies. If they are independent, they will form "data islands", which cannot support the general standards of the industry. The government can take the lead in formulating data collection standards, and car companies can access key information such as operation logs, sensor data and system decision records, and establish a privacy protection mechanism to desensitize sensitive information.
After the completion of the database, it is necessary to clarify the framework of responsibility division, and it is suggested that the legislature should formulate special laws, and at the same time clarify the criteria for identifying technical defects and dynamically divide the proportion of man-machine responsibility. In addition, we can also consider establishing a "data black box" system and handing over the right of data analysis to a third-party authority.
Viewpoint 4
Auxiliary driving will develop in the direction of equal rights.
Beijing News: Do you think the development of L3 or L4 autopilot will be hindered? What is the trend of assisted driving?
Lu Fang: The development of L3 or L4 may be hindered, but the overall trend is still forward. Although the existing technology has covered most conventional scenes, some extreme scenes (such as rainstorm, blurred road markings and sudden obstacles) are still the core obstacles for L3 and L4 to land. Coupled with the public’s lack of awareness of autonomous driving technology, the wide application of L3 and L4 technologies is limited to some extent.
However, at present, the industry has L3-level technical reserves, and the Ministry of Industry and Information Technology is also accelerating the improvement of the automatic driving standard system, promoting the safety baseline construction, and the industrial chain is also upgrading in coordination. These are all positive aspects.
Deng Chenghao: The current policy’s regulation of L2 function is essentially an "ice-breaking action" to pave the way for advanced autonomous driving. This will not hinder the development of L3/L4, but instead build three booster engines. The first is to establish the coordinate system of technological evolution. The latest national standard "Auto Driving Automation Classification" issued by the Ministry of Industry and Information Technology defines the responsibility dividing line (ODD operation design domain) between L2 and L3 for the first time, which will accelerate the shift of consumers’ cognition from "whether there is function" to "capability boundary" and clear the cognitive obstacles for the commercialization of L3/L4. The second is to open up the fast lane of technology verification, and the "car-road-cloud" three-dimensional verification system will shorten the technology maturity cycle by 30%. The third is to reconstruct the ecology of industrial cooperation, and institutional breakthroughs are more strategic than simple technical breakthroughs.
Zheng Zhuangcheng: From the perspective of four-dimensional map and safety redundancy, the future development trend of assisted driving will not be separated from the map. AI big model helps high-order intelligent driving to accelerate landing, but high-precision map is always an indispensable transcendental sensor to assist driving. The so-called "no map" will bring a series of problems such as increased cost and irregular data collection.
Zhang Ruixue: The large-scale landing of assisted driving mainly faces several problems: First, the perception and decision-making of complex scenes (such as non-signal scenes and complex environments) is still the core difficulty. The existing bicycle intelligence relies on multi-sensor fusion, but the hardware stack (such as laser radar) is expensive and cannot completely cover all long tail scenes. Second, the division of responsibility for autonomous driving accidents in the current law is not clear, and the insurance system also lacks an adaptation plan. Third, the data closed-loop ability is insufficient. High-order autopilot relies on the collection, labeling and model training of massive scene data, but the existing test area is limited in openness, resulting in insufficient data diversity.
Zhuang Jinggan: The landing of L3/L4 autopilot faces three core challenges: first, the long tail scene processing, that is, the reliability of the system is insufficient in extreme or rare scenes; Secondly, the law lags behind; Finally, the lack of infrastructure coordination, especially the slow progress of vehicle-road coordination (V2X). In the future, whether it is L2, L3 or L4 level assisted driving, it will develop in the direction of equal rights. This is mainly due to the rapid decline in the cost of the core components of assisted driving, and the technology is breaking through the high-priced barrier and moving towards large-scale application.
关于作者