Whether to use Tesla's "visual training method" or lidar for the optimization of intelligent driving has been debated in the new energy vehicle market for many years. Under the pressure of the price war, the departure and retention of lidar once wavered.
Entering 2024, when NOA (navigation assisted driving, also known as "high-end intelligent driving assistance" by some car companies) has become the key to the wrestling of new forces, the fate of lidar has also begun to reverse, and the confrontation between "Tesla faction" and "lidar faction" has become increasingly fierce.
On the occasion of the Beijing International Auto Show, Tesla, which has not participated in the exhibition for many years, broke the news of CEO Elon Musk's visit to China without warning.
According to CCTV News, Musk is visiting China at the invitation of the China Council for the Promotion of International Trade. According to multiple reports, this time Musk's core appeal is to promote the implementation of Tesla's FSD (Full Self-Driving software) in China, and hopes to transmit the car driving data collected by Tesla in China to foreign countries for autonomous driving algorithm training.
For many years, the person who bombarded LiDAR the most fiercely was none other than Musk. He has openly declared that those who use lidar solutions are doomed. However, from the 2023 Guangzhou Auto Show, models such as the new M7, Xpeng X9, Li MEGA, and Zeekrypton 007 with urban NOA will all be equipped with lidar.
Industry experts point out that China's traffic situation is more complex than in other markets, and visual training will gradually encounter bottlenecks in the process of autonomous driving from L2 to L3.
▲ (Researchers test LiDAR products)
In the face of the fierce attack of Chinese players and the decline in Tesla's sales, Musk had to start facing the pressure brought by LiDAR, seeking further expansion of training data and algorithm optimization. At the beginning of this year, FSD pushed the V12.3 version, marking a new stage for Tesla's vision-only self-driving system.
Musk has just ushered in a good thing: Tesla and other six car companies have passed all 4 requirements for China's car data security. On April 28, the China Association of Automobile Manufacturers issued the "Notice on the Testing of 4 Safety Requirements for Automotive Data Processing (the First Batch)", and Tesla is among them.
If the Chinese market is further opened to Tesla's FSD in the future, it is likely to be like Tesla's entry into Shanghai Lingang in 2018, which will stir up the new energy vehicle market in the short term, but also open up the prosperity of the industry in the long run. According to Guotai Junan's report, if FSD successfully enters China, it will become a strong competitor for domestic automakers' autonomous driving solutions. The competition for the domestic autonomous driving highland will become more and more intense, which is conducive to the vigorous development of the autonomous driving industry.
It seems that Tesla and LiDAR believers are ready to walk into the same river again.
01. The spring of lidar has just arrived
The development of China's new energy vehicle industry has once again ushered in a historic node. According to the data of the Passenger Association, from April 1 to 14, the retail sales of new energy passenger vehicles in China accounted for 50.39%, surpassing traditional fuel passenger vehicles for the first time.
In 2020, China set a target of more than 50% penetration of new energy vehicles by 2035. Under the guidance of policies, China's new energy penetration rate has ushered in an explosive increase.
When the market is reshaped, the entire automotive industry chain is also turned upside down. Lidar is one of the emerging vendors that are accelerating.
Lidar is a sensor that uses a laser beam to calculate the distance from an object to the surface of a target, and calculates the distance, velocity, and position information to the target object by measuring the time difference between the laser beam emitted to the target and the signal reflected back.
In the sixties and seventies of the last century, lidar was mainly used in the field of scientific research. With the commercialization of the product, its application field has gradually expanded to the industrial and early autonomous driving fields.
After the 21st century, lidar began to gradually enter the fields of mass production vehicles and service robots, and manufacturers in the industry ushered in a wave of listings. Velodyne and Luminar in the United States, and Hesai Technology and Suteng Juchuang, China's leading lidar companies, have all begun to land in the capital market.
In January 2024, Suteng Juchuang, which had just landed in the Hong Kong stock market, handed over its first annual report after listing in March. According to the financial report, in 2023, Suteng Juchuang's revenue will be 1.12 billion yuan, a year-on-year increase of 111.2% from 530 million yuan in the previous fiscal year.
The significant increase in revenue was mainly due to the significant increase in shipments. In 2023, the shipment volume of Suteng Juchuang LiDAR products will increase significantly from 57,000 units in 2022 to 259,600 units, an increase of 355.4%.
At the same time as the revenue growth, the gross profit of Suteng Juchuang also turned positive last year, reaching 93.64 million yuan, and the gross profit margin turned positive from -7.4% in 2022 to 8.4% in 2023.
Although shipments, revenue and gross profit have increased significantly, due to the change in fair value of -3.5 billion yuan brought about by the conversion of preferred shares into ordinary shares last year, the net loss of Suteng Juchuang in 2023 will further expand, from 2.086 billion yuan in 2022 to 4.33 billion yuan last year.
▲ (Auto Shanghai 2021, RoboSense, LiDAR for autonomous vehicles)
Hesai Technology, which landed on the U.S. stock market a year earlier than Suteng Juchuang, also achieved explosive growth in performance last year. According to the financial report, in 2023, Hesai's revenue will be 1.877 billion yuan, a year-on-year increase of 56.1%. Similar to the reason for Suteng Juchuang's revenue growth, Hesai's performance growth was also driven by shipments. In 2023, Hesai's LiDAR product shipments will be 222,100 units, a significant increase of 176% over the same period last year.
However, like Suteng Juchuang, Hesai's net loss is also expanding, with a net loss of 476 million yuan in 2023, an increase of 58.2% from the previous year.
Overall, although the financial reports are mixed, industry insiders told the "City Boundary" that the sharp rise in the performance of manufacturers is due to the same hero - ADAS (Advanced Driver Assistance System).
In the intelligent driving perception scheme, it is mainly divided into two technical routes: pure vision and multi-sensor fusion, the former is represented by Tesla, and domestic new energy vehicle manufacturers generally use multi-sensor fusion scheme.
In 2023, it is considered to be the "first year of urban NOA", and lidar will also usher in large-scale mass production.
According to the monitoring data of the Gaogong Intelligent Vehicle Research Institute, in January ~ December 2023, 440,300 new cars with pre-standard lidar for passenger cars (excluding imports and exports) in the Chinese market were delivered, with a total of 570,900 units, a year-on-year increase of 341.19%.
According to the financial reports of the above two companies, as of the end of last year, Hesai and Suteng Juchuang have obtained more than 60 mass production vehicle designations, and ADAS lidar products have become the main products.
In 2023, Hesai's ADAS lidar product shipments will be 194,900 units, a year-on-year increase of more than 200%, accounting for 87.8% of shipments. During the same period, the shipment of Suteng Juchuang ADAS lidar products also increased from 36,900 units in the previous year to 243,000 units, accounting for nearly 94% of the shipments. In addition, the vast majority of ADAS lidar products of the two manufacturers are supplied to domestic automakers.
02. From unanimous external to head-to-head
At present, China is already the most active and important market in the global automotive lidar solution market. Thanks to this, most of the global market share of automotive lidar has gone into the hands of Chinese manufacturers.
Speaking of which, the two rivals, Hesai and Suteng Juchuang, had a sense of unity in their early years.
Around 2014, when Hesai and Suteng Juchuang were founded, the main customers of LiDAR in the automotive field came from autonomous driving companies. As the self-driving craze sparked a wave of start-ups, related lidar startups sprung up, and the industry leader Velodyne didn't pay attention to fledgling Chinese manufacturers.
Among Chinese vendors, most startups have opted for a mechanically rotating route similar to Velodyne's. Among them is Hesai Technology, which was co-founded by Dr. Kai Sun from Stanford, Shaoqing Xiang from Stanford, and Yifan Li, Ph.D. from the University of Illinois. There is also Qiu Chunxin, who graduated from Harbin Institute of Technology with a doctorate degree in mechanical and electrical engineering and automation, and his younger brother Qiu Chunchao, who graduated from Hong Kong Chinese University with a master's degree in business administration, founded Suteng Juchuang.
However, due to their different entrepreneurial backgrounds, Hesai and Suteng Juchuang chose to enter from different markets. Hesai is focusing on high-end L4 autonomous driving customers, while Suteng Juchuang has chosen low-line LiDAR products to sweep the low-end market.
With its quick response and high cost performance, Hesai and Suteng Juchuang caught Velodyne's head, and the latter chose to shrink strategically and voluntarily withdrew from the Chinese market. In the early days, many lidar companies realized that car companies would become major customers in the future. However, Velodyne judged that it would be European, American, Japanese and South Korean car companies that could take the lead in putting lidar on mass-produced models. This misjudgment also gave Hesai and Suteng Juchuang a chance to overtake.
Since 2020, Chinese car companies have begun to launch smart cars equipped with lidar one after another, and at the same time, there has been a wave of investment in lidar companies. For example, Xiaomi Group has invested in Hesai, and BYD, Geely and other car companies have invested in Suteng Juchuang.
From the perspective of the current situation, it is from that time that Hesai Technology and Suteng Juchuang have begun to go head-to-head from unanimous to the outside world.
In order to enter the passenger car market, mechanical rotary radar, which was previously the mainstream of the Robotaxi market, could not be used, and manufacturers are developing hybrid solid-state LiDAR and solid-state LiDAR products suitable for passenger cars.
Depending on the scanning method, hybrid solid-state lidar includes rotating mirror, prism, and MEMS micro-galvanometer, and solid-state lidar products include Flash and OPA (optical phased array).
Hesai chose the rotating mirror solution, while Suteng Juchuang chose the MEMS microgalvanometer solution.
In 2020, Suteng Juchuang was the first to release the automotive-grade LiDAR M1, and a year later, Hesai's first hybrid solid-state LiDAR AT128 was also officially released. The Xiaomi SU7 Pro and Xiaomi SU7 Max, which have been popular recently, are also equipped with Hesai's ultra-high-definition long-range LiDAR AT128.
However, in that year, Suteng Juchuang, which seized the opportunity, harvested dozens of fixed-point projects in more than a year, and Hesai Technology was left behind.
According to the "2021 LiDAR Application Report for Automotive and Industrial Fields" released by consulting firm Yole Developement, in 2021, Suteng Juchuang ranked second in the world with a market share of 10%, while Hesai Technology ranked fifth with Huawei, Tudatong and Velodyne with a market share of 3%.
Fortunately, Hesai still seized the dividends of intelligent driving, and got an order from Li Auto at the beginning of 2021, and then successively became a supplier of lidar for Lotus, Changan, SAIC and other car companies, realizing the overtaking of Suteng Juchuang.
According to data from Yole Intelligence, a market research agency, in 2022, Hesai Technology ranked first in the global automotive lidar list with a market share of 47%, while Suteng Juchuang ranked fourth with a market share of 9%. In the global market, China's automotive lidar companies account for more than 75% of the global market share.
In 2023, Hesai's market share is still leading. According to Yole Intelligence data, in 2023, Hesai and Suteng Juchuang will rank first and second in the global passenger car lidar market with a shipment share of 41% and 29%, respectively.
At this time, Velodyne, a star American lidar company, was still spinning in mechanical lidar after merging with Ouster and having changed its name to Ouster.
03. How to reduce the price?
Although Hesai and Suteng Juchuang have taken most of the market share, the auto market is full of uncertainties, and they also have their own anxieties.
In the development of the lidar industry in recent years, not only is the technology upgrading, but the price is also declining. According to Hesai Technology's prospectus, in the first nine months of 2019~2022, the average selling price of Hesai Technology's lidar was US$17,400, US$12,700, US$7,700, and US$3,100 respectively.
According to the company's 2023 annual report, the product revenue is 1.767 billion yuan, the shipment volume is 222116 units, and the average price is 7,955 yuan (about 1,099 US dollars).
In the same period, the average selling price of a single ADAS lidar will drop from 4,300 yuan in 2022 to 3,200 yuan, and the price of 3,000 yuan can basically represent the average level of the industry, but this is not low enough for car companies.
"We are now being forced by OEM customers very tightly, that is, his cost is very limited, and I hope that the cost of sensors and domain control chips will be as low as possible. On April 19, at the Hesai Technology Spring Communication Conference, Li Yifan, co-founder and CEO of Hesai Technology, said.
Since 2023, the development of intelligent driving has ushered in an inflection point, but the demand for cost reduction from car companies is also increasing day by day. When the pure visual route represented by Tesla has been proved in stages, the route has also begun to penetrate into domestic car companies.
Last year, Xpeng and NIO announced their respective second brands, MONA and Ledao, respectively, and clearly adopted a purely visual route. Zhijie S7, which set off again in April this year, replaces the original Pro rear-drive long-range version with the new Pro version, directly cancels the lidar, and directly realizes the intelligent driving pilot assistance and intelligent parking of national highways and urban expressways with vision.
Immediately after April 12, GAC Group officially launched the pictureless pure vision intelligent driving technology. According to the official statement, the high-speed and urban NOA functions in the core urban area of Guangzhou have basically realized pure vision without pictures.
Therefore, there are now two voices: the positive side believes that the pure vision solution can achieve the purpose of low cost and high efficiency, and the lidar can be canceled; the opposite side believes that the current lidar and intelligent driving safety have been equated to a certain extent, and it is still necessary to exist within the scope of cost control.
Will lidar disappear in the end? For a while, it has become a high-frequency question for lidar manufacturers to answer. Li Yifan once joked that he woke up every day with 100 WeChat messages on his mobile phone, 99 of which were asking how to look at pure vision.
"In my opinion, there is an obvious difference between Tesla and ordinary car companies, it is a full-stack self-developed chip, it has a lot of data, and the entire algorithm capability is very leading in the industry, if you take it into account, you will know that the world is unfair, if we simply imitate Tesla, can it work? This is a big problem. Li Yifan said frankly.
Qiu Chunchao, CEO of Suteng Juchuang, has also been cued by the media many times to similar problems. He believes that no matter what kind of technical solution is used by traditional cameras, it is still difficult to distinguish between things with consistent colors, or there is a risk of inaccurate identification in strong light and low light environments.
Now, lidar manufacturers are also thinking about how to change their products from "functional parts" to "safety parts", so that users can find that the version with lidar is safer than those without lidar.
The AEB (automatic emergency braking system), which was brought on fire by Huawei's Yu Chengdong some time ago, changed the battle situation of lidar in Li Yifan's view, "It gives us a new opportunity".
Specifically, AEB has two key performance indicators: one is missed triggering and the other is false triggering. The core is that all the detected targets need to be detected. Lidar can improve the positive trigger rate, stop when it should be stopped, and also reduce the false triggering rate.
However, in Li Yifan's view, these are not the most critical, and the use of lidar is not only a technical problem, but also a cost problem. This is basically the consensus in the industry, rather than worrying about whether automakers will ban lidar, it is better to make the cost of products to the extreme.
Suteng Juchuang, which is already cost-effective, released a new generation of medium and long-range lidar "MX" on April 15. According to Qiu Chunchao, the cost of this product will be less than 200 US dollars, and the final price will drop to 1,000 yuan in the next 1~2 years. With the help of "MX", Suteng Juchuang wants to enter the 150,000 yuan model market.
How to promote cost reduction, Suteng Juchuang and Hesai have the same idea, that is, chip-based.
Lidar is mainly composed of four systems: transmitter, scanning, receiving, and information processing, and through chip reconstruction, hundreds of components can be integrated into several centimeter-level chips, thereby greatly reducing the number of internal components and reducing costs, and the integrated lidar also saves manufacturing costs due to the simplification of product structure.
Coincidentally, in April last year, Hesai released a long-range LiDAR "ET25", and in April this year, Hesai released an ultra-wide-angle long-range LiDAR "ATX". These two are the same as the "MX" of Suteng Juchuang, both of which are 25mm ultra-thin, low power consumption, and flexible deployment.
However, the cost has already been achieved in the thousand-dollar range, and Hesai has not yet announced the price of the "ET25" and "ATX", and Zhang Wei, vice president of Hesai Technology's Asia-Pacific automotive business, only revealed that "we must provide a very competitive price".
In fact, the pressure on manufacturers can also be seen from the gross profit margin of Hesai Technology and Suteng Juchuang. Although Hesai's gross profit margin is higher than that of Suteng Juchuang, it has also slowly declined in recent years, from more than 70% at the earliest to 35.2% in 2023.
"If you think that 100+ lines are enough, then it will be a matter of time before it drops to a few hundred yuan. An industry insider told the "City Boundary". Li Yifan also thinks that if LiDAR only needs about a few hundred to a thousand yuan in the future, consumers will definitely not refuse.
As for the value of lidar, the above-mentioned industry insiders believe that it is not determined by the creator, but by the user. "At present, the public's acceptance of intelligent driving is only transitioning from 'curiosity' to 'early adopt', whether it is really practical, useful and easy to use, and car companies and solution providers need to work hard to do a good job. ”
Author | Liu Dongxue
Edit | Tian Yanlin
Operations | Liu Shan