Due to the aggressive strategies of Chinese independent brand car companies on the high-end smart driving track, the trend of the global lidar industry is undoubtedly affected by the Chinese market.
This week, Hesai Technology released its third quarter report for 2023. The quarterly lidar delivery volume was 47,440 units, a year-on-year increase of 125.5%, of which ADAS lidar delivery volume was 40,593 units. Accounting for more than 85%.
Chart from Hesai Technology’s third quarter report
At the same time, the company's total lidar delivery volume in the first three quarters of 2023 was 134,380 units, a year-on-year increase of 307.9%, of which ADAS lidar delivery volume reached 114,482 units.
However, Judging from the quarter-on-quarter data, Hesai Technology’s lidar delivery volume in the third quarter fell by 8.95% quarter-on-quarter; the quarter-on-quarter delivery volume of ADAS lidar fell by even more. 11.16%.
As we all know, in the Chinese market, the current scale of lidar pre-installation on cars mainly comes from the contribution of a small number of brand models. For example, Ideal, NIO, Xpeng, Huawei (Wenjie, Avita).
Monitoring data from the Gaogong Intelligent Automotive Research Institute shows that taking the passenger cars in the Chinese market from January to September this year as an example, lidar from the above four brands accounted for more than 90% of the vehicle sales. Among them, the two brands of Lideal and NIO have more than 100,000 units on the road.
Judging from the actual delivery data, from January to September this year, more than 300,000 pre-installed standard lidars were delivered. Hesai, Tudatong, and Sagitar Juchuang rank among the top three, with a combined market share of more than 85%.
As of the end of September this year, 33 models of models on sale in the Chinese market offer lidar configurations (including optional equipment), accounting for less than 5% of the total number of models on sale in the market.
Judging from the fixed-point volume, it seems to be sending out positive signals.
Taking Hesai Technology as an example, the company has reached pre-assembled mass production cooperation with 14 mainstream car companies and Tier 1, and has obtained pre-assembled mass production appointments for more than 50 models. At the same time, the company has negotiated cooperation with six leading car companies in North America and Europe in overseas markets, and has made phased progress.
In addition (as of the end of March this year), Sagitar Juchuang has obtained fixed-point orders for pre-assembled mass production of 52 models from 21 automobile OEMs and first-tier suppliers. At the same time, the company has been selected by nine of China's top ten automobile OEMs ranked by 2022 sales.
However,fixed point does not mean the realization of mass production delivery.
For example, the lidar used by Jidu in the concept car is Hesai Technology's AT128. When the mass-produced car, Jidu ROBO-01, was unveiled, the lidar configuration was two. But in the end, the Jiyue 01 that was actually mass-produced directly canceled the lidar.
In JiYue’s view, “Lidar can quickly improve system performance in the short term, but there is an upper limit; in the long term, pure vision has advantages in system cost, iteration speed, data closed loop, and even end-to-end development.”
Prior to this, Huawei's ADS 2.0 version also took the initiative to "reduce" lidar (from 3 to 1); in addition, the facelifted Xpeng P5 (one of the earliest models equipped with lidar in China) ) even directly canceled the lidar configuration.
In addition, Livox, which was the first to achieve mass production of domestic passenger cars, has also been "abandoned" by its sister company DJI Automotive (Zhuoyu), which currently focuses on low-cost pure visual perception solutions.
Obviously, from the perspective of cost reduction, lidar has become a priority victim. A year ago, the automotive industry was still competing to see whose new cars were equipped with the most lidars.
At present, some new cars still use lidar to realize high-end smart driving, but the resulting increase in selling prices seems to be more for the purpose of "marketing selling points."
For example, the price of the Hechuang V09, which has just been launched, is 373,800 yuan for the sub-top model, while the top model equipped with 1 lidar rises sharply to 438,800 yuan.
As Volvo's lidar supplier, Luminar is involved in a class action lawsuit because investors believe that the company's board of directors made materially false and misleading statements about the company's business, operations and prospects.
Since it has not been able to achieve large-scale pre-assembled mass production and delivery so far, Luminar's stock price has shrunk by nearly 90% compared to its historical high. Obviously, the pre-installation process of lidar is still significantly lower than investors’ expectations.
There are two factors behind this. One is that the existing application scenarios of high-speed NOA are “sufficient” with the support of cost-effective solutions based on vision + millimeter wave radar; on the other hand, the cost of lidar is not enough for car companies. Said, the pressure is still huge.
At the same time, for suppliers, cost reduction and efficiency improvement mean continued R&D investment; for example, Hesai Technology’s R&D expenses in the second quarter of this year were 161 million yuan, an increase of 49.1% compared with the same period last year.
Previously, Luminar's CEO predicted that the company's orders would reach US$3.4 billion by the end of 2022, and at least another US$1 billion by 2023. But most of these revenues are still years away.
Meanwhile, Luminar's chief financial officer said investors shouldn't expect Luminar to break even before the end of 2025. Even Hesai Technology, which has achieved large-scale delivery, still has hidden worries in its financial situation.
Data show that Hesai Technology’s net loss in the third quarter of this year was 142 million yuan (RMB), an increase of 100.6% year-on-year; at the same time, as the delivery volume of ADAS products continued to rise, the gross profit margin in the third quarter dropped to 30.6% (2022 The same period last year was 37.1%).
Chart from Hesai Technology’s third quarter report
Obviously, in the short term, this track will continue to burn money, and the more it sells, the more it will lose money. For most companies that have not yet launched (pre-assembled large-scale mass production), the risk is even more amplified.
Just in September this year, the global auto parts giant Bosch announced that "taking into account the complexity of the technology and time to market, Bosch decided not to invest more resources in the hardware development of lidar sensors."
Currently, among the traditional foreign-invested Tier 1 companies, only Valeo is still developing the next-generation lidar, including peers such as Continental and ZF, which have all chosen external cooperation.
However, from the perspective of lidar suppliers, the industry still needs high-performance lidar; because the more stable the system is, it is actually more dangerous if many long-tail problems cannot be solved.
In the view of Bao Junwei, CEO of Tudatong, the resolution of pure visual solutions can be high enough, but in many scenarios due to the problem of limited training data sets and limitations of physical performance, the safety redundancy of the system is still far away. not enough.
At the same time, from the perspective of car company R&D, the current selection of lidar and 4D imaging radar is still in the early verification stage. This point has more say for ideal cars that have already implemented these two sensors on the car.
Previously, in a public interview, Lang Xianpeng, vice president of intelligent driving of Li Auto, said that lidar currently has two main functions: first, safety redundancy at night; second, truth value annotation of sensory samples. "Whenever these two functions no longer exist, they can be removed."
In his opinion, if lidar can be removed, car companies will want to reduce this large BOM. "However, 4D radar is not very effective at present because the resolution is too low. We have not publicized it too much and still regard it as a normal radar to meet the functions of the car."