Visual China.
Text |HIEV Garlic Grain Vehicle Research Institute, author |Zhang Xiangwei, editor |Dexin.In the movie "Volunteers: Heroes Attack", Mr. Peng always has a line that makes people's blood boil: You and I were born in this era, sacrifice is the price we must pay, if you don't pay, your son will pay, and your grandson will pay. Our generation, with blood and mud, let's pay for it. 」
As an observer of the automotive industry, it was found that "a body of blood, two feet of mud" is also very accurate to describe this wave of intelligent driving companies.
They want to break into the automobile industry chain, on the surface, it is written that they can command various operations of the car, but in fact, this is a group of hard-working new car people who follow the car company with their heads and want to settle in the car factory for a year or two.
China's smart driving players have risen inIn 2016A large number of autonomous driving companies have been established, including Pony.ai, WeRide, Momenta, Zhixing Technology, Maxieye, Nullmax, Foretek, etc.
These companies are divided into two ways, all the way to learn Waymo to do leapfrog L4, and put autonomous vehicles on the road for actual testing;Learn all the way to cruise along the progressive route starting from L2 and provide solutions for car companies.
Later, Progressive gained the upper hand and ushered in three big players:
One is Huawei, a mobile phone company, which established a smart car solution BU;DJI, a company that makes unmanned aerial vehicles, established a cross-border DJI automotive business unitAnother is its own brand Great Wall Motors, which has separated the company's intelligent driving forward-looking department and established Momo Zhixing.
In this process, autonomous driving is more rigorousIntelligent drivingSubstitution.
By 2023, most of the L4 players who focus on leapfrogging will fail hurdles, and gradual will almost become the only way for intelligent driving to evolve. It's a little regrettable that most of the smart driving players have not yet achieved positive profits, and they have to rely on external financing and the support of car companies.
China's intelligent driving company, which has developed for less than ten years, calls car companies a big brother and it is natural that the intelligent driving industry is still a buyer's market.
Throughout 2023, the smart driving company gives people the feeling that they haven't made money yet.
Since January this year, the financing news of intelligent driving companies began to appear, and it was not until the end of the year that Huawei packaged the car bu**, and the outside world found that most intelligent driving companies did not achieve positive profitability and needed continuous financing.
Why can't most of the intelligent driving industry, which has been developed for seven years, be profitable?
One is:R&D investment is high, and revenue cannot keep up
From its establishment in 2019 to the end of 2022, Huawei has invested a total of US$3 billion, equivalent to about RMB 20 billion, in its intelligent vehicle solutions. In terms of revenue, in 2022, Huawei's intelligent vehicle solution business achieved revenue of 2 billion yuan. In the first half of 2023, the revenue will be 1 billion yuan.
Huawei's car BU's main business is autonomous driving, and its investment in the past few years has been much greater than its revenue, and it is difficult to see any hope of profitability in the futureHuawei's only loss-making division, that's the price of a new business.
There is also a group of companies that do not invest as much in R&D as Huawei, but they need a longer cycle to make a profit.
Founded in 2016 with about 200 employees, Nullmax has no operating income on the books from January to June 2022 and 2023, with a net profit of -7525590,000 yuan, -3712990,000 yuan.
In this regard, Nullmax's investor, Iwasan Technology, explained that the revenue mainly consists of a one-time development fee and a per-vehicle licensing fee. Although these charges have been received, they are provisionally classified in the liability account as the project has not yet been closed. 」
There is a certain amount of income, but it still takes a few years to break even, which is the status quo of most intelligent driving companies.
Xu Lei, founder and CEO of Nullmax, said that we have passed the stage of surviving, and the next step is to think more about how to live better. It is expected to break even by the end of the year, and the future revenue at the software level will increase from tens of millions to hundreds of millions. From a financing perspective, we plan to bring in some more big strategic partners by the third quarter of next year. 」
There is also a category where the gross profit margin of products is not high, such as Zhixing Technology.
Zhixing Technology hasIt is known as the first stock of autonomous driving in Hong Kong stocksAccording to its prospectus, in 2020, 2021, 2022 and the first half of 2023, the company will achieve revenue of 476550,000 yuan, 17.8 billion yuan, 13$2.6 billion and $54.3 billion yuan, and the loss during the year was 538270,000 yuan, 46.4 billion yuan, 34.2 billion and 997480,000 yuan.
ZEEKR is the largest customer of Zhixing Technology, and the supervision of Zhixing Technology and Mobileye will provide it to ZEEKR. However, the lion's share of supervision's profits is in the hands of Mobileye.
Whether it is Huawei, which is as strong as thousands of people, or as small as more than 200 people, intelligent driving companies need to make profits and sell the products that car companies need.
Entering 2023, the price of the smart driving solution1,000 yuanIt is the keyword that the intelligent driving company shouts the most.
At the beginning of the year, Li Xiang, founder and CEO of Li Auto, introduced the intelligent driving cost of his own model at the spring communication meeting, including computing platforms, sensors, etc., which is about 4,000 US dollars, equivalent to 2 yuanAbout 80,000 yuan.
The thousand-yuan solution of the intelligent driving company is an order of magnitude lower than the high-end intelligent driving solution currently used by new car-making forces such as Ideal.
Integrated driving and parking schemeIt is the market that was first hit by involution.
Hongjing Zhijia has released an integrated driving and parking solution based on a single horizon J3, based on 5R5V configuration, with a computing power of 5TOPS, which is basically the smallest computing power on the market, and may be the lowest cost solution, with a total cost of 3000 yuan.
In addition, there are companies in the field of intelligent driving, such as Foretek, which are pursuing a thousand-yuan level and focusing on a high cost performance.
Subsequently, BEV-based solutions are also highly competitive.
Momo Zhixing has given three thousand-yuan schemes: 3,000 yuan-level high-speed no-map NOH, 5,000-yuan urban memory driving, and 8,000-yuan urban full-scene noh. Only when the product achieves the ultimate cost performance can it meet the threshold for entering the car company. Zhang Kai, chairman of the board of directors, said.
Compared with a large company like Huawei, Momo is more cost-effective. It has more experience in mass production engineering than smaller companies. A person in the intelligent driving industry told HIEV.
Shen Shaojie, head of DJI Vehicle, believes that the total cost of the L2+ intelligent driving system accounts for 5% of the vehicle price, and 8% is a reasonable range. Less than 5%, the performance is too low to use;Anything higher than 8% will bring challenges to cost control and purchase intentions. Both of these situations are not conducive to the popularization of intelligent driving. 」
Based on this principle, the Chengxing platform released by DJI Automotive adopts the "" with a computing power of 32TOPS and a pure visual configuration of 7V and 9VStrong vision**Real-time perception, no dependence on high-precision maps, no dependence on lidarThe strategy realizes L2+ functions including urban memory driving and urban pilot driving. The Chengxing platform has been installed on models under the Wuling Baojun and Chery iCar brands.
Compared with companies with nearly 1,000 employees, such as Momo Zhixing and DJI Automotive, the 200-person intelligent driving company not only provides low costs, but also emphasizes the hard core of its own technology.
MaxiPilot 20 lite, mainIt is the only one in the industry that supports BEV deploymentThe cost-effective computing power platform solution of the forward-looking all-in-one machine is positioned as a thousand-yuan product covering more than 70% of daily driving scenarios. According to HIIEV, the price of this package will be within 4,000 yuan.
The main thousand-yuan program is relatively consistent with the layout of mainstream car companies for intelligent driving.
Aside from several new full-stack self-developed car-making forces, mainstream independent brands mainly adopt two ways to deploy intelligent driving:
Most of its high-end brands will choose to develop their own intelligent driving solutions, and a very small number of models such as AVATR and Zhiji will use the capabilities of Huawei and MomentaIts low-end brands are also a blank area that has no time to take care of for the time being, and mainly adopts the method of cooperation with external intelligent driving companies. The price of DJI Automotive, Momo Zhixing, Maxieye, Fretech, Zhixing Technology and other models on the car is basically 200,000 yuan, or more than 100,000 yuan or even lower. Don't think that getting on the car at the level of 200,000 yuan and below will be very low. The main sales volume of the automobile market is in this range, and the first step for the intelligent driving companies in the market is to obtain the recognition and profits of car companies from these cars.
Why can the intelligent driving company launch a thousand-yuan program, so that car companies are willing to hand over the soul of some models?It all comes down to technology.
Let's see how the intelligent driving company reduces the cost to the level of 1,000 yuan.
One trick is to provideChip solutions based on lower computing power。Among the new car-making forces, the computing power that can realize the whole scene without map NOA will basically use two Orin X chips as the computing platform, and the computing power will reach 508TOPS.
In the 8000 yuan scheme of Momo Zhixing, the computing power provides 72TOPS and 100TOPS two chips, this set of solutions can realize the whole scene without a picture NOH, NOH is the same intelligent driving function as NOA, but the name is different.
If you only realize the NOA of high-speed and urban expressways, the computing power of the Momo Zhixing solution is 32TOPS. Most of the model solutions of new car-making forces to achieve similar functions use a single Orin X, which is 254TOPS, and the computing power is still down one gear year-on-year.
Not only is the computing power lower than that of the new car-making forces, but even among the intelligent driving companies, they are also competing for lower-cost solutions.
Xu Lei said one thing, this year, when the main engine factory tendered for the intelligent driving program, more than a dozen intelligent driving companies went to bid for it, and the plan was basically the same, while Nullmax was walking in the intelligent driving rivers and lakes with a non-homogeneous and lower-cost solution.
In order not to do a homogeneous and lower-cost solution, NullMax did not follow the trend when selecting chips, but chose overseas TI chip TDA4.
Because the heterogeneity of TDA4 is very good, and there is no need for an external MCU, the cost will be lower than that of the same level of intelligent driving solutions on the market. Compared with the TDA4 solution of our competitors, our solution is characterized by lower computing power. Xu Lei said.
If you calculate carefully, NullMax and DJI Automotive are also based on TDA4 chips, and the sensor solution is even further reduced than DJI Vehicle, using a lower-cost monocular camera.
Another trick, that isLaunched the self-developed BEV algorithm solution
Today, Tesla has proven that BEV+ Transformer is a good way to the endgame of intelligent driving**, and this technology also allows players in the intelligent driving track to have full confidence in the future.
After Tesla proposed BEV, Maxieye rapidly expanded its own team to develop technologies in the same direction from 10 to 100 people, and now has more than 200 employees to develop solutions based on BEV.
Last month, He Xiaopeng, founder and CEO of Xpeng Motors, lamented on his personal social platform that high-level automatic assisted driving, or someone with full autonomous driving, can see a clear road map, but completely unmanned driving cannot see the complete logic, and I even think that it may need to find another way. 」
In the end, the smart driving company may go to the same end, follow the rhythm of Tesla, and win on the basis of low-cost smart driving solutions with software algorithms.
The century-old automobile industry has followed a relatively closed set of play, the relationship between the main engine factory and the leading business is relatively stable, and the car companies have a high right to speak. Entering the era of smart electric vehicles, car companies even have to work for ** companies, for exampleIndependent brand and battery leader CATLrelationship.
Some car companies are getting rid of CATL, BYD, GAC, and Great Wall have their own battery companies. NIO, one of the new forces in car manufacturing, has laid out the R&D and manufacturing business of power batteries in the early morning, and is unwilling to give up R&D projects even if it is ready to do the best.
In the field of intelligent driving, will there be another CATL?It's not very likely now.
Huawei originally wanted to be the Bosch of the era of smart electric vehicles, providing incremental parts for intelligent networked vehicles for car companies, but in the end it only aggregated a few brands such as Changan and Chery, even if BAIC and JAC are added, it cannot change the reality that Great Wall, Geely, and BYD are all controlling the fate of intelligent driving in their own hands.
To conquer car companies, the difficulty of other intelligent driving companies can be imagined. Those players who want to tame car companies at the beginning after entering the market as a new tier 1 will find that they have to call the car company big brother first.
This is a buyer's market, and the foundation laid by car companies for so many years of development determines that they can pick and choose the intelligent driving solutions on the market, and even boil up the intelligent driving company.
At this stage, the investment circle is looking at how many orders and when to make profits, and the industry situation is worse than in 2019. If Tesla's FSD can enter the domestic market next year, it will be okay to bring up a wave. If you don't bring it up, there will be a lot of companies that will gradually quit. The current OEMs are eager to consume a smart driving solution company, waiting for their valuation to drop. An early executive told Hiev.
In 2023, we have indeed heard that several smart driving companies are in contact with selling their bodies, and the founding teams of many companies are still star technology bosses who have been sought after by capital, which also highlights the cruelty of player competition on the field.
It is not easy to be a Bosch in the new era, and after experiencing the implementation of technology in 2023, the smart driving company will find that it is still necessary to take care of the mood of car companies and keep up with the rhythm of big brother.