China's smart car industry chain is secretly completing the "historical handover".
Huawei, DJI, and Baidu, the three giants born in the ICT, robotics, and Internet industries, are taking the car BU, DJI, and Baidu Apollo as the starting point to conquer the city and win orders for high-end intelligent driving.
In 1886, when the young German Robert Bosch followed the trend of electrical and electronics to found Bosch, Continental developed the first treaded car tire for automobiles, and ZF solved the problem of machining accuracy of transmission gears for Zeppelins.
You don't expect these companies to become Tier1 giants for a century.
Of course, the Tier1 giants would not have thought that after the "electrocution" of the car, the iron curtain of the global automotive industry was opened, and a hundred years later, the Tier1 "New Three Heroes" of smart cars were born in the far East.
01. Why Huawei, DJI, and Baidu?
The old king abdicated and the new king rose.
In the era of fuel vehicles, automobiles revolve around the technical standards of high-end precision industrial products, pay attention to rigorous processes and advanced technology, and the advantages of Tier1 for a hundred years are fully displayed.
In this process, Tier can meet the standardized needs of car companies for mass production, and at the same time, due to Tier1's deep technical accumulation, integration capabilities, and engineering experience, it has more advantages in mass production.
For example, Bosch and Continental each have a global share of about 20 percent of the global automotive ECU market share and 70 percent of the world's diesel injection systems.
To this day, ZF still ranks first in the global transmission market with a market share of 18.7 percent.
The core competitiveness of traditional Tier1 is the combination of machinery and electronics, but the speed at which they migrate their core competitiveness is not as fast as the arrival of automotive intelligence.
In recent times, the Tier1 giants have changed dramatically:
- Bosch headquarters announced one month ago that it plans to lay off around 1,200 employees in the software and electronics division by the end of 2026.
- ZF is considering closing two German plants;
- The mainland also plans to lay off about 5,000 employees.
A relevant person who has worked in Bosch for more than 10 years told the heart of the car that Bosch China alone is actually quite localized, but the problem is that the core R&D team is still in Germany, the chain is too long, and the response to the intelligence of Chinese automobiles is still not fast enough, especially the platform-level technology is difficult to "move" for the Chinese market alone.
In the past, Tier1 has been a standard turnkey solution, with fast ramp-up prioritizing. However, in today's context of intelligent driving, which emphasizes an open ecology and is highly customized, this approach is not pleasing to the public.
Although today's "Global" giants are accelerating the transformation of intelligence, such as Bosch has just announced the reorganization of the automotive business into the "Bosch Intelligent Mobility Group", in the face of the wave of smart cars, the new Tier1 not only has no historical baggage, but also can integrate the product thinking of hard technology in the past - product power, brand power, and technology power into the car.
Traditional direct-to-consumer car companies are also aware of the need for change, and the current pain points of car companies can be divided into three categories:
- The first type of intelligence is weak;
- The second group sees new opportunities for intelligence, but suffers from the lack of methodology to cut into the intelligent track;
- The third category has completely eaten the dividends of intelligence and completed the self-revolution.
To push China's smart car industry forward, new players need to stand up.
The new forces of automotive intelligence, represented by Huawei, DJI Automotive, and Baidu, have seized the opportunity at the junction of the wave of fuel vehicles and smart cars.
Huawei, through three modes: HI mode, parts mode, and smart car, has attracted different car brands to enter the market.
This is not only due to its strength in self-research from software and hardware, but also due to its national-level influence in consumers, so that Tier1 has become a full-process service from a simple supply, and some people compare it to a 2C-2B-2C service.
If you think about it, Huawei is more like a powerful reformer.
DJI Automotive, after winning the designation of three models of Wuling Baojun, has made rapid progress, and then won the Chery iCar 03, especially the 1,000-yuan cost-effective intelligent driving solution to play the signboard, and now DJI Automotive is expanding from the "cost-effective" label to a more flexible supply + service model.
Therefore, from the perspective of DJI's in-vehicle style, it is more like an evangelist of intelligent driving, providing solutions according to how much needs are needed.
Baidu, as an autonomous driving Whampoa Military Academy, relies on more than ten years of intelligent technology accumulation, and also has a unique advantage among Tier1 manufacturers in involution.
At that time, the most wary of mainland and Bosch were none other than those emerging L4 autonomous driving companies, but the reality is that the top 1 giants in their respective fields have entered the automotive industry one after another.
This is because cars a hundred years ago were essentially huge precision instruments, while smart cars have more attributes of technological products.
Looking back, the relationship between Tier1 and car companies can be divided into three forms: parental, informational, and explainative.
In the era of traditional fuel vehicles, the two mostly exist in a paternalistic and information-based way.
Earlier, the Tier1 giants were the patriarchal type for car companies, these Tier1 giants are the professional authority on key parts of the automobile, have a certain monopoly position, and have technology and mass production experience.
When it comes to information-based, car companies have more initiative, and Tier1 has begun to explain the pros and cons of the "ABC Three Schemes" to car companies.
But this is essentially a retail relationship, with Tier1 being the expert and the car company being the payer. At this point, Tier1 jobs are providing the latest knowledge and technology, and they are still on the sidelines.
Now that the era of smart cars has fully moved towards an explanatory relationship, Tier1 must have the ability to provide professional technology for car companies, and at the same time, they must also work together with car companies to polish products, discuss and consult information with car companies, so that car companies have the ability to grasp the final decision.
It is obvious that the new three heroes of Tier1, represented by Huawei, DJI, and Baidu, are reshaping the relationship between Tier1 and car companies in three different ways: reformers, evangelists, and reservoirs.
02. Huawei, a strong reformer
Since 2018, when Huawei announced its entry into the automotive field, Huawei has contributed a lot of technologies and buzzwords to the industry.
For car companies, Huawei is like a strong reformer.
This strength is reflected on two levels – technology and strategy.
The outside world can think that its style is high-profile, but it cannot be denied that it has built a complete automotive intelligent technology map.
If we want to talk about Huawei, we have to talk about the car BU, the technology source behind the HI model, the smart car model and the parts supply model.
With an annual investment of RMB 10 billion in capital and talents, Huawei's automotive BU has developed seven areas: automotive intelligent driving solutions, automotive intelligent cockpits, intelligent vehicle digital platforms, intelligent electric vehicles, intelligent vehicle clouds, AR-HUDs, and intelligent vehicle lights.
An insider introduced to Auto Heart: "If customers only buy lidar, it can only be regarded as a traditional component model, and at least 1-2 solutions must be purchased to be considered a large component model."
To make a simple distinction:
- Traditional component model: 1-2 car BU non-core solution;
- Large component mode: 3 or more non-core solutions or 1-2 vehicle BU core solutions (intelligent cockpit or intelligent driving solutions) are adopted;
- HI mode: the full stack adopts seven major solutions for the vehicle BU;
- Smart mode: Huawei is responsible for sales and brand building while adopting the seven major solutions of the car BU in the full stack.
At least with the support of these solutions, Huawei can at least guarantee that there are car companies, regardless of whether they want it or not.
Industry insiders have lamented that ADS2.0 shows the real thing about Huawei's car BU. ADS1.0 is only supported by the intelligent driving assistance system for the difficulties encountered in daily commuting, while ADS2.0 is the breakthrough of the difficulties in all scenarios. At present, ADS3.0 is still in the pipeline, but it is expected to provide point-to-point intelligent driving, greatly reducing the proportion of human takeover.
ADS is not only a software-level solution, but also permeates the intelligent driving brain.
As we all know, there is also Huawei's MDC behind ADS as a computing power support, and the MDC610 with a dense computing power of 200TOPS has been widely used in various cooperative models with Huawei. Including but not limited to the M5, M7, M9, AVATR 11, 12 and other models.
Huawei's support for car companies also stems from strategy. Especially in the cooperation with Cialis, it has brought about the reform of the whole chain from the manufacturing end to the sales end. The results were also immediate, with more than 130,000 units ordered for the new M7 in four months after its launch in September.
Today, Cialis is the first among traditional car companies to fully understand the benefits of intelligence.
But at the same time, we must also see the soulless side of Cialis. Recently, Cialis announced that it expects a loss of 2.1 billion yuan to 2.7 billion yuan in 2023, compared with a loss of 3.83 billion yuan in the same period in 2022.
Another problem is that the recent organizational restructuring of the car BU split and independent will also bring uncertainty to car companies, and even increase the game between car companies and car companies.
An industry insider once described Huawei's internal relationship as follows: "The car BU has the technology, and sometimes the relationship between the smart car and the car BU is more like that of Party A and Party B. Obviously, the car BU is Party A, and the smart car is Party B.
Finally, Huawei's intelligent driving is not very advantageous in terms of cost.
The price of Huawei's HI cooperation models is more than 300,000 yuan. The underlying logic is that the cost of Huawei's intelligent driving is high. Some industry insiders estimate that the cost of Huawei's intelligent driving system is about 20,000 yuan, Xpeng's is in the early 10,000s, and Tesla's ADAS system costs about 10,000 yuan.
More than 300,000 models are the work of big reformers, but in the face of the mass market of 200,000 models, there are too many competitors.
Whether to hand it over to Huawei, car companies are still hesitating.
03, DJI Automotive, a high-level intelligent driving evangelist
More than 300,000 models are the work of big reformers, but what about the market below 300,000? According to the data of the Passenger Association, this part of the market accounts for about 80% of the entire market. At this time, I have to mention the Tier1 also from Shenzhen - DJI Automotive.
Speaking of investment alone, it has been revealed that DJI Automotive has recently received clear investment from BYD and FAW Group, with a valuation of about 10 billion;
The wrestling between Huawei and DJI Automotive is not only at the technical level, but compared with Huawei, DJI Automotive can give car companies more sense of security.
"Fast on the car, few problems", this is the industry's evaluation of DJI's vehicles.
Car companies choose suppliers based on technical strength and resource relationships, but in the final analysis, profitability is the sharp knife hanging over everyone's heads.
The Tier1 team settled in the car company and worked overnight, and they also had to bring the chief scientists to work in the factory together.
A roll of words seems to be light and light, but behind it is the supplier's rough skin and thick "durable" skills.
Therefore, to gain the trust of car companies, Tier1 must first climb in the merit price ratio.
This is the logic of DJI's cost-effective vehicle performance in the early stage, hoping to use a set of practical solutions that take into account cost and performance to popularize high-end intelligent driving in the market below 200,000 yuan.
Shen Shaojie, the head of DJI Automotive, once mentioned in his speech that DJI wants to be a popularizer of intelligent driving, and hopes to make intelligent driving cover more people to the greatest extent at the lowest or most reasonable cost.
In fact, DJI Automotive has established a cost-effective label through the basic version of the Chengxing platform, and two production models have been launched on the Chengxing platform: Baojun (Yunduo) and Chery (iCar 03) in cooperation with Wuling.
Both the Cloud and iCar 03 are equipped with the Chengxing platform 7V solution.
DJI's in-vehicle platform lowers the threshold for intelligent driving to a 100,000-level platform. In particular, the 7V solution equipped with iCAR03 has evolved into a more radical strategy - on the basis of the pure vision solution, the binocular bionic pure vision is realized, that is, the millimeter-wave radar and ultrasonic radar are completely removed, and only a pair of inertial navigation stereo binocular cameras, a monocular camera and four surround-view fisheye cameras are retained, and only 32TOPS is needed to achieve L2+ high-level assisted driving.
The solution without redundant sensors breaks down the cost performance of intelligent driving.
At the 100-person meeting in April 2023, Shen Shaojie believed that the cost of an L2+ intelligent driving system should have a range for the cost of the whole vehicle, with the upper limit of this range being 5% and the lower limit being 3%. Based on the current average price of the car model, the cost of the DJI 7V solution is about 6,000-3.5,000 yuan. In comparison, the cost of this solution is only one-fifth of that of Huawei's ADS solution.
The cost-effectiveness of "bringing down the price of intelligent driving" is the simplest and most straightforward explicit label to describe DJI vehicles, and the more flexible mode allows high-end intelligent driving solutions to be matched in a variety of ways.
At this year's CES exhibition, Qualcomm exhibited a demo of the city pilot jointly developed with DJI Vehicle, which conveyed a strong signal of the flexible combination of the high-end version of the Chengxing platform.
Compared with the basic version of the Chengxing platform, the hardware and functions in the demo have been upgraded - the urban pilot demo is based on a single Qualcomm SA8650P chip.
According to the heart of the car, the first-generation Ride SoC launched by Qualcomm is SA8540P, and the INT8 computing power is 60TOPS, so it is speculated that the SA8650 embedded in the second-generation Ride can reach the 100TOPS level.
The computing power chip is already doomed, and this solution is different from the "cost-effective" label that DJI Automotive has previously been labeled as "cost-effective" - with the flexible and expandable characteristics of the Chengdu platform, DJI Automotive has gradually opened a more flexible supply + service model for car company customers, and launched a high-end intelligent driving evangelism action to the 200,000-300,000 level market.
The so-called supply is to take the product as the core, provide a platform for software and hardware integrated intelligent driving solutions, and make flexible program adjustments according to the needs of car companies on the basis of inertial navigation stereo binocular eyes.
And service refers to the transition from providing only solutions to enablement of a full set of processes. Xie Tiandi, head of DJI automotive marketing, explained that the heart of the car can solve various problems encountered when driving intelligently.
"Smart driving development tools are all within the scope of normal services, but DJI is still expanding its services, for example, if the host lacks smart car marketing capabilities, we can even provide advice. 」
This time, Qualcomm's urban pilot demo released at CES showed at least three characteristics: DJI's flexible in-vehicle functions, flexible models, and flexible solutions.
After the end of CES, DJI Automotive officially took the initiative to claim this Qualcomm Urban Pilot Demo video, which was tested on urban roads in Shenzhen, and realized point-to-point pilot driving assistance on urban roads without relying on lidar and high-precision maps for laneless line intersections, traffic light recognition, and pedestrian avoidance and detour.
Combined with the chip, the whole set of solutions should no longer take the cost-effective route, but emphasize functional upgrades and flexible collocation.
Interestingly, the demo car in the video is a fuel vehicle, which also means that DJI's in-vehicle solution can support high-end intelligent driving in a variety of power modes.
According to people familiar with the matter, now the interior of DJI vehicles not only emphasizes the flexibility of models and program configurations, but also can even do old car remodelling in addition to new car research and development.
Compared with Huawei's big reformers, DJI's automotive style seems to be more inclined to "Be water", entering car companies in a more low-key and flexible way, and positioning itself as an evangelist for car companies to popularize intelligence through the new Tier1 of "service + supply".
I have to say that in the involution of Tier1, DJI has taken a more differentiated and flexible route, more like a "smart driving sieve", which can carry relatively complete intelligent car companies on the screen surface, and at the same time undertake a set of solutions under the screen surface, catching the traditional car companies that have not yet tasted the sweetness of intelligence at the bottom, but can come up with facelifted models to try.
It is very likely that this style of play can provide a window to open a Tier1 for DJI's vehicles.
04. Baidu Apollo, intelligent technology reservoir
Compared with Huawei and DJI, Baidu Apollo has at least three special features:
- The earliest enterprises that have contributed the most entrepreneurs to the intelligent driving industry chain;
- The only Internet technology company;
- It is the only company that makes cars and suppliers in parallel.
These three characteristics also mean the corresponding long board and short board.
As the earliest company to invest in intelligent driving, Baidu has contributed the most technology and talents to automotive intelligence, and the scientists who were born in Baidu are today the founders and core management of unicorns in the intelligent driving industry chain and enterprises in subdivided fields.
Baidu Apollo has a strong enough technical starting point.
Build a car, do a Robotaxi, or be a Tier1 eating a bigger piece of the pie?
Baidu Apollo has chosen to take it all, so its evolution is like a passive choice for a genius player.
The biggest longboard of a "genius" player is his technical accumulation.
Baidu is more like a huge technology reservoir, transporting intelligent driving resources in different directions.
Baidu Apollo's intelligent driving technology for passenger cars is reduced from L4 autonomous driving technology, and AVP parking is the entry point - extended to the driving domain to form an ANP series of solutions (ANP1.0, 2.0, 3.0), and the latest generation of ANP3.0 has intelligent driving function modules that support urban, high-speed, parking, and global scenarios.
Compared with Huawei and DJI, two Tier1s with experience in high-end hardware manufacturing, Baidu, as an Internet-born company, is not so close to the manufacturing industry, which also leads to the overall car rhythm is not outstanding and lacks some explosive power.
To sort it out, Baidu Apollo actually has more car company customers than expected, and a group of car companies that cooperated with Baidu AVP (autonomous parking) in the early days:
GAC, Great Wall, and the other batch of models that cooperate with the ANP series (L2 + assisted driving) solution: VOYAH FREE, Jiyue 01.
It is understood that in the next 2-3 years, there will be 8 models in mass production.
In the final analysis, on the one hand, the models cooperating with Baidu Apollo are still in the sales ramp-up period, and on the other hand, the process of Baidu positioning itself as a Tier1 is also relatively slow, with technical support but no immediate impact.
In 2022, Baidu Apollo got its first car company customer, which was VOYAH, and the cooperation plan finalized by the two parties was officially an upgraded version of the ANP2.0 (high-speed NOA+AVP) solution - replacing dual TAD4 with one TDA4.
At that time, Baidu Apollo did not directly position itself as a Tier1, but very euphemistically called itself a "automotive intelligent process accelerator".
It wasn't until April 2023 that Baidu Apollo generously announced its entry into Tier1, opening up four capabilities to car companies: "experience definition, independent experience evolution, full-cycle OTA services, and co-creation and growth with car companies".
Baidu Apollo is a potential stock for intelligent driving, but if the technology reservoir is to have an explosive period, then it is necessary to turn the resource curse into a resource tilt.
There is a view that Baidu Apollo has too many missions in the early stage, from assisted driving, unmanned driving, smart map, and smart cabin to seize the position, and it is cursed by resources, but it loses the opportunity to deepen the cultivation of a certain vertical field.
If Baidu Apollo can tilt enough resources into the passenger car field and build its service capabilities as a Tier1, it may also become a very explosive dark horse.
05. The new three heroes urgently need to solve the "incompatibility" of the integration of new and old industries
Intelligence is a cutting-edge technology based on AI, which is what Xinsanjie is good at.
However, when intelligence becomes automobile intelligence, it must be considered that the traditional automobile industry combined with intelligence has existed for a hundred years.
So, who embraces whom?
Returning to the most simple business logic, from the perspective of the relationship between Party A and Party B, it is Tier1 that actively integrates car companies.
Huawei's strong reform, DJI's intelligent driving evangelism, and Baidu's reservoir-like existence are all adding new elements to the smart automotive industry.
Each has its own characteristics, but also has the same dilemma, and the typical problem is all kinds of inadequacies arising from the integration of new industries and old industries.
Car companies feel that Tier1 Sanjie is too aggressive, while Tier1 feels that traditional car companies are too conservative.
Nowhere is this more evident than in the talent and organizational structure – to borrow the phrase of an industry media person, "Apple engineers go to work at Foxconn".
For example, one of the reasons for Huawei's split car BU is that it will have the opportunity to restructure its talent and organizational structure and adjust it to a model closer to the manufacturing industry to serve OEM customers.
In the past, most of the autonomous driving algorithms made by Huawei, DJI, and Baidu were PhDs, but from R&D to mass production, it is an obvious mismatch of needs for scientists to become front-line engineers.
The logic behind it is that only the two gears of intelligence and manufacturing can only run and rotate when they are stuck to the most accurate point with each other.
Therefore, every indication of automobile intelligence makes the industry closer to the car factory and in line with the manufacturing industry.
The run-in that China's Tier1 new Sanjie and traditional car companies are going through, while deepening their understanding of each other, will have more opportunities to solve the inappropriate problems of intelligent transformation of car companies.