Under the epidemic situation, the annual industry event , the China Electric Vehicle 100 Forum, was held online. Wan Gang, vice chairman of the National Committee of the Chinese People's Political Consultative Conference and chairman of the China Association for Science and Technology, said that there are three major development trends of new energy vehicles in the world: continuous maintenance of high-speed growth, continuous improvement of the market penetration rate of new energy vehicles, and intelligent development.
With the change of the automobile industry entering the second half, the "intelligent" drama in this year's forum has increased significantly. Throughout the conference, "intelligent", especially "intelligent driving", has always been a high-frequency vocabulary, reflecting the industry's interest, expectations and a hint of anxiety and uneasiness about the landing of intelligent driving. Miao Wei, deputy director of the Economic Committee of the National Committee of the Chinese People's Political Consultative Conference, said that with the large-scale commercial application of L2 level automatic driving in passenger cars, it is necessary to play a good job in the second half of intelligent networked vehicles.
This is indeed the case. In the past two years, intelligent driving is becoming a crossover of love and hate emotions of car companies at a rapid speed.
Love is that after the bubble period of 2017-2018 and the cooling off period of 2019, the industry tends to be pragmatic, the landing is gradually reliable, users are willing to pay, and as electrification becomes the "new normal", intelligent driving is a new opportunity point.
What is hated is that under the climate of "no wisdom and no joy", the whole industry is flocking to the top, and differentiation is becoming more and more difficult. Two years ago, ACC and AEB, which were quite good, have not yet become the standard of 100,000 levels, and the car companies that are bent on picking out are under pressure.
Once the anxiety is pervasive, it must lead to the "inner volume". Therefore, the spelling power, pile hardware, hold the "self-research", smoke everywhere.
As Chen Qingtai, chairman of the China Electric Vehicle 100 Association, emphasized at the meeting, "Software will be the foundation and core of future automotive intelligence." ”
Lively to lively, really like Tesla to play "self-research", after all, very few. Once the new track of intelligent driving is opened, although the car companies are dedicated to the field, there will be intelligent driving Tier 1 outside the stadium to escort them.
How to get a piece of the pie from this tempting incremental market, the players in the field, each has its own posture.
Why break through, only the "inner volume"?
It is inevitable that the intelligent driving track will enter the white heat at such a fast speed.
The "money" scenery and prospects are so impressive, it is no wonder that the track has been slightly crowded, and it is no wonder that players have switched to the "inner volume" mode in order to seize the position.
Tesla's NOA is a start for L2+, Weilai, Xiaopeng and Ideal have followed up, and a group of independent players such as the Great Wall and SAIC have formed the third echelon.
Players compete with each other, and the original focus is more or less on user experience. As the environment becomes more noisy, it gradually evolves into a rough ratio configuration, spelling number, and even spelling voice. As a result, the former "high-end goods" of lidar have changed from one on the Weilai ET7 to two on the Xiaopeng P5, and then to four on the salon "Mecha Dragon", as if to become "road goods". At the same time, various "first", "first" and "breakthrough" propaganda calibers are flooding, which makes people feel quite bored.
Even the "sacred" words such as "full stack self-research" have been shouted by various families more, and there are some bargains. However, I understand that in addition to Tesla's qualifications, other car companies, including the full-stack self-research of "Wei Xiaoli", have to limit the scope or scene, not to mention those players in the follow-up echelon.
Of course, the so-called "full-stack self-development" is not the only method, and it may not be a better method. After all, the time cycle and cost input, placed there, can withstand and consume, after all, there will not be much. In contrast, it is also a good way to bind one or several excellent intelligent driving Tier 1s and land the function beautifully.
It's easy to make money, it's hard to make money
Judging from the supporting facilities of intelligent driving so far, the "old brands" of European and American parts are still singing the protagonist. From the design and manufacture of key sensors such as ultrasonic radar, millimeter-wave radar and cameras, to the integration of chip design to sensor integration to the integration of ADAS software and hardware systems, almost all of them hold the main share of overseas giants.
Of course, when discussing the market penetration and related share of intelligent driving, it refers more to the functional level of L2 ADAS, and there is of course no conclusion on the parts involved under the concepts of L2+, L2.9 and so on that appear from time to time in the market, and the difficulty between L2 and L2.5 or L2.9 is increasing exponentially. In addition, as far as the industry pattern is concerned, the market share is far from being an ironclad piece.
First, intelligent driving is an incremental market, and in the continuous increment, there are a large number of "open land" for new players to enter the competition; second, China as an intelligent frontier, the foundation is solid, innovation is active, the conditions of local players are unique, and it is not a fool's dream to see the opportunity to leverage the industry pattern.
If nothing else, when it comes to cases where the industry pattern is loose, it's easy to think of Mobileye's loss of power.
In recent years, relying on the integrated solution of sensor + chip + algorithm binding, Mobileye has greatly met the needs of car companies for convenient deployment and rapid mass production, and was once the absolute protagonist of the mid-to-high-end ADAS market.
However, with the intensification of intelligent driving competition, car companies have a higher pursuit of differentiation, and mobileye of the "black box" model quickly fell out of favor, giving way to various open solutions. Although he still holds the "life-saving straw" of extreme krypton in his hand, the status of the industry can no longer be said to be the same.
In contrast, the local Tier 1 has been active in the past period of time.
Among them, the one who earned the most vocal volume was None other than Huawei. In April last year, a test video that burst the circle of friends pushed Huawei's intelligent driving to the front of the stage, and since then it has been transformed into a traffic bearer. Huawei's golden signboard, the ability to integrate software and hardware, coupled with the executives' "we are definitely the first" grand statement, properly contracted the traffic of the automotive and technology media in the past year.
When the thunder is louder, the raindrops will inevitably appear smaller. Sitting on a full set of capabilities from chips to radar, from systems to software, the cooperation between Huawei and car companies is a bit high or low. Big brands are worried about Huawei's appetite and taste it; small brands are willing, but the sense of existence is a hard wound. After a year of tossing, in the end, Huawei had to personally go down and launch the questioning boundary. What the prospects are, for the time being, is uncertain.
Last year, there was also DJI car. Strictly speaking, DJI started in 2016, but the flag that officially made The Tier 1 was at last year's Shanghai Auto Show. The first partner announced is SAIC-GM-Wuling, which is also working with Volkswagen China to develop intelligent driving. According to the official introduction, DJI Has developed three sets of intelligent solutions for highways, urban expressways and parking lots, covering automatic up and down ramps, lane change overtaking, anchored at L2+ level.
Momenta, also a local new Tier 1, had a good year last year, raising a total of $1 billion in Series C, the highest in the year. Moreover, big names of investors are gathered, and car companies, parts giants, and large institutions are readily available. So favored, basically because of Momenta's "flywheel" L4 implementation path, taking into account the current mass production landing and long-term technological breakthroughs, the business logic is more self-consistent. This year, with the launch of the first model, the Zhiji L7, which is equipped with its landing plan Mpilot, the real test is coming.
Of course, there are "veterans" like Baidu. As the earliest technology factory involved in autonomous driving in China, Baidu has explored the development route of "bicycle intelligence" and "vehicle-road collaboration" and the large-scale landing solution of automatic driving, and has achieved many eye-catching achievements, so every move has attracted much attention.
In the past, Baidu's more eye-catching operations are L4 commercialization attempts and cooperation with Geely to build cars. Relatively less eye-catching is the frequent layout of intelligent driving after the decentralization of L4 technology. After the release of the "LEGO-style solution" for automobile intelligence including ANP and AVP in 2020, Baidu's cooperation object quickly expanded from angel customers such as WM to head car companies such as GAC, Great Wall, and BYD. At present, including GAC's Safety Department, Haval Divine Beast and the next BYD new products, Baidu intelligent driving has entered a stable release period.
The choice of the main engine factory: to be reliable
Today's Tire1 in the intelligent driving arena can be described as a hundred flowers, but at the other end, the "fate" of intelligent driving players is still in the hands of the main engine factory. After all, what kind of players do OEMs prefer?
The answer is simple: be reliable.
The so-called reliable, first of all, software capabilities, in the context of soft and hard decoupling, especially to have strong software development capabilities and sufficient room for extension. After all, who wouldn't want to one day get on L4?
In addition, it also means that there is a need for a full set of systematic architecture capabilities from smart cabins to smart driving, which can provide more integrated intelligent driving capabilities, and integrated management is more efficient for OEMs.
In addition to the above technical capabilities, reliability actually has another layer of equal importance, and even more important in the current climate, that is, mass production experience.
From this perspective, mass production of experienced tier 1 means a more mature technical solution and a sufficiently flexible approach and attitude to work together, which may be more critical. At present, Baidu's mass production results are relatively excellent, and the AVP autonomous parking function has been mass-produced in Great Wall, GAC and Weima. Like Huawei and DJI, they have also accumulated technical advantages for many years. But after all, intelligent driving is not to buy a finished product, which involves architectural integration, technical research and teamwork, which are complex projects, and have high requirements for Tier 1's project capabilities, openness, flexibility, etc.
For the main engine factory, the most worrying thing is to invite a "big Buddha", which is not smooth everywhere when it lands, and finally can only be discounted and hastily ended. The worse situation is that the other side is too strong, the other side is controlled everywhere, and the final result is released, but there is nothing to do with it. Those models that do not go well and frequently skip tickets in mass production have more or less such troubles behind them.
Choosing a Tier1 with a sufficiently open posture, a sufficiently "soft" body, and actual mass production experience may be a more reliable choice for OEMs.
Of course, on the track of intelligent driving, the Tier 1 competition has just begun, and it is still difficult to predict who will eventually run out. But it is not difficult to judge that the sooner you switch to mass production mode than "saving big moves" or "pulling big flags", the better you have. After all, the story of electrification has clearly told us that the market is not waiting for anyone.