On the evening of March 20, 2018, Vasquez, an Uber security officer, drove a Volvo SUV to a desolate road between the center of the city and the desert.
This section of the road in Arizona, she has followed the test car 72 times before. But on that day, a pedestrian pushing a bicycle across the road was misdescribed by Uber's self-driving system, eventually causing the vehicle tested by Vasquez to crash 7 meters away and die from his injuries.
Uber's test car was in fully autonomous mode, leading to the world's first self-driving car fatality.
In the years that followed, Uber was pressured to test autonomous driving in closed venues. At the same time, many self-driving companies test millions of kilometers on public roads every year, gradually leaving Uber behind.
In 2020, the pandemic became the last straw that crushed Uber's self-driving car. Uber, whose profitability waned, began selling its loss-making businesses, including autonomous driving.
The deal was finalized on December 8 of the same year for $4 billion, and self-driving startup Aurora took over.
Coincidentally, Lyft, another ride-hailing platform in the United States, also sold its self-driving business in April of the following year to the purchase of car giant Toyota for $550 million.
Unlike the ride-hailing giants on the other side of the ocean who have abandoned their autonomous driving business, Didi is still persisting after a series of difficulties such as removal, review, and delisting.
Transcripts
After 1,020 days, Didi Autopilot handed over multiple answer sheets in one go.
DiDi Neuron, a future service concept car, Beiyao Beta LiDAR, a three-domain converged computing platform "Orca", Didi's autonomous driving freight KargoBot, and mass-produced models will be connected to Didi network hybrid dispatch in 2025.
This is 7 years since the establishment of Didi Autonomous Driving.
But Didi is no longer the Didi it was 7 years ago, and autonomous driving is no longer the autonomous driving it was 7 years ago.
Just recently, the original price reduction of Tesla triggered car companies to think about the cost of automatic driving, and was publicly bombarded by Wang Chuanfu.
"It's all, it's all flickering, it's an emperor's new clothes, automatic driving is just a concept wrapped up and hyped by capital, and ultimately it is an advanced assisted driving, automatic driving."
One stone stirs up a thousand waves.
After all, except for the media and bloggers, few people have really experienced automatic driving. But the car companies and supply chains that depend on it obviously hate Wang Chuanfu.
On April 16, He Xiaopeng, founder of Xpeng Motors, publicly stated that XNGP is the ultimate form of intelligent assisted driving, and "making unmanned driving 'nonsense' is not". This disguised response was still relatively mild, but then Yu Dazui was not so polite.
"Or don't know the industry; Or deliberately say so, or he has not done a good job in this area, deliberately hit the industry. ”
He Xiaopeng was also limited to his own products, and Yu Chengdong suddenly put the entire industry on the opposite side of Wang Chuanfu.
These lips and swords were originally far away from Didi, how to lie dormant for three years, Didi took the initiative to jump out to show off its muscles, although the online car platform is not a direct competition of car companies, but after all, Didi and BYD have cooperated to build cars.
Being indirectly bombarded by the once intimate cooperation partner may not have been expected by Didi anyway. What makes Didi even more unexpected may be the sharp turn of automatic driving.
Frank, a practitioner of lidar, mentioned in the article "The Price War of the Avalanche, the Life and Death of the Intelligent Driving Industry Chain" that in 2021, autonomous driving companies such as Momenta, Heduo, Qingzhou Zhihang and WeRide Zhixing that once focused on L4 have switched to the L2+ market.
This L4 escape will finally be verified in early 2023, with a series of news of self-driving company bankruptcy, bankruptcy, and board infighting in Silicon Valley, including star companies such as Argo.ai and Tucson Future.
The embarrassment of L4 is that after being sought after by capital too high, it has become a track without a way back, and high-investment and high-manpower R&D work has become the norm, in the past you can write an L4 data and technology feed L2+, two-legged walking commercialization story, but now, it is difficult for you to tell the sustainable commercialization of L2+ to feed a bottomless L4, no one will pay [8].
And Didi stood up at this time to show the report card, not only did not boost the morale of the wailing autonomous driving industry, but also showed the world that he had no way back.
Catch up
In July 2021, Didi removed 25 apps due to information security and IPO issues. Cheng Wei, Liu Qing, and Zhu Jingshi were investigated by regulatory authorities, Didi's public image fell to the bottom, and several major investors also left the market.
Three months later, Didi announced that it had officially launched its delisting on the New York Stock Exchange and said it would go public in Hong Kong. Six months later, Didi was delisted from the main board of the US stock market, and its share price was fixed at US$2.29, down 84% from the issue price of US$14[2].
During this period, Meituan, AutoNavi and Haro all launched an impact on the ride-hailing market, and Didi's market share quickly fell from 82% before the takedown to 74%[1].
18 months after being removed, Didi immediately opened multiple subsidies after the rectification ended, worked hard to save market share by 2 percentage points, and began to regain its presence in the industry little by little, the first foothold is automatic driving.
Didi's road to autonomous driving began in 2017 with the establishment of the American Research Institute. In the following year, Didi successively obtained the California road test qualification and Beijing autonomous driving road test license, and established a "torrent alliance" with 31 automotive industry chain enterprises, including Bosch, BAIC, BYD, CATL, NavInfo and other automobile manufacturing/spare parts manufacturing/new energy/digital map/Internet of Vehicles enterprises.
In 2019, Didi will separate its autonomous driving business unit, with Zhang Bo, CTO of Didi Chuxing, concurrently serving as the CEO of Didi Wobud, and Jia Zhaoyin and Zheng Jianqiang serving as the heads of the US R&D team and the Chinese R&D team, respectively, both reporting to Zhang Bo.
Since independence, Woya has received a total of 3 financings with a total of more than 1.1 billion US dollars, which is one of the important reasons why the industry is bullish on Didi Woya at that time.
Even if Jia Zhaoyin left in June 2020, the same period when Didi announced SoftBank's investment, including the later withdrawal of Zheng Jianqiang, it did not affect the pace of Didi Woya's progress. The reason behind it was that some voices pointed to Didi's grand IPO at that time.
But all this was interrupted indefinitely in July 2021, until this month, Didi once again announced the progress of its business.
And if you consider the decline of the autonomous driving industry at this time, you can't help but wonder: Why did Didi stand up and tell a story at this time?
The answer is probably the one that most people will ignore, Didi is out of money.
As we all know, autonomous driving is an extremely expensive business. One of the root causes of the pain of many autonomous driving companies today is that L4 does not create higher than expected value for car companies, but has a tendency to be locked into a small black room because of cost issues.
How much does L4 really cost?
For test vehicle modifications alone, Pacific Securities has compiled cost data for Waymo and Cruise, which are $150,000 and $200,000, respectively.
In actual operation, the actual single kilometer of Robotaxi with safety officers has become as high as 1.77 yuan, higher than the 1.43 yuan of traditional ride-hailing [3].
See also Uber's data. In 2019, a court document revealed that Uber burned US$20 million, or about 134 million yuan, per month for its autonomous driving business, or 4.467 million per day for 30 days in January[4].
The court document, written during Uber's lawsuit with Waymo and filed in September 2017, details Uber's push for self-driving technology.
This may explain why Didi still insists on flexing its muscles in the downturn of the industry. In addition, the core data such as the number of autonomous driving units, the scale of operating vehicles, and the number of operating miles were not disclosed.
Security
In a sense, Didi is the most insecure type of platform company.
After surviving the taxi war, Didi hoped that the scale effect would bring revenue and profits to rise, but the continuous blurring of Internet boundaries has eroded its ride-hailing business from other platforms.
In 2018, Didi's market share reached 91%, and then continued to decline, and in 2021, the market share of order caliber fell to 74%, while the adjusted EBITDA still lost -39.317 billion yuan.
The crux of this is that Didi cannot obtain pricing power through monopoly position, so it cannot obtain monopoly profits, and it also has to pay the high competitive costs of other platforms when they erode.
This unattainable sense of security forces Didi to constantly defend, attack, and extend its borders to achieve a relatively safe situation.
In operation, it is the online car war between Didi and AutoNavi / Meituan / Haro, the takeaway war with Meituan, the battle with community group buying with e-commerce, and the car building plan and now autonomous driving.
But apart from the war between Didi and other platforms around the taxi business, other wars can basically be summed up in two words: burn money.
For example, Orange Heart Preferred, which began operations in May 2020, announced its closure after only one year and seven months. According to Didi's Q3 2021 financial report, Orange Heart's loss reached 20.8 billion yuan.
Another example is building cars.
In 2017, Didi found the ideal and BYD hope to cooperate in the development of exclusive ride-hailing, BYD is D1, ideal is the larger D1 Plus.
According to Li Xiang's revelation, Didi's vision of D1 Plus at that time was to have no passenger seat and trunk, and the luggage was placed in the co-pilot's position; There are two independent seats in the second and third rows, and the door on the left rear side has been eliminated, completely for ride-hailing and ride-sharing scenarios.
To this end, in 2018, Didi and Ideal established Judian Mobility. But then Ideal took Meituan's investment, and Meituan (cutting into the taxi business) and Didi began to compete, so Ideal and Didi terminated the cooperation "tacitly" [5].
As a result, in 2020, Didi only disclosed the D1 cooperation with BYD. However, Didi officially only revealed that the sales volume in the first quarter was 5,000 vehicles, and there was no sound in the follow-up.
Through public channels, the sales of this car in 2021 and 2022 are 10,176 and 994 units, respectively. Perhaps Didi has given up strategically, so at the end of 2022, BYD D1 began to be sold to individual consumers.
In this context, the progress of Didi's re-release of its autonomous driving business three years later is a little more intriguing.
New changes
It is safe to say that Didi has not yet got rid of the chaotic situation in the travel market, and even the ride-hailing market has given birth to more variables during the period when Didi was censored and removed.
On the one hand, in addition to Meituan and AutoNavi, more aggregation platforms have eroded Didi's living space.
For example, Douyin opened up its platform service provider in December last year, Huawei launched Petal Travel in July last year, and Tencent launched a taxi service in March last year.
According to the data of the National Ride-hailing Regulatory Information Interactive Platform, from July to December 2022, the proportion of orders on aggregation platforms in the national order volume increased from about 22% to about 26%.
On the other hand, car companies are also increasing their travel platforms.
From 2019 to 2022, China's mobility sector completed RMB5.27 billion, RMB10.23 billion, RMB16.66 billion and RMB2.74 billion respectively[6]. Among them, ride-hailing platforms such as Cao Cao, Sunshine Travel, and T3 have repeatedly received new financing, with a total financing amount of more than 10 billion yuan.
Capital has become a bargaining chip for these second- and third-tier travel platform players to fight back against Didi.
According to Guotai Junan data, from July 2021 to March 2022, the market share of T3 Mobility and Cao Cao Mobility increased by 6.9% and 1.4% respectively. Correspondingly, Didi's market share fell by 7.7%.
The essence of car companies' layout of travel services is a huge demand for cars in the short term.
At present, there are more than 5 million licensed ride-hailing drivers nationwide, and the actual number is said to be tens of millions. E-hailing is not basically replaced every three years, which means that at least 5 million new sales opportunities will be added to the market every three years.
Originally, the cooperation between Didi and BYD could be copied, but apparently the cooperation between Didi and car companies has reached a certain stalemate. Speculation is that beyond vehicle sales themselves, more problems may come from travel data.
Mobility data is related to the research and development of autonomous driving, and this is the purpose of GAC Group's investment in Didi Woya. At the beginning of April this year, while launching a customized car, Cao Cao also announced a cooperation with the startup Black Sesame Intelligence, which is to some extent the gap between car companies and Didi.
End
Didi's former rival, Uber, which has now withdrawn from the Chinese market, generated $8.6 billion in revenue in Q4 last year, with ride-hailing, food delivery and trucking contributing $4.1 billion, $2.9 billion and $1.5 billion, respectively.
After getting rid of the cash quagmire of autonomous driving, although there are no surprises in growth, Uber has used efficiency control fees to improve profit release.
Today's Uber can still be used as a reference for Didi, whether it is business composition, refined operations, or attitude towards the self-driving business.
More importantly, the disenchantment of the autonomous driving industry is actually a manifestation of a return to rationality. Wang Chuanfu actually has a discussion behind the "self-driving":
"There are currently millions of deaths in car accidents worldwide every year, but as long as the vehicles comply with the requirements of laws and regulations, today's automakers will not be complained about car accidents. But as long as there is a car accident in autonomous driving, then your car cannot be sold. [7]”
Wang Chuanfu understood, but what about Cheng Wei?
Resources
[1] Media Internet Industry: A New Order of Online Ride-hailing Development from the Perspective of Products, Competition and Policy, CICC
[2] Can automatic driving save Didi when the road is broken? Parity
[3] Smart Car Unicorn Series Report: Smart Traveler: GM L4 Architecture Enables Multiple Scenarios to Facilitate the Progressive Popularization of Autonomous Driving, Pacific Securities
[4] For autonomous driving, Uber burns 130 million qubits per month
[5] Li Xiang revealed that he had developed hybrid: "PHEV is not difficult", and also disclosed the real reason for breaking up with Didi, smart car reference
[6] China Mobile Financing Data List in 2022, NetEconomics
[7] Wang Chuanfu's bombardment of automatic driving is, blowing up all kinds of bigwigs, IT Times
[8] Avalanche's Price War, Life and Death of the Intelligent Driving Industry Chain, Tokawa Research Institute