Text | Bounded UnKnown, Author丨Camellia, Editor|Qianjiang
This will be an event that will go down in history. ”
On November 11, Tesla's "WE ROBOT" launch event was held at Warner Bros. Studios in California, United States. This was Musk's high-profile remarks before the press conference.
But after the launch, it would be more appropriate to say that it was not so much an event that went down in history as it was a "late" event.
According to the original plan, the conference was supposed to take place in August this year, but Tesla postponed it to October. Before the press conference, an accident delayed the event by another hour, from the original 10 o'clock to 11 o'clock.
In addition to the timing delay, an even bigger delay lies in the release of Robotaxi.
According to the atmosphere set by Musk, coupled with the current situation that other self-driving companies in the market have landed, many people originally thought that Tesla would launch a full version of the Robotaxi business this time.
But in fact, Musk only took 20 minutes to briefly introduce the two new cars, and then fried a bowl of cold rice for humanoid robots, leaving the audience in front of the screen looking at each other, sitting on the Cybercab and leaving in style.
After the press conference, Tesla's stock price plunged 9% on the same day. Obviously, many people, both at home and abroad, are dissatisfied with this "history-described" press conference.
Two years ago, Musk frequently mentioned the imminent realization of full self-driving. But to this day, this high-profile "wish" is still far away.
This time, Musk drew another big pie in front of people all over the world.
特斯拉Robotaxi,姗姗来迟
The biggest highlight of the whole press conference was the release of two new cars, Cybercab and Robovan.
First up was the Cybercab, a model designed by Tesla specifically for the Robotape.
In terms of appearance, the Cybercab is very similar to Tesla's classic Model 3, but the color and tail have a very strong Cybertruck style, with sharp edges and corners, with Tesla's "ancestral" stainless steel, and scissor doors that are usually used only by sports cars, which is very sci-fi.
请Cybercab，图源Ars Technica
Inside, the Cybercab has only two seats. According to the interaction with netizens on the X platform before Musk's press conference, only two seats are designed because 90% of the mileage of United States vehicles is only single or double.
Cybercab，图源Ars Technica
In addition to this, the Cybercab does not have mirrors, no steering wheel and pedals (brakes, accelerator), and no charging plug because the car is equipped with inductive wireless charging.
According to the plan, the Cybercab will be fully operational by 2060. It is reported that the Cybercab will be produced at the Gigafactory in Texas using a new "boxless" manufacturing technology, which is expected to cut production costs in half.
Musk said, "It's costing up to about $1 per mile right now, but we think it can be reduced to about $0.2 per mile as the Cybercab business rolls out further." ”
The second car, the Robovan, is a self-driving minibus.
Robovan's appearance is somewhat similar to the previous ideal and much-criticized MEGA, like a polished building block, without the front and rear windshields commonly used in buses, five light bands surround the body, and the wheels cannot be seen from the outside, giving the whole car a sense of "suspension".
Cybercab，图源Ars Technica
Inside, Robovan has no cab, the seats are symmetrically arranged front and rear, and the whole car can carry 20 people, as well as transport cargo. According to Musk, Robovan can reduce the cost of travel to about 5 to 10 cents per mile.
显而易见,Cybercab和Robovan构成了当前特斯拉Robotaxi业务的两大支柱。
At the launch event, Tesla prepared 50 cars equipped with unsupervised self-driving, including the Cybercab, Robovan and Model Y, for them to drive back and forth on the streets of Warner Bros. Studios.
Sci-fi vehicles, with movie-quality sets, the scene is just like the cyberpunk scene in "Blade Runner" imagined by Musk.
But in any case, Tesla's Robotaxi is clearly late. Currently, the fastest mass production date for the Cybercab is 2026, and the mass production date of Robovan is still far away.
However, in China, Robotaxi has not only been widely operated, but domestic companies have gradually begun to expand overseas.
Judging from Baidu's financial report, as of July 28 this year, Radish Express has provided more than 7 million travel orders for autonomous driving services for the public. And it is expected to break even in Wuhan by the end of this year.
At the same time, Radish Express has also begun to lay out overseas markets. Ahead of Tesla's press conference, the "Science and Technology Innovation Board Daily" quoted sources as reporting that Baidu Carrot Express is planning to launch a Robotaxi service in Hong Kong, and is also in discussions with foreign automakers, ride-hailing platforms and other technology companies to deploy Carrot Express overseas through local partners.
Radish run, source Baidu
In addition, autonomous driving companies such as Pony.ai and WeRide are also accelerating the implementation of Robotaxi at home and abroad.
For example, Pony.ai CEO Peng Jun mentioned in an interview in August, "Pony.ai has been operating in many cities such as Beijing, Shanghai, Guangzhou and Shenzhen, among which Guangzhou has launched more than 100 Robocasis, with an average of almost 15 orders per car per day." ”
Pony.ai CTO Lou Tiancheng also said, "At present, there is no technical obstacle to autonomous driving, and Pony.ai will be likely to achieve positive bicycle profitability next year." ”
Overseas, Pony.ai has also signed memorandums of understanding with local enterprises and governments in Korea, Luxembourg, Saudi Arabia, U.A.E., Singapore and other places since the beginning of this year to accelerate its overseas layout.
WeRide's Robotaxi in China has also been in operation for nearly 5 years. In September this year, WeRide reached a cooperation with Uber and plans to launch a batch of autonomous vehicles through the Uber app in Abu Dhabi, the capital of the U.A.E., by the end of this year. Local passengers will then be able to use Uber taxis to call WeRide's Roboride.
In terms of autonomous unmanned minibuses, Qingzhou Zhihang's autonomous minibuses have started normal operation in 10 cities including Shenzhen, Wuhan, Beijing, and Wuxi.
In July this year, WeRide's driverless minibus began operation in Shenzhen. Prior to this, WeRide had sold 147 driverless minibuses from 2021 to 2023. As of September this year, it has received an order of interest for about 2,000 driverless buses.
It can be said that whether it is Robotaxi or Robobus, Chinese companies are far ahead.
The production of Cybercab is just the beginning
Of course, the competition in the operation of autonomous vehicles is not entirely reflected in the speed of landing. Behind this, there are many elements of competition between enterprises, such as vehicle costs, operating costs, user scale, and so on.
In terms of vehicle cost, according to Musk, the cost of the Cybercab after large-scale manufacturing will be less than 30,000 US dollars, or about 210,000 yuan. However, at present, the procurement cost of the latest sixth-generation unmanned vehicle of Baidu Radish Express is about 204,600 yuan, and the expected use cycle is 5 years.
So in terms of vehicle cost, the Cybercab doesn't have any advantages.
The second is the operating cost, although Robotaxi is driverless, it does not mean that people are not needed.
Taking Baidu Radish Express as an example, its full-time operation team behind the scenes is equipped with a variety of positions, including autonomous driving operation and maintenance engineers who do system upgrade and deployment, software and hardware tuning and troubleshooting for Robotaxi; Ground safety officers who dispatch, collect, wash, and charge Robotaxi vehicles; As well as collecting data for Robotaxi and testing scenarios, driving test safety officers, etc.
In this regard, there is no more information as the Tesla Robotaxi has not yet started operations. But before the launch, Tesla had filed a patent for a self-driving taxi's ability to automatically clean the cockpit.
From the introduction, this technology can allow the vehicle to achieve automatic cleaning and cabin disinfection while driving.
For example, when a vehicle detects a cough or sneeze from a passenger, the vehicle can use the air flow to expel droplets to the outside of the vehicle. At the same time, the technology can also eliminate certain viruses by using infrared and ultraviolet rays emitted at specific wavelengths, using air conditioning to raise cabin temperatures, and generating disinfectant steam.
Of course, if the vehicle detects that it cannot clean itself, it will also automatically drive to a dedicated cleaning facility, where it will clean itself using infrared illumination, contact heat or steam.
Tesla Robotaxi self-cleaning cockpit function, source Bloomberg
Judging from this patent, it seems that Tesla plans to build a highly automated Robotaxi operation system, so that its efficiency and cost may be more optimized. But this matter is also quite difficult, and it faces a huge upfront investment.
In addition to this, Cybercab also faced a more important challenge, which was to build a huge ride-hailing platform and dispatch system.
Before the press conference, Tesla released a 20s video of a woman calling Cybercab with one click on her mobile phone and adjusting the temperature of the air conditioning in the car before getting into the car.
For Tesla, it is not difficult to create a ride-hailing software similar to Didi or Uber, but the real difficulty lies in how to form a user scale and user habits.
In the more than a year since Didi was taken off the shelves, the market has fully proved that the moat built on user habits and huge operational data is not easy to break. In the 18 months since Didi was taken off the shelves, competitors led by AutoNavi have only seized less than 20% of the market share.
Therefore, whether it is Baidu, WeRide, or Pony.ai, when they leave the local market, they do not regard self-built taxi platforms as the first choice. Even in China, Pony.ai and WeRide have chosen to cooperate with ride-hailing platform Ruqi Mobility to provide Robotaxi services through Ruqi Travel.
Again, technology isn't everything in the case of Robotaxis. For Tesla, the mass production of Cybercab by 2060 is obviously only the first step, and the competition for users is more fierce hand-to-hand combat.
Some of Musk's flatbread
In addition to Cybercab, Musk actually has a lot of ideas about Robotaxi. One of the most important keywords is "sharing".
As early as 2023, Musk said in a high-profile manner at the World Artificial Intelligence Conference that an average passenger car is only used for about 10 to 12 hours a week, and most of the time it is parked in the parking lot.
However, a car with full autonomous driving ability can be used for 50~60 hours, and the utilization rate is 5 times that of ordinary passenger cars.
As a result, Musk believes that vehicles equipped with self-driving vehicles can be put into public transportation to make money when not in use.
And at the latest press conference, Musk exaggerated this logic again. He mentioned that if fully autonomous driving is used, the utilization rate of the car can be increased by 5~10 times on the basis of 10 to 18 hours, so that each car can be used for more than 100 hours per week, or even more than 160 hours.
During this time, car owners can fully entrust their vehicles to the platform and earn income through operation.
Cybercab，图源Ars Technica
Many people will definitely sigh when they hear this news, and finally when my car grows up, I can make my own money to support me. But if you also have plans to buy a car and let it work to earn money to support you, you may have to wait.
Because there are still many bugs and difficulties for this story to become a reality.
The first is in terms of time of use. Musk's so-called busyness and idle time of vehicles are actually in conflict with each other for most people.
For example, during the morning and evening rush hours or holidays, when you need to use a car yourself, it is also the peak period of public transportation. And late at night or during working hours, when your car is idle, there are not many people who need a car.
The second is in the use of psychology. For most people, a private car is the most intimate place outside of the room, and many people are reluctant to even let strangers ride in their own private vehicle, let alone put the car into the market in the absence of the owner. Therefore, it is clear that this change in psychology will also take time.
Finally, it is really in the legal position. In most countries, the regulatory requirements for commercial vehicles are stricter than those for private vehicles, including insurance, age and annual inspection requirements.
In China, for example, the scrapping standard for taxis is 250,000 kilometers, or 8 years of use. However, as long as a private car can pass the annual examination, there is generally no scrapping limit.
Therefore, how to legally position private vehicles involved in autonomous driving operations has become a major problem. If it is characterized as a commercial vehicle, the owner does not agree. However, if it is defined as a non-commercial vehicle, the operation safety of the vehicle cannot be guaranteed.
So Musk's plan is good, but it will obviously take a long time to actually land.
In addition to car sharing, another layer of "sharing" means computing power. This is another business model that Musk has envisioned for self-driving cars.
Like the logic of idle vehicles, in Musk's view, every car equipped with self-driving has a huge amount of computing power, but these cars are not calculated all the time.
Therefore, if the computing power of these cars is collected in idle time, it can also play a huge value.
At present, the computing power of Tesla's FSD HW 3.0 computing platform is 144TOPS, and its latest HW 3.0 computing platform may reach 216TOPS.
特斯拉HW3.0 图源:electrek
"If a car is driven for more than 50 hours a week, there are more than 100 hours to utilize, and if there are 100 million vehicles in the fleet, the computing power will reach 100GW." Musk said.
Compared with "shared cars", the threshold for sharing computing power is relatively low, but continuous high-intensity computing will also wear out hardware equipment and bring potential security risks.
Around 2010, when cryptocurrencies first became popular, CPU mining was popular in China, and many mining software would secretly mobilize personal computing devices to mine during idle time.
In this process, not only will the electricity bill increase, the hardware will accelerate aging, and the computer will become stuck because the computing resources are occupied. In addition, due to the security of the mining link network, the computer is often infected with malware and subject to cyber attacks.
Unlike PCs, the safety needs of automobiles are of a higher standard, as the potential risks and damages are clearly only greater.
Musk has always been the one who is very good at telling stories. But in the business world, people are obviously more concerned about something more realistic.
So the market needs Musk to come up with something tangible faster, instead of drawing cakes again and again.