Musk's super bet: bet on trillions of Tesla and go all out to mass produce Robotaxi!

On April 22, Beijing time, Tesla has made frequent recent big moves:

1. Large-scale layoffs: more than 10% of employees worldwide;

2. Big price cuts in China and the United States: China cut prices by 14,000 yuan and the United States by 2,000 US dollars;

3. Resignation of senior executives: power battery executives, including senior vice presidents of energy engineering and powertrain, have resigned one after another;

4. FSD price reduction: The price of FSD in the United States has been reduced from $12,000 to $8,000, EAP has been cancelled, and the reading subscription fee has been reduced from $199/month to $99/USD;

5. Adjust the mass production model plan: cancel the mass production plan of cheaper models than the Model 3;

The core of all this action points to Musk mobilizing all the company's resources to do something unprecedented and unprecedented, that is, Musk's announcement on April 6 that "Robotaxi will be released on August 8". Musk's official announcement: Robotaxi will be released on August 8th!

Musk bet on Tesla, and if it is not successful, Tesla will be a car company and a robot company. At present, Tesla's latest market value is 450 billion US dollars, about 3.25 trillion yuan! Musk is betting all in this time.

Moreover, it is a model that is completely born for unmanned driving, "no steering wheel, no brake pedal, no rearview mirror" and other equipment that serves human drivers. This is also the heaviest fully driverless vehicle released by an American tech giant since Waymo's predecessor released the driverless prototype "Firefly" released in 2014, and GM Cruise released the driverless vehicle Origin in 2022.

firefly

"Firefly" is a driverless vehicle exploration by a software giant, "Origin" is an unmanned vehicle exploration by an automobile giant, and Tesla's Robotaxi is a real artificial intelligence giant (Musk has always believed that Tesla is an artificial intelligence company rather than a car company, and the market valuation is the same) A big gamble on the mass production of driverless vehicles.

The previous 2008 Roadster, 2013 Model S, and 2018 Model 3 are all in to solve the mass production problem of electrification, and this time, Musk wants to fully start the intelligent war, which is also the best way to jump out of the involution of China's new energy vehicles!

The problem is that the mass production of completely unmanned vehicles requires both software (autonomous driving algorithms) and hardware (vehicle platforms) to be prepared, and the inability of "Firefly" to mass produce is obviously due to the fact that the vehicle hardware is not ready, and the inability of "Origin" to be mass-produced can be regarded as a shortcoming in software. So, is Tesla ready for both software and hardware?

01

$10 billion to purchase FSD for computing power training

In order for Robotaxi to achieve mass production, the first thing to solve is the problem of autonomous driving algorithms.

At Tesla, there are three levels of autonomous driving algorithms: AutoPilot, Enhance AutoPilot, and FSD:

1. AutoPilot (AP): Standard configuration on every Tesla production car, which can realize L2-level autonomous driving functions;

2. Enhance AutoPilot (EAP): 32,000 RMB in China, which can realize L2+ autonomous driving functions, including high-speed ramp changes, etc.;

3. FSD, full-self-driving: completely unmanned function, additional cost, 64,000 yuan in China, once as high as 12,000 US dollars (more than 86,000 yuan) in the United States, in March, FSD Beta was also renamed FSD Surprised!

It is worth mentioning that Tesla has recently canceled the EAP level in the United States. At present, it is unclear whether Tesla will delegate the EAP function to all Tesla owners for free in order to obtain more training data. At the end of March, Tesla announced a month of FSD free for Tesla models across the United States.

In order to train the driving model of FSD, especially the FSD that adopts an end-to-end large model, Tesla has built a supercomputing center Dojo, and according to Tesla's latest plan, it will invest $10 billion in 2024 to purchase computing power for training FSD, that is, to stock up computing chips for the supercomputing center Dojo.

In Q1 2024, Tesla bought 90,000 Nvidia H100, and now the hottest computing chip in the AI field, Tesla may have paid a price of $3 billion, more than the sum of 2016-2023. And the $10 billion invested in chips in 2024 alone will exceed all of Tesla's $8.9 billion in capital expenditure in 2023.

Such a strong investment may be able to hoard a computing power scale comparable to that of Microsoft and Meta, the world's largest computing power companies, for Tesla.

Computing power is especially important for training FSD. Training data is another dimension of demand, and Tesla is completely self-sufficient in this area.

Musk once said in the financial report that data is extremely important for end-to-end large models, "1 million video clips - barely enough, 2 million - slightly better, 3 million - just feel wow, 10 million - this is an incredible amount of data" This is the demand for video data fragments for the end-to-end large model Scaling Law.

As of Q1 2024, Tesla has produced more than 6 million vehicles, and on April 22, Tesla's 10th anniversary of entering China, it has also accumulated more than 1.7 million car owners. With such a large stock of Tesla, it is possible to contribute video footage data of training FSD to Tesla every day, of course, the premise of desensitization, or compliance with local regulations exists locally.

02

Build a Robotaxi vehicle platform

Tesla plans to invest $10 billion in 2024 to purchase computing chips and provide video footage of training FSD through Tesla's stock vehicles, so as to train fully driverless robot drivers.

The next step for Tesla is to mass-produce a vehicle platform suitable for Robotaxi.

Although Tesla has produced more than 6 million pure electric vehicles, it is another matter to use the Robotaxi vehicle platform. Here are some of the features of the Robotaxi vehicle platform:

1. Safety redundancy: Previously, all models produced by Tesla, human drivers were the last safety redundancy of the FSD system, and the Robotaxi vehicle platform will no longer have human drivers as the last safety redundancy, which requires a new design, especially in terms of safety redundancy.

2. Durability: The intensity of use of Robotaxi models exceeds all of Tesla's previous models, therefore, Tesla needs a battery pack to meet 1 million miles, that is, 1.6 million kilometers, which is about 100 years of driving history for human drivers commuting to and from work.

3. Remove all designs that serve human drivers: including steering wheels, acceleration and brake pedals, rearview mirrors, etc., all the designs that serve human drivers are removed, but at least double in terms of safety redundancy;

In 2021, Tesla's vehicle factory VP Lars and chief designer Franz persuaded Musk to design a cheaper model for both a manned and unmanned version. However, in February 2024, a new round of meetings will only be able to choose one of the above two options.

In the end, since ancient times, 2B and 2C have served two completely different objects, and they can't have both!

Musk decided to all in the driverless version, that is, the mass production of Robotaxi models, the car intelligence blindly guessed, the name of the vehicle may be Tesla Robotaxi, Tesla's Robotaxi service is Tesla Go!

It is worth mentioning that Tesla also hopes that the Robotaxi model platform can be reused in the two best-selling models, Model 3 and Model Y, in order to seek common parts and further reduction in manufacturing costs.

Tesla is so determined to engage in Robotaxi, or Musk is so determined to engage in Robotaxi, Musk may see that it is completely unable to compete with Chinese new energy vehicle companies in terms of volume cost, after all, the US bill requires everything to be produced back in the United States, which naturally causes the cost to not be reduced!

Then, it is better to jump out and directly start the next war - intelligent war.

03

Start intelligent warfare

Intelligent warfare, completely unmanned driving is definitely the core of the core, the foundation of the foundation, the vast majority of the business model of automotive intelligence is based on unmanned driving, and Robotaxi is the best and largest application scenario of unmanned driving.

Musk believes that Robotaxi can push Tesla's market capitalization to $10 trillion.

For the mass production of Robotaxi, Tesla needs to comply with the supervision of Robotaxi in the United States, in addition to its own FSD training and mass production of vehicle platforms.

At present, many places in the United States have realized the complete commercial closed loop of Robotaxi that can be unmanned, manned, and charged, Waymo is also actively carrying out Robotaxi operations, and Cruise is also trying to recover after being stopped after being stopped for the fatality accident.

However, Tesla's process in this regard still does not go that way. It is reported that Tesla is in active communication with the California Department of Motor Vehicles (DMV), hoping to obtain the same license as Waymo to operate Robotaxi in limited areas within 2024.

In fact, the regulations of the California DMV are there, and Tesla just follows suit.

It's like Musk dissed Lei Jun compared the physical data of the su 7 and Model 3 at the pixel level at the press conference, saying, "The real innovation is to push the limits of physics, not just surpass the competition." ”

For Tesla, the cost can't be rolled up, so let's start the next intelligent war! Continue to use technology to earn high returns.

Driverless technology will maximize the residual value of a car!

I hope Musk, this time the bet, can win!

Humanity needs the commercialization of destructive innovative technology - unmanned driving, so as to enter the next stage of development!