HomeArticle

Tesla doesn't believe in fuel tanks and is gambling heavily on driverless technology in a difficult situation.

韩永昌2024-10-11 17:25
Whether Tesla can reverse its situation from the bottom is related to the entire automotive industry.

Written by Han Yongchang and Li Anqi

Edited by Li Qin

Chinese electric vehicle companies, deeply embroiled in the price war quagmire, are collectively retreating and equipping fuel tanks to seize sales and survival opportunities. Tesla, conversely, is taking an aggressive step to the opposite end - launching driverless taxis to seek a breakthrough.

On the evening of October 10 local time, Tesla CEO Elon Musk got into a highly sci-fi vehicle that has no steering wheel, brakes, or driver and possesses the ability of driverless driving.

This is one of the main characters of the day's conference, Cybercab (autonomous driving taxi). Simultaneously, Tesla also launched Robovan (autonomous driving transporter).

Tesla Autonomous Driving Taxi Cybercab, Source: Tesla

Tesla claims that in 2025, it will initially launch the UnsupervisedFSD (unsupervised fully autonomous driving) technology in California and Texas, USA, utilizing the Model 3, Model Y, and Cybertruck models.

FSD stands for Full Self-Driving (fully autonomous driving), which is Tesla's core autonomous driving technology. Achieving unsupervised status is a crucial step to remove the driver and achieve an autonomous driving taxi.

Elon Musk stated that it is anticipated that the new autonomous driving taxi form, Cybercab, will be mass-produced and widely rolled out in 2026 and 2027.

"The future charging of Cybercab can reach $0.2 per mile, while the current ride-hailing is approximately $1 per mile," Musk said.

In another corner of the conference, Tesla's humanoid robot Optimus can interact and dance with the audience. Musk said that the price of Optimus after mass production is about $20,000 to $30,000. "It can take care of children, walk the dog, mow the lawn, buy groceries, and even serve as an assistant in social situations."

Tesla chose Hollywood as the conference venue, which is the American movie dream factory. It seemingly implies that science fiction will become a reality.

But unfortunately, in the only half-hour conference, Musk only briefly introduced the two new autonomous driving vehicles and the recent situation of Optimus, without revealing more content. Musk did not mention the technical implementation basis from Tesla's FSD (fully autonomous driving) software to autonomous driving taxis, and how to handle the complex operation and supervision.

Tesla is a firm supporter of pure vision and has explored an intelligent driving evolution route that does not rely on lidar.

But from assisted driving to fully autonomous driving, Tesla obviously needs to answer more questions. In addition to technical breakthroughs, the operation of Robotaxi also involves various issues such as supervision, fleet management, user charges, software failures, and liability division.

For Tesla's overall operation, the commercialization of FSD and Robotaxi is still premature, and Musk obviously has not provided a powerful answer on how to counter the current slow decline situation.

"Elon Musk's Robotaxi Dream"

In 2016, Elon Musk claimed that a Tesla electric vehicle could drive from Los Angeles to New York without the need to touch the steering wheel. If you want the car to drive back by itself, just click the summon on your phone.

Immediately, he began to push engineers to design a vehicle version that has no steering wheel, accelerator, or brake pedals.

Obviously, with the technical level at that time, such an idea was difficult to realize.

But Elon Musk has never given up the dream of Robotaxi. It is mentioned in "The Biography of Elon Musk" that Musk will always instill a sense of urgency in his employees: "This will be a historically significant and revolutionary product that will change everything and make Tesla a $10 trillion company. A hundred years from now, people will remember this moment."

At today's conference, Musk finally began to fulfill his promise. The design of Cybercab is filled with a futuristic sense, with no rearview mirrors, pedals, or steering wheels, and is equipped with scissor doors.

Tesla Autonomous Driving Taxi Cybercab Interior, Source: Tesla

Elon Musk's plan for Robotaxi is not limited to Cybercab. He said at the conference that a vehicle is only used for 10 to 100 hours per week. When fully autonomous driving is achieved, the value of the same vehicle can be increased by 5 to 10 times.

Although Elon Musk did not announce the operation mode of Robotaxi, it is evident that private cars can also participate in it.

In terms of cost, Elon Musk believes that the cost of autonomous driving will be very low. "Cybercab can reach $0.2 per mile in the future, and the current ride-hailing is about $1 per mile, and the vehicle cost will be less than $30,000."

As for the time, "Before 2026 and 2027, we will mass-produce Cybercab, and there will be Cybercab 2 in the future."

At the conference, Musk also demonstrated the patent of automatic cleaning. The robotic arm will suck away the garbage on the seat like a vacuum cleaner and wipe and clean the screen, etc. The unique inductive wireless charging method of Cybercab was also mentioned, but it was not discussed in depth.

In addition to Cybercab, Robovan also made an appearance at the conference. Musk compared it to "the vehicles in Star Wars."

Tesla Autonomous Driving Transporter Robovan, Source: Tesla

Robovan can carry 20 people and also transport goods. In Musk's plan, the travel cost of this vehicle in the future will be as low as 10 cents per mile. This is obviously disruptive.

Take Waymo Robotaxi, which has several years of operating experience in the United States as an example. The per-mile charge is about $13.3, which has a huge gap with Musk's expected $0.2 per mile, and is even further from 10 cents.

Even though Waymo announced that its weekly paid orders have exceeded 100,000 times, it has not generated considerable profits.

Tesla may use its large-scale manufacturing and supply chain capabilities to compress Cybercab to $30,000 and quickly optimize the operational financial model of Robotaxi.

But the problem remains how quickly Tesla's FSD technology can be implemented to get rid of human drivers and achieve driverless driving in reality.

"Tesla Autonomous Driving, A Bumpy Road to Monetization"

Elon Musk believes that FSD will be ten times safer than human drivers in the future, and he also has high expectations for the income brought by FSD.

"The future income of autonomous driving will make all gross margin figures look stupid."

There are two paths for FSD to generate profits. One is a subscription service like Apple's iPhone, and the other is to be installed on Robotaxi to save labor costs and transform the existing travel market form. But both paths are full of challenges.

Take the subscription model as an example. The gross margin of Apple's hardware products such as iPhones has remained at around 40% for many years. Although sales have declined, software services have supported the company's profits. Its software products such as the App Store, Apple Music, and iCloud have a gross margin of more than 70%.

In the third quarter of this year, Apple's hardware revenue was $61.6 billion, and its software revenue also reached $24.2 billion. As long as the product retention is sufficient, software can continuously generate profits. This path has been verified in Apple.

Tesla previously disclosed that the gross margin of FSD is as high as 90%. All Tesla models produced after 2021 with HW3.0 can purchase the FSD service.

From 2021 to the third quarter of this year, Tesla has delivered more than 5.35 million vehicles. The most favorable subscription price of FSD is $99 per month. If the business model of FSD can be like Apple's, and it is globally open for subscription, even if only 10% of the subscriptions are made, it can contribute $140 million in profits to Tesla every quarter.

FSD has been open for subscription in the United States, Canada, and other regions. Elon Musk also revealed not long ago that Tesla plans to apply to the regulatory agencies in Europe, China, and other countries for the approval of its FSD supervision version in the V12.5 or V12.6 version of the FSD software, and it is expected to be approved by the end of the year.

At today's conference, Musk also mentioned that Tesla will first achieve fully unsupervised FSD in the Model 3/Y and Cybertruck, "and it will be first launched in Texas and California next year."

But to increase the subscription volume of FSD, the key still lies in the product experience and technical capabilities.

Some industry insiders said that after Tesla established a closed loop of end-to-end intelligent driving, massive data, and super-large cloud training computing power, its FSD capability has made a significant leap. "The technical upper limit is higher than the traditional autonomous driving solution."

However, there is still a gap between the product ceiling and the actual performance. Multiple data show that users still need to frequently take over the vehicle when using FSD.

The evaluation report of the third-party testing agency AMCI Testing shows that the latest released FSD v12.5 version requires human intervention once every 13 miles (about 21 kilometers) on average. Pony.ai, a domestic autonomous driving company, also told 36Kr that it has conducted road tests in the relatively busy Bay Area urban roads in the United States, and the test results show that it needs to be taken over once every less than 10 kilometers on average.

An FSD user told 36Kr that the main reason for the manual takeover of the V12.5.4 version he is currently using is that in some road construction scenarios, the lane lines are more complex. "People can basically understand it quickly, but the car may not be able to understand it."

This is still far from the performance of a mature Robotaxi. According to the data released by the California Department of Motor Vehicles (DMV) in 2023, Waymo requires intervention only once every 27,900 kilometers.

Going directly from L2 assisted driving to L4 autonomous driving, Tesla's FSD also needs to face more problems.

Zhang Ning, an executive of the domestic Robotaxi company Pony.ai, told 36Kr that Tesla's FSD performs well in assisted driving (L2), but there is still a big gap from the technical and safety indicators required for driverless (L4).

"Tesla uses pure visual sensors and relatively low map dependence. FSD has achieved a relatively good assisted driving human-like experience in North America and many regions, but some facts about FSD may not be completely consistent with the public perception."

Kyle Vogt, the founder and CEO of the American autonomous driving company Cruise, listed 15 key points for the landing of Robotaxi before the Tesla conference, including traffic congestion, collision detection, the coverage of artificial intelligence, supervision and licensing, etc. Tesla may not be an exception to these difficulties.

In other words, at the current stage, whether Tesla charges for FSD subscription for car owners or operates Robotaxi, it is difficult to quickly generate considerable profits.

Selling cars well is still Tesla's inevitable survival destiny.

"The New Model Y, Entering a Fierce Battle"

The long-rumored low-cost Tesla Model Q has still not appeared. Before the Model Q is launched, Tesla's sales and profits can only be supported by the Model Y.

36Kr learned from several informed sources that the new Tesla Model Y will be launched in the first quarter of 2025. Compared with the current model, its endurance will be increased by about 10%, and the battery capacity will be larger. Since Tesla is a platform development strategy, after the battery pack of the new Model Y is modified, it will also drive the endurance of the Model 3 to further increase.

Not only that, "The appearance of the new Model Y will be greatly changed, the chassis will be more comfortable, the power consumption will be lower, and the interior can refer to the refreshed Model 3," an informed source said.

Tesla has also planned a six-seat version of Model Y for the Chinese market, but this car is "adding a seat to the Y. The third row is a small bench, and the interior space will not be very competitive."

Model Y is Tesla's meritorious model. In 2023, this car ended the 37-year sales championship of the Toyota Corolla and became the world's best-selling model with sales of more than 1.23 million units. At Tesla's delivery center on working days, a Tesla is picked up every 6 seconds.

Entering 2024, Tesla's sales growth has begun to show signs of weakness, with year-on-year declines in the first two quarters. Although there was a slight improvement in the third quarter, it still did not meet market expectations.

36Kr Chart

After the sales decline, profits also began to decline. In the second quarter, Tesla's operating profit was $1.61 billion, a decrease of $790 million compared with the same period last year; the net profit was $1.48 billion, a decrease of $1.23 billion, with a decline of up to 45%.

In the semi-annual view, Tesla's net profit in the first half of 2024 is almost only half of that in 2023 and 2022, and the gross margin has also dropped to 17.6%.

36Kr Chart

More and more new cars are also taking the opportunity to aim at the market share of Model Y and launch a fierce attack. Zhijie R7 supported by Huawei, NIO Ledao L60, Zeekr 7X, and the new carmaker Xiaomi will also launch SUV models next year.

These models have made detailed comparisons with Model Y at the conference and have all received tens of thousands of large orders in a short period of