Text: Liu Junhong
Edited by Wang Yisu
Musk may indeed want to "party" too much.
At 11 a.m. Beijing time on October 11, amid the anxious waiting of technology enthusiasts around the world, Musk, who was an hour late, appeared at the Warner Bros. studio in a Tesla Robotaxi ride.
Subsequently, at the Tesla "We, Robot" press conference, which was previewed as "going down in history", Musk took 20 minutes to speak, and then called the robot to hurry up and "dance and pour wine" for everyone.
Looking back at the event, Musk showcased a total of four products: Cybercab (driverless car Robotaxi), Robovan (driverless minibus), FSD latest progress and robot Optimus.
Maybe it's because Musk moved too quickly from speech to party. Investors, who did not hear the details of specific technology and commercialization, expressed their dissatisfaction with Musk's "big pie" by saying that Tesla's market value fell by more than 9% the next day.
After we take a closer look at Tesla's product design, we can still see Musk's ambition to achieve full self-driving. Relying on Robotaxi's accelerated iteration of intelligent driving capabilities, Tesla is hoping to break through from Baidu and Waymo as a car manufacturer.
However, since all the products in this demonstration are still "futures that can be cashed out for at least one year". Musk's layout may indeed have been announced a little earlier. Perhaps, Musk will have to wait a little longer to achieve the goal of fully autonomous driving. But in the face of increasingly fierce competition in the auto market, Tesla's time to maintain its technological leadership is running out.
Musk's big party, unmanned taxis are long overdue
The first is Cybercab, where Tesla has achieved a high degree of Robotaxi product completion.
Specifically, there is no steering wheel and pedals in the car, and with the small space that can only accommodate two passengers, it has the taste of "the ultimate means of transportation". In addition, it may be to facilitate operation and maintenance and reduce manpower participation in the energy replenishment link of Robotaxi. Cybercab has no plug and only supports wireless charging. In terms of autonomous driving technology, Cybercab is consistent with Tesla's products and is still a pure vision FSD solution.
In terms of manufacturing, Tesla currently expects to start production in 2027 at the latest. If users want to experience it first, they can experience driverless taxis from existing models after Tesla plans to launch fully autonomous vehicles in Texas and California next year. Finally, there is the cost, which is expected to cost the Cybercab about US$0.2 per mile (roughly equivalent to 9 cents per kilometer). For Cybercab, the current cost guidance is less than $30,000.
Next up is the Robovan, which can basically be understood as a larger Cybercab and looks a bit like a typical MPV. For specific planned uses, Tesla believes that Robovan can carry up to 20 people or transport goods. If used as a tourist minibus, the cost of travel is about 0.05-0.1 US dollars per mile (about 2 to 4 cents per kilometer).
Regarding the latest progress of FSD, Tesla did not disclose the news of FSD v13 as expected this time. In addition to Musk's belief that a completely unsupervised FSD is expected to be launched in Texas and California next year, all that remains is Tesla's massive training data, and FSD will eventually be safer than human driving.
Finally, there is the robot Optimus, and Musk gave an expectation of a future cost of $2-30,000. Musk did not mention anything about other developments. Combined with the performances of Tesla robots dancing, greeting, and pouring wine seen at the scene, it is generally not far from the demonstration in the first half of this year. Probably this appearance is really the "atmosphere group" of the party.
Perhaps the content of the presentation is too brief, and the four products mentioned are all "unredeemable futures within one year", or it may be that Musk has too many "vague" places in this press conference.
In short, investors were clearly not very happy with Tesla, and Tesla's market value fell by more than 9% at one point on the second day of the press conference.
At the same time, many analysts are worried that even if Tesla's Cybercab can solve a series of problems such as stainless steel shell production capacity, local supervision, and wireless charging station construction, and achieve large-scale mass production. But after three years of waiting, Tesla's operating space may have been swallowed up by other Robotaxi giants. Even, if you take into account the "urine" of Musk's almost every time he postpones the cashing schedule, it is difficult to imagine that the Robotaxi that Tesla has been mentioning since 2019 has really become "three years after three years".
Tesla's entry into the Robotaxi is really "chasing ducks to the shelves"?
Entering the Robotaxi comes from the perspective of car companies
Robotaxi has finally entered the stage where it is about to be commercialized on a large scale. In October of this year, the global Robotaxi giants invariably sounded the clarion call for expansion.
First, on October 4, according to Reuters, Waymo announced that it would add models from Korea's Hyundai Motor to its self-driving fleet. This cooperation means that Waymo has further expanded its influence among global automakers in the form of cooperation in the form of landing models after Tata Motors in India and Zeekr Motors in China.
Subsequently, on October 9, Nikkei reported that Baidu was planning to roll its Apollo Go robo-taxi service to markets outside of China. At the same time, in order to accelerate the progress of overseas deployment, Baidu has begun to actively seek cooperation with foreign automakers, ride-hailing platforms, and other technology companies to jointly promote the implementation of Robotaxi.
Now, with Tesla's official entry, the global Robotaxi industry is about to form a "three-legged" trend of Tesla, Waymo, and Baidu.
Among them, Baidu and Waymo are currently in a similar situation. Both are in the stage where they have already started operations and have initially verified their business models. The move of global expansion is essentially a commercial promotion from "pilot operation" to seeking economies of scale.
On Tesla's side, although Musk only showed a "prototype" this time, the specific landing time is not yet certain, saying that "Cybercab will be put into production in 2026, and it may not be until 2027 at the latest". However, the purpose of Tesla's release of Robotaxi this time is not similar to the expedited layout of Baidu and Waymo. In Tesla's view, perhaps the final winning logic of the Robotaxi track has to return to the car companies that are closest to the iteration of autonomous driving and car manufacturing.
Based on Musk's message at the "We, Robot" press conference as a premise. From a technical point of view, Tesla's Robotaxi first verifies that the "progressive" of autonomous driving can also run through the evolution path of L4.
This means that in order to realize Robotaxi, it does not necessarily require extensive investment in Baidu's vehicle-road-cloud collaboration, or Waymo's algorithm and sensor design for L4 "tuning". Car companies with strong intelligent driving capabilities can rely on a set of general intelligent driving development models (such as Tesla's end-to-end + pure vision), and only need to change the corresponding supporting computing hardware (Cybercab is expected to carry about 10 times the computing power of the AI5 platform), and then launch intelligent driving and Robotaxi products to the market respectively.
In other words, according to Tesla's thinking, the Robotaxi business is essentially another battlefield for intelligent driving players to compete for sales. After all, if Robotaxi shipments are measured only by operating break-even, Zhang Ning, vice president of Pony.ai and head of Robotaxi autonomous driving travel business, gave the result to Light Cone Intelligence that "a single city will be put on the order of 500-1000 units".
In this process, on the one hand, car companies have reaped higher revenue (operation or direct sale) through Robotaxi. On the other hand, the incremental driving data brought by Robotaxi will further accelerate the iteration of intelligent driving. The improvement of intelligent driving capabilities can also allow car companies to grasp a more favorable bargaining position.
Although, there are no cases of car companies that have done both Zhijia and Robotaxi. However, we may be able to learn from the changes in the sales structure of Li Auto this year and confirm the direct impact of the improvement of intelligent driving capabilities. "Since version 6.0 (upgrade), in the past 1-2 months, our AD Max has accounted for more than 50%, and there is a growth of more than 10% every month." Lang Xianpeng, vice president of intelligent driving R&D of Li Auto, introduced the light cone intelligence at a communication meeting in August.
In this case, Tesla is expected to obtain more FSD subscription revenue, and truly realize the commercialization of software (AI) revenue as a technology company.
How long will Musk's "AI" pie be painted
Although 80% of Tesla's revenue (2023 annual report data) comes from selling cars. But Wall Street has long believed that half of Tesla's nearly 5 trillion yuan market capitalization comes from AI valuations.
This is because, when Tesla became the global leader in new energy vehicles in 2020, it formed the dual attributes of the automotive industry (consumer discretionary consumer goods, traditional value stocks) + FSD subscription (software sales, technology stocks) with the advanced centralized automotive EE architecture design and Musk's pursuit of autonomous driving. Under this model, when the capital market uses the PS (price-to-sales ratio) indicator to simply calculate the valuation of the attributes on both sides, it forms the fundamental that Tesla sells cars only on the basis and AI has "very high valuation leverage".
To put it simply, as long as the car is not "collapsed", then Tesla's AI cash basically determines the direction of the company's market value.
The sharp drop in market value after Tesla's "We, Robot" conference reflects the dissatisfaction of the capital market with Tesla's failure to come up with anything new, and even Musk himself is not sure when the Cybercab will be mass-produced.
Perhaps, the capital market's "punishment" of Tesla is still somewhat "merciful".
According to Musk's expectations, the premise of mass production of Cybercab is that Model 3 and Model Y will achieve unsupervised autonomous driving next year. It is true that in the end-to-end intelligent driving development mode, Tesla has achieved the overthrow and reconstruction of most of the car's intelligent driving functions in the entire iteration of the FSD v12 version. The intelligence of FSD can even take He Xiaopeng, chairman of Xiaopeng Motors, to "walk all over the street" when there is a bug (no arrival point is set).
Considering the performance of the current version of FSD, combined with Musk's promise in 2016 that Tesla's car will be able to drive from New York to Los Angeles alone "next year". Whether Tesla will actually be able to achieve unsupervised autonomous driving in 2025 is still a question mark. Perhaps, Tesla's goal of achieving both sales and FSD subscription revenue growth by relying on Robotaxi is just another "reasonable" story told by Musk to the market.
But after all, there is still a limit to what the market can "eat".
Looking around the reality, in the market led by China's smart cars, Tesla's glorious days are long gone. In the field of automobile manufacturing, Tesla's global sales and gross profit margin of automobiles (Tesla excludes the impact of carbon credits) have long been defeated by BYD. On the other hand, in terms of intelligent driving capabilities, although Chinese intelligent driving players have not yet been able to achieve the height of "complete end-to-end". However, through the comparison of many Chinese manufacturers who are about to achieve a "door-to-door" intelligent driving experience in 2025, Tesla does not have too many advantages.
Perhaps, according to the "formulaic" plot design of "Elon · Musk". The goal of achieving full self-driving is still in the stage where Musk is "in a desperate situation and has been delayed many times". However, Tesla has returned to the world's first intelligent driving window with full self-driving, and there is obviously not much time left.