Tesla has pulled out another big move.
In 2016, Elon Musk said that in addition to selling electric vehicles, Tesla would also create an equivalent of "Uber + Airbnb".
Don't understand what it means?
It means that when owners' Teslas are idle, they can be used to earn charging fees. Tesla's believers were left stunned by this idea.
By 2024, this grand plan evolved into the Robotaxi concept.
As the "biggest disruptor" in the L4 autonomous driving field, Tesla launched the Cybercab last October. With a pure vision system and no steering wheel or pedals, it defines autonomous driving according to Tesla's engineering philosophy.
Ten years have passed. Tesla has made money, but this trillion - dollar opportunity is just getting started.
When will we be able to smell the aroma of this opportunity?
01 Why is Robotaxi Different?
Eight months ago, Elon Musk announced the arrival of the Robotaxi era at the grand "We Robot" event.
The CyberCab prototype without a steering wheel and pedals caught many people's attention. Once you hail a CyberCab, it means you have to fully trust Tesla's L4 - level autonomous driving capabilities.
However, the vehicles in this trial operation are not CyberCabs, but 20 modified 2025 Model Ys.
There are safety operators sitting in the front passenger seat to verify passengers' identities. The service is not open to all users, and the operation area is limited to a designated area in Austin, Texas, which is only about half the size of Waymo's service area.
After the official launch of Robotaxi, many videos of people taking test rides appeared on the X platform. From these videos, the vehicles can generally handle daily urban driving scenarios. For example, they can make turns and U - turns smoothly and slow down to avoid pedestrians and construction sites. However, there are also some individual cases where problems occur.
It seems like a show that has successfully attracted a lot of attention.
On one hand, Tesla adopted a conservative road - going strategy to meet regulatory requirements. On the other hand, using Model Ys left an impression on onlookers, as if to say, "Look, our Model Ys can already pick up passengers."
The most crucial point is that this trial operation demonstrated the most outstanding aspect of Tesla's autonomous driving approach - scalability.
Different from Waymo, which relies on hardware stacking and exhaustive scenario testing, Tesla is a firm advocate of the "data flywheel" approach. It uses only vision sensors and the supercomputing cluster Dojo, and through end - to - end AI model training, allows the algorithm to learn autonomously from massive amounts of scenario data.
The more problems you solve, the more proficient you become.
Without the need for additional hardware installation or map surveying every time it enters a new city, it means that as long as local operation approval is obtained, mass - produced Model Ys and Model 3s can directly become Robotaxis in the future.
There is no need for expensive professional equipment or large - scale map surveying. Just do it.
There is potential for a leap from a few thousand to millions of vehicles. As long as road - going rules permit, the growth of the Robotaxi fleet will be exponentially terrifying!
Moreover, Tesla's confidence in its FSD technology is evident as it directly uses L2/L3 mass - produced vehicles for L4/L5 operations.
Tesla has long claimed that its "Full Self - Driving" (FSD) function will eventually reach L4 or even L5 levels, meaning the vehicle can drive safely on its own in most scenarios.
Since 2023, Tesla has gradually been testing the unsupervised version of FSD in North America.
To achieve the transition from L2 assisted driving to L4 autonomous driving, efforts must be made in both algorithm and computing power.
Tesla focuses on computing power in hardware stacking. Through self - developed chip iterations, it increases the computing power ceiling to support more complex AI model operations.
In March this year, at the Tesla employee conference, Musk revealed that the company invested approximately $10 billion in artificial intelligence last year, nearly half of which was internal investment, mainly used for Tesla's self - developed AI inference computer, all vehicle sensors, and the Dojo supercomputer.
The current HW4 is the second - generation version of the FSD chip. It is reported that the total computing power of HW4 reaches 500 - 700 TOPS, and the computing power of the next - generation HW5 is said to exceed 2000 TOPS.
The AI architecture of "massive data - supercomputing - end - to - end large model" lays the foundation for FSD capabilities, and the OTA mechanism enables Tesla to form a closed - loop of "data - training - deployment - re - learning", continuously improving the algorithm in the real world.
Different from competitors who spend large amounts of money to build customized unmanned fleets, through OTA, Tesla can upgrade millions of its on - sale models to L4 capabilities at once.
On the 28th, science fiction became reality.
Musk excitedly said on social media that the first Tesla vehicle was delivered to a customer's home without any human intervention.
Moreover, Tesla's FSD has achieved fully automated delivery from the production line to the logistics shipping area and then to the customer.
If the implementation of Robotaxi proves to be successful, it may likely evolve into a full - scale upgrade of mainstream mass - produced vehicles in the future, making it no longer a joking idea for private cars to earn charging fees on their own.
Currently, Robotaxi charges a fixed price of $4.2 per order (about 30 yuan), which is slightly more expensive than Waymo and Apollo Go.
However, due to Tesla's cost advantage in car manufacturing and the fact that fully autonomous driving no longer requires human intervention for maintenance, with the power of scale, the operating cost of Tesla's Robotaxi has the potential to be much lower than that of its competitors.
At a previous earnings conference, Musk repeatedly emphasized that the target cost for Cybercab is $0.2 per mile.
Undoubtedly, the decline of the unit economic cost curve depends on the growth rate of the fleet operation scale and the proportion it can occupy in the ride - hailing market.
This also determines how many billions of dollars in market value Robotaxi can add to Tesla.
Is everything now in place, waiting for the right opportunity?
02 There is Great Potential for Scaled - up Profitability, but Not Yet
Tesla has extended from the automotive manufacturing industry to an integrated mobility operation platform, and together with players like Waymo, Apollo Go, and Pony.ai, it has opened up the Robotaxi market.
Next, Tesla plans to expand its service to regions with higher regulatory thresholds, such as California. Musk said that he hopes to increase the number of Robotaxis to thousands within "a few months" to achieve a real unmanned ride - hailing network.
In 2026, the Cybercab without a steering wheel and pedals will debut as the main operating vehicle, and the scale will be greatly expanded.
A new pattern has emerged where an automotive company directly operates a large - scale Robotaxi fleet and competes head - on with traditional taxi companies and ride - hailing platforms.
However, with its cost advantage, driving experience, and the concentrated effect of data scale, the Robotaxi market worth trillions of dollars is likely to be a winner - takes - all market.
Financial institutions also support Tesla, pointing out its two core advantages: the customized chip architecture reduces hardware costs by 35%, and the end - to - end AI training model enables "inferential adaptation" to new environments instead of relying on preset programs.
In 2016, Musk planned to launch one million autonomous taxis on the road. However, by 2025, the number of Tesla autonomous taxis actually on the road was less than 20.
Compared with the gradual iteration from L2, other players have been deeply involved in the L4 level from the beginning and have made more rapid progress.
For example, Waymo, a subsidiary of Google, is estimated to have a fleet of nearly 2000 vehicles based on the latest order numbers. Its service area includes cities such as San Francisco, Los Angeles, and Phoenix, and it may plan to expand to new cities like Austin and Atlanta in the future.
Last month, Waymo announced that it had completed its 10 millionth passenger order. On the other side of the ocean, Apollo Go, a subsidiary of Baidu, had more orders than Waymo, reaching 11 million.
Two Chinese companies, Pony.ai and WeRide, have deployed unmanned fleets in Beijing, Guangzhou and other places, and their scale is expected to expand to over 1000 vehicles in the future. This year, the most notable thing is their entry into the Middle East market and cooperation with the ride - hailing platform Uber, which is also expected to achieve a faster scale - up speed.
However, the total number of Rototaxis on the road globally is currently less than 10,000. Therefore, competition is not the current theme. Growth and penetration are.
Goldman Sachs predicts that the US Robotaxi market will expand at a compound annual growth rate of 90%, reaching a scale of $7 billion in 2030, and the number of commercial vehicles will soar to 35,000.
In terms of economics, Robotaxi has significant advantages over ride - hailing services, which allows unmanned taxis to attract users with lower prices.
However, at present, due to limitations in operation area and cost, although the order data of unmanned vehicles is rising rapidly, the proportion is still negligible compared with ride - hailing services with density advantages.
Moreover, people don't choose taxis just based on cost. The most important thing is that taxis are everywhere, easy to hail, and provide a safe and reliable riding experience.
The prejudice that Robotaxi needs to overcome comes from the latter, and it needs to bridge the gap of user trust.
The prerequisite for successful commercialization is not just being cheaper. Even if Tesla can achieve lower costs and faster expansion, it still needs to prove that it can achieve at least the same safety level as its competitors, which is essentially a battle between the pure vision approach and the multi - sensor fusion approach.
There are certain regulatory risks. If local markets set road - going standards for unmanned vehicles based on the driving performance of industry leaders, then Tesla's path to scale - up must first overcome regulatory hurdles.
Scaled - up profitability is a common challenge for the Robotaxi industry. The industry as a whole is still in the investment stage. Pony.ai's net loss in the first quarter increased by 80.77% year - on - year, reflecting the high cost of technology R & D and fleet expansion.
The accelerated rotation of the unmanned driving mileage scale flywheel will significantly reduce the cost per mile. However, it may still take a decade to approach the ultimate cost range.
Only when the industry accelerates its penetration can the capital market continue to price it.
Epilogue
While other automakers use intelligent driving as a selling point to promote new cars, Tesla has taken a different path, steadily upgrading from L2 assisted driving to L4 - level autonomous driving and gradually building the blueprint of an unmanned mobility platform.
However, this is just the beginning.
Currently, Robotaxi is undoubtedly Tesla's most commercially imaginative new story to support its trillion - dollar market value.
Imagine that when a car owner can choose to connect their idle vehicle to Tesla's Robotaxi network, and the AI automatically accepts orders, allowing the vehicle to earn money for you.
Of course, Musk has been painting this beautiful dream for 10 years.
Regulatory review, the safety boundary of autonomous driving technology, and the synergy between software and hardware... Every step forward is not easy.
In the next decade, the competition in the Robotaxi field will be a tough battle of commercialization capabilities - whoever can find the best balance between cost, safety, and user experience will be able to dominate the trillion - dollar mobility market.
Glonhui Statement: The views in this article are from the original author and do not represent the views and positions of Glonhui. It is specially reminded that investment decisions should be based on independent thinking. The content of this article is for reference only and is not intended as any actual operation advice. Trading risks are borne by the investor.