On October 23rd, after-hours trading in the U.S. stock market, Tesla released its third-quarter report. For a detailed interpretation of the financial report, please see “The 'King of Drawing Cakes' Tesla has finally returned as a champion!”. Tesla's automotive business gross margin exceeded expectations, and it announced a new affordable model (referred to as Model 2.5) ahead of schedule. Both the delivery timeline and the contribution to sales growth in 2025 surpassed market expectations, leading to a 21.9% increase in Tesla's stock price overnight.

Following this, with Trump's ascension to power, the realization of AI narratives became a new catalyst for Tesla's stock price, propelling it back to a market capitalization of over one trillion dollars. The stock price rose 67% since the release of the third-quarter results, leaving many other electric vehicle stocks far behind.

Regarding Musk's AI narrative "cake," hints were already evident during the third-quarter earnings call. In “Is Musk really going to tear up the $25,000 Model 2?”, it was pointed out that Musk's strategic thinking has shifted: he believes that simply launching a standard version model priced at $25,000 is meaningless, as future vehicles will be driven by autonomous driving.

Thus, the purpose of the upcoming affordable Model 2.5, set to launch in 2025, is evident. This vehicle is likely intended merely as a transitional product to stabilize deliveries and stock prices, and the cost reduction is not driven by technology but is more likely a selectively downgraded version.

However, during the third-quarter earnings call, Tesla prepared excess production capacity for the Robotaxi prototype Cybercab: the Cybercab is expected to go into mass production in 2026, with an annual production capacity of 2 million units. With multiple Gigafactories producing together, the final output could reach 4 million!

It is clear that compared to the Model 2.5 launching next year, Tesla's true strategic focus is on the Cybercab model. Since Cybercab is the prototype for Tesla's Robotaxi business, this research by Dolphin focuses on Tesla's Robotaxi business, filling in the gaps not covered in the previous deep dive trilogy on Tesla: Tesla: How Far Is Musk's "Trillion Empire Dream"?, When the Lion Meets the Pack of Wolves, Can Tesla "Watch the Home"?, FSD Smart Driving: Unable to Support Tesla's Next Valuation Miracle.

Dolphin's main concerns regarding the Robotaxi business are as follows:

① Tesla's entry into the ride-hailing market: Is shared mobility a good business? What is the potential market size?

② Robotaxi vs. ride-hailing: Can replacing drivers with FSD fundamentally change the essence of the shared mobility market?

③ Why does Tesla want to enter the shared mobility market? What is Tesla's true purpose?

④ Why has Tesla prepared excess capacity for the prototype Cybercab?

The following is a detailed analysis:

1. Is shared mobility a good business? What is the potential market size?

Since Robotaxi ultimately serves the mobility market, let's first look at what changes have occurred in this market in recent years, taking China, currently the largest market, as an example:

In 2023, the overall market size of the mobility market has reached 7.5 trillion yuan, with private car travel still being the main component of the mobility market, accounting for 85% of the overall mobility market in 2023.

According to the structural evolution of the transportation market, Dolphin Jun found that after ignoring the impact of the pandemic, the proportion of public transportation and traditional taxis has shown a downward trend, while the share of private car travel and shared mobility (online ride-hailing + taxis) continues to rise.

This reflects a shift in the choices of transportation methods among the Chinese people: transportation modes dominated by shared mobility and private car travel are forming a substitute for public transportation and traditional taxis.

Robotaxi represents a technological revolution, but essentially it is still shared mobility. The difference lies in the fact that traditional ride-hailing services have human drivers, while Robotaxi is a software service composed of code and computing power (AI drivers).

Therefore, in this article, Dolphin Jun will first take a look at the size of the shared mobility market (online ride-hailing + taxis) without considering the impact of Robotaxi. How large is the market scale? What is the expansion speed?

From the chart below, the market size of shared mobility in China in 2023 is less than 300 billion yuan, with a penetration rate of only 3.8% in the overall transportation structure. The compound annual growth rate from 2018 to 2023 is 6.4%.

Currently, the market growth rate is far below what Didi claimed at the time of its IPO—30% compound annual growth rate from 2020 to 2025, with the market size of shared mobility expected to reach 534 billion yuan in 2023, while Didi itself has completely stagnated in the domestic market—no growth, merely squeezing profits.

From the interpretation of the past few years, urban public transportation and offline taxis have indeed lost a market size of 550 billion yuan, but this lost share has mainly been captured by private car travel (78% share), while shared mobility has only taken a small portion (close to 14%).

Clearly, when conditions permit, owning a private car is the first choice for users' travel. Correspondingly, the car sales business is the vast ocean, while ride-hailing services do not provide the actual benefits in the transition of user travel as imagined.

Why has the shift in transportation methods occurred, with private cars being the biggest beneficiary and ride-hailing services merely taking a small share? Dolphin Jun shares his thoughts:

① The driver’s car has not only a tool attribute but also a face and enjoyment attribute (as the price of the model increases, the proportion of the car's tool/face attribute becomes higher). Currently, the upgrade of automobile consumption is a phenomenon that transcends economic cycles (see the chart below), while ride-hailing services can only satisfy the tool attribute of the car (and can only replace part of the tool attribute), failing to meet the face and enjoyment attributes of owning a private car

② From the perspective of the tool attributes that private cars can be most replaced by ride-hailing services, users who regard private cars as tools account for the largest proportion, and the price of the purchased models is all below 200,000 yuan. In other words, this group of users mainly consists of price-sensitive users.

However, from the business model of ride-hailing, the supply-side drivers and vehicles are the core resources providing the service, so the share given to drivers (including drivers' labor costs + vehicle usage costs) generally accounts for about 80% of GTV.

But compared to the use of private cars on the consumer side, there are also vehicle usage costs to be paid, so the core difference lies in the labor costs of drivers.

According to a simple estimated UE model (referencing Didi), the expenditure on labor costs paid to drivers, excluding the vehicle usage costs (purchase cost - vehicle depreciation/rental cost, usage costs - fuel/electricity fees, etc.), accounts for 40% of the unit price per ride.

In addition to driver costs, ride-hailing users also need to pay the travel platform (the role of the travel platform mainly lies in matching user travel needs with driver capacity) which accounts for 10% of the unit price per ride, so these two expenditures together account for about 50% of the unit price.

This also leads to the conclusion that, in terms of long-term input-output ratio, using ride-hailing services actually costs more than owning a private car: according to Didi, the cost is inherently higher than the driving cost of owning a car (driving oneself does not require paying labor costs), and Dolphin Jun has made an estimate of the final costs of the two:

a. For Didi, the actual cost for passengers to take a ride is about 3 yuan/km;

b. For private cars, based on (annual depreciation, fuel/electricity costs, insurance, etc.), the average annual expenditure for private car ownership is about 22,000 yuan, while the average annual mileage for private cars is about 15,000 kilometers, which translates to a usage cost of 1.46 yuan/km for private cars.

Currently, the usage cost of private cars is still 50% cheaper than the ride-hailing cost, which corresponds to the earlier statement by Dolphin Jun that 50% of the unit price is paid to drivers' labor costs and platform fees.

(Note: In Dolphin Jun's estimation of the usage cost of private cars, the time cost of driving oneself is difficult to quantify, so it is not included in the calculation of usage costs.)

Based on the tool-like attributes mentioned by Dolphin, private cars are most likely to be replaced by ride-hailing services, where users are primarily price-sensitive (with most purchases under 200,000 yuan). However, since the cost of ride-hailing is mainly driven by the driver's labor costs, it is inherently higher than the cost of owning a car (by 50%), making it difficult for ride-hailing to replace private cars.

③ Ride-hailing replacing taxis: Ride-hailing and taxis have a similar cost structure (both consist of driver labor costs + vehicle acquisition and usage costs + commissions paid to taxi companies/ride-hailing platforms), but in practice, the profit distribution is flexibly adjusted based on supply and demand, and single-platform operations have greater economies of scale.

The core issue is that this is also what ultimately presents itself: the traditional taxi market has indeed been partially occupied by ride-hailing, but a larger share has been taken by private car travel, which has lower travel costs.

In summary of points ①-③, whether considering face value attributes or tool attributes, shared mobility is actually difficult to replace private cars and can only penetrate a small portion of the traditional taxi market.

Based on the above logic, without considering the impact of Robotaxi launches, Dolphin optimistically assumes an average annual growth rate of 10% for the Chinese shared mobility market from 2024 to 2030 (market size = user scale * average transaction value * user frequency, assuming an average annual growth rate of 5% for user scale and user frequency, with average transaction value remaining unchanged). Compared to the average market size growth rate of 6.5% from 2018 to 2023, this is already optimistic. By 2030, the market size of shared mobility in China will only approach 560 billion yuan, which is still limited compared to private cars.

With the limited scale of the shared business, shared mobility platforms like Didi also have relatively limited ecological barriers (for a detailed discussion, refer to “Unpacking Didi's Mobility 'Utopia'”).

In summary, shared mobility itself is a market with limited scale, and the shortcomings of its business model keep the profit margins of mobility platforms low, making it difficult to maintain market share, so it is not considered a "good business."

II. Can Robotaxi fundamentally change the essence of the shared mobility market?

The biggest challenges currently facing the launch of Robotaxi are still technical and regulatory issues. Assuming that the technical and regulatory problems of Robotaxi are resolved (implying that the safety of Robotaxi has far surpassed that of human drivers), the introduction of Robotaxi will lead to changes in the business model. The platform providers will still offer point-to-point passenger services on the product side, and the essence of the product has not changed. The biggest change comes from the supply side, where the core factor shifts from drivers + vehicles to vehicles equipped with autonomous driving technology.

There are two sources of vehicles equipped with autonomous driving technology:

From the B-side: Autonomous driving technology providers collaborating with car manufacturers (such as Baidu's Apollo Go), or original equipment manufacturers (OEMs) with self-developed autonomous driving technology: for example, Tesla's self-operated Robotaxi, which uses its own vehicles in a heavy asset model for platform operation (similar to the current CaoCao Mobility);

From the C-side: OEMs with self-developed autonomous driving technology—such as Tesla selling vehicles equipped with autonomous driving technology to the C-side, or existing Tesla vehicles gaining autonomous driving capabilities after the iteration and maturity of FSD technology.

The change from drivers + vehicles to autonomous driving technology vehicles on the supply side fundamentally saves on driver labor costs. Previously, 40% of the ride-hailing fees paid by users for human-driven vehicles went to cover driver labor costs. What Robotaxi can fundamentally change is the redistribution of this saved 40%, with possible distribution methods as follows:

a. Part of the savings is passed on to consumers, making Robotaxi's pricing lower than that of current human-driven vehicles, thereby expanding market size;

b. Part of the savings is allocated to platform providers to improve their profitability;

c. Part of the savings is allocated to the providers of autonomous driving vehicles, regardless of whether the vehicles come from the B-side or C-side.

The potential market size for Robotaxi, assuming safety is fully guaranteed, fundamentally depends on how much of this 40% is willing to be passed on to the user side to accelerate the replacement of human-driven ride-hailing/taxi services with autonomous ride-hailing services.

However, even assuming that Robotaxi fully penetrates traditional taxis + shared mobility, by 2030, based on optimistic estimates, the market size for traditional taxis + shared mobility will only be about 800 billion yuan, accounting for only about 8% of the total penetration rate of the mobility market, which is still a small business compared to the private car mobility market

Therefore, for Robotaxi to expand its market scale, it fundamentally needs to penetrate the private car travel market. In other words, shared mobility ideally should replace the demand for users to buy cars, even if users own private cars, they should prioritize shared mobility as their main mode of transportation (but this creates a conflict of interest with Tesla's car sales business).

3. Why does Tesla want to enter the shared mobility market? What is Tesla's true purpose?

The market size of Robotaxi itself is not particularly large, and selling cars is clearly a much larger market with easier profit potential, so why is Tesla pushing Robotaxi so aggressively? Dolphin believes this can be considered from two aspects:

1. From the perspective of maximizing Tesla's own interests as a supplier, since the technology underlying Robotaxi comes from the iteration of FSD technology, if Robotaxi can be successfully launched, it also means that FSD technology can mature.

When Tesla holds mature FSD technology, what is the choice for maximizing interests when there may be conflicts between selling cars and becoming a mobility platform provider?

First, let's take a look at the car sales business. According to Mordor Intelligence's forecast, by 2029, the market size of China's new energy vehicle market is expected to reach USD 674.3 billion (approximately 4.8 trillion RMB). In terms of market size, the market for selling cars is far larger than that for becoming a mobility platform provider.

Looking at Tesla's market share/profit margin in China's new energy vehicle sector, the manufacturing barriers for new energy vehicles are not particularly high, and Tesla's current technological lead has basically been caught up or even surpassed, leading to a continuous erosion of market share/profit margin.

In a previous in-depth study on Tesla titled "FSD Smart Driving: Unable to Support Tesla's Next Valuation Miracle," Dolphin mentioned that although Tesla has formed a commercial ecosystem in autonomous driving (especially along the pure vision route), this high-barrier ecosystem is centered around intelligence, However, intelligence has not yet become the primary essential demand for users when purchasing cars, nor can it be said to be the core barrier of Tesla's car-selling business.

If Robotaxi is successful, the implicit premise is that FSD technology is mature enough to drive completely autonomously on any road without the need for driver assistance.

In this case, FSD would be fully available, and the primary demand for users may have shifted to intelligence. Tesla could then form a closed-loop ecosystem similar to Apple's hardware-software integration in smartphones, allowing its business to achieve higher gross profit margins.

At the same time, from the perspective of vehicle economy, although Robotaxi can redistribute 40% of labor costs, fundamentally, users still need to pay (fees to platform providers + autonomous vehicle providers), which only narrows the cost gap compared to owning a private car (depending on how much of that 40% the platform provider is willing to pass on to users, meaning how Tesla balances the two businesses of private cars and ride-hailing).

From Tesla's perspective of self-interest, when conflicts arise between the interests of ride-hailing platform providers and car sales (as ride-hailing platform providers need to penetrate the private car market to truly scale), Tesla will still focus on maximizing profits by selling cars and FSD software.

However, for price-sensitive users, the cost of private car travel is still lower than that of ride-hailing, and owning an autonomous vehicle can save users time. Therefore, Dolphin believes that the likelihood of ride-hailing replacing private car travel remains low.

Thus, whether from the supply side—Tesla's own interests—or from the demand side—user choices, Robotaxi may still struggle to effectively penetrate the private car market.

With this analysis, Dolphin believes that Tesla's goal in entering the Robotaxi business becomes clear:

No matter how large the market for Robotaxi becomes, the ultimate effect is to allow more users to experience autonomous driving technology through the ride-hailing platform, cultivating user mindset, which in turn promotes the real goal—selling cars and integrating FSD software.

4. Is the 4 million capacity Cybercab really going to eliminate Uber and capture the ride-hailing market?

Tesla has prepared excess capacity for the prototype Robotaxi, Cybercab, with an annual production capacity of 2 million units, and with multiple Gigafactories producing together, the final volume can reach 4 million!

In Dolphin's view, Cybercab is not about doing ride-hailing business, but rather the next generation of truly high-volume fully autonomous vehicles and "One More Thing":

① Timing: Mass production is scheduled for 2026, allowing Tesla ample time to prepare for the next generation of models;

② Capacity preparation: Tesla has prepared a final production capacity of 4 million units for this vehicle, which corresponds perfectly with the 5 million units prepared for the next generation Model 2/Q, fully aligning with the definition of a "hot-selling car." ③ In terms of technical preparation: In electrification technology, Cybercab adopts a new round of unboxed manufacturing strategy in production technology, which can further reduce production costs for Cybercab.

At the same time, the launch of Tesla's Cybercab in intelligent technology also means that Tesla's autonomous driving technology is basically reaching a mature state, which coincides with Musk's definition of the next-generation vehicle: Musk believes that the true next-generation vehicle must be driven by advanced autonomous driving technology, with a revolutionary technological advancement to promote a large product cycle.

④ In terms of pricing: The pricing of Cybercab at launch is below $30,000, and the cost reduction from next-generation manufacturing technology may allow for further price reductions. However, once Tesla's integrated hardware and software closed loop is formed and its technological leadership is consolidated, a price of $30,000 will be sufficient for users to make a purchase.

The basic premise of the above discussion is that FSD can achieve true fully autonomous driving, minimizing the number of takeovers. This is the key for Tesla to establish hardware-software synergy and realize the "App store" moment. Dolphin will bring a detailed analysis of Tesla's FSD technology in the next article, but before that, we will cover Pony.ai and conduct an in-depth study of the monetization business model of intelligent driving service providers starting from L4+. Stay tuned!

Tesla's "covert operations," is the Robotaxi story just a "smokescreen"?