Robotaxi: Market Prospects, Core Premises for Development, Economic Viability, and Industrial Chain Landscape

1. Three Core Prerequisites for Robotaxi Development

Hardware Costs: Significant Cost Reduction in Core Components and Vehicle Costs

The sharp decline in LiDAR costs is a crucial hardware breakthrough. Around 2020, the unit price of a 64-line mechanical rotating LiDAR reached $80,000, while by 2025, automotive-grade products like Hesai Technology's AT512 have fallen to approximately $200, directly reducing the hardware cost per vehicle from over CNY 1 million in early stages to a range of CNY 0.2-0.35 million. Leading enterprises have achieved remarkable results in cost control:

• Baidu's 6th-generation autonomous vehicle has an overall cost of only CNY 204,600, a 60% decrease compared to the previous generation.

• Pony.ai's 7th-generation autonomous vehicle has a cost reduced to CNY 270,000, a 73% drop from the first generation, with potential for a further 30-40% reduction over the next three years.

• WeRide's next-generation Robotaxi is expected to see a further 20% - 30% cost reduction.

Table: Autonomous Driving Hardware Configuration of Various Companies

Software Technology: Advancement in Safety and Generalization Capabilities Driven by Large Models ​

The core breakthrough in algorithms lies in the combination of large models and data closed loops. Baidu Apollo's ADFM (Autonomous Driving Foundation Model), launched in 2025, adopts an end-to-end architecture, achieving a safety level 10 times higher than that of human drivers and covering complex urban scenarios. Pony.ai's simulation platform can generate millions of variant traffic scenarios every day, significantly reducing the need for real-road testing and accelerating the iteration cycle.

Technical approaches exhibit diversified competition: Baidu emphasizes vehicle-road-cloud synergy, Tesla adheres to a pure vision solution, and Geely introduces a "one-stage integrated large model", which improves the iteration speed through mutual calibration between cloud and on-vehicle models. Although handling extreme scenarios (such as nighttime construction zones and strong sunlight conditions) remains a key challenge, the industry accident rate is significantly lower than that of human driving. ​

Policy: Local Pilots Take the Lead, National Framework Takes Shape

China has formed a policy promotion path of "local breakthroughs and national coordination". By September 2025:

• 54 provinces and cities have issued normative documents on open roads.

• 28 cities support autonomous driving exploration:

  • Shenzhen took the lead in issuing intelligent connected vehicle management regulations.

  • Beijing and Shanghai allow commercial operations without safety drivers.

  • Wuhan and Chongqing open remote testing permissions.

Regarding financial support, cities such as Shenzhen, Wuhan, and Suzhou have launched subsidies or rewards for vehicle modification, testing fees, and demonstration applications, lowering the threshold for early investment and accelerating vehicle iteration and scenario verification.

The national-level Autonomous Driving Law has not yet been promulgated, and cross-regional regulatory standards remain inconsistent, issues such as accident liability determination and the legal status of remote safety operators are still unresolved. However, local pilot programs have accumulated substantial experience. For example, Guangzhou has achieved 24-hour Robotaxi network coverage, and collaboration with overseas markets like Dubai has also provided references for policy innovation.

2. Economic Efficiency: The Underlying Logic of Robotaxis Replacing Traditional Taxis ​

Fundamental Reconstruction of Costs ​

Labor costs are core burden for traditional taxis (accounting for approximately 60% of operating costs), whereas Robotaxis can completely eliminate driver expenses. Assuming an annual operation of 100,000 kilometers, the annual salary of a traditional taxi driver is about CNY 120,000, while the average annual operating cost of a Robotaxi is only CNY 30,000, with maintenance cost at CNY 50,000, indicating a significant long-term cost advantage. According to the OnTime prospectus and Frost & Sullivan’s report, the operating cost of ride-hailing/taxis in 2023 was approximately CNY 1.8 yuan/km, while that of Robotaxis was about CNY 4.5 yuan/km. These costs are expected to be equivalent by 2026, and the operating cost of Robotaxis is expected to drop to CNY 1.0 and 0.9 yuan/km in 2030 and 2035, equivalent to 42% and 36% of the cost of human drivers.

Figure: Forecast of Travel Service Costs, in RMB per Kilometer, 2019 to 2030 (Estimated), Source: OnTime(Chenqi Technology Limited) Prospectus, Frost & Sullivan

Improvement in Service Efficiency and Riding Experience

Robotaxis have the following advantages:

A) 24/7 uninterrupted operation, which increases vehicle utilization by more than 50% compared with traditional taxis.

B) Algorithm-optimized route planning shortens the average driving time by 15% - 20%. Moreover, dynamic dispatching can be realized through vehicle-to-infrastructure (V2I) collaboration, reducing urban traffic congestion.

C) Standardized service processes eliminate industry problems such as detours and refusal to take passengers.

Furthermore, passengers not only benefit from standardized driving performance (which, as autonomous driving technology matures, will probably surpass that of human drivers), but also enjoy a private space, allowing them to effectively convert travel time into productive work or leisure activities.

3. Enterprises Involved in the Industrial Chain

The Robotaxi industrial chain is structured as "upstream (hardware) - midstream (vehicle/OEM/solutions) - downstream (operation platforms/infrastructure)," as detailed in the table below:

In terms of business models, the common one is the tripartite cooperation among technology companies, automakers, and operation platforms, such as Baidu (technology) +BAIC/JMC (automaker)+Apollo Go/Baidu Maps (operation/traffic), Pony.ai (Technology) +Toyota/GAC (automaker) +Cao Cao/OnTime (Operation), WeRide (Technology) + GAC/Nissan (automaker) +OnTime (Operation). In addition, Tesla employs an integrated model encompassing vehicle manufacturing, technical services, and operation, thereby building its own ecosystem.

Conclusion

Driven by the triple synergy of continuous decline in hardware costs, rapid software technology iteration, and policy system improvements, Chinese Robotaxis are expected to replace traditional taxis from an economic perspective in the long run. Companies throughout the industrial chain are expected to benefit from this industry trend.