Apollo Go's 2025 Outlook: Cost Optimization, Global Expansion, and the Path to Robotaxi Profitability

1. Single-Vehicle BOM Analysis: The Cost Revolution of the Sixth - Generation Model​

Through technological iteration, Baidu Apollo Go has achieved a substantial reduction in hardware costs. The BOM (Bill of Materials) cost of the sixth - generation model has dropped to CNY 205,000 yuan, a 60% decrease compared with the fifth-generation model. Its core composition is as follows:

A) Perception system (approximately 45% of BOM cost) ​

LiDAR: Domestic LiDAR has realized domestic substitution. By June 2025, the unit price had dropped to over CNY 1,000 yuan, a year-on-year decrease of over 50% compared to 2024. The sixth-generation vehicle utilizes four 128-line LiDAR units (approximately CNY 3,000 yuan each), enabling ultra-long-range detection up to 200 meters and 360° dead - angle - free perception.

Vision and Radar Combination: It includes 12 high-definition cameras (approximately CNY 2,500 yuan per camera), 12 millimeter-wave radars (approximately CNY 600–700 yuan each), and 12 ultrasonic radars (totaling approximately CNY 32,000 yuan). Combined with the third - generation domain controller, it realizes multi - modal data fusion.

B) Computing and Control System (approximately 25% of BOM cost) ​

Dual AI Multi-Core Chip: With a computing power of 1,200 TOPS (costing about CNY 28,000 yuan), it supports the end-to-end decision-making of the ADFM large model. ​

Wire-Controlled Chassis and Actuators:  It adopts a customized chassis from JMC (costing about 23,000 yuan), and the service life of the braking system is 2.3 times longer than that of the previous generation.

C) Vehicle Body and Basic Configuration (approximately 30% of BOM cost) ​

Customized Electric Vehicle Body: Lightweight design (approximately CNY 35,000 yuan), equipped with massage seats and other passenger features. ​

Redundant Systems: Dual redundant configurations for power supply, steering, etc. (approximately CNY 7,000 yuan), meeting ASIL D functional safety certification. ​

​

2. Business Model: The Path from Scale to Profitability ​

The core variables of Unit Economics include vehicle costs (depreciation + maintenance + electricity), labor costs (safety operator), and revenue (order volume, average transaction price, discounts, mileage, etc.).

As the costs of Robotaxi model decrease, the number of vehicles supervised per safety operator increases, and daily order volume and average price increase, Robotaxi is expected to achieve profitability. Based on the following assumptions, we have estimated Robotaxi profitability under pessimistic, neutral, and optimistic scenarios for 2025:

1) Revenue: As Robotaxi technology matures, the unit price and order volume are expected to increase. Assuming optimistic, neutral, and pessimistic scenarios, the average daily order volume for Robotaxi is 26, 22 and 18, respectively, and the price per order is 20, 16 and 12 yuan, respectively.

2) Cost: The introduction of new-generation Robotaxi models is expected to reduce the average vehicle unit price. As the proportion of these new models increases, it is assumed that under optimistic, neutral, and pessimistic conditions, the average price of Robotaxi models will be about 205,000 yuan, 287,000 yuan, and 480,000 yuan respectively. Furthermore, with technological advancements, the required number of safety operators per vehicle should decrease. We assume one safety operator can supervise 10, 3, and 2 Robotaxis under optimistic, neutral, and pessimistic scenarios, respectively. In terms of daily operating expenses, we estimate the average maintenance cost per vehicle to be 667 yuan per year, tire costs to be 2400 yuan per year, and brake pad costs to be 800 yuan per year, under optimistic, neutral, and pessimistic scenarios, which is equivalent to 10.6 yuan per day.

In summary, in optimistic, neutral, and pessimistic scenarios, the average daily total revenue per Robotaxi in 2025 will be CNY 520, 352, and 216 yuan respectively, with net profits of CNY 304.6, 32.9, and -249.3 respectively.

Table. Profitability Calculation of Robotaxi in 2025 under Optimistic, Neutral, and Pessimistic Assumptions

3. Current Operation Scale: The World's Largest Autonomous Driving Mobility Network​

City Coverage: By 25Q1, Apollo Go had covered 15 cities, deployed more than 1,000 vehicles and accumulated more than 11 million orders (with 1.4 million orders in the single quarter). By 25Q2, it had expanded to 16 cities, received 2.2 million orders in the quarter, and accumulated over 14 million orders, positioning among the global leaders. Wuhan has the world's largest autonomous driving service area, covering 3,000 square kilometers with a population of 7.7 million, with the longest open mileage in China.

Vehicle Deployment: Apollo Go has a global fleet of more than 2,000 vehicles, including about 600 vehicles deployed in Wuhan alone. It plans to increase to 1,000 vehicles in Wuhan by the end of 2025.

Fleet Composition: At present, the sixth - generation JMC customized models account for 60% of the fleet, the fifth - generation Arcfox models account for 40%. The fully unmanned operating vehicles account for 75% of the fleet.

Operation Data: Apollo Go’s safe driving mileage exceeds 200 million kilometers, and the accident rate is only 1/14 of that of human drivers.

4. Future Operation Plan: Parallel Domestic Consolidation and Global Expansion

Domestically, Apollo Go is transitioning from pilot programs to large - scale nationwide implementation. Baidu plans to cover 65 cities in China by 2025 and expand to 100 cities by 2030, fully accelerating the urban expansion of Robotaxis. Currently, it operates in first-tier cities such as Beijing, Shanghai, Guangzhou, and Shenzhen, as well as other large cities, and conducts fully autonomous driving mobility service tests in Wuhan, Beijing, Chongqing, Shenzhen, Shanghai.

Overseas, according to the Baidu’s 2Q25 earnings call, as of June 2025, Apollo Go's business had covered 16 cities globally. In July, Apollo Go announced a strategic cooperative partnership with Uber, with plans to integrate thousands of autonomous vehicles into Uber's global mobility network. In August, Baidu announced that Apollo Go had reached a strategic collaboration with the U.S. ride-hailing platform Lyft, and would enter UK and German via Lyft in 2026, gradually expanding the scale to thousands of vehicles in the European market to achieve large-scale deployment of autonomous driving. Previously, during 1Q25, Apollo Go achieved notable progress in international expansion by entering Dubai and Abu Dhabi, with open road validation tests commencing in Dubai in May.

Conclusion:

Apollo Go has established a closed loop of “low-cost vehicles (RT6 approximately CNY 204,600 yuan) + multi-sensor fusion+ cloud-based remote assistance to reduce labor costs.” The most significant BOM cost reductions per vehicle are attributed to LiDAR and domain controllers. Unit Economics, excluding depreciation on the operational side, are approaching breakeven. With further reductions in safety operator costs and vehicle costs, positive Unit Economics (UE) are anticipated around 2028. Currently, the global coverage has expanded from 15 cities in 25Q1 2025 to 16 cities in 25Q2, with a fleet size exceeding 1,000 vehicles, quarterly orders of 2.2 million, and cumulative orders of 14 million+. Apollo Go’s future overseas expansion strategy deserves attention.