In-depth analysis: Baidu Apollo's pure vision path

 

Analysis of Baidu Apollo's pure vision strategy

Strategic CoreBusiness ValueMarket competitionrisk assessment

A calculated leap

Baidu's Robotaxi business is shifting from multi-sensor fusion to pure vision solutions. This is not a simple technological change, but a strategic choice related to cost, scale and market outcome, aimed at winning the race against time.

"If we can't quickly capture the market and hone our technology, we may not have a chance when Tesla's pure vision approach matures."

— Robin Li

Strategic Core

Why the shift: From “technical perfection” to “commercial practicality”

The core logic of this move is that the winner of Robotaxi will be the company that first achieves scale and runs a successful business model. Therefore, the technical route must serve the highest business goal of "rapidly achieving scale".

Comparison of technical routes

Old: Multi-sensor fusionHigh cost and slow iteration

Camera

LiDAR

Millimeter wave radar

✓ Strong sensor redundancy, theoretically safer
✗ High hardware and maintenance costs
✗ Complex system and slow algorithm iteration

New: Purely visual solutionLow cost and fast expansion

HD camera

(small amount) millimeter wave

✓ The hardware cost is greatly reduced, which is conducive to scalability
✓ System homogeneity and fast algorithm iteration
✗ It has extremely high requirements for AI models and extreme scene processing

🎯 Strategic goal shift

From pursuing absolute technological leadership in a single vehicle, we have shifted to pursuing rapid expansion of the entire business network and data closed-loop efficiency.

💪 Technical reserve support

Baidu is not starting from scratch. It has accumulated a mature pure vision technology stack (BEV+Transformer) in the L2+ assisted driving business, and has used L4 high-quality data to feed back.

⏳ Overcoming the “scalability gap”

The bottleneck of multi-sensor solutions lies not in the technological ceiling, but in their high total cost of ownership (TCO) and slow iteration speed, which has become the lower limit of commercialization efficiency.

Business Value

The Economics of Vision: The Path to Profitability

The most direct impact of switching to pure vision is a significant reduction in costs, which is a critical first step in closing the Robotaxi business model loop, accelerating expansion, and ultimately becoming profitable.

The “triple jump” in vehicle costs

The pure vision solution is expected to reduce hardware costs by another 10-15% based on the sixth-generation model.

Profitability Roadmap: The Race Between Cost and Revenue

The model predicts that with continued cost optimization and revenue growth, gross profit is expected to turn positive in 2025.

11 million+

National cumulative orders

6 million+

Wuhan cumulative orders

97.1%

Five-star reviews from users

1000 vehicles

Wuhan team target by the end of 2024

Market competition

Reshaping the competitive landscape and strategic barriers

Turning to pure vision is not only an internal optimization, but also an adjustment of external competitive strategy, aimed at building new differentiated advantages and moats.

Advantages: "Dimensionality reduction" of cost and speed

• Cost structure advantage: pull the competition back from "technical parameter competition" to "commercial efficiency competition" and suppress high-cost competitors with low costs.
• Advantages of scale and speed: Lower vehicle costs and simplified systems support faster fleet deployment and urban expansion, seizing market windows.
• Technology stack synergy advantage: L2+ mass production data feeds back to the L4 model to form a unique data closed loop, which is not available for pure L4 manufacturers.

risk assessment

Challenges behind opportunities

The shift to pure vision is a high-risk, high-return strategic choice. Investors need to be aware of the challenges they face in terms of safety, R&D, and policy.

Visual blind spots: safety risks in extreme scenarios

+

Risk points: Severe weather such as heavy rain, dense fog, and sudden changes in lighting may affect camera performance and cause perception failure.

Response plan: 1) Limit the operational design domain (ODD) and suspend operations in severe weather; 2) Strengthen the robustness of AI models through massive special data; 3) Use "Vehicle-Infrastructure Collaboration (V2X)" as the most important systemic safety redundancy.

The cost of R&D: continuous capital investment

+

Risk point: Transferring system complexity from hardware to software requires continuous investment of huge R&D funds in AI large-scale model training, computing power, data platforms, etc., which may continue to affect the company's profits.

Contingency plan: Rely on cash flow from core advertising and AI cloud businesses to support long-term investment; the ultimate goal is to allow the revenue growth of the Robotaxi business to cover expenses and achieve self-sustaining revenue.

Policy threshold: regulatory acceptance

+

Risk point: There is uncertainty as to whether regulators will approve large-scale fully unmanned commercial operations of vehicles without lidar. If mandatory standards are introduced, it will be a disruptive risk.

Response plan: Actively communicate with regulators, demonstrate the reliability of the systematic safety solution of "smart cars + smart roads", deeply integrate vehicle safety with national-level smart transportation infrastructure, and strive for policy support.

Conclusion: A high-risk, high-reward strategic bet

Baidu is betting that the speed of AI algorithm evolution can make up for the lack of hardware sensors, and that scale and low cost can overcome technical perfectionism. If successful, Baidu Apollo Go is not only expected to become the world's first large-scale profitable Robotaxi business, but is also likely to reshape the entire smart travel industry.