China is betting that the future of robotaxis depends less on what's inside the car and more on what's built into the city itself. The country's new automotive roadmap, released as part of its 15th Five-Year Plan, reveals a starkly different vision for autonomous vehicles than the one Waymo and Tesla are pursuing in the United States. Rather than relying primarily on standalone vehicle sensors and onboard computing power, Chinese cities will deploy connected roadside sensors and 5G vehicle-to-everything infrastructure, sharing the computing burden between the car and centralized municipal cloud systems.

This architectural choice matters because it signals how China plans to scale robotaxis after a sobering setback. In the spring of 2026, robotaxi permits were frozen for nearly three months following an outage involving Baidu's Apollo Go fleet in Wuhan. The freeze underscored the risks of rapid deployment without adequate safeguards, and the new roadmap reflects a more cautious, infrastructure-first approach.

What's the Difference Between China's Approach and Waymo's?

The contrast is fundamental. Waymo's robotaxis rely on a vehicle-centric philosophy: cameras, lidar sensors, and powerful onboard computers process the driving environment independently. The car makes decisions based on what it perceives and learns from its own data. China's model, by contrast, distributes that cognitive load across the city itself. Roadside sensors feed real-time data to cloud systems that help guide vehicles, reducing the need for expensive sensor stacks and complex onboard computing on every single car.

The infrastructure-first model also signals a shift in how autonomous driving software gets built. China's roadmap favors end-to-end artificial intelligence (AI) models designed to reduce reliance on costly high-definition mapping and elaborate sensor arrays over time. This could theoretically make robotaxis cheaper to manufacture and deploy at scale, though it transfers complexity and cost to cities rather than automakers.

How to Evaluate China's Autonomy Strategy

• Safety Requirements: The plan requires autonomous vehicles to achieve higher safety levels than human drivers, though it stops short of setting a specific numerical target for deployment itself, reflecting caution after the Wuhan outage.
• Infrastructure Investment: Cities will deploy connected roadside sensors and 5G vehicle-to-everything systems that share computing between vehicles and municipal cloud systems, shifting the cost model away from individual cars.
• Software Architecture: End-to-end AI models are favored to reduce reliance on expensive high-definition maps and complex sensor stacks, potentially lowering per-vehicle costs but requiring massive city-level infrastructure investment.

The timing of this roadmap is significant. China's new energy vehicle (NEV) market has already far outpaced government targets. The country's previous five-year plan aimed for 20 percent NEV market share by 2025, but China reached 54 percent last year. By August 2026, NEVs accounted for 65 percent of car sales in China. The new roadmap targets 70 percent NEV penetration by 2030, a figure that may prove conservative given current trajectory.

Yet beneath this growth story lies a crisis of overcapacity. Car sales in China declined 21 percent over the first eight months of 2026, and a brutal price war has eroded margins across the industry. The roadmap explicitly pushes for industry consolidation, discourages overinvestment, and strengthens antitrust enforcement to eliminate inefficient production. This context matters for robotaxis: the freeze on permits and the new infrastructure-first approach reflect not just technical caution but also a desire to prevent another wave of speculative investment in autonomous driving startups.

Will China's Model Actually Be Safer Than Vehicle-Only Autonomy?

That remains an open engineering question. The infrastructure-first approach could accelerate safe deployment by distributing decision-making across multiple systems and reducing the burden on individual vehicles. Alternatively, it could simply shift cost and complexity from the car to the city, creating new failure points and dependencies on municipal systems that may not be equally robust everywhere.

The divergence also reflects deeper differences in how China and the West approach technology deployment. China's roadmap emphasizes centralized coordination, municipal infrastructure, and end-to-end AI models that learn from data rather than explicit rules. The West's approach, embodied by Waymo and Tesla, emphasizes vehicle autonomy, distributed decision-making, and detailed sensor fusion. Neither approach has proven definitively superior yet, but the stakes are enormous. Whichever model scales first and most safely could reshape the entire robotaxi industry globally.