WeRide Completes Cross-Continental Road Tests for L2++ Autonomous Driving Across Germany, France, and Japan

On July 14, WeRide systematically disclosed for the first time the latest progress of its global road validation for the WRD 3.0 L2++ production-level assisted driving solution. The company's end-to-end core algorithm has completed public road testing and validation in Germany, France, and Japan, covering three fundamentally different driving environments.

Germany's unlimited-speed autobahn presents extreme demands on AI real-time decision-making, where high-speed overtaking and rapid lane changes are common, requiring the system to complete each decision in fractions of a second. France's complex multi-exit roundabouts test the system's ability to continuously judge traffic participant behavior — when to enter, when to wait, when to yield — scenarios with no textbook answers. Japan's right-hand-drive, left-hand-traffic system demands adaptation to an entirely different set of driving rules, from driver positioning to intersection organization to road signage.

Crucially, WeRide did not develop separate versions for different countries. Instead, it used the same technical architecture across R&D, testing, validation, and commercial deployment. The company views the surface-level differences as fundamentally different data distributions, arguing that mature L2++ should not rely on having 'seen' a scenario before, but rather on understanding the world well enough to handle novel situations.

WeRide's L2++ capabilities are built on years of L4 autonomous driving technology accumulation, including its end-to-end architecture, the WeRide GENESIS world model, and multi-compute-platform adaptability. The company has formed a virtuous cycle where L4 serves as the technical source for L2++, while L2++ mass production data feeds back to improve L4 iteration. In Q1 2026, WeRide reported 114 million RMB in revenue, up 58% year-over-year, with a 35% gross margin. Over the past year, it has won six consecutive rounds of China's Smart Driving Competition and secured production orders for over 30 vehicle models.

The global road validation reflects a shift in the competitive logic of autonomous driving. As more players achieve similar feature sets, the genuine differentiator is no longer 'who can build it' but 'who can perform reliably in unfamiliar environments' — the generalization capability of AI. By driving the same model across three distinct 'driving worlds,' WeRide is answering a fundamental question: when AI encounters a completely unfamiliar environment for the first time, is it merely repeating learned rules, or has it developed the ability to truly understand the world?