The race in physical AI has entered a new phase. On June 25th, the winners of the "2026 AI Breakthrough Awards" were announced. WeRide Inc. (NASDAQ: WRD, HKEX: 00800) saw its self-developed world model, WeRide GENESIS, honored as the "Overall Gen-AI Solution of the Year." This recognition places it alongside globally renowned tech firms like AMD, Qualcomm, and Dell, marking significant international industry validation for its generative AI capabilities.
This award follows closely on the heels of another major achievement. Just a day prior, on June 24th, WeRide GENESIS was awarded the "Simulation Innovation Award" at the 2026 Automotive Testing Technology International (ATTI) Awards. This consecutive recognition from authoritative bodies further underscores the technical value of GENESIS as a foundational model for physical AI.
In recent years, generative AI has reshaped how information is produced in the digital realm. However, a far greater industrial opportunity lies within the real, physical world. For AI to operate effectively in these real-world scenarios, it must comprehend physical laws, causal relationships, and spatiotemporal changes, and then execute actions that are safe, stable, and reliable. Consequently, physical AI has emerged as the central challenge for the next generation of artificial intelligence development. This dynamic signifies that the entity capable of constructing the underlying infrastructure connecting the digital and physical worlds will be poised to define the coming era.
Among all sectors applying physical AI, autonomous driving stands out as the first to achieve large-scale commercial validation and is currently the only scenario that simultaneously realizes a large-scale, real-world data feedback loop and sustainable, revenue-generating operations. As an autonomous driving company with one of the broadest global operational footprints, WeRide is positioning itself as a builder of this physical AI foundation. Its world model, WeRide GENESIS, serves as a bridge between the real physical world and virtual simulation environments, accelerating the large-scale development, training, and commercial deployment of autonomous driving technology.
The capabilities of the WeRide GENESIS physical AI model can be understood across four distinct layers. The first layer involves reconstructing the physical world, enabling the pixel-level generation of highly realistic road scenes within minutes to accurately replicate rare, long-tail scenarios difficult to capture frequently in reality. The second layer focuses on simulating physical traffic interactions, learning the underlying physical principles behind complex situations like pedestrian movement, slippery road surfaces, occlusion risks, and collision boundaries. The third layer is dedicated to reasoning about causal logic, allowing the AI to understand the chain reactions linking events such as a leading vehicle's sudden braking, a pedestrian jaywalking, a neighboring vehicle yielding, and the host vehicle slowing down. The fourth layer involves previewing spatiotemporal evolution, predicting multiple possible outcomes over the next few seconds within continuously changing road environments to help the autonomous system select safer, smoother, and more rational driving strategies.
Leveraging these capabilities, WeRide GENESIS establishes a logical closed loop from perceiving the world to understanding it. This allows autonomous driving systems to undergo training, verification, and iteration within massive, high-fidelity virtual environments. The process compresses what would be millions of kilometers of road testing into just a few days, while reducing data collection and labeling costs by over 75%.
The ultimate value of a world model must be validated in the real physical world, and Level 4 fully driverless operation represents the most stringent and direct proving ground for physical AI capabilities. Powered by the GENESIS engine, WeRide is driving its global expansion. Its autonomous driving products are now operational in over 40 cities across 12 countries, successfully completing commercial validation across diverse cities, nations, and road environments.
Currently, WeRide's Robotaxi service is conducting fully driverless commercial operations in cities like Guangzhou and Beijing in China, as well as Abu Dhabi and Dubai in the UAE. It is offering public services in Singapore and Riyadh, with plans to launch fully driverless public operations soon in locations including Madrid, Spain, and Zurich, Switzerland.
Simultaneously, the world model capabilities of GENESIS are gaining traction in the L2++ mass-production advanced driver-assistance systems (ADAS) market. At the second China Intelligent Driving Competition, the Chery Exeed Sterra ES model, equipped with WeRide's one-stage, end-to-end ADAS solution WRD 3.0, achieved an unprecedented and singular "six consecutive championships" in the event's history. Across urban roads, highways, and parking scenarios, WRD 3.0 consistently demonstrated stable and coherent assisted driving capabilities. It outperformed competitors with significantly larger R&D budgets, fundamentally rewriting the traditional competitive playbook of "throwing more people, more vehicles, and more rules" at the problem. This achievement represents a "ChatGPT 3.5 moment" for the autonomous driving field.
To date, WRD 3.0 has secured design wins for over 30 vehicle models from OEMs including Chery and GAC. As these mass-production collaborations advance, the world model capabilities of WeRide GENESIS will rapidly translate into enhanced product competitiveness for the passenger vehicle market.
The universality of physical laws means a world model adept at understanding urban traffic possesses a foundational ability to extend into numerous other physical interaction scenarios. As more vehicles, cities, and scenarios connect to this unified model base, WeRide GENESIS is poised to continuously amplify data value and boost R&D efficiency. A positive feedback loop—from data accumulation and model training to simulation verification and commercial deployment—has now been established.
Looking ahead, WeRide will continue to utilize WeRide GENESIS as its core physical AI model foundation. The company aims to propel AI beyond merely understanding the physical world towards acting upon it at scale, thereby accelerating the global commercial adoption of autonomous driving technology.
Comments