XPENG Unveils VLA 2.0, Targets 2027 Global Deployment with Volkswagen

XPENG announced its VLA 2.0 intelligent driving system, confirming 2027 global delivery plans and naming Volkswagen as its first launch partner in China. Learn more.

XPENG has outlined the next phase in the evolution of its intelligent driving technology, confirming that global deliveries of its VLA 2.0 system are scheduled for 2027, while public road testing of Robotaxi vehicles equipped with the new architecture has already begun in China. Trial operations are expected to start later in 2026.

The announcement was made during the company’s VLA Media Experience Day, where XPENG presented both the technical structure of the system and its global deployment roadmap. In the Chinese market, Volkswagen has been named as the first launch partner for VLA 2.0 integration.

At the core of VLA 2.0 is a unified AI foundation model that integrates perception, reasoning and action within a single end-to-end vision-to-action architecture. Unlike traditional modular autonomous driving stacks, which translate visual data into intermediate language-based reasoning before generating commands, XPENG’s system removes this step, aiming to improve response speed and processing efficiency.

According to the company, early testing shows stable performance in dense urban traffic and mixed road environments. In measured tests conducted during evening rush hour in Guangzhou, traffic efficiency improved by 23% compared with traditional L2 intelligent driving systems and existing Robotaxi models. Performance levels were described as comparable to experienced human drivers.

VLA 2.0 remains a supervised intelligent driving system, and further commercialization of higher automation levels will depend on regulatory frameworks. XPENG did not disclose a fixed timeline for fully autonomous commercial deployment, although CEO He Xiaopeng stated that he believes full autonomy could arrive within one to three years.

The system is powered by XPENG’s proprietary Turing AI chip, delivering up to 2,250 TOPS of computing power per chip. The Robotaxi configuration deploys up to four such processors, enabling total onboard computing capacity of up to 3,000 TOPS. The company says its optimization of the chip–operator–model pipeline allows large-scale AI models with billions of parameters to run in production vehicles.

The Robotaxi rollout comes as China’s autonomous mobility sector continues to expand, with projects such as Baidu’s Apollo Go and Pony.ai already operating in multiple cities. XPENG plans to introduce three Robotaxi models in 2026 and has named Amap as its first global ecosystem partner.

Beyond passenger vehicles, XPENG positions VLA 2.0 within a broader Physical AI strategy. The underlying architecture is designed to extend across Robotaxi fleets, humanoid robots and modular flying vehicle platforms, reflecting the company’s ambition to build a unified embodied AI foundation.

The technological cooperation with Volkswagen adds strategic weight to the initiative. The German automaker previously acquired a 4.99% stake in XPENG and entered into agreements to jointly develop electric vehicles for China. The partnership now expands into next-generation intelligent driving systems.

If deployment timelines hold, 2027 could mark a significant milestone for XPENG in the global race for advanced intelligent driving. However, the pace of higher-level automation will continue to be shaped not only by technical progress but also by evolving regulatory environments worldwide.