XPeng Pushes Robotaxi Plans From Demonstration to Employee Trials

XPeng has started employee testing of its Robotaxi platform, according to the company, eight months after the initiative was first unveiled at XPENG AI Day 2025. The test, the company says, demonstrated a connected service chain running from ride-hailing and autonomous dispatch through to passenger transport and journey completion. The platform is powered by XPeng's self-developed Turing AI chip and its VLA2.0 model, running on infrastructure the company describes as proprietary, and is presented as one of its most complete Physical AI applications to date, combining intelligent-vehicle, autonomous-driving and embodied-AI work under a single system.

For automotive leaders, the detail that matters most is architectural rather than ceremonial. XPeng states that the Robotaxi platform shares its technology foundation with both its L2 driver-assistance and L4 autonomous systems, and that it is designed to avoid lidar-heavy sensor suites and high-definition maps. If that approach holds up operationally, it points to a deployment model built around software and compute reuse rather than bespoke, sensor-dense robotaxi hardware, which has implications for how quickly and cheaply such fleets could be replicated across cities with differing regulatory and mapping conditions.

The move also sits inside a broader pattern at XPeng, which has separately said its chief executive will lead a robotics unit targeting mass production of humanoid robots by the end of the year. Taken together with the Robotaxi programme, this suggests a company positioning itself across multiple Physical AI applications rather than treating autonomous ride-hailing as an isolated bet, a strategy that will invite comparison with domestic rivals such as BYD as competition in China's premium and technology-led segments intensifies.

What deserves close attention now is the path from employee testing to public operation. XPeng has said it intends to complete trial operations and establish regular demonstration services during 2026, using Guangzhou as a reference city intended to generate operational experience for replication elsewhere. The company has also confirmed it is exploring Robotaxi partnerships in Europe, the Middle East and Southeast Asia, though no agreements have been detailed. Automotive leaders should watch how the lidar-light, HD-map-free architecture performs once exposed to public passengers and unfamiliar regulatory regimes, since that transition, rather than the employee-testing milestone itself, will be the real test of the platform's claimed portability.