Xpeng has created a dedicated robotaxi business unit as it steps up its work on autonomous driving.

The newly established division will sit as a tier-one organisation within the Chinese electric vehicle maker.

Discover B2B Marketing That Performs

Combine business intelligence and editorial excellence to reach engaged professionals across 36 leading media platforms.

Find out more

It will be responsible for coordinating product definition, project integration, R&D testing and day-to-day operations.

The unit will run with a lightweight structure, drawing on shared resources from across the company’s platform-based ecosystem.

The company is aiming to start real-world passenger trials in the second half of 2026.

Speaking at Xpeng’s Q4 and full-year 2025 earnings call, chairman and chief executive He Xiaopeng said the business plans to begin passenger-carrying demonstration operations for its robotaxis in the second half of this year to assess the technology, user experience and commercial model.

Xpeng expects its robotaxis to be able to operate without safety operators next year, He Xiaopeng  added.

The company has also been reorganising to support what it describes as a shift from L2 to L4 autonomous driving and to advance its global approach.

Over the past year, Xpeng said it has adjusted its organisational structure, reworking its autonomous driving technology architecture, R&D approach and internal set-up, and moving to a framework led mainly by AI and software.

In February, Xpeng combined its autonomous driving centre and smart cockpit centre, forming a general intelligence centre.

Earlier this year, the company released factory test footage of robotaxi vehicles using its second-generation VLA (Vision-Language-Action) system. Xpeng said the vehicles have cleared third-party closed-track tests and are now taking part in regular Level 4 autonomous driving trials on public roads.

According to the company, its robotaxi vehicles are built for L4 autonomy and include dual hardware redundancy for safety.

They use four Turing AI chips, which Xpeng said provide up to 3,000 TOPS of computing power, and run on the second-generation VLA system.

The vehicles are also set to include external interaction features designed to show pedestrians the car’s intended manoeuvres—such as turning or looking for parking—via displays on the front windscreen.