XPENG, a prominent player in China’s electric vehicle and autonomous driving landscape, has officially commenced internal employee testing of its advanced robotaxi platform, as reported by Gasgoo on July 10, 2026. This strategic move underscores the company’s commitment to accelerating the development and deployment of its self-driving technology. The initiative saw CEO He Xiaopeng personally undertake the inaugural end-to-end journey as the first internal passenger, signaling the company’s confidence in its proprietary AI-driven solutions.
This development positions XPENG not merely as an automotive manufacturer but as a comprehensive technology provider and ecosystem enabler in the burgeoning field of autonomous mobility. The company’s approach diverges from some industry peers by focusing on a scalable, less hardware-intensive architecture to achieve broader, faster market penetration. The internal testing phase is crucial for refining the system’s performance, safety protocols, and user experience before potential public rollout.
Proprietary AI: The Core of XPENG’s Robotaxi Ambition
At the heart of XPENG’s robotaxi platform lies its self-developed URIng AI chip, a testament to the company’s deep investment in in-house technological capabilities. This chip powers the sophisticated VLA2.0 AI model, which has now entered formal mass production. The VLA2.0 model represents a significant leap in autonomous driving intelligence, designed to process complex environmental data and make high-precision, reliable real-time driving decisions. The integration of a custom-designed chip with an advanced AI model provides XPENG with a distinct advantage, allowing for optimized performance and greater control over its autonomous driving stack.
The efficiency gains from the VLA2.0 training process are particularly noteworthy. XPENG reports a staggering 1,010% increase in training efficiency per GPU and an even more impressive 4,360% increase per job. Furthermore, GPU utilization has improved substantially, rising from 40% to 90%. These metrics highlight the effectiveness of XPENG’s AI development methodology, enabling faster iteration cycles and more robust model training, which are critical for the rapid evolution of autonomous driving systems. Such advancements are vital for handling the vast amounts of data needed to train AI models for diverse, unpredictable urban driving scenarios.
A Strategic Shift: Technology Provider, Not Fleet Operator
XPENG’s strategy for its robotaxi platform is clear: to serve as a technology provider and ecosystem enabler, rather than directly operating large-scale autonomous fleets. This business model allows XPENG to leverage its technological prowess without incurring the significant operational complexities and capital expenditures associated with managing a vast fleet of robotaxis. By offering its advanced autonomous driving solutions to other operators, XPENG aims to foster a broader ecosystem of self-driving services, potentially accelerating the adoption of autonomous mobility across various sectors. This approach could also mitigate regulatory hurdles and insurance complexities that often accompany direct fleet operations.
The platform’s design emphasizes rapid deployment, a critical factor for scaling autonomous services. Notably, XPENG’s system is engineered to function effectively without relying heavily on LIDAR-intensive architectures or high-definition (HD) maps. While LIDAR and HD maps offer certain advantages, their cost and maintenance requirements can hinder widespread deployment. By developing a system that is less dependent on these components, XPENG aims to create a more flexible and economically viable solution for autonomous driving, capable of operating in a wider range of environments with less preparatory infrastructure.
Trial Operations and Future Outlook
XPENG has ambitious plans for the remainder of 2026, intending to complete trial operations and establish regular demonstration services for its robotaxi platform. Guangzhou, a major metropolitan hub in China, has been designated as the model city for these initial deployments. This choice of city is strategic, offering a complex urban environment that will rigorously test the platform’s capabilities in real-world conditions. The insights gained from these trials will be instrumental in further refining the technology and preparing it for broader commercial applications.
The company’s commitment to innovation extends beyond robotaxis. XPENG is also making significant strides in humanoid robotics, with its IRON humanoid robot targeting mass production by the end of 2026. This parallel development underscores XPENG’s vision of becoming a comprehensive physical AI company, integrating advanced AI across various domains, from autonomous vehicles to intelligent robotics. This broader strategy positions XPENG at the forefront of the next generation of intelligent technologies, aiming to redefine mobility and automation.
The internal employee testing of the robotaxi platform represents a pivotal moment for XPENG, showcasing its dedication to advancing autonomous driving technology through proprietary innovation. The company’s strategic focus on being a technology provider, coupled with its impressive AI model efficiency gains and plans for rapid deployment, sets a clear path for its future in the competitive global AI and automotive markets. As trial operations commence in Guangzhou, the industry will closely watch XPENG’s progress as it transforms its ambitious vision into tangible, widespread autonomous mobility solutions. The successful integration of its URIng AI chip and VLA2.0 model, alongside its less hardware-dependent approach, could indeed set a new benchmark for the future of robotaxi development.