Gasgoo Munich- XPENG recently held its first all-hands meeting on the Robotaxi business, announcing the unit has officially entered employee testing. Management made one thing clear: Robotaxi operations will serve as the core vehicle for the group's physical AI and "robotic vehicle" strategy.
Chairman He Xiaopeng offered a telling forecast at the gathering: over the next decade, every form of embodied intelligence will essentially become a robot. For XPENG, Robotaxi represents a critical leap from new-energy vehicles to "robotic vehicles" — and a major strategic move in the physical AI landscape.
With this positioning, XPENG is not focusing on building its own fleet or offering ride-hailing services directly to consumers. Instead, it's pivoting to become a global hardware and software supplier. The plan is to package the vehicle platform, autonomous driving algorithms, and backend dispatch systems into a turnkey product for overseas mobility firms or local partners, who will then handle operations and regulatory compliance.
This approach aligns with the industry's prevailing "golden triangle" model, which pairs smart-driving tech firms, automakers, and mobility platforms. TrendForce forecasts the Chinese Robotaxi market will hit $44.5 billion by 2035. As the Middle East emerges as a hotspot for global expansion, China's top players are accelerating their presence there — and XPENG's model is well-positioned to capture opportunities both at home and abroad.
The internal testing launched this week serves as a crucial technical validation step within that strategic framework. Leveraging controlled environments like corporate parks and public roads, XPENG will collect real-world driving data to continuously refine the algorithm's adaptability. Supporting systems — including vehicle dispatch, human-machine interaction, and emergency response — will also be stress-tested during these trials. The technical and operational insights gained here will lay a solid foundation for future pilot operations and the delivery of overseas solutions.
Robotaxi Elevated to Strategic Pivot
By positioning Robotaxi as a core strategic vehicle, XPENG is signaling that its ambitions extend far beyond a simple autonomous ride-hailing service.
Traditional automakers typically view autonomous driving as an add-on feature for passenger cars, keeping their business focused on vehicle sales. XPENG sees it differently. It defines Robotaxi as a scalable scenario for mobile embodied intelligence — a central hub in the group's physical AI expansion.
The business logic behind this distinction is straightforward: resources from vehicle platforms, autonomous driving algorithms, and general AI models will converge on Robotaxi. The multidimensional data generated during continuous operations — covering road conditions, interactions, and dispatching — will then flow back into the iterative loop of the group's general AI capabilities.
Take XPENG Motors as an example: its newly built cloud factory and data infrastructure have increased vehicle data upload capacity by 22 times and boosted multimodal data reading and decoding bandwidth by 15 times. Backed by a multi-layer computing architecture that includes a cloud computing platform, the company has quintupled model training speeds — effectively using operational data to accelerate AI evolution.
Image Source: XPENG Motors
In other words, operating Robotaxis isn't just about making the ride-hailing service work; it's about providing a continuous, high-density real-world training environment for AI models.
In terms of resource allocation, this strategy acts as a powerful anchor. By treating Robotaxi as the landing point for physical AI, XPENG is removing any ambiguity from its technological roadmap. All R&D investment, data feedback, and product definition will now advance along a single track focused on embodied intelligence, rather than being scattered across disconnected business units.
The commercialization path is equally clear: XPENG's primary focus is technology export. It delivers complete solutions — a trinity of vehicle, algorithm, and backend — rather than just providing ride capacity. The core competitive advantage here lies in leveraging its mature manufacturing capabilities and full-stack proprietary technology to standardize complex systems, then adapting them to fit specific market needs.
A Fork in the Road
The Robotaxi track is crowded, and development paths have clearly split into two directions. These divergent routes reflect trade-offs based on each company's unique resources — neither is inherently superior.
The first is the self-operated fleet model. Here, companies purchase their own vehicles, build on-the-ground operations teams, and secure operating permits city by city to offer Robotaxi services directly to consumers.
Baidu Apollo Go, Pony.ai, and WeRide have all adopted this route. The advantage is complete control over the end-user experience and the data loop — but the costs are obvious. Heavy investments in vehicle procurement, facility maintenance, and staffing continuously burn cash. Moreover, cross-border expansion is often hampered by significant differences in traffic regulations and licensing requirements, slowing the path to scale.
Image Source: XPENG Motors
The second is the hardware and software solution export model — the path XPENG has chosen. Leveraging its mass-production capabilities and proprietary autonomous driving tech, XPENG packages its platform, algorithms, and dispatch systems for partners, who then handle operations. This approach avoids the burden of heavy fleet assets and the need to navigate regional licensing barriers individually. By adapting standardized solutions to different markets, the company can achieve a lighter, faster global expansion.
This choice aligns closely with XPENG's core strengths. The combination of proprietary chips, VLA world models, and vehicle manufacturing gives it the foundation to modularize and standardize complex systems. A technology export model maximizes the economies of scale from vertical integration, rather than wasting resources on fragmented operational tasks.
Yet the long-term challenges on this path are just as real. Road signs, traffic laws, and mobility habits vary significantly across countries, requiring localized adaptation of the entire solution for every target market. Furthermore, overseas partners have varying requirements for safety standards, maintenance costs, and system stability. The precision and efficiency of these adaptations will directly dictate the pace of global expansion.
The strategic value of the real-world road data gathered during XPENG's employee testing lies precisely here — improving the algorithm's generalization across different scenarios and providing a proven basis for future overseas deployments.
Clearly, the signal from XPENG's all-hands meeting extends beyond a single business unit. In its strategic blueprint, Robotaxi serves as both the landing ground for physical AI and the vehicle for technology exports. While most industry players are still trying to balance fleet building with algorithm development, XPENG has chosen a differentiated path: handing operations to partners while focusing its own efforts on standardizing and exporting solutions.
Yet the effectiveness of this path remains to be proven by the market and practice. Key milestones to watch include the actual progress of algorithm iteration, whether overseas partners materialize on schedule, and how well standardized solutions adapt to different markets.
