AI News Feed
Market watch
Companies

Zelos Launches City-Level Physical AI Strategy After 30,000 RoboVans

Zelos launched a city-level physical AI strategy in Guangzhou, adding a rental platform and unmanned capacity franchising to its 30,000-vehicle RoboVan fleet.

At the event, Zelos introduced the "Zelos Rental" mini-program for express logistics, commercial distribution, and urban short-haul transport. It also opened unmanned capacity franchising under Zelos Rental and announced exclusive preferential policies for the first group of franchise partners.

Zelos signed cooperation agreements with industrial partners including GAC Lingcheng, Guangzhou Public Transport Group, Guangzhou Environmental Investment Group, Guangzhou Ouweide, Shangyuan, and Zelos Jijing. The cooperation covers unmanned vehicle manufacturing, urban services, and industry-academia-research collaboration.

The move follows five years of development. The company started with RoboVan, a product form aimed at the large demand for box trucks in urban logistics, rather than passenger robotaxis. In 2023, Zelos launched what it described as the world's first RoboVan and pushed it into public road operations. Since then, it has iterated on vehicle products, autonomous driving systems, and scaled operations.

Safety has been a threshold for scaling, Zelos CEO Kong Qi said at the event. "Safety is not a problem to be solved after scaling. It is a capability that must be built from day one," he said. Zelos uses a data closed loop to feed real-road operating experience back into algorithm and system iteration. It also designs redundancy for vehicle control and communications, and runs thousands of tests on stability and reliability. This year, the company established the industry's first safety committee.

Zelos said its vehicles now operate routinely in more than 300 cities worldwide, with cumulative operating mileage exceeding 250 million kilometers. Its commercial model includes vehicle sales, financial installment services, and autonomous driving subscription services. The company said its RoboVan fleet has exceeded 30,000 vehicles, serving customers such as China Post, SF Express, the "three Tongs and one Yida" express companies, Mengniu, New Hope, and Zhongce Rubber, along with many small and medium-sized clients. It said its market share ranks first.

Still, Zelos identified demand in logistics that cannot be met by simply selling vehicles. In early 2025, a user in Kunshan identified as A Gang bought four Zelos unmanned vehicles and began renting them to clients with delivery needs. His fleet grew from four to dozens, and now to more than 200 vehicles. He also pooled orders and time to reorganize idle vehicle hours so the same vehicles could serve different customers at different times.

Zelos tested the rental model in 45 cities and 300 customer scenarios, completing more than 1,500 rental orders and running over 1 million kilometers. The company said many customers with fluctuating transport needs care less about buying a vehicle than about securing stable transport capacity. "What customers really care about may not be what kind of vehicle to buy, but whether they have stable transport capacity every day and whether the vehicle can work according to their needs," Kong said.

The Zelos Rental platform offers daily, monthly, and annual rental options. Zelos provides vehicles and platform capabilities such as vehicle management and intelligent dispatch, while local partners use their resources in logistics, commerce, and transport to deploy unmanned vehicles in express, supermarket, and commercial distribution scenarios. The company described the shift as a move from selling equipment to providing transport capacity.

Zelos has also positioned its "Zelos Brain" as an AI capability hub for the physical world, not just driving software. It is standardizing L4 capabilities validated in real operations and, through Zelos Inside, outputting autonomous driving systems, AI algorithms, software and hardware, and scaled operational experience to more automakers and industrial partners. The company has previously worked with Dongfeng and is now extending cooperation to more vehicle models, cities, and scenarios. It is building three city-level capabilities: a capacity management platform, an operations monitoring platform, and a government regulatory platform.

The company said the strategic upgrade reflects a shift from a product company to an infrastructure platform. In its view, physical AI must do more than complete a single autonomous drive; it must connect vehicles, transport capacity, industries, and cities, and form capabilities that can be replicated at scale in real environments.