IN Brief:
- Pony.ai expects 500–1,000 fourth-generation autonomous heavy trucks to enter service in China within two to three years.
- Initial operations cover long-haul freight, bulk commodities, and port logistics, including activity at Mawan Port in Shenzhen.
- Utilisation, remote support, maintenance, and operating costs will determine whether the vehicles progress beyond controlled deployments.
Pony.ai plans to deploy between 500 and 1,000 fourth-generation autonomous heavy trucks in China over the next two to three years.
The deployment will focus on long-haul freight, bulk commodity transport, and port logistics. Production of the company’s fourth-generation heavy trucks is under way, with batches expected to enter commercial service over the coming months.
Shenzhen’s Mawan Port is among the first locations identified for deployment, where dozens of vehicles are planned for commercial operation. Ports provide a relatively controlled environment with repetitive routes and defined loading and unloading points.
The battery-electric heavy truck supports single-vehicle autonomous operation and Level 4 platooning, depending on the operating environment. Pony.ai says the cost of its autonomous driving hardware has fallen by approximately 70% compared with the previous generation.
He Xing, vice-president of Pony.ai and head of its Robotruck business, said: “We had been waiting for the right moment.” The company argues that lower component costs, automotive-grade production, and stronger commercial partnerships have made larger deployments more practical.
Pony.ai entered autonomous trucking in 2018, when its early vehicles were largely hand-built prototypes. Subsequent generations have been developed with truck manufacturers as the programme has moved towards repeatable production.
By November 2025, the company operated approximately 200 trucks and said the fleet had transported more than one billion tonne-kilometres of freight. Robotruck services generated US$10.2 million in revenue during the first quarter of 2026, an increase of 31% from the previous year.
The operation remains small beside China’s road freight market, but it provides commercial data from working routes rather than closed-course testing alone. Fleet growth will increase the amount of evidence available on utilisation, interventions, maintenance, and cost per movement.
The fourth-generation truck has been designed for 20,000 operating hours or up to one million kilometres. Redundant systems cover steering, braking, communications, power, computing, and sensing, allowing the vehicle to move towards a safe stop if a critical system fails.
Heavy trucks create different technical demands from passenger vehicles. They have longer braking distances, different weight transfer, trailer behaviour, and less tolerance for control errors when operating at speed or around loading equipment.
Commercial deployment will also depend on the surrounding freight operation. Autonomous vehicles still require maintenance, charging, dispatch integration, remote assistance, insurance, regulatory approval, and enough predictable work to keep costly assets moving.
Pony.ai is working with vehicle manufacturers, including SANY Truck, and logistics groups such as Sinotrans. Its operating models range from transport services managed more directly by the technology company to arrangements where partners own and operate the fleet.
That structure allows logistics providers to retain control of customer and transport operations, but also divides responsibility between the vehicle manufacturer, autonomous system developer, and fleet operator. Contracts must make clear who handles software incidents, roadside failures, remote interventions, and maintenance decisions.
Ports, mines, and regular hub-to-hub corridors are likely to provide the clearest early applications because the operating environment can be more tightly defined. Similar reasoning supports autonomous yard truck deployments in private logistics facilities.
Long-haul and bulk operations introduce mixed traffic, changing weather, longer distances, and more variable road conditions. Freight may also need to transfer between autonomous trunk routes and human-driven first- and last-mile services.
Pony.ai is separately developing a Level 4 light commercial vehicle for urban and intercity distribution. Developed with CATL, the vehicle provides about 18m³ of cargo space and is being tested in express, retail, and cold chain operations.
The company has set a longer-term target of 100,000 autonomous light commercial vehicles by 2030. The light and heavy programmes can share elements of software, fleet support, and data infrastructure, although their commercial routes are likely to differ.
Urban vehicles may have access to a larger number of potential routes, while heavy trucks could scale first in locations where movements are repetitive and the operating area can be tightly controlled.
Customers and regulators will expect evidence covering safety, cyber risk, software updates, incident response, and mixed-fleet management. Vehicle production numbers alone will not show whether those systems work consistently in daily freight operations.
A fleet of up to 1,000 heavy trucks would move Pony.ai beyond pilot scale. The decisive measures will be utilisation, safety performance, remote-support demand, maintenance cost, and the number of revenue-generating movements completed without shifting new costs elsewhere in the operation.
