Reporter |
Editor | Wen Shuqi
Compared to its Robotaxi (autonomous ride-hailing) business, which has been expanding steadily in recent years, Pony.ai has kept a relatively low profile when it comes to Robotruck (autonomous freight trucks). Now, the company is ready to push its autonomous freight ambitions back into the spotlight.
At a media briefing on August 3, He Xing, head of Pony.ai’s truck division, announced that the fourth-generation autonomous heavy-duty truck will move into mass production in the second half of this year. Meanwhile, the first-generation autonomous light-duty truck, which debuted at the Beijing Auto Show in April, has already entered the pilot testing phase with partners. Over the next two to three years, Pony.ai plans to deploy 500 to 1,000 autonomous heavy trucks across three key scenarios: highway trunk routes, bulk commodity corridors in the northwest, and port logistics.
Robotruck has long been a significant revenue pillar for Pony.ai. According to its Q1 2026 earnings report, Robotruck service revenue hit $10.2 million, outpacing the $8.57 million generated by Robotaxi services. At one point, the trucking arm was actually the company’s top revenue generator.
However, when stacked against the rapid growth of Robotaxi, the Robotruck segment has clearly lagged behind. To unlock new growth, the company needs to scale up mass production and expand commercial operations.
Pony.ai’s renewed emphasis on Robotruck isn’t because the autonomous driving tech for trucks just recently matured.
He Xing told reporters that the single-vehicle autonomous capabilities for trucks had already reached a fairly high level two to three years ago. But back then, the cost of building a full L4-capable truck was still steep — “we were waiting for the right moment.”
The real game-changer came from Robotaxi: after achieving mass production at the thousand-unit scale, Pony.ai began adopting more automotive-grade sensors and solutions, and gained hands-on experience integrating autonomous systems directly into production vehicles. That know-how simultaneously brought down costs, streamlined the supply chain, and strengthened the overall vehicle capabilities needed to scale up Robotruck.
According to Pony.ai, the autonomous heavy trucks and Robotaxis now share nearly identical components — from sensors to computing platforms. Thanks to this synergy, the fourth-generation heavy truck has directly benefited from Robotaxi’s manufacturing scale, slashing the hardware cost of the autonomous driving kit by roughly 70% compared to the previous generation. By the company’s calculations, autonomous light trucks share over 90% synergy with Robotaxi in terms of autonomous tech, scaled operations, policy pathways, and supply chain, while heavy trucks share more than 80%.
For city logistics, the autonomous light trucks primarily leverage Robotaxi’s long-accumulated urban driving capabilities, sensor configurations, and fleet management systems. Heavy trucks, on the other hand, can share algorithms for overtaking, lane changing, and traffic negotiation, but still require custom tuning for vehicle control and operational scenarios unique to trucking.
Pony.ai has zeroed in on three priority use cases for the fourth-generation heavy truck. On the highway front, the company is continuing its partnership with Sinotrans and also running trials with China Post. For ports, the first batch of fourth-gen trucks will be deployed at Shenzhen’s Mawan Port, starting with a fleet of a few dozen units. Meanwhile, the company has launched a bulk commodity corridor in the Golmud region of the northwest.
The “Hub to Hub” model has emerged as the practical backbone for connecting this logistics chain. In this setup, autonomous light or heavy trucks handle the feeder routes — from warehouses to hubs and from hubs to retail outlets — while the long-haul trunk segments between hubs are primarily served by autonomous heavy trucks.
Building on this, Pony.ai has designed two business models for Robotruck. Under the TaaS (Trucking-as-a-Service) model, Pony.ai owns and operates the vehicles, providing transportation services directly to cargo owners. Under the ADaaS (Autonomous Driving-as-a-Service) model, logistics companies or asset partners own and operate the trucks, while Pony.ai supplies the factory-installed autonomous driving system and charges a technical service fee.
He Xing explained that the company plans to first validate the Hub-to-Hub routes through TaaS, and once safety and economics are proven in pilot lines, gradually transition toward the asset-lighter ADaaS model.
As previously reported, Pony.ai currently uses a “1+N” platooning formation for unmanned freight operations. In this setup, the lead truck has a safety driver on board, while the following trucks operate fully driverless — all equipped with L4-level autonomous capabilities. The number of trailing trucks (“N”) depends on real-world cargo scheduling and loading considerations.
But rolling off the production line doesn’t mean the trucks can immediately start generating revenue. Xiao Ping, product head of Pony.ai’s truck division, noted that even though the previous generation models already obtained platooning permits for driverless operations, the fourth-generation trucks must still go through the regulatory process from scratch — road testing, pilot operations, and commercial deployment. The first-gen light trucks are likewise still in the road-testing and data-collection phase, with plans to apply for driverless operation permits down the line.
On the other hand, Pony.ai’s light trucks are jumping straight into the urban roads where Robotaxi has been running for years, targeting city logistics like parcel sorting, supermarket restocking, and cold-chain food delivery.
This is getting to be a pretty crowded field. Low-speed autonomous delivery vehicles (Robovans) have already expanded from campus and last-mile delivery into broader urban distribution. Players like Ninebot (Juesi Intelligent) and Neolix are scaling up to ten-thousand-unit annual deliveries, while autonomous driving suppliers such as Momenta, LightYear (Qingzhou Zhihang), and Yoke Intelligence, along with major OEMs, are piling into the space.
That means autonomous light trucks are also facing potential competition from the already-scaled low-speed delivery vans. When asked whether the two vehicle types could replace each other, Xiao Ping pointed out that there are roughly 14 million commercial vehicles in China’s urban logistics fleet, of which close to 8 million are light trucks — making them the dominant workhorses in the current city delivery system. Pony’s light truck offers a cargo volume of about 18 cubic meters, whereas most low-speed delivery vans with road access rights are limited to 5 to 6 cubic meters. Plus, the former can operate on city roads and expressways at truck speed limits.
Beyond that, autonomous light trucks can help fill the driver shortage. He Xing noted that they can step in during off-peak hours like early mornings when reliable drivers are hard to find, boosting vehicle utilization and shaving transit times across the logistics network.
Competition in commercial-vehicle autonomous driving is now spilling over into capital markets. CiDi (Xidi Intelligent) listed on the Hong Kong Stock Exchange in December 2025. Since the start of 2026, DeepWay (Shenxiang Technology), Tage Idriver (Tage Zhixing), and Zero One Automotive have all filed for Hong Kong IPOs, while Trunk Technology re-submitted its application in June after its earlier filing lapsed. Some players focused on closed-scenario autonomous driving are also trying to steer their driverless trucks onto open highways.
As world models, factory-installed mass production, and platooning become the standard playbook for more and more players, Pony.ai needs to prove what sets it apart. He Xing acknowledged that when the fleet size is still small, “the moat of single-vehicle technology is pretty limited.”
Pony’s answer lies in the scaled operational capabilities honed through its Robotaxi business. He Xing said the company’s edge isn’t just about whether a single truck can drive itself — it’s about whether the vehicles and systems are ready for thousand-unit-scale driverless operations. That includes factory-installed mass production, closed-loop data systems, redundancy design, plus the remote assistance, safety operations, and fleet management experience accumulated from Robotaxi.
Robotaxi’s thousand-unit mass production and operations have indeed solved some of the toughest cost and engineering challenges for Pony’s Robotruck business. But trucks aren’t just a simple copy-paste of passenger-car autonomous tech. Policy hurdles, logistics efficiency, and other factors will ultimately determine whether this business can truly take off.
