Gasgoo Munich- Commercial vehicles—once tipped to be the first to commercialize autonomous driving—are finally seeing a tech influx from the passenger car sector.
On October 29, Zhuoyu ("ZYT") officially launched its commercial vehicle brand, ZYT Prax Drive System ("Zhuoyu Panxing"), detailing partnerships with seven leading industry players. Hundreds of heavy trucks equipped with its driver-assistance systems are already in service on express delivery routes.
That same day, the first mass-production models of the T9 ROBOTRUCK—an L4 autonomous heavy truck jointly developed by GAC New Energy Commercial Vehicle Co., Ltd. and Pony.ai—rolled off the assembly line.
One player is pivoting from passenger car driver assistance to heavy trucks; the other is applying its tech stack from robotaxis and autonomous trucking to L4 vehicle development. Behind these divergent paths lies a singular shift: the race for commercial vehicle intelligence is attracting contenders from entirely different backgrounds.
Yet, the market these passenger car veterans face is hardly an untapped frontier.
Commercial AV Players Are Already on the Road
Around 2017, the autonomous driving industry pinned high hopes on long-haul highway logistics.
Highway conditions are relatively standardized, freight routes are fixed, and logistics companies are acutely sensitive to labor costs and efficiency. Compared to city streets, heavy trucks were once seen as the most likely candidates for early high-level autonomous driving commercialization.
FAW Jiefang outlined a "three-step" autonomous strategy back then: complete L3 development by 2018, L4 by 2020, and L5 by 2025.
Reality has proven far more complex than those early visions.
Variable loads, braking distances, and trailer dynamics mean autonomous technology can't simply be copied from passenger cars. Moving to L4 introduces further hurdles: vehicle redundancy, safety validation, regulatory approval, and operational logistics.
The long-haul unmanned freight revolution has yet to arrive at scale.
What has scaled first is driver-assistance systems that still require human supervision.
In September, Inceptio Technology announced that its commercial driver-assistance operations had surpassed 1 billion kilometers.
What does that figure actually mean? STO Express's operations offer a clearer view.
Image Source: Inceptio Technology
As of September, STO's fleet running Inceptio's system had logged over 150 million kilometers. Inceptio reports that driver assistance was used for more than 90% of highway driving, cutting energy consumption by 3% to 7% per 100 kilometers.
For drivers grinding out miles on the highway every day, driver assistance is no longer just an add-on feature—it has become an essential tool of the trade.
For logistics firms, 150 million kilometers signals that the system has endured real-world conditions, amassing critical data on vehicle usage, energy management, and operational reliability.
Inceptio also claims it holds over 90% of the domestic truck autonomous driving market. But beyond that self-reported figure, the deep customer relationships and engineering expertise forged through long-term operations tell a more compelling story about its competitive footing.
Another veteran, PlusAI, has charted a different course.
PlusAI previously partnered with FAW Jiefang and Rokin Logistics to push smart truck production, and has delivered more than 100 driver-assisted heavy trucks to SF Express in 2025.
But its shifts haven't just been operational—they've been capital-driven too.
In 2023, PlusAI split its domestic and overseas operations. By July 2025, Full Truck Alliance had acquired an 80.8% stake in PlusAI's Chinese unit for roughly 1.376 billion yuan, while the overseas business continues on an independent path.
That means PlusAI China, once an independent tech developer, is now folded into a new equity and resource structure.
While Inceptio expands its operational footprint, PlusAI undergoes structural and capital adjustments. Commercial autonomous driving firms have clearly diverged in their paths over the past few years.
Crucially, neither has stopped expanding its reach.
Inceptio unveiled a "Physical AI for Freight" strategy this year, aiming to extend its capabilities from highways to regional routes, urban distribution, and last-mile logistics.
As passenger car players enter the commercial arena, these freight-focused incumbents already hold the cards: real-world operational data, deep logistics relationships, and battle-tested product lines.
Newcomers aren't starting from scratch; they're fighting to find their footing in a market where incumbents have already achieved scale.
How Much Can Passenger Car Experience Shorten the Path?
Zhuoyu's answer: make passenger and commercial vehicles share the same underlying technology stack.
At the Panxing launch, Zhuoyu CEO Shen Shaojie revealed the company now serves 22 passenger car customers, with over 60 mass-produced models and more than 110 designated models in the pipeline.
Those numbers aren't just about scale; they signify that Zhuoyu has already built a comprehensive ecosystem for passenger cars—covering model R&D, data processing, hardware and software development, and mass production delivery.
Now, it aims to deploy that ecosystem for heavy trucks, buses, and light commercial vehicles.
Liu Sikang, Zhuoyu's VP for commercial vehicles, breaks the technical path down into three stages: unified foundation, scenario enhancement, and engineering adaptation.
In short: take the passenger car base model, layer in truck-specific data, tweak sensor layouts and control strategies, and adapt the same tech stack to different vehicle types.
According to Zhuoyu, Panxing inherits over 3 billion kilometers of passenger car data while incorporating 400 million frames of truck-specific data for training.
The promise of this approach is avoiding the need to build a separate tech ecosystem for every vehicle class.
The real challenge is determining which capabilities can be shared—and which must be redeveloped from scratch for commercial applications.
Acceleration and braking response vary drastically between a loaded and an empty truck. When a tractor-trailer over a dozen meters long negotiates a turn, the system must account for the inner wheel difference, trailer swing, and rollover risk. Longer braking chains also demand earlier decision-making.
To address this, Zhuoyu developed the Jimu 2.0 perception system, with specific adjustments for load estimation, trailer attitude recognition, grade prediction, and braking distribution.
In terms of product roadmap, the Panxing PDS 4 focuses on highway NOA for heavy trucks and is already in operation. The PDS 5 plans to leverage native multimodal foundation models to extend capabilities to city NOA for buses and light trucks.
The real story here isn't whether Zhuoyu can get highway NOA working on a truck.
Inceptio and PlusAI have already proven that this function works in real-world freight.
What Zhuoyu must prove is whether its passenger car legacy allows for shorter development cycles and lower redundant investment, enabling mass production across multiple commercial vehicle models.
This is a clash of two different philosophies on technology accumulation.
Long-time freight specialists possess data and engineering experience tailored to heavy-duty conditions, and they better understand how logistics companies buy, use, and manage fleets.
Zhuoyu's opportunity lies in the model training, hardware and software R&D, and production engineering resources already sunk into the passenger car market. If those capabilities can be reused, development costs for new commercial models could be significantly amortized.
Yet, the extent of that reuse depends on actual investment. Truck-specific data, dynamics adaptation, vehicle validation, and ongoing maintenance won't simply vanish because the base model is shared.
Incumbent commercial vehicle players are also pursuing platformization and cost reduction through scale; they aren't stuck developing a unique tech stack for every single model either.
So, Zhuoyu doesn't offer a guaranteed low-cost advantage so much as a technical pathway that could accelerate model coverage.
Partnerships with seven commercial brands provide the proving ground for this approach.
Pony.ai, which also debuted that day, represents a different approach.
Pony.ai isn't new to the heavy truck space. It has long pursued a dual strategy in robotaxis and autonomous trucking, pouring resources into algorithms, system safety, and vehicle engineering.
Image Source: GAC Group
The T9 ROBOTRUCK, produced in partnership with GAC New Energy Commercial Vehicle Co., Ltd., integrates Pony.ai's fourth-generation L4 system with GAC's fully redundant steer-by-wire chassis, incorporating redundancy in steering, braking, power, and communications.
Unlike Zhuoyu's focus on cross-model reuse, Pony.ai prioritizes deep integration of the L4 system with the vehicle platform itself, laying the groundwork for mass production in unmanned transport.
One offers highway NOA to existing logistics networks; the other pushes forward with L4 autonomous freight. They operate at different commercialization stages and aren't direct competitors in the same product tier.
Zoom out, and the shift is clear: mass-production experience from passenger car driver assistance, autonomous tech from robotaxis, and the freight engineering expertise of commercial incumbents are all converging on the same market.
Commercial vehicle OEMs now have more options for technical collaboration, while incumbent suppliers face fresh competitive pressure.
But the outcome of this competition won't be decided by product launches alone.
Seven Partnerships: How Far From a Market Shakeout?
Zhuoyu's seven partners include XCMG, Shacman, Sinotruk, FAW Jiefang, Foton, Dongfeng Commercial Vehicle Co., Ltd., and Yutong.
The list is impressive, but the only project with significant operational scale so far is the Shacman Delong X5000.
Mass production began in June, and hundreds of these trucks are already in service on express delivery routes.
Zhuoyu projects that by December, NOA will be used for over 90% of highway driving. That signals a goal to boost actual driver adoption beyond just getting units on the road.
Other projects need time to deliver results.
XCMG's Hanjing electric truck is set for delivery by year-end. Sinotruk's Sitrak and Dongfeng's Tianlong KH are slated for mass production, while Foton's Auman Galaxy 5 is expected to launch. Additionally, Dongfeng's Tianlong KL has secured a contract designation.
At FAW Jiefang, Zhuoyu has won designations for the J7, J6 Pro, Yingtu, and Yinghang. The J6 Pro is among the first batch for mass production, though delivery volumes remain undisclosed. Yutong's Yuwei E9 PRO is planned to feature the PDS 5, capable of both highway and urban NOA.
For an autonomous driving supplier, getting on seven OEMs' partner lists is one thing; getting those OEMs to actually install and deliver systems at scale is another.
And further down the line, the real test is whether logistics clients actually use the systems once the trucks are in their hands.
Commercial driver assistance is already in regular use at some logistics firms, where actual mileage and system usage rates are becoming key metrics for success.
For Zhuoyu, partnerships with multiple OEMs lay the groundwork for scaled delivery. The question now is whether these projects can sustain installation volumes and win customer approval in real-world operations.
It is also trying to win over customers with operational savings.
At the launch, Zhuoyu released fuel-saving data: tracking its mass-produced trucks, which have logged over 20 million kilometers, the company reported a fuel efficiency improvement of 4.5% to 7.5% using an AI reinforcement learning algorithm.
The algorithm dynamically adjusts driving strategy by analyzing vehicle status, road grade, traffic conditions, and braking system data, minimizing unnecessary acceleration and braking.
Zhuoyu calculates that a truck running 800 kilometers a day could save 20,000 to 30,000 yuan annually—potentially recouping the cost of the driver-assistance system within a year.
In an industry with thin margins, that savings is a powerful incentive.
Still, fleet procurement decisions are based on long-term total cost of ownership. Payloads, routes, energy prices, mileage, and system costs all influence the actual return on investment.
And fuel efficiency is no longer Zhuoyu's unique selling point. Other companies have already achieved measurable savings in real-world operations.
The data sets aren't directly comparable due to different conditions. For logistics clients, the deciding factor for long-term partnerships will be who can consistently save money across more routes—and whose system is more reliable and easier to use.
By that logic, 2027 could be a critical window for assessing Zhuoyu's success in the commercial vehicle sector.
Under current production schedules, models from XCMG, Sinotruk, Foton, and Dongfeng are set to begin deliveries in the coming years.
If these projects achieve stable installation volumes, Zhuoyu will have a chance to prove: passenger car technology can indeed help latecomers fast-track their entry into the commercial vehicle market.
Conversely, if designations don't translate into sufficient deliveries, or if operational results fall short of expectations, a long partner list won't be enough to reshape the competitive landscape.
On the alternative tech path, Pony.ai and GAC New Energy Commercial Vehicle Co., Ltd. also have miles to go.
The initial production run of the T9 ROBOTRUCK provides a fresh vehicle foundation for their L4 ambitions. The plan calls for a deployment of 100 units by 2027, scaling up to thousands thereafter.
While not directly comparable to Zhuoyu's operational highway NOA trucks, this progress faces the same scrutiny regarding delivery scale, operational reliability, and cost.
Looking back at this cycle of development in commercial vehicle intelligence, the industry has moved past the stage of winning attention with tech demos alone. Some have accumulated hundreds of millions or even billions of kilometers in operational data; some are migrating passenger car models to trucks; and others are driving deeper integration of L4 autonomy with mass production.
The influx of new tech power offers more options to the market—and forces incumbents to confront new pressures in R&D and cost structures.
For Zhuoyu, seven brand partnerships prove it can win OEM approval. Now, it must prove it can convert its passenger car legacy into higher delivery volumes, greater adoption rates, and sustained operational savings.
Commercial vehicle autonomy has finally welcomed the passenger car players. The real question is whether they can build a scale advantage without having to retrace every step of their predecessors.
