The Vans That Beat the Robotaxis
Not in robotaxis or flying cars, but in delivery vans humming through China’s city streets.
Content
Free section
Paid section
Things That Caught Our Attention
Introduction
China’s surge in autonomous delivery wasn’t driven by faith in technology. It was a practical response to mounting logistical pressure. Dense urban networks demanded higher efficiency, falling sensor and computing costs made automation affordable, and gradual regulatory easing allowed cities to experiment.
From 2008 to 2023, parcel volume rose from 1.5 billion to 132 billion—an 88-fold increase, or roughly 34% annual growth in China. Over the same period, the courier workforce grew only tenfold. The widening gap underscored a labor shortage and made unmanned delivery vehicles a natural solution to improve last-mile productivity.
Between 2020 and 2024, the sector entered a phase of technical discipline. Regulators and investors demanded proof that vehicles could operate safely, reliably, and at manageable cost before allowing large-scale rollout. That restraint slowed expansion but strengthened fundamentals. Manufacturing processes became standardized, pilot programs matured into policy templates, and cost control gained as much weight as software design. By the end of this period, the focus had shifted from building smarter algorithms to building systems resilient enough to support them.
In 2025, competition enters a new phase. The question is no longer whether the vehicles can drive, but whether the businesses behind them can endure. Beneath the rhetoric of autonomy, the contest is clear: the future of driverless logistics will be decided not by who drives best, but by who builds the most resilient business to keep driving.
We hope you enjoy this article, and that it helps you see China’s autonomous delivery vehicles in a new light—a field often overshadowed by the hype around Robotaxis and eVTOLs, yet arguably the most commercially advanced and economically grounded segment of the self-driving industry.
Yours sincerely,
Rita Luan, Tech Buzz China Researcher
When the Last Mile Leads the Way
In China, the “last mile” of logistics has quietly become the country’s most successful proving ground for Level-4 autonomy—a setting defined by low speeds (under 40 km/h), fixed routes of 5–50 km, and short, repeatable trips. Those constraints, far from limiting innovation, have made China an ideal environment for scaling the technology.
By the end of 2024, more than 6,000 driverless delivery vehicles were operating across over 100 city zones, completing hundreds of millions of deliveries. What began as a showcase for engineering prowess has turned into part of the daily logistics routine, signaling that the technology has shifted from laboratory testing to real-world productivity.
The economics are straightforward. On repeatable, high-frequency routes, per-order costs fall about 35%–45% as robovans swap a driver’s wage for a fixed software-and-maintenance fee and lower energy use. A case study from Wuhan highlights the economic upside. One logistics hub replaced manual dispatches with three autonomous vans serving five nearby stations, handling 5,000–6,000 orders per day. This setup cut operating costs by 40–50%, saving ¥15,000–20,000 (≈US$2,100–2,800) per month, while redeployed human couriers contributed to 20–30% growth in pickup volumes. Such field data validate the financial logic that sustained utilization directly drives profitability.
Across two to four daily dispatches, robovans consistently undercut human-run vans on a per-parcel basis. On a monthly view, replacing a small EV plus driver (≈¥9,600) with a robovan subscription (≈¥3,766) reduces operating expense by roughly 60% and pushes unit cost from about ¥0.32 to about ¥0.126 at ~30,000 parcels per vehicle.
Operationally, round-the-clock runtime and “string of pearls” routing, with fewer back-and-forth trips and more direct links between pickup points, further reduce idle and handling time, pushing hub-level savings toward 40%. Analysts expect 2025 to be the first true scale-up year as fleets reach breakeven and expand profitably. Robotaxis and eVTOLs, by contrast, remain in regulatory limbo, constrained by stricter safety thresholds and more capital-intensive hardware.
Cost Breakdown of Terminal Human-Driven Delivery
Sources: Wind; Yicai Research; Changjiang Securities Research
Policy has played a decisive role. Since 2023, cities from Beijing and Shenzhen to Hefei and Qingdao have granted broader road access under “vehicle-road-cloud” pilot programs—integrating vehicles, infrastructure, and cloud networks into unified data systems. These initiatives have allowed fleets to move beyond industrial parks and gated campuses onto open urban and suburban roads, extending driverless operations across both urban cores and suburban supply loops.
Financing has supported these trials and early operations. In 2024, China’s autonomous-driving sector raised RMB 40 billion (≈ US$5.5 billion), with roughly one-third directed to low-speed logistics and urban distribution. Leading internet and manufacturing players including Meituan, JD Logistics, SF, Alibaba Cainiao, Baidu Apollo and Haomo, are piloting driverless delivery across multiple cities, using on-road data to iterate algorithms and refine system design.
International pilots are progressing in parallel. Several Chinese delivery-AV companies have launched projects in the Middle East and Southeast Asia—for example, Cainiao with Saudi Post in Riyadh, Haomo with local partners in Indonesia for light Robovan services, and Meituan and JD with trials in the UAE and Thailand, providing early evidence on the transferability of China’s L4 operating model.
Forecasts indicate that the autonomous-delivery supplier revenue pool (hardware plus autonomy software) will grow from RMB 35.2 billion (≈US$4.9 billion) in 2026E to RMB 373.4 billion (≈US$51.9 billion) in 2030E. Within this total, hardware is projected to increase from RMB 18.9 billion to RMB 138.0 billion, while software subscriptions rise from RMB 16.3 billion to RMB 235.3 billion, becoming the primary driver of growth. For context, in the United States, Grand View Research estimates the autonomous last-mile delivery market at US$632.3 million in 2024, reaching US$1.98 billion by 2030 (21.4% CAGR).
The Real Race
Autonomous driving is often framed as a technology arms race. Over the past five years, China’s driverless delivery vans have been in that first phase. Companies competed on their sensors such as lidar, short-range radar, and cameras used together, as well as on software that can keep track of where the vehicle is while it builds a map (SLAM) and drive smoothly and accurately. They focused on low-speed, fixed routes of about 5 to 50 kilometers where it is easier to operate reliably. At the same time, they raced to win city testing permits and better parts supply deals, gaining early access to programs where cars communicate with roadside equipment and cloud systems, and pushing down the cost of automotive-grade hardware.
Platform giants such as Meituan, JD.com, and Cainiao were the first to scale across campuses, parks, and residential communities, accumulating millions of safe kilometers that set safety thresholds and regulatory baselines. In that phase, technology and safety were the entire answer to getting on the road.
Around 2023, the turning point became clear. Technical moats began to erode as algorithms converged, hardware costs plunged, and pilots normalized. Prices of lidar, domain controllers, and compute platforms dropped by more than half in two years, pushing unit costs to roughly ¥50,000 (≈ US$6,900). The question shifted from Can it run? to Can it earn? As the technological threshold flattened, competition moved to economics—pricing, cash cycles, and scalability, ushering in a second half defined by business models.
Two Paths, Two Playbooks
China’s autonomous delivery market is splitting along two tracks. Platform players like Meituan, JD.com, and Cainiao are embedding driverless vehicles within their own fulfillment systems, using them as tools to replace human couriers on predictable last-mile routes. The approach emphasizes control and compliance: data remains internal, testing occurs in controlled environments, and every deployment doubles as a safety validation. Yet this path is bounded by each platform’s delivery network and capital budgets. Expansion often resembles cost optimization, not new market creation. The pilots prove reliability, not yet profitability—but they lay the groundwork of public trust that the rest of the industry depends on.
Independent manufacturers such as Neolix, Zelos, and Rino.AI are driving the next phase of growth. Their model reframes the delivery robot not as proprietary hardware but as shared logistics capacity—rentable, reusable, and increasingly standardized. The innovation lies in the financing, not the sensors: combining low-cost hardware with software subscriptions converts capital expenditure into operating expense. For logistics clients, that shift is decisive. With sufficient route density, the payback period averages about 15 months; for smaller operators, the savings appear from the first month of use.
Unlike robotaxis, delivery vehicles operate at low speed and on semi-closed roads, facing fewer regulatory and public-sensitivity risks. Their demand—groceries, parcels, restocking—is real, repetitive, and resilient. Subscription and on-demand leasing models are turning autonomous vehicles into plug-and-play capacity. Today, companies already rent fleets on monthly or short-term contracts; the next stage is cloud-like elasticity, where fleets can be deployed within days as permitting, mapping, and remote-operation systems become more standardized.
This shift is possible because two forces have converged: supply-chain deflation and policy acceleration. Falling component prices, longer hardware life, and a more predictable “vehicle-road-cloud” infrastructure have, for the first time, turned route-level cash flow positive at leading operators. The industry’s competitive edge is migrating from who codes better to who runs denser, faster, and cheaper.
Still, margins hinge on route density and road access. Fixed suburban-urban corridors enable 24/7 scheduling and night-time peaks, maximizing utilization. Fragmented demand or tighter local oversight quickly erodes the cost advantage. In short, profitability depends less on algorithmic sophistication than on sustaining utilization and cash-flow velocity at scale.
As both models mature, a symbiosis is emerging. Platforms retain control over data and safety, while investing in or sourcing fleets from manufacturers that bring speed and cost discipline. Cainiao, for instance, develops its own vehicles but also procures from Neolix and Zelos; Meituan and JD.com follow similar hybrids. Platforms lighten their asset load, manufacturers gain steady orders, and both refine their business models through live operations.
Industry concentration is accelerating. Two manufacturers now command close to 90% of China’s professional autonomous-delivery market, not through flashier algorithms but through validated economics. Neolix anchors itself in Tier-1 and Tier-2 cities, emphasizing service reliability, regulatory compliance, and cash-flow stability; Zelos scales through ultra-low pricing and subscription-based, quick-payback packages. Together they are transforming the business from selling vehicles to selling logistics capacity. China’s autonomous delivery landscape is being redrawn. In the real race ahead, victory won’t belong to whoever builds the smartest machine—but to whoever builds the most resilient model.
Major Autonomous Delivery Vehicle Companies in China:
