NeuroAI NEUROAINEUROAI.SITE
China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories
ESC

The driverless van that quietly took over China's last mile

Neolix has put more than 10,000 autonomous delivery vehicles on roads across 300-plus cities, but the model still leans on cheap hardware and granted road rights.

2026-11-08 · 1142 words · NeuroAI
The driverless van that quietly took over China's last mile

On a weekday morning in Beijing's Yizhuang district, a small, cab-less white box on wheels rolls up to a community gate, slows for a pedestrian, reads the traffic light, and continues to a convenience store's back door. Nobody is inside. This is not a trial — it is a delivery run, one of thousands happening daily for Neolix (新石器), a company that has quietly become the largest supplier of autonomous delivery vehicles (无人配送车) in China.

For the people who live behind that gate, the change is mostly invisible and mostly quiet. The van arrives without a knock, waits at the curb, and leaves without a conversation. There is no driver to tip, no phone call to miss, no missed-delivery slip to chase. In a country where e-commerce has made same-day and next-day parcels a baseline expectation, the autonomous leg quietly fills the final stretch between the depot and the doorstep. A neighbourhood may not notice the van at all, and that is precisely the point: the technology succeeds when it disappears into the daily routine of receiving things.

From zero to ten thousand in seven years

Neolix was founded in 2018. In 2019 it opened what it describes as the world's first L4 autonomous-vehicle factory with ten-thousand-unit capacity. In 2021 it received one of China's first road permits for driverless delivery. The inflection came in 2025: the company says its 10,000th vehicle rolled off the line on 23 September 2025, and by late that year it reported cumulative deliveries above 10,000 units deployed across more than 300 cities and regions, with footprints in around 15 countries.

A October 2025 Series D round of about US$600 million (roughly HK$4.7 billion) — described by state media as the largest private fundraising in China's autonomous-driving sector to that point — underscored how seriously capital now treats the "last mile."

What the vehicle actually does

Neolix's flagships, the X3 and X6 RoboVans, are purpose-built for intra-city logistics rather than passenger transport. They serve express delivery, retail replenishment, fresh food, cold chain, and auto-parts runs. In Qingdao alone, the company states more than 1,200 units are in operation. Customers include major couriers such as SF Express and ZTO Express.

For a resident, the difference shows up in small, familiar ways: a cold-chain order of groceries that arrives still chilled, or a convenience store shelf that never goes empty because the RoboVan top-ups the stock before the lunch rush. The vehicle is not a science project; it is a small, repetitive workhorse that does the dull, fixed-loop trips a human courier would otherwise repeat dozens of times a day, leaving the unpredictable, door-to-door handoffs to people.

The economics are the point. A government-linked industry report estimates per-unit cost has fallen to around RMB 100,000 (about US$14,000 / HK$109,000) as car-grade hardware scales. At that price, a fleet can run "早7晚10" (7 a.m. to 10 p.m.) schedules without a driver's wage, and the vehicle becomes a fixed cost rather than a labour line item. A small merchant no longer has to staff a delivery run; it books a machine for a route, much as it might lease a freezer.

Why road rights, not sensors, are the real story

The harder barrier was never the camera. It was permission. Beijing's Yizhuang zone built one of China's first "vehicle-road-cloud" autonomous-driving demonstration areas, and Neolix moved its headquarters there in 2020 to get real road-test scenarios. By 2021 it held a national-first delivery-vehicle road permit. Today, progressive city policies that define road rights and operating zones are what separate deployable fleets from showpieces.

Neolix has also moved abroad, securing road rights in Singapore and what it describes as the UAE's first RoboVan license via an Abu Dhabi partner — a reminder that the same permission problem repeats in every new market.

The bigger pattern: logistics as robotics

China's intra-city logistics still runs on over 30 million vehicles, most of them human-driven. The addressable market for autonomous logistics vehicles has been estimated at around US$400 billion. Neolix's claim of over 60% cumulative share by shipments in China signals consolidation around the early mover.

Why it matters

The significance of Neolix is not that a robot can drive — it is that a piece of routine physical work is being unbundled from a human wage. When the last mile becomes a scheduled, driverless, fixed-cost service, small merchants gain a logistics option that used to require a staffed fleet, and cities gain a working testbed for how autonomous machines share pavement with people on foot and on bikes. The open question is whether the permission that unlocked all of this stays granted: road rights are granted by cities, and a change of policy in one zone can shrink a "national" footprint overnight. The technology is the easy part; the social licence is the fragile part.

Key takeaways

  • Neolix passed 10,000 cumulative autonomous delivery vehicles in September 2025, deployed across 300-plus cities and about 15 countries.
  • Unit economics hinge on a roughly RMB 100,000 (US$14,000 / HK$109,000) hardware cost and granted road rights, not on exotic sensors.
  • Customers include SF Express and ZTO Express; Qingdao alone hosts 1,200-plus units.
  • A US$600 million (HKD 4.7 billion) 2025 Series D marks the sector's capital inflection.

What readers can do now

  • If you run local commerce: evaluate driverless vans for fixed, repeatable routes (campus, industrial park, dense residential) where road rights already exist — that is where ROI appears first. Start by mapping your current routes for the traits these vehicles exploit: fixed stops, predictable timing, low-speed zones, and a single depot-to-door loop repeated many times a day. Pilot one loop before committing a fleet, and measure the saving against a staffed run on the same route.
  • If you invest: weight city policy and road-rights progress above sensor specs; deployment permission, not technology, gates scale. Read a city's autonomous-vehicle operating-zone rules before reading a company's spec sheet, because a permit won is a revenue line and a permit denied is a stranded asset.
  • If you live in a covered city: expect quieter, slower curb-side delivery as couriers add autonomous legs to human last-mile handoffs. Learn your city's published autonomous-vehicle operating zones, and if a vehicle behaves oddly, report it through the same channels you would use for any road hazard — the machine has no driver to query.

Honest limitations

Delivery counts, city totals, and market-share figures are drawn from company statements and a government-linked industry report, not independently audited here. "10,000 units" and "300 cities" reflect company-reported cumulative deployments; real active utilization may differ from cumulative deliveries. The RMB 100,000 per-unit cost is an industry-report estimate, not a published Neolix price list. Road rights remain city-by-city and can be revoked or stalled, making geographic expansion uneven. Safety performance across millions of kilometres is asserted via cumulative-mileage claims (reported above 50 million km) but not verified per incident here.

Related coverage

Cite this article

NeuroAI (脑机网), "The driverless van that quietly took over China's last mile," 8 November 2026, https://neuroai.site/a/na-robot-neolix-delivery-vehicle

Machine-readable: BibTeX · RIS · JSON · Markdown

More in “Humanoids & Robotics” → · Back to home · Markdown version

China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories