A white, driverless cart with no steering wheel threads through a Beijing suburb, reads the traffic light, and stops at a delivery point — then does it again, 24 hours a day, in the rain. This is not a trial behind a fence. It is one of more than 10,000 such vehicles that Neolix (新石器无人车) says it has put on public roads, making the once-futuristic "last-mile" robot a routine piece of city logistics.
Key takeaways
- Neolix says its 10,000th L4 autonomous delivery vehicle rolled off the line in September 2025, the first company to reach that scale.
- As of late 2025 it reported road permits in roughly 300 cities and deployments across 15 countries and regions.
- Its vehicles work fixed B2B routes for logistics giants including SF Express, JD Logistics, and ZTO.
- A 2025 Series D raised more than US$600 million, described as the largest financing in China's autonomous-driving sector to date.
From factory to fleet
Neolix was founded in 2018 and built what it describes as the world's first L4 autonomous-vehicle factory with 10,000-unit capacity back in 2019. The bet was always that driverless carts make economic sense on predictable, repetitive routes — factory to warehouse, warehouse to station, station to neighborhood — long before they make sense as roving taxis.
The scale milestone is specific. In September 2025, Neolix announced its 10,000th vehicle rolled off the line in Tonglu, Zhejiang, which it and Chinese state-linked media described as the first 10,000-unit-scale unmanned-delivery fleet anywhere. Reporting from the same period put its road permits at nearly 300 cities and regions, with about 10,000 vehicles delivered and a footprint across 15 countries and regions. Earlier in 2025, the company said it had secured a 20,000-vehicle order, and its own timeline notes nearly 10,000 units ordered by a head logistics customer in early 2024.
What the carts actually carry
The vehicles come in sizes for the job. The X3 model carries roughly 400 to 500 items per trip; the larger X6 carries 800 to 1,000, at road speeds up to about 50 to 65 km/h. Neolix emphasizes self-developed core technology and its own factory as the source of low unit cost. The economic pitch is direct: it claims to cut per-kilometer transport cost from 2–7 yuan down to under 1 yuan, which is why logistics operators — not consumers — are the buyers.
The partnerships are the proof of adoption. Neolix works with SF Express (顺丰), JD Logistics (京东物流), ZTO (中通), YTO (圆通), China Post (中国邮政), STO (申通), J&T (极兔), Yunda (韵达), and Deppon (德邦). During the 2025 Singles' Day peak, Neolix said it and its partners completed more than 10 million kilometers of delivery mileage in a single month. In Qingdao, a deployment with DiDi's freight platform passed 1,200 vehicles using a RoboVan-as-a-Service model where users book unmanned pickup and drop-off through an app.
The hidden infrastructure
The less visible part is the coordination layer. At Neolix's smart co-distribution center in Beijing's Yizhuang district — described as the world's first unmanned-vehicle-centric distribution hub — a cloud-edge-end platform coordinates hundreds of carts, runs around the clock in "black light" operation, and handles up to 100,000 sorted items a day. The robot is only as useful as the system scheduling it.
In October 2025, Neolix closed a Series D of more than US$600 million, led by UAE-based Stone Venture with participation from China Mobile's digital-economy fund, CICC, and others. Chinese media called it the largest financing in the country's autonomous-driving sector to date — a signal that investors see fixed-route logistics, not open-road robotaxis, as the nearer-term commercial win.
Why this is the real robotics story
Humanoids grab the headlines, but the quiet scaling of low-speed, fixed-route autonomous carts may matter more to logistics economics this decade. They avoid the hardest part of autonomy — unpredictable open-road driving with pedestrians — by sticking to planned lanes and known stops, and they attack a real cost: end-of-line distribution is a large share of total logistics spend.
According to Li An, chief scientist at BrainNet (脑机网), the Neolix model shows that robotics commercialization often arrives first where the problem is boring and bounded, not where it is flashy.
Honest limitations
Most figures here come from Neolix's own announcements and Chinese state-linked or industry media that cite the company; the 10,000-vehicle and 300-city numbers are company-stated and not independently audited. "Delivered and deployed" can include vehicles in storage or partial rollout, so the active on-road count may differ. The cost-saving claims (under 1 yuan per kilometer) are Neolix's own and depend on route density and utilization. The competitive field is moving fast and crowded — rivals such as Ninebot's autonomous unit and Cainiao have also pushed low-cost vehicles, and Meituan's own outdoor autonomous fleet, while smaller in active scale, reached over 5.5 million delivery orders by end of 2025 per its corporate social responsibility report, a reminder that "scale" looks different by metric. Road-permit counts shift as cities approve or restrict deployments, so any single number is a snapshot.
What readers can do now
- When evaluating autonomous-delivery claims, separate "vehicles delivered" from "vehicles actively earning on public roads" — the second is the number that matters.
- Watch fixed-route B2B logistics (warehouse-to-station) as the most likely place robotics hits real unit economics before humanoids do.
- Track Neolix's overseas permits and the Stone Venture-led funding as a sign of whether China's low-speed autonomous logistics is exporting or staying domestic.