NeuroAI NEUROAINEUROAI.SITE
ESC

China's Neolix quietly put 10,000 delivery robots on the road — then 15,000

Neolix (新石器无人车), a Beijing-based autonomous-delivery-vehicle maker, became the first company in the world to deploy a 10,000-strong L4 driverless fleet, hitting that mark in September 2025. By the end of the year it said the number had passed 15,000 vehicles across more than 300 cities, quietly making sidewalk-scale automation one of China's most deployed robot categories.

2026-09-28 · 900 words · NeuroAI
China's Neolix quietly put 10,000 delivery robots on the road — then 15,000

Most robot stories are about a machine that can do one impressive thing once. Neolix (新石器无人车) is the opposite kind of story: a boxy, driverless cart that does one boring thing — move parcels from a depot to a doorstep — tens of thousands of times, every day, in hundreds of cities.

You probably have not heard of it. That is exactly the point.

The milestone that actually mattered

On 23 September 2025, Neolix announced the delivery of its 10,000th vehicle, at a ceremony in Tonglu, Zhejiang. The company and state-linked reporting called it the world's first L4-level autonomous delivery fleet to reach 10,000 units — the moment unmanned delivery stopped being a pilot and became infrastructure.

Neolix did not stop there. By late December 2025, Beijing's E-Town administrative zone — where the company is based — reported the global fleet had passed 15,000 vehicles, deployed across more than 300 cities, with over 70 million kilometers of cumulative safe autonomous driving.

Those are big numbers for any robot category, let alone one most consumers only notice when one rolls past on a sidewalk.

What the robots actually are

Neolix builds low-speed autonomous delivery vehicles, not humanoids. The two workhorses:

  • X3 — about 520 kg payload, 3 m³ of cargo space, 100 km range, suited to deliveries within a 15 km radius.
  • X6 — about 1,130 kg payload, 6 m³ of space, 200 km range, covering a 60 km radius for town-and-country routes.

Both run Neolix's self-developed L4 "map-free" (无图) autonomous-driving stack, meaning they do not depend on pre-built high-definition maps of every street — a design choice that makes cross-city and cross-border deployment far faster. The company says its single-kilometer operating cost is more than 50% lower than a traditional delivery vehicle.

Why China is the right lab

Neolix was founded in 2018 in Beijing's Yizhuang district, which became China's first "vehicle-road-cloud" autonomous-driving demonstration zone in 2020. That policy opening gave the company something rare: real roads to test on, and a path to road-rights permits (路权) in nearly 300 cities.

Its customers are the giants of Chinese logistics — SF Express, JD.com, and China Post — where the economics are brutally simple. In one Chongqing county, China Post used Neolix vehicles on a ~60 km daily rural route moving about 1,400 parcels per vehicle per day, cutting delivery time to towns by roughly 30%. When the world's largest parcel market needs to move packets the last three kilometers cheaply, a ¥40,000–50,000 cart starts to look less like a gadget and more like a line item.

The overseas push

Neolix is also one of the more globally active Chinese robot exporters. In October 2025 it signed a strategic partnership with K2, a UAE state-linked AI company, to build a large-scale RoboVan deployment in the Middle East, and it has secured an autonomous-vehicle license in Abu Dhabi. The company has said overseas orders stood at around 5,000 units heading into 2026.

That same month, Neolix closed a Series D financing of more than US$600 million (≈ HK$4.68 billion), led by UAE's Stone Venture (磊石资本) with Chinese state and corporate investors — reported by Xinhua as the largest single financing in China's autonomous-driving sector to that point.

What readers should keep straight

The "10,000" and "15,000" figures are company announcements carried by Xinhua and a Beijing municipal channel, not an independent audit. "Deployed" can mean owned-and-operating, or placed-with-partners-on-trial — the two are not identical. And the 50%-lower cost claim is the company's own benchmark against traditional delivery, not a neutral study.

None of that dulls the core fact: this is a robot that has left the lab at scale, which is rarer than any viral demo.

What readers can do now

  • If you run last-mile logistics, benchmark Neolix's per-kilometer cost against your own fleet; the China Post rural case (≈1,400 parcels/vehicle/day, ~30% faster) is the concrete reference point.
  • If you follow robotics commercialization, treat deployment count and road-rights permits as the real scoreboard — not gait videos. Neolix's 300-city permit footprint is the moat.
  • If you watch export controls or trade, note that Neolix is already shipping autonomous vehicles to the Middle East; low-speed delivery is a category where Chinese firms lead on volume today.

Honest limitations

Verified facts: Neolix (新石器无人车) founded 2018 in Beijing's Yizhuang district — from the company and Xinhua. The 10,000th-vehicle delivery on 23 September 2025 (Tonglu, Zhejiang), described as the world's first 10k-scale L4 delivery fleet, is reported by Xinhua's Economic Information Daily (经济参考报, 23 September 2025). The >15,000 vehicles / 300+ cities / 70M+ km figure is from the Beijing E-Town government portal (29 December 2025). Vehicle specs (X3: 520 kg, 3 m³, 100 km; X6: 1,130 kg, 6 m³, 200 km; L4 map-free stack; >50% lower cost) are from the company site and Beijing E-Town coverage of the Saudi CoMotion summit (December 2025). The China Post Chongqing case (≈1,400 parcels/vehicle/day, ~30% faster) is from Beijing E-Town (29 December 2025). The >US$600M Series D led by Stone Venture and the UAE K2 partnership (October 2025) are from Xinhua (3 November 2025). Unit price 40,000–50,000 RMB is from the company's site, converted to ≈ US$5,600–7,000 / HK$43,600–54,500. Deployment counts are company-reported via state media, not independently audited. Analysis is current to 28 September 2026 and is not investment advice.

Related coverage

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