See The Factory Where Robots Make Robots

Photo: evgeniy jamart / Shutterstock

China has crossed a psychological threshold in robotics: not just deploying robots by the millions across factories, but standing up facilities where robots, digital twins, and automated logistics help build the next generation of robots at industrial scale.

At a Glance

UBTECH has commissioned a purpose-built humanoid robot factory in Liuzhou, Guangxi, designed for 10,000-plus annual unit capacity across its Walker S and Cruzr product lines.
The plant integrates automated guided vehicles (AGVs), closed-loop intralogistics, and end-to-end digital management, with a stated takt time of one humanoid every 10 minutes.
The site’s footprint and line configuration mark a shift from one-off demos to mixed-model assembly for commercial units, not lab prototypes.
The move fits a broader national arc: China hosts the world’s largest base of industrial robots and is channeling that capability into humanoid production at scale.

What UBTECH has actually built in Liuzhou

UBTECH’s Liuzhou facility is presented as a “super smart factory” configured for mass production of its Walker S humanoid and Cruzr service-robot series. Reports place the plant’s floor area around 14,000 square meters with a tall bay (circa 13.8 meters) and mixed-model lines—an industrial layout chosen to support assembly, calibration, verification, and packaging of multiple SKUs without constant retooling. The manufacturing system is described as digitally managed end-to-end, spanning component kitting and subassembly through test, warehousing, and outbound logistics. Within that flow, mobile robots conduct material moves, and process data feed a central control layer intended to cut changeover time and stabilize quality at volume.

Crucially, the factory is tied to named, shipping product families rather than research platforms. Walker S variants (including S2) target industrial humanoid use cases; Cruzr models (such as S2 and Y1) occupy service and commercial roles. Anchoring the line to defined platforms and variants matters: it forces tooling, fixtures, test protocols, and supplier relationships into the discipline required for repeatability—everything that separates a show-floor demo from a production operation.

Throughput claims and the mechanics of “robots building robots”

Multiple accounts cite a takt time—one humanoid every 10 minutes—achieved by orchestrating a closed-loop intralogistics system with parallelized subassembly and final integration stations. In manufacturing terms, a takt claim of this kind implies three conditions: balanced work content across stations; reliable in-station test automation to prevent downstream bottlenecks; and a logistics spine robust enough to keep kits, harnesses, and mechatronic modules arriving just in time. Digitally twinned stations and AGVs are the critical enablers here: the twin lets engineers pre-validate line changes and fault responses; AGVs decouple material flow from fixed conveyors, allowing the factory to expand or re-balance lines without civil works.

“Robots building robots” is not marketing poetry but an architecture choice. When mobile robots handle intralogistics, collaborative robots perform certain fastening, dispensing, and inspection tasks, and machine-vision cells validate alignment and tolerances, the human role concentrates in exceptions, complex harness routing, and final acceptance. The practical payoff is steady beat-time and predictable quality drift, which are prerequisites for any 10,000-unit class target, regardless of whether the mix leans industrial humanoids or service platforms.

Why Liuzhou, and why now

Liuzhou’s industrial pitch is straightforward: a manufacturing-centric city aligning local policy, land, and utilities with national priorities to claim a share of the humanoid economy’s value chain. Official and state-linked reporting frames the plant as a benchmark step—Liuzhou “entering a new stage” of large-scale robot manufacturing—and folds it into a broader transformation narrative in Guangxi. The symbolism matters internally, but the practical logic rests on supply-chain geometry. The Chinese robotics ecosystem, especially Greater Bay Area suppliers, can deliver actuators, reduction gears, controllers, sensors, and harness assemblies with short lead times and dense vendor redundancy—a precondition for stable ramp.

That supply-chain density is quantifiable. China hosts the world’s largest installed base of industrial robots—just over two million in factories as of mid-decade—and leads annual installations by a wide margin. In such an environment, mechatronics competency, factory automation integrators, and test-equipment vendors form a thick labor and vendor market that reduces risk in commissioning new lines.

From demos to deployment: what changes at 10,000-unit capacity

Humanoid robotics shifts from spectacle to industry when the build system industrializes: serialized parts, statistically controlled processes, field-replaceable modules, and aftersales logistics. A 10,000-unit design capacity forces those disciplines. It requires a supplier-qualification regime, automated end-of-line testing with fault-code traceability, and firmware provisioning workflows that can update hundreds of units a day without bricking devices. It also reshapes cost: yield improvement and learning-curve effects become as valuable as any breakthrough actuator or controller; every tenth of a percent in first-pass yield turns into real dollars at volume.

On the demand side, humanoids are graduating from pilot lines in automotive and electronics to broader factory functions—monotony, danger, and ergonomic pain points first. China’s manufacturing base is unusually well-suited to this migration: high robot density, mature PLC-driven processes, and extensive digitalization make it easier to carve out humanoid-ready tasks without halting the entire line.

How this fits the national trajectory in robotics

The Liuzhou plant is not an isolated flourish. It is one node in a national effort to translate a commanding lead in industrial robots into leadership in embodied AI and humanoid platforms. Media access to other Chinese humanoid lines, from Shanghai to Guangdong and Beijing, shows similar design patterns: modular final assembly, machine-vision inspection, and mobile-robot intralogistics, all wrapped by digital management layers that shorten iteration cycles and aggregate real-world data for controller training.

The broader backdrop is decisive scale. China’s robot economy compounds through three flywheels: component cost-down via volume, systems integration talent trained by thousands of brownfield retrofits, and data loops from extended deployments that feed motion-control and perception stacks. Each new factory that moves from prototype to mixed-model line accelerates those loops, which is why commissioning announcements—when tied to named products and configured lines—often mark genuine capability inflections rather than mere PR.

What to watch next: signals of real operating maturity

Commissioning a factory is the start, not the finish. The operating signals that matter over the next year are concrete and measurable: sustained takt adherence without expensive overtime; field reliability metrics for Walker S and Cruzr units built in Liuzhou; turnaround times for service parts; and evidence that line balancing absorbs new variants without disrupting output. If those indicators trend the right way, Liuzhou will read not just as an emblem of policy ambition, but as proof that China can industrialize humanoids with the same relentlessness it applied to smartphones, EVs, and traditional industrial robots. A brief caveat applies to all early-stage ramps: capacity targets describe design intent; maturity is proven by months of stable shipments and field performance.