Two days after the World Humanoid Robot Games, the PLA Daily called for robotic combatants. A Reuters review of 100+ Chinese military records shows how a dominant civilian industry is feeding that research.
Robots at the World Humanoid Robot Games in August danced, boxed, and fell over in front of a Beijing crowd. Two days after the final event, the PLA Daily, the People's Liberation Army's official newspaper, called for those same kinds of machines to be developed faster and named robotic combatants as a target.
The scene and the editorial are not two stories. They are the public and policy halves of one pipeline: a civilian robotics industry that already builds the overwhelming share of the world's humanoids is now visibly feeding PLA combat research.
The strategic anchor is a market-share number. BofA Global Research estimated that Chinese companies shipped about 95 percent of the world's humanoid robots in 2025, according to the Reuters review. That is not a battlefield metric. It is an industrial metric. The PLA is not building a bespoke weapons program from scratch. It is sitting on top of the world's dominant commercial humanoid base, with the same actuators, sensors, and large-model software that companies are already selling to factories and warehouses.
Reuters reached that conclusion by reviewing more than 100 Chinese military documents and other records: procurement notices, academic papers, patents, government documents, official publications, and material from defense companies. Research accelerated in 2025 and 2026, with a focus on robot perception, manipulation, and the large training datasets that teach both. The record shows a military establishment reading the same civilian literature its domestic industry produces and steering procurement toward capabilities the commercial sector is already building.
The PLA Daily editorial, published two days after the Games closed, is the most public signal of intent. It called for faster movement of advanced robotic technologies out of research labs and into military training and referred explicitly to the potential development of robotic combatants. The editorial, not Reuters, framed the long-run target as combat-ready humanoids.
The PLA is also treating the war next door as a reference conflict. China's military is closely studying the Russia-Ukraine war for lessons applicable to future fights, the Reuters review found. That framing, peer war as a procurement and training template, is the PLA's own, and it tells the reader what kind of doctrine the robotics research is being shaped against. A walking manipulator that can clear a room, carry a load, or emplace a charge is the kind of platform that gets a long look when small drones and urban combat are the running examples.
The Games were not a parade of deployable combat systems. Some of the machines moved quickly; others fell or struggled with simple movements. Reuters documents procurement notices, academic research, patents, and government publications, not fielded systems in service. The PLA Daily editorial names an ambition, not a fleet. The 95 percent figure is a 2025 vendor-research estimate of global humanoid shipments, not a verified count of combat-capable units.
What the documents do show is a military establishment preparing to convert commercial momentum into research output. That is the dual-use pattern. A civilian industry ships the volume; the military funds the perception, manipulation, and training-data work that turns volume into capability. The next datable moment is the first procurement notice that names a humanoid platform for a training-range role rather than a research lab: the move from documents to a real robot under a real soldier.