China’s PLA Pushes Humanoid Robots Toward Combat as Beijing’s 95% Share of Global Shipments Raises New Threat to U.S. Forces
China’s humanoid-robot boom is moving rapidly from exhibition halls and factory demonstrations toward military planning. Days after robots boxed, danced and competed in martial arts at Beijing’s World Humanoid Robot Games, the official newspaper of the People’s Liberation Army called for researchers to accelerate the transfer of advanced robotics from laboratories to military training grounds for future robotic “combatants.” A Reuters investigation based on more than 100 Chinese military procurement notices, academic papers, patents, government records and defense-industry materials shows that Beijing’s defense establishment is actively studying how humanoids could operate alongside soldiers, conduct reconnaissance, infiltrate hostile areas and eventually participate directly in combat.
For the United States, the most consequential part of this development is the combination of military ambition with China’s extraordinary commercial manufacturing scale. Chinese companies accounted for roughly 95% of global humanoid shipments in 2025, according to BofA Global Research data cited by Reuters. That industrial base gives the PLA access to an enormous ecosystem of motors, batteries, sensors, actuators, perception software, artificial intelligence, training data and mass-production expertise. China does not have to build an entirely separate military robotics sector from scratch. It can draw from a commercial industry already producing robots at a scale American competitors have struggled to match.
That is precisely the kind of military-civil fusion dynamic Washington has spent years warning about. The Defense Department describes China’s Military-Civil Fusion strategy as a system designed to ensure the PLA can obtain advanced technologies and expertise developed by Chinese companies, universities and research institutions that may outwardly appear civilian. Humanoid robotics is a near-perfect example. A machine developed for warehouse work, factory inspection or entertainment already contains many of the same capabilities a military would need: vision systems, environmental mapping, balance, dexterous hands, autonomous navigation, AI decision tools and the ability to move through spaces originally designed for humans.
Chinese military planners are already examining how those capabilities could translate into combat. An August 2025 study by researchers at a National University of Defense Technology test center modeled an urban assault force in which six “combat robots” — including humanoids, robot dogs and unmanned vehicles — would operate alongside soldiers to clear an enemy-held building floor by floor and room by room. The researchers projected that equipment capable of supporting the scenario could become available within five to ten years. That type of mission is particularly significant because urban warfare remains one of the most dangerous forms of combat for human soldiers. Robots that can climb stairs, open doors, inspect rooms and move through narrow interiors could give the PLA new ways to reduce its own casualties while maintaining offensive pressure.
The PLA is also studying roles that extend beyond simple logistics. In 2025, an NUDT competition required robots to simulate “infiltration behind enemy lines,” enter an area, map the environment and locate targets. Other scenarios included identity checks, security patrols and military-base inspections. Around the same period, PLA procurement notices sought a humanoid robot with technical training and an “embodied humanoid robot intelligent perception and dexterous operation system.” In June 2026, another military procurement document budgeted approximately $300,000 for a system designed to collect and label camera, radar and motion data for humanoids, including information about the terrain they could traverse. These records show that Chinese military interest has advanced from conceptual discussions into the perception, manipulation and training-data infrastructure needed to make robots useful outside controlled demonstrations.
China’s state-owned defense industry is moving in the same direction. Norinco, one of the country’s major defense conglomerates, has developed a full-size humanoid known as Fuxi. Company materials describe potential functions including sentry duty, all-weather reconnaissance, intelligent patrol and other dangerous missions. Fuxi can also operate through a teleoperation system, allowing a human operator to control the machine remotely. That capability is strategically important because it allows the military to deploy a humanoid into hazardous areas without waiting for fully autonomous artificial intelligence to become reliable enough to make every battlefield decision independently.
The connection to Taiwan should concern American military planners. In December, the PLA Eastern Theater Command — the command responsible for operations around Taiwan — released an AI-generated montage depicting military robots, including humanoids and quadrupeds, overwhelming Taiwan’s defenses in a hypothetical future conflict. Chinese military commentary has also discussed humanoid missions evolving from combat support toward primary combat roles, including circumstances in which a robot might fire after a human has verified the target. The imagery does not establish that such systems are operational today, but it demonstrates where parts of China’s defense establishment believe robotics may fit into future warfare.
That matters directly to the United States because a Taiwan conflict would place American forces against a Chinese military operating close to its own industrial base and supply networks. Humanoids and quadrupeds could eventually help the PLA scout buildings, inspect tunnels, transport supplies, conduct security patrols, expose ambushes or enter areas too dangerous for soldiers. Even limited robotic capability could force U.S. and allied forces to expend ammunition, surveillance resources and time against machines that Beijing may be willing to sacrifice far more readily than trained personnel. Once robots become sufficiently inexpensive and available at industrial scale, quantity itself becomes a military advantage.
China’s manufacturing dominance therefore deserves as much attention as the robots’ present technical limitations. Reuters found no evidence that an armed humanoid has yet been deployed with an operational PLA unit, and current systems still struggle with battery life, reliability and unpredictable environments. Those limitations are significant today, but they should not encourage complacency. Military technologies often become dangerous through iteration, falling costs and mass production long before they reach science-fiction levels of autonomy. China’s ability to manufacture enormous numbers of motors, reducers, batteries, sensors and complete robots gives it a powerful platform for that iterative improvement.
Unitree illustrates how that industrial advantage can intersect with American innovation. Reuters reported in August that technologies underlying the Chinese company’s highly successful robot dogs closely resemble innovations developed through U.S. military-funded research at MIT. The research was openly published, allowing Chinese engineers to build on it legally, and Unitree subsequently achieved far greater manufacturing scale. Its robots have appeared in PLA-related contexts, while the company has since been identified by the United States as connected to China’s military ecosystem. The strategic lesson for Washington is uncomfortable: American laboratories can pioneer a technology while Chinese companies industrialize it faster and eventually make related capabilities available to the PLA.
This pattern extends beyond any single robotics company. The Pentagon has repeatedly warned that China’s Military-Civil Fusion system seeks to channel civilian research and commercial technology toward PLA modernization. In robotics, the civilian and military pathways are particularly difficult to separate. Better factory robots produce better actuators. Better delivery machines improve navigation. Better consumer humanoids generate training data. Better AI models improve perception and planning. Every advance in China’s commercial robotics ecosystem can potentially shorten the distance between an awkward demonstration robot and a useful military platform.
The United States is also exploring militarized humanoid capabilities. The U.S. Army launched a competition involving potential reconnaissance, security, obstacle-clearing, hazardous-material and urban combat applications. That makes this an emerging technological competition rather than a capability unique to China. The imbalance arises from industrial scale. A country capable of producing most of the world’s humanoid robots has opportunities to test more units, gather more training data, reduce costs more quickly and support a much larger supply chain than competitors operating at smaller volumes.
Washington should therefore treat advanced robotics as a strategic industrial capability alongside drones, semiconductors, shipbuilding and artificial intelligence. America needs domestic and allied manufacturing capacity for actuators, motors, reducers, sensors, batteries and embodied-AI systems. U.S.-funded robotics breakthroughs with obvious military relevance also deserve more careful evaluation before technical knowledge spreads into ecosystems where American companies may later face heavily subsidized competitors. Protecting research alone will not solve the problem; the United States must also become better at commercializing and manufacturing its own inventions.
The PLA’s current humanoid programs are still experimental, yet the direction is increasingly clear. China’s military universities are testing battlefield scenarios, procurement offices are buying hardware and training systems, state-owned defense companies are building purpose-designed humanoids, and military publications are openly discussing robots advancing from support missions toward combat. Chinese researchers themselves acknowledge that movement, perception, intelligence, cost and mass production still need improvement. One Academy of Military Sciences researcher nevertheless predicted that, once those barriers fall, humanoid robots will begin flowing onto battlefields.
Americans should pay attention now, while the technology is still developing. China’s advantage in robotics is no longer confined to dancing machines, factory demonstrations or viral videos. Beijing possesses the manufacturing ecosystem, military research institutions and state-directed industrial structure needed to push civilian robotics toward battlefield use. If the PLA succeeds, U.S. troops could eventually face machines built by an industry that already dominates global humanoid shipments and has demonstrated an ability to absorb innovations pioneered elsewhere. The strategic danger lies in allowing China to turn commercial scale into military scale faster than the United States can build a competitive response.
