In a BMW factory in Spartanburg, South Carolina, a humanoid robot made by California-based Figure AI began handling basic logistics tasks during a pilot program earlier this year. Roughly 6,000 miles away, Chinese startup Unitree shipped its G1 humanoid to research labs and light industrial clients at a price point under $16,000. These parallel developments capture a widening contest between the United States and China over machines that walk, grasp, and reason in the physical world.
The competition is no longer theoretical. Both governments are pouring resources into humanoid robotics, but their strategies diverge sharply. Washington is layering safety evaluation and international standards on top of a fast-moving private sector. Beijing is executing a state-directed industrial plan that treats humanoid robots as a named priority and rewards speed to market. The outcome will shape factory floors, hospital corridors, and elder-care facilities for decades.
China’s state-driven acceleration
China’s push into humanoid robotics builds on a foundation laid nearly a decade ago. The State Council’s New Generation Artificial Intelligence Development Plan, issued in July 2017, framed AI as a strategic national priority and set a goal of global AI leadership by 2030. That document covered the full spectrum of intelligent systems, from autonomous vehicles to smart manufacturing.
Beijing sharpened its focus in October 2023, when the Ministry of Industry and Information Technology released guidelines that singled out humanoid robots by name and called for mass production breakthroughs by 2025. The directive identified humanoid machines as a “new track” for economic growth and instructed local governments and firms to build integrated supply chains for actuators, sensors, and AI control systems.
Chinese companies have responded aggressively. Unitree, Fourier Intelligence, UBTECH, and Agibot have all demonstrated bipedal prototypes at trade fairs and begun limited commercial shipments. Unitree’s G1, priced to undercut Western competitors by a wide margin, drew global attention when demonstration videos went viral in 2024. Fourier’s GR-2 entered pilot deployments in rehabilitation clinics. The sheer number of Chinese startups working on humanoid platforms now exceeds two dozen, according to industry trackers, a density of effort that reflects coordinated state and venture capital investment.
The U.S. standards-first approach
The American strategy looks different. Rather than issuing a top-down production mandate, Washington has focused on building the evaluation infrastructure that advanced AI systems, including those powering humanoid robots, will eventually need to clear before reaching sensitive environments.
In November 2024, the Departments of Commerce and State launched the International Network of AI Safety Institutes, a multilateral body described in a NIST fact sheet as establishing shared risk assessment principles and a common scientific basis for evaluating frontier AI. The network is designed to shape governance norms across borders, covering the large foundation models that humanoid robots rely on for perception, planning, and movement.
The U.S. AI Safety Institute Consortium, also housed at NIST, held its first plenary meeting to set research priorities around evaluation methods for advanced AI. That work feeds into the technical groundwork needed before humanoid systems can be certified for deployment in warehouses, hospitals, or homes. The consortium brings together government agencies, companies like Microsoft and Google, and university researchers to develop shared testing benchmarks.
However, the trajectory of U.S. AI governance has not been smooth. Executive actions in early 2025 shifted federal AI policy toward lighter regulation and faster commercial deployment, raising questions about the Safety Institute’s long-term authority and budget. As of April 2026, the institute continues to operate, but its role in regulating embodied AI remains a work in progress rather than a settled framework.
Private sector momentum in the U.S.
Where Washington has been cautious on mandates, American companies have moved fast on hardware. Figure AI secured over $1.5 billion in funding and placed its Figure 02 robot in real factory environments. Agility Robotics opened a manufacturing facility in Salem, Oregon, dedicated to producing its Digit robot for logistics clients. Tesla continues developing Optimus, with CEO Elon Musk projecting commercial availability in the coming years, though timelines have repeatedly slipped.
Apptronik, a startup spun out of the University of Texas, partnered with Mercedes-Benz to test its Apollo robot in automotive assembly. Boston Dynamics, long a leader in legged robotics, has pivoted its Atlas platform to an all-electric humanoid design aimed at commercial use. The common thread across these efforts is venture capital and corporate R&D funding, not federal grants, driving the pace of development.
This creates a tension at the heart of the U.S. approach. The private sector is building capable machines quickly, but the public-sector evaluation layer meant to certify those machines for sensitive settings is still being assembled. If NIST’s standards work falls behind the deployment curve, American humanoid robots could reach market without the rigorous safety testing that the international network was designed to provide.
What this means for workers and industries
For companies in manufacturing, logistics, and elder care, the U.S.-China humanoid race carries immediate practical stakes. Firms that build or deploy these machines will need to navigate whichever safety and certification standards emerge from the NIST-led international network, or they will compete against Chinese-made systems that may arrive faster under a different regulatory philosophy.
Workers face a split outlook. In environments where strong evaluation requirements take hold, humanoid robots are more likely to fill collaborative roles alongside humans, with retraining programs and oversight built into deployment plans. In markets where speed dominates, displacement pressures could arrive sooner and with fewer guardrails.
The clearest takeaway from the available evidence is that this contest is not only about which country builds a better robot. It is about competing governance models that will determine how safely, how quickly, and for whose benefit these machines enter the real world. Businesses and policymakers tracking this space should watch two streams closely: NIST’s consortium outputs for signals on U.S. certification timelines, and any new directives from China’s MIIT for signs that Beijing is tightening or loosening its production targets. Those documents will set the rules of engagement for the next wave of embodied AI.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
