How US curbs could split the robot world 

On 28 July, the US Federal Communications Commission (FCC) added “foreign-produced advanced robotic devices” to its Covered List, marking the first broad US market-access restriction targeting advanced robots as a standalone product category on explicit national-security grounds. Unless granted conditional approval, new foreign-produced models can no longer receive the FCC equipment authorisation required for US import and sale. While the measure remarkably applies to foreign-produced devices regardless of nationality, its implications for China are hard to miss.

From chips to robots

The decision came just three weeks before Unitree, one of China’s best-known humanoid and quadruped robot makers, listed in Shanghai on 19 August. China accounted for more than 80% of global humanoid robot installations in 2025. The juxtaposition makes the FCC action a watershed in Washington’s technology policy: restrictions are moving from the technologies that make intelligent machines possible to the machines themselves.

This did not begin with humanoid robots. In 2018, Washington’s Section 301 tariffs explicitly identified robotics among the industries benefiting from “Made in China 2025”. From 2022, attention moved upstream: US export controls began targeting advanced computing chips and semiconductor manufacturing capabilities that underpin frontier AI. While robotics was not named, increasingly capable embodied intelligence depends on precisely those sanctioned computing infrastructures.

The 2025 National Defense Authorization Act, introduced in 2024, explicitly included “mobile robotics” and “humanoid robots” in its definition of unmanned ground vehicle systems when requiring the Pentagon to examine systems produced by entities from China and several other countries. 

The policy widened again in 2025 when the Commerce Department opened a Section 232 national-security investigation into imports of robotics and industrial machinery. Then, in June this year, Unitree was added to the Pentagon’s Section 1260H list of Chinese military companies. From 30 June, companies on the list became subject to a direct Pentagon procurement ban under Section 805.

The July FCC action is therefore the latest, and qualitatively different, step in this progression. Earlier measures targeted trade practices, enabling technologies, defence procurement or individual companies. The new decision attaches the security concern to the product category itself. The machine has become the security object.

Why is the US restricting Chinese robots?

National security is Washington’s most direct justification. Advanced robots combine cameras, microphones, positioning systems and other sensors with network connectivity, remote maintenance and software updates. In sensitive sites, this creates both data and physical security risks: compromised systems could collect information, disrupt operations or alter how machines behave. Similar concerns already underpin US restrictions on connected vehicles from China and Russia.

Robotics is also becoming a central arena of technological and industrial competition. Robots bring AI into the physical world, combining models and computing power with computer vision, motion control and advanced manufacturing. Washington is especially concerned that China’s manufacturing scale could become a systemic advantage: large-scale deployment generates operational data and engineering feedback, lowers costs and enables further deployment. US strengths in frontier AI, software and advanced chips therefore do not automatically translate into leadership in embodied AI. The US AI Action Plan’s emphasis on next-generation manufacturing reflects this concern.

Rules and institutions governing the robotics industry are still taking shape. Questions over data collection and localisation, remote access, software updates, technical disclosure, vulnerability management and liability already fall across data protection, cybersecurity and product safety regimes. Yet countries still differ significantly in regulatory scope, technical requirements, enforcement and the allocation of responsibility.

If the US can turn technical and governance standards that favour its interests into widely accepted international norms, Chinese companies may face higher certification and compliance costs, or even exclusion from some markets. From this perspective, US restrictions on robots from certain countries are not only about protecting the domestic market, but also about shaping international standards, coordinating with allies and competing for access to third-country markets.

Can US restrictions achieve their goals?

In the short term, Chinese robotics companies will face real costs. US measures can reduce orders, complicate certification and financing, limit access to some technologies, and force firms to redesign products or supply chains. But these restrictions are unlikely to remove the foundations of China’s robotics industry. 

China remains the world’s largest industrial-robot market and has a dense manufacturing ecosystem spanning batteries, motors, reducers, cameras, lidar, automotive electronics and precision machinery. This gives robotics companies domestic customers, supply chain depth and engineering feedback even when access to the US narrows. External restrictions may slow international expansion, but they are less capable of breaking this internal innovation cycle.

The US will also bear costs from restricting Chinese robots. First, the US risks raising automation costs in manufacturing, warehousing and logistics. Restricting imports of Chinese robots may temporarily reduce foreign competition, but it does not help a US manufacturing industry facing shortages of skilled labour and high production costs. In June 2026, there were around 481,000 job openings in US manufacturing. These figures suggest that labour shortages remain significant and provide a practical reason for firms to use automation to ease recruitment pressures.

Second, overly broad restrictions could raise research costs for US universities and laboratories. Robotics research often depends on affordable experimental platforms, open software interfaces, reproducible hardware environments and international academic exchange. William Paterson University, for example, used funding from a US Department of Education grant to purchase a Unitree G1 Edu humanoid robot for teaching and research in AI, robotics and natural-language processing. Cutting access to such platforms could make experimentation more expensive and narrow the range of technologies available to US researchers.

The longer-term effect will depend heavily on whether Washington can internationalise its approach. This may prove harder than in semiconductors: the FCC measure applies to foreign-produced robots regardless of nationality, potentially catching Japanese, Korean and European suppliers as well. US allies therefore have less obvious reason to follow it wholesale.

Avoiding two robot worlds

US-China competition in robotics is not itself a problem. Healthy competition can accelerate innovation and give developing countries more technological choices. The real concern is that geopolitical rivalry could divide the robotics industry before the global market has properly formed. Humanoid robotics is already becoming unusually bipolar: China leads in manufacturing and deployment, while the US remains the strongest system-level challenger through its advantages in frontier AI, semiconductors, software and capital. Yet most countries are still only beginning to adopt these technologies.

This timing matters because robotics is likely to generate strong network effects. Installed hardware attracts developers; developers build models; deployment produces data; and firms organise supply chains around platforms they use. If these systems become entrenched early, switching between them will become increasingly costly. Political separation today could therefore turn into technological lock-in tomorrow.

For Chinese companies, resilience should mean more than simply finding alternative export markets. Products built around transparent software updates, auditable remote access, local or on-device data processing and credible cybersecurity certification will be easier to defend in markets where legitimate security concerns are growing.

The larger responsibility lies with governments and international organisations. While the industry is still at an early stage, greater dialogue and common rules are essential. Legitimate security concerns need to be addressed, but national security should not become an ever-expanding justification for excluding foreign technologies. Otherwise, robotics risks becoming another source of protectionism and technical fragmentation. China would have much to gain from supporting multilateral, risk-based governance rather than answering one origin-based barrier with another. 

This is also in the interest of Europe, Southeast Asia, the Middle East and other emerging markets, which benefit from keeping access to competing technologies open and the costs of switching between them low.

Competition between China and the US is unavoidable. Two mutually incompatible worlds of intelligent machines are not. The urgent task is to build enough common ground on safety, data, cybersecurity and interoperability before political bifurcation becomes technological bifurcation — and, through network effects, begins to reinforce itself.

Related: China’s fluid humanoid stuns — but US still rules robot brains | [Big read] Sanctioned but surging: How China is building its AI empire