China shipped 12,750 of the 15,000 humanoid robots sold worldwide last year.
That is not a market. That is a monopoly.
On July 28, 2026, the Federal Communications Commission updated its Covered List to include foreign-produced mobile robots—humanoids, quadrupeds, and connected power inverters—citing national security risks, according to Manufacturing Dive. The ban means no equipment authorization, no import, no sale. The International Federation of Robotics confirmed the restrictions apply to any mobile, ground-based robot that can operate away from a human operator with sensors, software, and connectivity for navigation or data collection.
The justification, laid out in a National Security Determination document cited by The Register, points to vulnerabilities like the UniPwn flaws that let attackers take over Unitree's humanoid robots. The White House convened an interagency body of national security agencies to make the call, though the FCC declined to name which agencies or officials participated.
But the security argument is the curtain, not the play. The real story is industrial desperation.
Not who, but where
The ban targets where a product is made, not who makes it. The IFR noted that foreign-owned companies manufacturing in the U.S. are exempt. Hyundai-owned Boston Dynamics, which builds its machines stateside, can keep selling. A Chinese-owned factory in Ohio would pass. A U.S.-owned factory in Shenzhen would not.
This is a preemptive strike on the next generation of robotics. Industrial robots—articulated arms, SCARA, cartesian, parallel—are explicitly excluded. The FCC is not cleaning up the current factory floor. It is trying to wall off the future one.
The scope is narrow by design. The AP News report on July 29, 2026, confirmed the ban includes quadruped robots—robot dogs—alongside humanoids. Beijing immediately accused the U.S. of protectionism and promised "all measures necessary" to protect its businesses. The ban is the latest in a series: TikTok, Hikvision, Dahua, consumer routers, and now the machines that walk among us.
But here the sequence breaks. Those earlier bans targeted established products with domestic alternatives. The humanoid robot ban targets a market the U.S. does not yet have the factories to supply.
The trap inside the ban
The FCC's ban forces a painful reallocation of capital from Silicon Valley venture firms to domestic manufacturing infrastructure over the next 18 months.
The ban creates an immediate supply vacuum. U.S. startups—Agility Robotics, Figure, Apptronik—will announce plans to build domestic manufacturing facilities within 12 to 24 months. They have no choice. The conditional approval loophole the FCC left open is a temporary band-aid, not a strategy.
But scaling will be brutal. There is no domestic supply chain for the actuators, sensors, and compute modules that go into a humanoid robot. The ban covers finished robots, not components. A U.S. assembler can still import Chinese servo motors, Chinese lidar, Chinese motor controllers. The robot will be "American-made" in the same way an iPhone is "assembled in India"—final integration on domestic soil, guts sourced from the same Shenzhen ecosystem the ban is trying to escape.
This is the trap. The ban does not decouple the U.S. from the Chinese robotics supply chain. It obscures the dependency. Every U.S.-assembled humanoid will carry a bill of materials that traces back to the same factories in Guangdong and Zhejiang that produce components for Unitree and Fourier Intelligence. The difference is that those components will now arrive through intermediaries, with markup, without the quality feedback loops that come from direct supplier relationships.
Meanwhile, Chinese manufacturers will not stop building. They will pivot to Southeast Asia and Europe, where they will capture more market share and scale advantages. Every robot Unitree ships to Vietnam or Germany drives down unit costs through scale economics that U.S. startups, locked into a smaller domestic market, cannot match. The cost gap widens. The technology gap follows.
The global market bifurcates. A China-dominated low-cost segment keeps its 85% share and grows. A smaller, premium U.S.-made segment serves defense, critical infrastructure, and a handful of well-capitalized enterprises. Total U.S. humanoid robot adoption lags behind China by at least two years as supply chains struggle to scale.
This is the contrarian call: the ban will hurt U.S. robotics adoption more than it hurts China. The metric to watch is not the number of U.S. startups that announce factory plans. It is the unit cost curve of a Chinese humanoid versus a U.S.-assembled one in 2028. If the Chinese unit cost drops below $15,000 while the U.S. equivalent stays above $50,000, the ban will have failed.
The decoupling the U.S. is not ready for
China's industrial policy treated robotics the way Beijing once treated solar panels and EV batteries—dump capacity, crush margins, own the supply chain. The 12,750 units China shipped in 2025 were not accidents of the market. They were the output of years of state-backed scaling.
The U.S. response is a ban without a build. The CHIPS Act took two years to start printing checks. The robotics supply chain is more complex than semiconductors, with more moving parts, more materials, more firms that need to coordinate. An 18-month reallocation of venture capital to manufacturing infrastructure is the optimistic scenario. The pessimistic one is that U.S. robotics firms burn cash on factory retrofits while Chinese competitors iterate through a dozen hardware revisions in the same period.
The ban also quietly benefits one constituency: defense contractors. The national security framing ensures that the first large-scale domestic orders for humanoid robots will likely come from the Pentagon, not from Amazon or Walmart. That shapes the product. A robot built for a DOD logistics contract has different requirements than one built for a Foxconn assembly line. The U.S. humanoid robot may be born in a defense acquisition program, and it will carry that DNA—ruggedized, secure, and expensive.
What this means for operators
For manufacturers waiting on humanoid robots: expect a two-to-three-year window of scarcity. Plan for manual processes to persist longer than your automation roadmap assumes.
For venture capital: stop funding software-first robotics plays. The bottleneck is no longer perception stacks or manipulation algorithms. It is physical production capacity. The next fundable robotics company is not the one with the cleverest control system. It is the one with a lease on a factory floor and a supply agreement with a domestic actuator supplier that does not yet exist.
For startups: the path to market now runs through domestic production. The conditional approval loophole is a bridge to nowhere if you do not have a manufacturing plan behind it.
The starting gun
China shipped 12,750 of the 15,000 humanoid robots sold last year. The ban does not change that number. It means those robots will not come here.
The real race is not about security. It is about who can build the assembly lines first. The starting gun just fired. The U.S. is still looking for the track.
