Read the two sentences back to back and the whole policy comes into focus.
The first is from the government's own security determination: importing foreign-controlled, high-autonomy mobile robots could become "akin to importing foreign soldiers and deploying them in American homes and factories."
The second is buried in IEEE Spectrum's analysis of the same rule: "Because China also produces a large fraction of robot components, even for robots assembled in the United States, it will have strong leverage in any related negotiations until U.S. robotics companies further diversify their supply chains."
Those two claims cannot both hold and still add up to what Washington says it bought. If the foreign soldier is the finished robot, you can turn it away at the port. If the foreign soldier lives in the actuators, batteries, and sensors, then you're still importing him — you're just letting an American company screw a chassis around him first.
If the foreign soldier lives in the actuators, batteries, and sensors, then you're still importing him — you're just letting an American company screw a chassis around him first.
That gap is the story. On July 28, 2026, the FCC added "advanced robotic devices" — mobile, networked robots reportedly over 4.4 pounds, per ien.com — to its Covered List, the same lever it pulled on DJI drones and Huawei gear. New foreign-produced models can no longer get the equipment authorization required to reach the U.S. market.
The reactions arrived within the week, and every one of them was a mirror.
Agility Robotics called it a step that "rewards companies that have invested in building trusted domestic supply chains from the ground up" and ensures its Digit robot, "engineered and assembled here in the US, can be deployed safely, securely, and at scale."
The top reply under that post: "so why don't you ship anything then?"
Boston Dynamics' policy VP Brendan Schulman was more expansive on LinkedIn: "I sense that this is just the first round in a series of policies that will define the success and growth of the industry for decades to come."
And then there was Faraday Future, which turned the ban into a press release about itself.
In its 66th weekly investor update, CEO YT Jia announced that FF EAI Robotics shipped 152 units in July — "another monthly record" — bringing the year to 394 units against a full-year target of 2,000.
Read that again. The company celebrating a national-security moat sits at 20 percent of its own annual shipment goal, in a market where global mobile-robot shipments reportedly crossed 200,000 units in 2025, per ien.com.
Jia's framing is nakedly self-interested and, on its own terms, correct: he lists "six key areas" of advantage, the sixth being that the policy will "accelerate the flywheel effect of FF's first-mover advantage." When you ship 152 robots a month, a rule that freezes your competitors' new models is the best product you have.
That is what a protected market looks like from the inside. The ban is real, immediate, and enforceable. What it protects is thinner than the cheering suggests.
Start with what "foreign-produced" actually means, because the whole industrial-policy case rides on it.
The rule restricts foreign-produced models, not models made by foreign entities. A U.S. company manufacturing abroad can be covered; a foreign-owned company manufacturing in America is not.
Then comes the loophole that eats the policy. For commercial-off-the-shelf products, the only test to escape "foreign-produced" status is U.S. assembly. No domestic-content floor.
The Foundation for American Innovation, which supports the ban, says the quiet part plainly: "Superficial final assembly (screwdriver assembly) should not qualify." Should not. There is, at present, no body that even determines whether a product qualifies as COTS for import-control purposes.
So the plausible near-term outcome is not a reshored supply chain. It's a relabeled one — imported bills of materials assembled stateside to clear the gate, exactly the "screwdriver plants" FAI is begging regulators to preempt.
The perverse twist is who gets punished. A standardized foreign-designed robot assembled in volume in America may qualify as COTS and skate through on assembly alone.
An early-stage American humanoid built in small quantities may be classified non-COTS, which triggers the Buy American standard: U.S. assembly plus 65 percent domestic content by cost. The genuinely American startup faces the harder test than the relabeled import.
This is the load-bearing contradiction. A rule sold as reshoring can, on its face, advantage a well-capitalized firm bolting together Chinese parts over a scrappy domestic one that hasn't yet found non-Chinese actuators.
And it hasn't, because there mostly aren't any. The determination's own evidence base — Unitree vulnerabilities, Chinese dominance of humanoid manufacturing — describes a component concentration in actuators, batteries, and sensors that the ban does nothing to relocate.
You cannot ban your way to an actuator supplier that doesn't exist. That is the exact leverage Beijing now holds.
China's Ministry of Commerce has already called for withdrawal and warned of countermeasures, framing the "non-discrimination" language as cover to "discriminate against and suppress Chinese enterprises." Counterpoint's Marc Einstein named the two cards: "further restrict rare earth sales to American companies and further restricting Chinese market access for American companies like Tesla and Nvidia."
Both cards are drawn from the deck the ban leaves in China's hand. Rare earths feed the magnets in the motors; market access is the leverage of a country that still makes the parts American robots are built from.
The ban tightens the same unit-economics vise that squeezes every humanoid maker — and it does reportedly dent Chinese IPO plans, per Einstein — but that math belongs to another conversation. What matters here is that the negotiating table was set by the U.S. and the strong hand was dealt to Beijing.
Which brings us to the reshoring question the cheerleaders keep skipping: can the United States physically build the capacity to replace what it just banned?
The most recent construction data says no, not on this trajectory. U.S. real manufacturing construction is reportedly down more than 25 percent from its 2024 highs, as CHIPS Act projects complete, Inflation Reduction Act projects get cancelled, and tariffs weigh on new investment, per Chris Paxton.
The one-line verdict from that data, from roboticist Chris Paxton: "We are not going to ban or tariff our way out of this, it needs large and sustained investment."
A ban is a switch. A supply chain is a construction schedule. You can flip the switch overnight; you cannot pour foundations, qualify battery lines, and stand up domestic actuator production on the same timeline — and right now the foundation-pouring is trending the wrong way.
The mechanics reinforce the delay. The rule bites only new models seeking authorization; already-approved robots, including Chinese ones like the Unitree G1 research platform, keep selling, with firmware updates reportedly allowed until at least January 2029, per ien.com.
So the immediate effect is not scarcity. It's a slow freeze on next-generation foreign hardware while the domestic base it's supposed to nurture shrinks in real dollars.
None of this means the security case is fake. A networked machine that senses homes, receives remote commands, and moves through the physical world is a genuinely different risk than a router, and the drone precedent shows why waiting until the platform is embedded is worse.
But a security gate and an industrial policy are different instruments. The FCC built the former and let everyone project the latter onto it.
Agility gets a quieter competitive market. Faraday Future gets a moat around 394 robots. Boston Dynamics gets "the first round." FAI gets a policy hook and a list of fixes it hopes regulators adopt before the screwdriver plants arrive.
What none of them get, on current spending, is the domestic actuator, battery, and sensor base that would make "Built in USA" mean more than the label on the crate.
Until that capital shows up — sustained, not seasonal — the ban relocates the sticker, not the leverage. And the leverage is still sitting in Shenzhen.