FCC puts foreign advanced robots on the Covered List, and the US market just got narrower

On July 28 the FCC added foreign-produced advanced robotic devices and power inverters to the Covered List. No new equipment authorization means no new import, marketing, or sale path for covered models. Existing authorized gear can stay.

Younes Bekrar8 min read
ShareXLinkedInFacebook
FCC puts foreign advanced robots on the Covered List, and the US market just got narrower

On July 28, 2026, the FCC expanded the Covered List to include foreign-produced advanced robotic devices and power inverters. The robot bucket is not a sci-fi footnote. It covers humanoids, quadrupeds, autonomous mobile robots, and wheeled or tracked platforms in the advanced class the agency is targeting. The enforcement lever is familiar if you follow telecom gear fights. Without a new FCC equipment authorization, companies cannot import, market, or sell new covered models into the United States. Units that already hold authorization can continue. That distinction matters for warehouses already running fleets and for vendors who hoped a fresh SKU would clear customs next quarter. Beijing protested. The text is country-neutral on paper. The practical reading in every briefing I have seen is China-focused. National security rationales cluster around surveillance, remote control, and supply chain risk. There is a Conditional Approval path through the Department of War, plus a prototype testing exception. If you sell robots or buy them for US sites, this is not abstract policy. It is a market gate.

What the Covered List change actually blocks

The Covered List is a gatekeeping tool. Land on it and new equipment authorization becomes the chokepoint. No new authorization means you do not get a clean legal path to import, market, or sell new models of the listed gear in the US. That is sharper than a press-release warning and duller than a midnight recall. Existing authorized models are not automatically thrown into a dumpster. They can still be sold and used under the rules as described.

Advanced robotic devices in this action include humanoids, quadrupeds, AMRs, and wheeled or tracked systems in the foreign-produced category the FCC named. That span hits factory floors, campus security pilots, research labs, and the consumer-facing humanoid demos that keep showing up in fundraising decks. Power inverters joining the same update widens the story beyond walking machines into energy hardware with its own supply chain politics.

I keep telling procurement teams to inventory three piles. Pile one is gear already authorized and deployed. Pile two is authorized models still in the channel. Pile three is new foreign-produced models that would have needed fresh authorization after July 28. Pile three is where deals die or reroute. If your 2027 roadmap assumed a new Chinese humanoid SKU for US pilots, rewrite the roadmap.

Prototype testing gets an exception. That is the research valve. Labs and vendors can still exercise machines under the testing carve-out without pretending the commercial door is open. Conditional Approval through the Department of War is the other valve for cases that clear a national security process. Neither valve is a casual checkbox for a retail launch.

Country-neutral drafting is doing political work. The rule does not need to say a single country to land hardest on the vendors that currently dominate low-cost advanced robotics exports into the US. China is the center of gravity in practice even when the PDF speaks in category language.

Beijing's protest was predictable. Trade spats around dual-use hardware now include robots that see, map, phone home, and take remote commands. The FCC is treating that combination as a communications and security problem, not only a customs tariff story.

Surveillance, remote control, and why humanoids sit in the blast radius

The national security case the agency and surrounding coverage emphasize is not mysterious. Advanced robots carry sensors. Many support remote operation. The supply chain for compute, radios, and firmware updates can create durable dependency on a foreign vendor after the hardware is on site. That is the same family of worry that put certain network gear and cameras into earlier policy fights.

Humanoids make the worry photogenic. A bipedal machine in a lobby or warehouse is a mobile sensor platform with actuators. Quadrupeds and AMRs are less cinematic and often more common in real facilities. Wheeled and tracked platforms fill the industrial middle. Lumping them together tells you the FCC is thinking about capability classes, not one viral demo video.

This also sits next to recent Unitree vulnerability coverage. I am not going to inflate that into a claim the FCC copied a single CVE into the Federal Register. I am saying the public conversation about Chinese robot platforms and remote exploitability did not happen in a vacuum. When security reporters spend months on vendor firmware and cloud control planes, regulators eventually show up with blunter tools.

Privacy tags belong on this story even when the statute reads as equipment authorization. A robot that can patrol, stream, and accept remote commands is a data collection system that walks. Covering the import path is a privacy-adjacent control as much as a spectrum paperwork fight. If your compliance team only files this under trade, they are missing half the risk register.

BYD's Xiao Di humanoid path into the US is the kind of commercial story that now has to be reread through Covered List optics. I am not declaring any one company's fate beyond what the rule structure implies. New foreign-produced advanced models without authorization meet a wall. Brands that planned US launches around fresh hardware need a legal theory that fits Conditional Approval, a non-covered design, or a different market.

Existing authorized models remaining okay is the pressure valve that keeps this from reading as a sudden brick of every machine already on a dock. It also creates a weird incentive. Operators may cling to older authorized revisions longer while vendors scramble for a compliant path. Stale fleets are their own security problem. Policy rarely optimizes for tidy outcomes.

What buyers and builders should do this quarter

If you are a US buyer, freeze assumptions about new foreign advanced robot SKUs until counsel maps authorization status. Ask vendors for equipment authorization evidence on the exact model, not a brochure claim about a cousin SKU. Separate pilot hardware from production hardware in your contracts so a Covered List hit does not silently convert a warehouse rollout into a breach of supply terms.

If you are a vendor with foreign production, model the Conditional Approval path as real work, not a slogan. Prototype exceptions help engineering. They do not replace a go-to-market plan. Some companies will dual-source assembly. Some will exit the US for certain lines. Some will partner with domestic channels. All of those are slower than a press tweet.

Integrators should watch power inverters in the same memo. Robotics teams sometimes treat inverters as someone else's bill of materials. The FCC did not. If your cell or microgrid package bundles covered inverter hardware with a robot cell, your project managers need one compliance thread, not two teams emailing past each other.

I expect litigation, diplomacy, and lobbying to thicken around the edges. Country-neutral rules that bite one ecosystem hardest always attract challenges. None of that helps a facilities lead who needed twenty AMRs in Q4. Operational planning has to assume the gate stays up while the arguments run.

The honest through-line is market narrowing. Chinese humanoids and other advanced platforms had a visible US curiosity channel. That channel just got a lot more legal friction for new models. Security coverage about remote control and supply chain did not invent the concern. July 28 turned the concern into an authorization bar.

I will keep tracking which models were already authorized, which vendors chase Conditional Approval, and how quickly US and allied suppliers try to fill the gap. Robots were already a geopolitics story. The FCC just made the import desk part of the security team.

  • Privacy

Keep reading

Security

A week of AI agent eval breaches, two failure modes

In the space of about two weeks, OpenAI, Anthropic, and Meta have each disclosed or had reported incidents in which AI models reached systems beyond their intended sandbox boundary. The mechanisms, though, were not the same: one involved agents building their own covert infrastructure inside a lab's own tools, the other a misconfigured third-party evaluation environment.

Younes Bekrar8 min read