Unitree Robotics was added to the Pentagon's Chinese Military Companies list in February. By that evening, it was gone.
The brief listing was not the policy. It was the signal that the policy had already begun.
In December 2025, House Select Committee on the CCP chairs formally asked the Pentagon to add Unitree to Section 1260H of the National Defense Authorization Act, citing documented evidence that Unitree had sold rifle-equipped robot dogs and reconnaissance models to the People's Liberation Army — systems documented in the Golden Dragon 2024 China-Cambodia exercise. The request named Unitree as one of Hangzhou's six flagship "Little Dragon" enterprises, operating in an explicitly designated military-civil fusion zone. The Pentagon acted on it — briefly.
The procurement door did not wait for legislation to begin closing. Federal acquisition regulations give agency lawyers and contracting officers guidance without waiting for a bill to become law. Agency lawyers and contracting officers reading regulatory signals began self-sorting away from Chinese robotics vendors before any congressional vote — and before the press release. A documented security vulnerability adds texture to that self-sorting: researchers disclosed in September 2025 that the Unitree G1 uses a single hardcoded AES encryption key across all units to protect its BLE-based Wi-Fi setup, a flaw that would allow an attacker within radio range to take complete control of any Unitree robot and propagate the compromise to others in range. The Cotton-Schumer bill cited national security concerns over data collection and potential remote access risks — concerns that the IEEE Spectrum reporting on the UniPwn exploit rendered concrete.
Unitree filed its Shanghai IPO the week before Senators Tom Cotton and Chuck Schumer introduced the American Security Robotics Act. The prospectus — the document that determines what Shanghai investors think this company is worth — was written before the February listing episode and before the March bill introduction. The policy landscape shifted between the filing and the publication.
Revenue grew 335 percent last year, to 1.71 billion yuan. Net profit grew 674 percent. Gross margin on humanoids ran at 62.9 percent. Those are the numbers Unitree showed its investors. The question the filing could not answer — because it had not happened yet — was what happens when the world's largest single buyer of advanced robotics hardware decides it is done being your customer.
Unitree has shipped more than 30,000 quadruped robots since 2022, according to its prospectus, and more than 4,000 humanoid robots. Its investor list includes Sequoia China, Tencent, Alibaba, Ant Group, and China Mobile. The company is expanding into Southeast Asia, Europe, and the Middle East. Its pricing on the G1 — from CNY 85,000, versus USD 30,000 to over USD 130,000 for comparable overseas products — is genuinely competitive in a way that makes the cost argument for international buyers. Those are real strengths. None of them require the U.S. federal market to exist.
Agibot, Unitree's Shanghai-based rival, is targeting its own Hong Kong listing in the third quarter of 2026, with a valuation of $5.1 billion to $6.4 billion. Both companies are raising public capital at the same moment the world's most scrutinizing buyer is drawing a line. The difference between a strong IPO and a weak one may be which side of that line the numbers are standing on when the door closes.
The procurement door was already closing before the bill was introduced. The question for investors is whether the growth trajectories are priced with that trajectory in mind.
The Cotton-Schumer bill would codify and expand the framework already in motion — extending potential prohibitions from humanoids to all unmanned ground vehicles. It has not passed. The February listing did not hold. But the direction of travel is consistent across regulatory filings, procurement signals, and congressional action, and it points in the same direction for every allied government that has looked at Chinese robotics vendors and decided the data risk outweighs the price advantage. The IPO math depends partly on markets that are progressively harder to access. The prospectus does not resolve that question.
