The Trump administration is preparing new restrictions on Chinese-made equipment used inside American data centers, extending Washington’s technology controls to the infrastructure powering the artificial intelligence industry.
The Federal Communications Commission is drafting a measure that would prevent imports of new Chinese optical-transceiver models into the United States. These components transmit data through fiber-optic cables and connect the servers, processors and networking systems operating within data centers.
FCC Could Publish Restrictions This Year
Four people familiar with the discussions shared that officials hope to publish the FCC measure in 2026, when it would also take effect. However, the sources cautioned that regulators could still revise or abandon the proposal before it becomes final.
Reuters said the administration wants to prevent Chinese companies from using data-center equipment to steal information, install malicious software or interrupt services. The facilities contain the advanced chips used to train AI models and deliver their services to users, making their internal hardware increasingly important to national-security planning.
The Next Web noted that enforcement may be complicated by equipment sold through resellers, subsidiaries or relabeled supply chains. The publication said drafting effective rules may require closing those potential routes rather than merely identifying prohibited manufacturers.
Washington Seeks to Avoid Another Huawei Problem
Officials want to prevent Chinese equipment from becoming deeply embedded in critical American infrastructure, as Huawei telecommunications products previously did. Removing Huawei equipment has proved slow, costly and incomplete in parts of the US communications system.
Divyansh Kaushik, an AI policy expert at Beacon Global Strategies, shared that transceivers present a security risk and that the supply chain should be protected as the data-center expansion accelerates.
The FCC has already imposed similar restrictions involving Chinese drones, routers, robots and power inverters. Under the proposed approach, regulators could initially prohibit all new transceiver models before exempting many products supplied by non-Chinese companies.
Innolight Could Face the Largest Impact
The restriction would likely affect Zhongji Innolight, one of the world’s leading data-center transceiver suppliers.
Reuters reported that Innolight controls approximately 27% of the global market and generates 90% of its revenue outside China. The Pentagon added the company to its list of businesses allegedly backed by China’s military in June, although placement on that list does not itself impose an import ban.
The proposed restriction could increase expenses for American cloud providers such as Amazon Web Services by forcing them to purchase equipment from alternatives including Coherent and Lumentum. A Foundation for American Innovation report cited by Reuters said the two US-based suppliers offer competing technology but currently lack the scale needed to replace Chinese vendors completely.
The Next Web similarly warned that excluding cheaper Chinese equipment could raise costs or delay construction when US operators cannot secure enough comparable products at the required price and volume.
China Threatens a Response
The Chinese Embassy in Washington shared that the United States should stop threatening Chinese companies with sanctions and listen to business communities in both countries. Beijing said it would take necessary measures if US action caused material harm to Chinese interests.
The proposal reflects a broader effort by both Washington and Beijing to gain domestic control over the machinery supporting artificial intelligence. As each country introduces new restrictions, technology supply chains built across decades are being separated one product category at a time.
The draft ban remains subject to change, but its direction is clear: the US government increasingly views even small networking components as strategic assets when they sit inside the data centers powering the AI economy.
