US Moves to Ban Chinese Optical Transceivers from AI Data Centers

The US government is moving to purge Chinese-made optical transceivers from domestic data centers as part of a strategic effort to secure the underlying hardware of artificial intelligence. The Trump administration and the Federal Communications Commission (FCC) are drafting proposals to limit or ban the import of these critical components, with officials targeting the end of 2026 for the ban to take effect.

These optical transceivers are essential for high-speed data transmission within cloud networks and AI data centers, acting as the vital links that allow GPUs to communicate. The FCC aims to publish the formal proposal before the end of 2026. This regulatory push follows a broader pattern of security tightening; for instance, the Department of Defense added Zhongji Innolight to its 1260H list of Chinese military companies in June 2026.

The Battle for AI Plumbing

For years, the US-China tech conflict focused primarily on high-end semiconductor chips. However, the focus has now shifted to what experts call the "plumbing" of AI—the networking hardware that facilitates massive data flows. This shift is driven by a desire to reduce cybersecurity risks, specifically concerns regarding malware and the potential for data exfiltration through hardware backdoors.

Executing this decoupling is a significant logistical challenge because Chinese suppliers, including firms like Eoptolink and Innolight, currently hold a dominant share of the global optical transceiver market. The reliance on these vendors means that a total ban would require a systemic overhaul of how US data centers are built and maintained.

Industry Implications

Removing Chinese components from the supply chain will force US AI infrastructure operators to undergo a massive requalification of their suppliers. This transition is expected to be both slow and expensive. The potential ripple effects of an FCC ban demonstrate that any decoupling of US-China tech will be a costly endeavor.

While the ban creates significant headwinds for data center build-out speeds and increases immediate costs, it opens a substantial market window for non-Chinese manufacturers. Companies based in the US and Malaysia are positioned to fill the vacuum left by Chinese vendors as the US seeks to diversify its critical infrastructure.

What to Watch

Industry observers are now waiting for the FCC to release the formal proposal, which will clarify the exact scope of the restrictions and whether any exemptions will be granted for existing installations. The primary uncertainty remains the timeline for implementation and whether the logistical difficulty of replacing dominant Chinese suppliers will lead to modifications in the final regulatory language.