Trump Plans to Ban Chinese Optical Transceivers from US Data Centers
The Trump administration is drafting plans to bar Chinese-made optical transceivers from US data centers by end of 2026. Here's what that means for AI infrastructure.
The Trump administration is drafting plans to bar Chinese-made optical transceivers from US data centers by the end of 2026, according to Reuters, which cited four people familiar with the matter. Optical transceivers are the devices that transmit data over fiber-optic cables inside large data centers, and they have become critical to connecting the thousands of AI chips needed to train models like ChatGPT and Claude. The Federal Communications Commission could still change or drop the restrictions entirely.
What happened
| Detail | Fact |
|---|---|
| Target component | Chinese-made optical transceivers |
| Planned deadline | End of 2026 |
| Regulating body | Federal Communications Commission (FCC) |
| Source | Reuters, citing four people familiar with the matter |
| White House response | Did not respond to a request for comment |
Administration officials are drawing up plans to prohibit imports of Chinese optical transceivers, the devices that convert electrical signals into pulses of light and back again to move data through fiber-optic cables at high speed. Inside modern AI data centers, these components provide the optical links between thousands of AI chips that work together to train large language models.
The FCC has not finalized anything. The sources who spoke to Reuters noted the agency could still modify or abandon the reported restrictions before they take effect.
Why does the US want to ban these components?
US officials have raised two distinct concerns. The first is cybersecurity. American intelligence agencies have documented Chinese state-backed hackers accessing critical US infrastructure and consumer devices, which officials say demonstrates a capability to collect intelligence or disrupt key systems during a future crisis.
Adam Segal, director of the Council on Foreign Relations’ Digital and Cyberspace Policy Program, told Newsweek that the transceiver concerns center on “data collection, disruption, and long term dependence on Chinese suppliers.”
The second concern is supply chain leverage. Dmitri Alperovitch, chairman of the Silverado Policy Accelerator think tank in Washington, told Newsweek: “Banning imports of Chinese-made optical transceivers would further help secure the AI supply chain and reduce dependencies on China that they are capable of weaponizing against us, as they have repeatedly done with rare earths.”
China tightened export controls on rare earth minerals during last year’s trade tensions with the US, and more recently restricted exports to Japan following Prime Minister Sanae Takaichi’s defense buildup and comments about Taiwan. That pattern of using export controls as geopolitical leverage is driving the push to reduce reliance on Chinese components across the AI supply chain, not just chips.
Why it matters
Until now, the US-China technology competition has focused heavily on advanced semiconductors. Export controls already restrict the sale of high-end AI chips to China. This move signals the battleground is widening to cover the physical infrastructure that runs AI workloads, including components that are currently sourced largely from Chinese manufacturers.
For businesses that depend on cloud computing or are building out AI infrastructure, the downstream effects could be real. Segal warned that the ban “raises costs and uncertainty for hyperscalers,” referring to companies like Amazon that run massive cloud infrastructure. Higher component costs or supply disruptions would likely filter through to cloud pricing over time.
China’s embassy spokesperson Liu Chang called the restrictions “market-distorting acts of bullying” and urged the US to revoke them. Beijing has consistently framed these kinds of controls as damaging to global supply chains.
The broader pattern here connects directly to the open-weight AI safety debate and the question of where AI infrastructure sits geographically. As we covered in our look at the White House AI cybersecurity framework, the US government is making a series of interconnected moves to tighten its grip on the full AI stack, from models to the hardware that runs them.
Our take
The security logic is sound. A data center full of hardware with potential remote-access vulnerabilities is a real risk, and supply chain dependence on a strategic rival is a documented problem. The rare earth playbook China has already run makes the concern concrete, not hypothetical.
The cost question is also real. Optical transceivers are not exotic custom chips. They are commoditized components, and Chinese manufacturers have competed aggressively on price. Replacing them at scale, inside data centers that are already being built at record speed to meet AI demand, adds procurement complexity and cost pressure at the worst possible moment for the industry.
For most businesses, this will play out invisibly in cloud pricing rather than as a direct procurement headache. If you are a company running significant workloads on hyperscaler infrastructure, this is worth tracking. If you are planning a private or on-premises AI deployment, the component sourcing landscape may look different by the time you’re ready to buy hardware.
Our AI integration work increasingly involves helping clients think through infrastructure decisions, not just which model to use. Supply chain resilience is becoming part of that conversation.
What to do about it
- Monitor FCC announcements for any formal rulemaking on optical transceiver import restrictions.
- If you procure data center hardware directly, ask your supplier about the origin of optical transceiver components in their equipment.
- Factor potential cloud price increases into AI infrastructure budgets planned for 2026 and beyond.
- Track how hyperscalers respond publicly to this policy, as their procurement shifts will signal how quickly costs move.
The ban is not final, but the direction of travel is clear: the US is expanding its technology controls well beyond chips.
Frequently asked questions
What are optical transceivers and why are they used in data centers?
Optical transceivers are devices that convert electrical signals into pulses of light and back again to transmit data through fiber-optic cables. Inside AI data centers, they provide the high-speed links connecting thousands of AI chips needed to train large language models.
When would the Trump ban on Chinese optical transceivers take effect?
According to Reuters, administration officials are targeting the end of 2026. However, the FCC could still modify or abandon the restrictions before they are finalized.
Why does the US want to ban Chinese optical transceivers specifically?
US officials cite two concerns: cybersecurity risks from components that could enable data collection or disruption, and supply chain dependence on a strategic rival that has previously used export controls on rare earths as geopolitical leverage.
How would a ban on Chinese transceivers affect cloud computing costs?
Analysts warn the ban could raise costs and create uncertainty for large cloud providers (hyperscalers) like Amazon in the short term, which could eventually affect the pricing of cloud services businesses rely on.
