- Proposed Chinese transceiver import ban could squeeze AI data center supply chains and cloud buildouts
- Analysts warn optics shortages may spike costs for hyperscalers including Amazon, Microsoft and Meta
- FCC security push risks creating new hardware bottlenecks as AI infrastructure demand keeps surging
A move by U.S. officials to ban imports of Chinese-made data center transceivers would create a huge bottleneck for cloud companies and drive up already ballooning prices for AI infrastructure, analysts told Fierce.
Dell’Oro Group VP Jimmy Yu, who specializes in tracking optics for metro and long-haul transport networks outside the data center, told Fierce, “This is a terrible time to limit access to components in data centers.”
Yu pointed out that the industry is deep in the AI super-cycle and already dealing with shortages around sever key components.
“My understanding is that transceivers are currently in tight supply. Hence, any disruption in the supply chain will drive transceiver prices up sharply across the industry, not just limited to hyperscalers,” he said.
Why the transceiver ban?
Transceivers are critical components in data center networks, sending and receiving signals to transmit data across fiber optic cables. Chinese optical module manufacturers currently make up around two-thirds of the global transceiver supply and account for 60% of global revenue, according to data from Counterpoint Research.
Reuters reported that the White House and Federal Communications Commission (FCC) are mulling a measure that would block imports of Chinese-made transceivers. Officials are aiming to roll out a ban sometime this year, Reuters’ sources said.
Such a move would be consistent with recent FCC efforts to ban foreign-made communications equipment from certain vendors deemed to be a national security threat. The FCC notably clamped down on consumer-grade Wi-Fi routers in March and has long targeted mobile network gear from Chinese vendors Huawei and ZTE.
In an order passed last month, the FCC sought to extend the reach of its so called “Covered Entity” list, aiming to ban any “logic-bearing hardware components” produced by companies on the list.
“If the Commission prohibits the authorization of certain equipment produced by Covered List entities, the Commission should similarly prohibit authorization of any equipment that contains Covered List entity-produced logic-bearing hardware components, whether by SDoC or by certification,” the order reads. Optical transceivers were listed as examples of such components.
It is not clear whether the order and proposals therein are related to the ban Reuters reported.
Terrible timing for a transceiver ban
Counterpoint Research’s team noted a ban would create a “significant near-to mid-term operational bottleneck” for U.S. cloud companies like Amazon, Alphabet, Microsoft and Meta.
Here’s why: Yu said that Innolight, Coherent Corp., and Eoptolink are market leaders in the transceiver space. Both Innolight and Eoptolink are Chinese vendors. But Counterpoint said Coherent and fellow Western vendor Lumentum “currently lack the cleanroom capacity, automated packaging infrastructure, and yield scale required to absorb Innolight and Eoptolink's volume within a 12-to-24-month horizon.”
That means supply shortages and delay, which could, in turn, lead to delays in turning up AI data center clusters (and thus, delay revenue recognition) and increase costs for companies already facing inflated bills of materials. Amazon, for instance, recently had to ratchet up its 2026 capex forecast by $20 billion dollars due to rising memory prices related to shortages in that market.
“The global AI ecosystem remains heavily reliant on Chinese optical module vendors for scale execution,” Counterpoint concluded. “While policy initiatives seek to insulate critical AI supply chains, sudden regulatory shifts risk creating hardware bottlenecks that could slow down deployment schedules for the world's largest cloud operators.”
Read more about the bottlenecks AI data centers are facing here:
Amazon's AI infrastructure bill just got $20B bigger
Mobile ecosystem is facing major memory problems
Hyperscalers have a new build vs buy dilemma: electricity
AI power demand hits regulatory tipping point
Copper demand on the rise thanks to data center boom