FCC's Draft Ban on Chinese Optical Transceivers Is Simpler in Markets, Harder in Reality
ChinaTalk · rss · 2026-10-07
ChinaTalk dissects the FCC's draft ban on new Chinese optical transceivers and the supply-chain reality behind it.
Timeline
- Aug 4: Reuters reported an FCC draft barring import authorization for new Chinese transceiver models, citing data theft, malware, and disruption risks; targeting year-end effect but could be shelved
- Within a week, Innolight (world's largest vendor) fell 10% while Coherent rose 17%
- FCC previously restricted drones, routers, inverters, robots; ITI (representing hyperscalers) urged the FCC not to add transceivers to the Covered List
Technical breakdown
- Modules contain an InP laser, a DSP consuming half the power budget, TIA, MCU; passive optics are 10–20% of module BOM
- "Copper if you can, optics if you must": copper within racks, fiber between racks
- NPO/CPO architectures bypass DSP loss but cost more to service
Core argument
- Of the three claimed harms, only service disruption is plausible; espionage claims are unfounded — transceivers blindly convert bits to light, and hyperscalers encrypt
- China dominates the top of the stack but depends on Western firms in the middle; at the bottom (materials/metals), Chinese control is ingrained and an FCC rule can't shift it
- The Covered List doesn't cover the biggest Chinese vendors, so a production-location category rule would be needed — leaving many gaps
More from Infra
- Java Vector API: Writing SIMD Directly Since JDK 16 to Unlock Single-Core Performance — lemire · 2026-10-07
- Temp 0 doesn't guarantee deterministic LLM output — batch shape and kernels shift logits — JFPuget · 2026-10-07
- Open-source LLM reverses and rewrites Tuya Zigbee firmware in 20 minutes after Claude trips safeguards — ngxson · 2026-10-07
- cmpunlocker v0.5 Unlocks ECC + 4 SMs on CMP170hx: 1.5TB/s for Half a DGX Spark's Price — chemist_slime · 2026-10-07
- Musk denies slowdown: SpaceX accelerating AI data center buildout, exploring orbital compute — DimaZeniuk · 2026-10-07
- VEDA Sparse Attention cuts MiniMax H3 video gen time in half in ComfyUI with no visible quality loss — robomar_ai_art · 2026-10-07