FCC Optical Module Restrictions May Be Phased In Starting at 3.2T; 65% U.S. Content Threshold Benefits Laser Suppliers
nashnova research
Morgan Stanley learned from Washington meetings that the FCC's restrictions on Chinese-made optical modules will likely phase in starting with 3.2T, with action possible as early as October; the 65% U.S.-content threshold means the real squeeze falls on laser sourcing, not overall module share.
What exactly does this rule restrict?
The FCC is more likely to use its Covered List mechanism, phasing in restrictions starting with 3.2T optical modules, with action possible as early as October.
The key threshold: Chinese-assembled modules can still be imported as long as 65% of bill-of-materials value comes from U.S. companies.
This means → the rule does not shut Chinese module makers out; it requires them to use more American core components inside each module — especially lasers and DSPs.
This is still at the policy-discussion stage. No formal rule text has been published; threshold percentages and phasing could change.
Why is the near-term revenue impact limited?
AI data centers are currently deploying 800G and 1.6T at scale. Morgan Stanley expects 3.2T ramp to begin only in 2028, with broader adoption around 2029.
In plain terms = the rule targets 3.2T — like setting emissions standards for a car that won't ship for two years. Today's products are untouched.
Lumentum previously told the FCC the same thing: 60%–65% of modules come from Chinese makers today; an abrupt switch would severely disrupt supply-demand balance.
This reflects a roughly two-year preparation window for the supply chain. Module makers' 2026–2027 revenue remains anchored to 800G and 1.6T.
How does the 65% threshold work — and whose pricing power shifts?
Morgan Stanley's specific combination: Marvell's DSP plus a Lumentum or Coherent laser could push a Chinese-assembled module past the 65% U.S.-value threshold.
This means → DSP and laser account for the bulk of a module's value; all other components and assembly combined make up only about a third.
For Chinese module makers such as Zhongji Innolight (中际旭创) and Eoptolink (新易盛): they need not surrender large share, but the room to switch to Chinese-made lasers in U.S.-bound 3.2T products shrinks sharply.
In plain terms = Chinese module makers used to hold a bargaining chip — "we can swap in a domestic laser." That card is now largely unplayable, and pricing power tilts toward Lumentum and Coherent.
Why not simply require U.S.-made modules?
Applied Optoelectronics is the only U.S. module maker with meaningful scale. Its 800G/1.6T monthly capacity was close to 200,000 units at Q2-end, with plans to reach roughly 650,000 by year-end 2026.
Zhongji Innolight and Eoptolink's combined annualized capacity in H1 2026 is roughly 101 million units — more than 8× Applied Optoelectronics.
This means → the U.S. simply lacks the domestic capacity to replace Chinese assembly. "Raise U.S. content" rather than "switch manufacturers" is the only realistic path.
Applied Optoelectronics management expects demand to exceed capacity through mid-2027, meaning new capacity is already absorbed by existing orders — leaving even less room for redirected demand.
What does this mean for investors?
Morgan Stanley's judgment: the rule's transmission runs stronger through high-speed laser pricing and Lumentum/Coherent component share than through any broad reshuffling of module share.
For U.S. hyperscalers: they will likely keep buying from the same module suppliers, with no supply disruption. What changes is the source of lasers and DSPs inside each module.
Morgan Stanley rates both Lumentum and Coherent "Equal-weight" and did not adjust ratings or price targets, but characterized the rule's direction as favorable to both.
This reflects a more immediate impact: reducing the competitive uncertainty Chinese laser makers posed for next-generation products — whether this plays out ultimately depends on when the FCC's formal rule text lands and its exact scope.
市场有风险，内容仅供研究参考，不构成投资建议。