Last night, the optical interconnect sector collectively strengthened, and the core catalyst should be a research report on FCC and optical interconnect by Da Mo yesterday, which provided a detailed judgment that "if FCC ultimately adopts new restrictions on Chinese optical transceivers, policy pressure is more likely to start from the next generation 3.2T. The 800G and 1.6T that have already been deployed on a large scale are likely to leave a buffer. It is equivalent to saying that the previous trading in the market was a "ban on Chinese optical modules", and the framework provided by Da Mo's latest research has been refined into "3.2T intergenerational switching+an increase in the value of core components in the United States". Sort out a few key points: 1. Why might the United States choose the timing of 3.2T? At present, Chinese manufacturers account for a very high proportion of 800G and 1.6T optical modules. If the United States suddenly cuts off Chinese made 800G and 1.6T optical modules, the first to encounter problems will be the large number of optical modules required by the AI clusters of its own hyperscalers (AWS, Meta, Google, Microsoft). However, the domestic supply chain in the United States does not have enough capacity in the short term to make up for all Chinese supply. 3.2T is currently in the development, validation, and import stages. If the United States writes the rules into it now, cloud vendors still have time to redo supply chain certification, and optical module companies also have time to adjust BOM, production lines, and production locations. This is actually a very typical industrial policy window: changing the supply chain rules in advance before the next generation of products is truly mass-produced. 2. The most noteworthy aspect of this report by Da Mo is the "American Content" Da Mo believes that there may be a compliance path for the 3.2T policy in the future regarding the US value content, with the market circulating figure of 65%. Da Mo's judgment is that if a Chinese optical module extensively uses American DSPs and lasers, the American value of the core BOM itself may already be quite high. If a high-speed optical module is disassembled, there are roughly these things: DSP, Laser/EML/CW, TIA/Driver, SiPh/Optical Engine, PCB, connector, and back-end (packaging, coupling, assembly, testing). If the United States requires more core components from American or trusted supply chains for 3.2T in the future, the importance of DSP, Laser, TIA, and Driver will significantly increase. So there was a strong reaction from Lumentum and Coherent stocks yesterday. 3. Which targets benefit more according to the framework of Da Mo? This is the most worth watching, Da Mo emphasized a judgment this time: if the policy really starts from 3.2T, the impact on laser pricing may be more important than the redistribution of market share in the entire optical module market. 3.2T is entering the era of 400G/lane, and whether we choose EML or silicon laser+CW laser in the future, laser is a very core bottleneck. If the US policy adds another condition: it must come from the US or a trusted supply chain. That production capacity will be even more insufficient. 1）Lumentum Lite has been expanding its production capacity of optical chips and lasers in the United States this year. Including domestic InP manufacturing capabilities in the United States, high-power CW lasers, and lasers for the next generation of optical interconnects in AI data centers. If the FCC really establishes credible supply chain requirements around 3.2T, these production capacities will have an additional value: policy compliance value. 2) The logic of Coherent is similar. The characteristic of COHR is that the entire system is more vertical, covering many links such as InP, EML, CW laser, silicon photonics, transceiver, etc. It has also extended its product route to 3.2T. (Coherent Inc) So whether the policy is more focused on core components or more on complete modules, it has room for participation 3）AAOI AAOI has been expanding its domestic optical module manufacturing capabilities in the United States. If the FCC ultimately requires stricter regulations, including the location of the final assembly, the value of domestic module manufacturing capacity in the United States will significantly increase. So in the end, it depends on which level the FCC ultimately regulates? 1) If we mainly look at the sources of core components such as DSP and Laser, the logic of companies like LITE, COHR, Broadcom, and Marvell will be more direct. 2) If we continue to manage to the final assembly and production site, AAOI's strategic position will continue to improve. If future US policies require a higher proportion of American core components, Chinese module manufacturers may need to purchase more American DSPs in order to continue supplying American hyperscalers. 4. The core bottleneck is still: InP The United States can request more American lasers. The problem is that many high-speed lasers rely on InP, which is indium phosphide. The global supply chain of InP substrate is highly centralized. AXT is headquartered in the United States, but its production system has a deep relationship with the Chinese supply chain. That's also why Da Mo listed InP substrate supply as a highly observable variable when discussing the 3.2T FCC risk. 5. Looking back further 3.2T may just be the beginning, once the United States introduces the source of core components, BOM disclosure, production location, and supply chain certification on this generation of 3.2T products These rules are gradually established, and the subsequent CPO and Optical Engine can continue to execute along the same set of logic. We need to understand this FCC discussion as an important observation window: it may determine how the next round of supply chain division for AI optical interconnection will be carried out.