Why did the optical communication sector fall so badly? Disassemble Rosenblatt's latest report and discuss my own judgment Where do market concerns come from? The main reason is the combination of three factors: One reason is that the market is concerned that Chinese manufacturers are rapidly expanding their production capacity of indium phosphide (InP) lasers, fearing a price war due to future supply surges. Secondly, there are rumors in the market that CPO (co packaged optics) may be delayed for about a year before large-scale deployment, which has made many people doubt whether the demand for optical communication in the next few years will also be delayed. Thirdly, many bears are not really bearish on the fundamentals of this track. They are just creating panic by taking advantage of these news and the gap period before the financial report is released. Even a considerable number of bears plan to gradually replenish their short positions before the financial report is released at the end of July or early August. Some people even plan to buy the entire sector again in 2027, laying out the CPO super cycle from 2028 to 2030 in advance. Is China's expansion of production a real threat? This is the most concerning issue in the market, but I think the worry is a bit excessive. Firstly, even if Chinese manufacturers continue to expand production, it is expected that there will still be a gap of about 40% between the global supply and demand of InP lasers by 2030, and the industry as a whole will still be out of stock. Expanding production will not lead to oversupply in the short term. Secondly, China's current main energy producers are low-power continuous wave lasers and 100G EMLs, with applications concentrated in relatively mature product lines such as silicon optical transceivers and 800G modules. However, the most profitable 200G EMLs and high-power continuous wave lasers in the next few years do not yet have mature mass production capabilities in China. These two are precisely the core light source devices of CPO and NPO, which means that European and American manufacturers will still have significant technological advantages in the high-end segment in the coming years. Thirdly, the supply chains between China and the United States are gradually decoupling. Due to political factors, supply chain security, differences in technological routes, and differences in customer certification systems, European and American customers are unlikely to switch to purchasing products from Chinese manufacturers. Currently, many products from Lumentum and Coherent have been scheduled for production until 2027 or even 2028, with many orders coming from Nvidia's pre lock orders. Has CPO really been delayed? I tend to believe that there is no substantial delay. CPO can be roughly divided into two stages: The first phase is small-scale construction, starting in the second half of 2026 and shipping in 2027. Currently, the progress is still progressing according to plan, and it is expected that Nvidia and Broadcom will each produce around 30000 to 350000 sets in 2027. The second stage is the true large-scale volume increase, which will take place in 2028. In 2027, NPO (Near Package Optics) will serve as an intermediate transition solution to meet a portion of the demand. Will changes in network architecture reduce the demand for optical modules? The market is also concerned about one thing, whether the emergence of new network technologies such as MRC and RNG will reduce the number of optical modules required for each switch. It is indeed possible that the number of GPUs will decrease from 6 to 7 to 3 to 4, but the growth rate of GPU count is much faster than this decrease, and overall optical demand is expected to continue to grow, which is an unnecessary concern. Highlights of Several Companies' Financial Reports Lumentum is facing the greatest financial pressure this time, as it has already provided the most aggressive long-term guidance in the market. The market is concerned about whether it can further raise its 2027 target, especially whether its quarterly revenue can reach $2 billion (annualized $8 billion) and whether its operating profit margin can exceed 40%. Coherent is a favored company among market institutions this time, as the proportion of self-produced 200G EML and high-power lasers continues to increase, coupled with the recent acquisition of Google's approximately $1 billion annual 1.6T module order and Nvidia's long-term procurement agreement from 2027 to 2030, the profit margin is expected to continue to improve in the future. AAOI's financial report this time may be a report card with both revenue and guidance exceeding expectations, mainly driven by 800G orders from Amazon and Oracle. It is expected that the revenue in the second quarter will exceed $200 million, and the guidance in the third quarter will be close to $300 million. The gross profit margin will also be significantly higher than the level of about 30% in the first half of the year. ANET's growth momentum remains strong, with major clients covering Meta, Microsoft, Oracle OpenAI、Anthropic、 Google, among which Google may become a major customer with a revenue share of over 10% in the future, is expected to recognize a large amount of deferred revenue starting from the second half of 2026, driving a revenue growth rate of over 30% in the second half of the year and over 40% in 2027. The core conclusion of this report In summary, China's expansion of production is indeed a real phenomenon, but it is not enough to change the global supply and demand pattern in the short term. The 200G EML, high-power lasers, and CPO related products that truly have technological barriers are still leading among European and American manufacturers. 2028 is still the time point for CPO to truly increase production on a large scale, and 2027 will be taken over by NPO to meet transitional demand. At present, the current pullback in the optical communication sector is mostly driven by emotions and expectations, and the fundamentals have not deteriorated. The upcoming financial reporting season from late July to early August will be an important window to verify whether demand in this track remains strong. After reading this report, I would like to add my own judgment I believe that short-term shortages and future structural surpluses are not contradictory (we are very optimistic about 2026 and 2027, but we should pay attention to them after 2028). These two things can be established simultaneously, and the key is to see where the surplus will first appear. Now the entire industry has seen the demand for AI and is actively expanding production. Cloud vendors also hope to introduce more suppliers to reduce costs and diversify supply chain risks. If there are more and more suppliers of 800G and 1.6T modules in the future, and the decline rate of ASP starts to be faster than the cost decline rate, even if the shipment volume and revenue continue to grow, the gross profit margin may be under pressure first. However, from the current perspective, the industry is far from reaching this point. The information disclosed by various companies still points to tight supply: high-end laser production is scheduled for the next few years, backlog orders remain high, export restrictions, yield rates, customer certification, and testing capabilities are still the main bottlenecks, and what is truly lacking is qualified high-end production capacity. If the industry enters a supply-demand rebalancing in the future, I believe that the first to face price pressure is likely to be the highly standardized optical module segment, such as 400G, gradually maturing 800G, and 1.6T products entering the mainstream stage in the future. Manufacturers such as AAOI, Zhongji Xuchuang, Xinyisheng, and Source Photonics, which mainly focus on optical modules, need to pay more attention to changes in ASP and gross profit margin in the future. However, this is a risk that is more worth paying attention to after 2028. In my opinion, the industry will still be in a stage of supply shortage from 2026 to 2027, and companies such as AAOI will still benefit from order release, capacity ramp up, and performance realization. In contrast, I am more optimistic about high-end component companies with core device barriers, such as Lumentum（LITE）、Coherent（COHR）； And system layer companies such as Ciena (CIEN) and Nokia (NOK), which rely more on technology, network architecture, and system integration capabilities to compete. As for equipment and testing companies such as Veeco (VECO), Aehr (AEHR), Viavi (VIAV), etc., as long as the industry is still expanding, they are expected to continue to benefit from the capital expenditure cycle. On the timeline, my judgment is basically consistent with this report: from 2026 to the first half of 2027, there will still be a shortage of supply; From the second half of 2027 to 2028, as new production capacity is gradually released, the market will begin to reprice the supply-demand relationship. It is too early to discuss the overall overcapacity in the optical communication industry. In the next few years, different products, technology routes, and companies are likely to enter completely different supply and demand cycles, and the entire industry will not experience oversupply together.