CITIC SEC: Clear trend in the AI industry, optimistic about the optical communication sector.

date
09:50 27/01/2026
avatar
GMT Eight
CITIC Securities is optimistic about the development potential of leading domestic optoelectronic communication companies.
CITIC SEC released a research report stating that the demand for AI computing power has surged, driving the upgrade of the optical communication industry. Overseas cloud vendors and TSMC continue to expand their capital expenditures, confirming the strong demand for AI infrastructure and resulting in a flourishing demand for high-speed optical modules. Although there is a short-term gap in the supply of high-speed optical chips and other materials, upstream manufacturers are actively expanding production, and the increasing penetration of silicon photonics solutions is expected to alleviate supply chain bottlenecks. In terms of technological iteration, NPO, with its advantages of low power consumption, high bandwidth, and maintainability, is becoming a key transitional solution at the Scale up level. Leading companies have already made strategic moves, driving the industry towards a more efficient optical interconnect architecture. The industry believes that as technological advancements progress rapidly, leading companies can leverage their technological accumulation and innovation capabilities to continuously strengthen their technological advantages and lead the industry in upgrading directions, with a focus on recommending domestic and foreign leading companies in optical communications. The communication industry maintains a "stronger than the overall market" rating. CITIC SEC's main points are as follows: The high prosperity of the AI industry continues, with expanding capital expenditures confirming strong demand. Recently, major manufacturers such as OpenAI and Google have accelerated the iteration of large models, such as GPT-5.2 and Gemini 3.0 Pro, achieving breakthroughs in multimodal and reasoning capabilities. According to the financial reports of various manufacturers, the capital expenditures of the top four overseas cloud vendors (Google, Microsoft, Meta, Amazon) in 2025Q3 increased by +74% year-on-year, and collectively revised upwards their guidance for 2026, highlighting their long-term confidence in investing in AI. TSMC's revenue in 2025Q4 was $33.7 billion, up 25.5% year-on-year; the mid-range of revenue guidance for 2026Q1 reached $35.2 billion, with a gross margin increase of +1.7pcts to 64%; the full-year CAPEX guidance for 2026 increased by 27%-37% to $52-56 billion, further validating the strong demand for terminal AI chips, with optical modules as core components of computing networks to directly benefit. Upstream optical chip manufacturers are actively expanding production, and the tight supply situation of materials is expected to ease. In its FY2026Q1 financial report conference call, Lumentum stated that due to a short-term shortage of high-speed EML laser chip capacity, the gap had widened to 25%-30% in 2025Q3, but overseas leading manufacturers of upstream optoelectronic chips have taken proactive measures to expand production: Lumentum stated that the production capacity of its indium phosphide wafer factory in Thailand had increased by 40%, with the proportion of high-speed rate EML optical chips continuing to rise, with orders locked in until 2027; in addition, the increasing penetration of silicon photonics solutions is expected to alleviate material shortages. According to FY2025Q3 financial reports, Tower Semiconductor added $300 million to expand production of silicon photonics chips, with factories in Israel's Fab 2 and the U.S.'s Fab 9 continuing to expand production, with production capacity expected to triple in the second half of 2026. As production capacity for high-speed EML chips is gradually being released, combined with the accelerated penetration of silicon photonics solutions, the tight supply of high-speed optical module materials is expected to gradually ease in 2026. NPO technology can meet various needs and is expected to become the direction of evolution for the next generation of optical interconnects. Copper interconnects are facing physical bottlenecks at the Scale up level of AI clusters, gradually transitioning to high-speed rate optical interconnection solutions. Compared to traditional pluggable optical modules, NPO (Near Packaging Optics) solutions reduce power consumption by 20%-30%, increase port density by over 50%, and are more suitable for short-distance high-speed applications; compared to CPO (Co-Packaging Optics) solutions, maintainability is greatly improved, and compatibility with existing ecosystems makes resistance to adoption lower. Currently, both Broadcom and NewPhotonics have launched NPO solutions, with domestic manufacturers leading in the silicon photonics NPO field and expected to achieve bulk commercialization by 2027. The rapid iteration of optical communication technology will further solidify the market position of leading enterprises. Investment Strategy Against the backdrop of the sustained high prosperity of the AI industry, optical interconnects are gradually replacing copper connections, becoming the key to achieving high-performance, high-bandwidth, low-latency AI networks. Among them, the emerging optical module market at the Scale up level is in a high-speed iteration stage, with NPO, CPO, and other solutions evolving simultaneously. In the long term, CPO remains the industry's major trend, but in the short term, factors such as industry maturity, NPO is gradually becoming the main battlefield for the next generation of applications. The industry believes that as technological advancements progress rapidly, leading companies can leverage their technological accumulation and innovation capabilities to continuously strengthen their technological advantages and lead the industry in upgrading directions, with a focus on recommending domestic and foreign leading companies in optical communications. The communication industry maintains a "stronger than the overall market" rating. Risk Factors AI investments fall short of expectations; intensified market competition; delays in the progress of new products such as 1.6T; technological route risks; geopolitical risks; and rapid release of production capacity.