CICC's outlook for the communication equipment industry in the second half of 2025: Acceleration of AI commercialization, focusing on computing power and policy opportunities.

Zhongjin releases research report, stating that looking forward to the second half of 2025, with the continuous enhancement of large model capabilities and the increasing number of application scenarios, the demand for AI reasoning computing power is expected to continue to explode, driving the growth of AI hardware demand. Zhongjin believes the following trends are worth noting: 1) the certainty of the growth of domestic and foreign computing power chain demand is increasing; 2) the importance of networks in AIDC is expected to increase; 3) AI applications such as Agent are accelerating landing. In addition, the "14th Five-Year Plan" policy signals, digital infrastructure, marine economy, military communication, and other directions are worth paying attention to. 1H25 Review: As of August 8, 2025, the SW communication equipment sector has increased by 31.4% compared to the beginning of the year, outperforming the broader market. In the first half of this year, the computing power chain remains one of the main investment themes in the market, with fluctuations in the trends of domestic and overseas computing power chains, mainly affected by factors such as the release of DeepSeek R1, retaliatory tariffs, and the boost in North American AI expectations. Since mid-April, the capital expenditure of major overseas cloud service firms has exceeded expectations, gradually dispelling the pessimistic narrative of "computing power contraction", and the stock prices of core suppliers have hit new highs. The main investment themes for AI suggest focusing on three directions. 1) New hardware technologies: Zhongjin believes that as the certainty of AI application scenarios increases, the penetration rate of cost-effective and energy-efficient AI ASICs is expected to increase. The demand for high performance and low power consumption drives the acceleration of AI hardware technology iteration, with liquid cooling, silicon photonics, high-core count MPOs, and other technologies accelerating commercial deployment. 2) Localization: The restart of H20 delivery and the continuous improvement of the domestic cloud-based AI chip supply chain are expected to support the demand for domestic AI servers, while the corresponding domestic AI hardware ecosystem is expected to usher in growth opportunities. 3) AI applications: Terminal innovation is accelerating, and edge AI is driving the growth of IoT device connections, with opportunities in the communication module market expected to continue to grow with the ongoing upgrade of edge intelligence. Progress in large models empowers the security IoT, driving the hardware products, system software, and overall solution upgrades of security IoT enterprises. Telecom capital expenditure continues to tilt towards the computing power network direction, focusing on opportunities such as 5G-A, optical communication fiber, and optical fiber access. According to the financial reports of the three major operators at the beginning of the year, the capital expenditure guidance of the three major operators in 2025 is expected to decrease by 9.1% year-on-year to 289.8 billion yuan, with overall investment spending continuing to shrink and the investment focus continuing to tilt towards the computing power direction. According to data from the Ministry of Industry and Information Technology, in the first half of 2025, cumulative mobile Internet traffic reached 186.7 billion GB, a year-on-year increase of 16.4%, with a month-on-month increase since 2025. Recommendations include: 1) Accelerated deployment of 5G-A, and faster development of 6G technology; 2) AI-driven growth in demand for new types of fiber optics within/between data centers, with accelerated application of multimode fibers; 3) Mature deployment of optical fiber access, steady progress in gigabit optical network broadband construction, and beginning of upgrades to "10-gigabit" networks. Risks AI progress falls short of expectations; optical communication/telecom capital expenditure falls short of expectations; competition intensity exceeds expectations.