Intelligent Finance News APP learned that CICC pointed out in a research report that in the first half of 2024, AI computing power is still the core investment theme. AI hardware core suppliers' performance and market performance resonate upwards, driving up the market cap of the sector since the beginning of the year. Looking ahead to the second half of 2024, it is recommended to focus on the five major opportunities in the communications equipment industry: 1) Domestic computing power accelerates, driving long-term empowerment of new quality production factors for the AI industry; 2) AI large models evolve towards multimodal direction, focusing on engineering optimization and long-term inference implementation to drive high demand for overseas cloud hardware; 3) Operator capital expenditure is flat, focusing on structural opportunities such as 5G-A and backbone network upgrades; 4) Accelerate domestic satellite construction; 5) Enterprise going abroad leads new growth.

The marginal changes of domestic computing power are rich, and the upgrading of domestic AI technology strength in the medium and long term is expected to deeply empower new quality production factors.

From Wenxin Yanyan, Kimi, to DeepSeek-V2, the number of domestic large models is accelerating, evolving towards large parameters, long texts, and multimodal. On the infrastructure side, the policy level actively promotes the construction of domestic smart computing centers, and the uncertainty of the global supply chain further expands the demand space for domestic computing power. CICC believes that with the continuous upgrading of domestic GPU computing power and the improvement of ecological construction, the development space of domestic computing power is vast. It is recommended to focus on AI servers, data center switches and chips, and domestic optical modules.

In the long run, the evolution of AI can reduce industry costs and increase efficiency, better empower industrial intelligence transformation, promote the development of new quality productivity, and promote solution providers for enterprise intelligent transformation, as well as IoT module companies that serve as key components of data entry into the network.

Overseas demand for AI hardware continues to be highly prosperous, and low power consumption requirements lead technological innovation.

CICC pointed out that looking at the entire AI industry, innovation of large models explores the two directions of Scaling Law and engineering optimization together, and multidimensional pursuits jointly promote the continuous upward trend of cloud hardware demand. At the same time, it is noted that there is an accelerating trend in cloud network iteration. Nvidia's Ethernet hardware planning upgrades once a year, and a 3.2T network commercial is expected to appear by 2026. Against this background, on the one hand, the industry replacement is accelerating, which is good for existing leading suppliers to play a first-mover advantage; on the other hand, low power consumption needs are becoming increasingly urgent, and solutions such as liquid-cooled servers and silicon-based, LPO optical modules may accelerate landing, leading to structural investment opportunities.

Target: It is recommended to focus on computing hardware suppliers Zhongji Innolight (300308.SZ), Accelink Technologies (002281.SZ), Shengke Communication (688702.SH), ZTE (000063.SZ), Dawning Information Industry (603019.SH), Unisplendour Corporation (000938.SZ), Inspur Electronic Information Industry (000977.SZ), Ruijie Networks (301165.SZ); solution providers Hikvision Digital Technology (002415.SZ), Zhejiang Dahua Technology (002236.SZ); communication module vendors Quectel Wireless Solutions (603236.SH), Fibocom Wireless Inc. (300638.SZ); optical cable & submarine cable vendors Jiangsu Zhongtian Technology (600522.SH), Hengtong Optic-Electric (600487.SH); overseas target Yealink Network Technology (300628.SZ), Ezviz Network (688475.SH).

Risk factors: infrastructure investment is lower than expected; uncertainty of new technologies such as AI; trade and technology frictions.