Tianfeng Securities: Domestic demand for computing power is on the rise, focusing on opportunities in computing power leasing and AI infrastructure investment
I'm LongbridgeAI, I can summarize articles.Tianfeng Securities released a research report indicating that domestic computing power demand continues to grow, with a surge in demand for computing power leasing and intelligent computing centers. In 2022, China's intelligent computing power scale was 259.9 EFLOPS, and it is expected to reach 1117.4 EFLOPS by 2027, with a compound annual growth rate of 33.9%. It is recommended to pay attention to targets such as Hainan Huatie, Gansu Engineering Consulting, DYRS, and NB CONSTRUCTION, emphasizing investment opportunities in AI infrastructure
According to the Zhitong Finance APP, Tianfeng Securities released a research report stating that domestic computing power remains highly prosperous, with a surge in demand for computing power leasing and intelligent computing centers. In 2022, China's intelligent computing power scale was 259.9 EFLOPS (based on FP16 computation), and it is expected to reach 1117.4 EFLOPS by 2027, with a five-year compound annual growth rate of 33.9%. High-end computing resources are scarce, and computing power leasing offers high returns. Recommended stocks include Hainan Huatie and Gansu Engineering Consulting, and it is suggested to pay attention to Dongyi Risheng (002713.SZ) and NB CONSTRUCTION (601789.SH). The demand for intelligent computing centers is surging, and it is recommended to focus on AI infrastructure investment opportunities and the post-restructuring value reassessment of Chengdi Xiangjiang (603887.SH).
The main points of Tianfeng Securities are as follows:
Domestic computing power remains highly prosperous, with a surge in demand for computing power leasing and intelligent computing centers
Domestic computing power remains highly prosperous, with China's intelligent computing power scale at 259.9 EFLOPS (based on FP16 computation) in 2022, expected to reach 1117.4 EFLOPS by 2027, with a five-year compound annual growth rate of 33.9%. The top-level design of the national integrated computing power network system is continuously improving, and the "East Data West Computing" project accelerates the overall construction of the national computing power network. In February 2022, four ministries jointly issued a notice agreeing to start the construction of national computing power hub nodes in eight regions: Beijing-Tianjin-Hebei, Yangtze River Delta, Guangdong-Hong Kong-Macao Greater Bay Area, Chengdu-Chongqing, Inner Mongolia, Guizhou, Gansu, and Ningxia, and planned ten national data center clusters in Zhangjiakou, Yangtze River Delta, Wuhu, Shaoguan, Chongqing, Tianfu, Gui'an, Hohhot, Qingyang, and Zhongwei. As of the first half of 2024, more than 250 intelligent computing centers have been built or are under construction nationwide, with 791 bidding events, and over 20 cities have built intelligent computing centers.
High-end computing resources are scarce, and computing power leasing offers high returns
Computing power leasing essentially involves renting out computing power resources. The rise of intelligent computing power leasing business is due to two main reasons: first, the short-term supply capacity of intelligent computing resources is insufficient, especially the shortage of AI chips represented by GPUs; second, the investment scale for building intelligent computing centers is large, and the operational and maintenance capabilities required are high, resulting in high comprehensive costs for computing power usage. Against the backdrop of tightening restrictions on high-end chips, domestic chips are accelerating self-development but still have some gaps. Therefore, the core competitiveness of the computing power leasing business lies in high-end chip resources, funding sources, and core customer resources.
In terms of return rates, a machine equipped with 8 H100 graphics cards, with a five-year lease term and a monthly rent of 70,000 yuan, is estimated to have a net profit margin for equipment leasing of 32%, 34%, 35%, 36%, and 37% from the first to the fifth year using the average depreciation method; if calculated using accelerated depreciation, the first year is expected to incur a loss of 51,000 yuan, with net profit margins of 14%, 35%, 55%, and 75% in the second to fifth years. The equipment recovery period is around 3 years, and the overall return rate for computing power leasing is at a high level. Recommended stocks include Hainan Huatie and Gansu Engineering Consulting, and it is suggested to pay attention to Dongyi Risheng and NB CONSTRUCTION The demand for intelligent computing centers is surging, emphasizing the importance of AI infrastructure prosperity
The "2024 China Intelligent Computing Center Industry Development White Paper" shows that the investment scale in this field reached 87.9 billion yuan in 2023, a year-on-year increase of over 90%; as of August 2024, there are more than 300 intelligent computing center projects nationwide, and it is expected that by 2028, the market size of China's intelligent computing centers will exceed 288.6 billion yuan, with a compound annual growth rate of 26.8% from 2023 to 2028. The operating models of data centers are mainly divided into retail and wholesale types. Taking the Hohhot computing power base of Guanghuan Xinwang as an example, the distribution system, building construction, and air conditioning system account for approximately 46%, 25%, and 17% of the total investment in the intelligent computing center, respectively. After the project reaches full load (with a rack rate of 95%), the net profit margin is about 8.19%.
Chengdi Xiangjiang has an IDC operation project with a total designed capacity of approximately 197MW. As of the end of March 2024, it has obtained approvals for a total of 115MW and has achieved cumulative sales of 35MW. The company is also planning a targeted private placement to introduce China Electronics Intelligent Computing as its controlling shareholder, integrating high-quality data center assets in the East China region to create a "national team" for computing power centers. It is recommended to pay attention to AI infrastructure investment opportunities and the value reassessment of Chengdi Xiangjiang after restructuring
