The company is a leading park-level data center with strong performance. Since its establishment in 2009, Runze Technology has focused on developing and operating large-scale, high-grade, efficient and high-performance data center clusters. By the end of 2023, 7 AIDC intelligent computing infrastructure clusters had been built in 6 major regions of the country. A total of about 61 intelligent computing centers and 320,000 racks were planned, and satellite cities mainly concentrated in first-tier cities such as Beijing, Shanghai, Guangzhou, and Shenzhen, have basically completed the national “integrated computing power center system” framework layout. In 2018-2023, the company's revenue increased from 628 million yuan to 4.351 billion yuan, corresponding CAGR of 47.27%. In 2019-2023, the company's net profit to mother increased from 127 million yuan to 1,762 million yuan, corresponding CAGR of 92.99%. The company has been in an intensive data center delivery period since 2023, and it is expected that as the listing rate gradually increases, the company's newly delivered data centers are expected to contribute rapidly to the company's profits.

AI is leading the growth in demand, and the popularity of the IDC industry continues to rise. IDC: With the implementation of emerging technologies such as big data and cloud computing, data center traffic and bandwidth are growing exponentially around the world. Information technology industries such as data center services and cloud computing are strategic industries encouraged and developed by the country. In recent years, they have become one of the key areas of new infrastructure determined by the country. Overall, under the guidance of the country's top-level policies, China's computing power infrastructure construction continues to accelerate, the country's “integrated computing power system” is rapidly being formed, and the prosperity of the IDC industry has improved. AIDC: 1) Increased demand for computing power: In recent years, China has always attached great importance to artificial intelligence development opportunities and top-level design. Under policy promotion and demand guidance, a number of outstanding big model companies have emerged in China, mainly including Internet technology vendors such as Ali, Baidu, Tencent, etc., and big model startups such as Smart Spectrum and Dark Side of the Moon. We believe that as more models with larger numbers of parameters are released one after another, demand for computing power will continue to be released, and intelligent computing centers are expected to usher in a new wave of construction. 2) Increased cabinet power: At the GPU level, the Nvidia H100/B200 chip currently consumes 700/1000W, corresponding to the power requirement of a single cabinet reaching 40/25kW, which is significantly higher than the power requirements of previous CPU-based cabinets.

Currently, there is a significant trend of increasing power consumption in CPU and GPU thermal designs. We believe that companies with liquid cooling technology and the ability to deliver high-power cabinets are expected to establish a first-mover advantage in the future.

Competitive advantage strengthens the company's leading position in park-level IDC. Location+listing rate+operation and maintenance+customer advantages establish a leading position in park-level IDC. 1) Location advantage: The company is based in Beijing-Tianjin-Hebei Langfang, Yangtze River Delta, Pinghu, Guangdong-Hong Kong-Macao Greater Bay Area, Foshan, and Huizhou, radiating the Cheng-Chongqing Economic Zone and Hainan to form six core regions, seven major computing power nodes, and a total layout of 61 intelligent computing centers and 320,000 racks, with remarkable location advantages. 2) Shelf rate: Adopting the construction model of an oversized park. In terms of structural structure, the company can connect each computer room through an integrated pipe gallery to avoid the need for redundant space in a single computer room. Under this model, the company's mature cabinet shelf rate exceeds 90%. 3) Operation and maintenance: Adhering to the “self-investment, self-construction, self-maintenance, self-operation and maintenance” operation model is conducive to creating a long-term, stable and reliable operating environment, thereby improving service quality and customer stickiness. 4) Customers: End customers, such as ByteDance, have high demand. The company takes a long time to sign contracts with operators, and the sales fee rate and accounts receivable repayment days are superior to other comparable companies in the same industry.

Card AIDC is a major trend, and the company's first intelligent computing center has been delivered. The intelligent computing center mainly uses liquid cooling in terms of technology. It provides high-speed connections between the CPU, GPU, and storage at the network level, and most of the data flows east-west. The effective computing power of the intelligent computing center cluster = GPU single card computing power × total number of cards × acceleration ratio × effective operation time, and the acceleration ratio/effective operation is determined by network performance/network reliability, respectively, and is the core factor that determines the cluster computing effect. Therefore, the intelligent computing center construction emphasizes the manufacturer's IB network track optimization ability, IB network troubleshooting ability, server-side tuning ability, and understanding of large model communication libraries. Only when multiple factors are combined and superimposed can the full link delivery capability of a large cluster be achieved. In 2023, the company delivered the A-11 of the Runze (Langfang) International Information Port Pure Liquid-Cooled Green Intelligent Computing Center, which starts at 21.5KW in the industry. In the future, the company's Langfang Area B and Pinghu Phase II are both intelligent computing center plans. Among them, the Langfang B District Intelligent Computing Center is a single intelligent computing center project that can gather 100,000 cards or more. AIDC is expected to achieve rapid growth and is expected to become the second growth curve.

Investment advice: The company is a leading park-level data center leader in China, and has multi-dimensional competitive advantages such as location layout, high availability rate, mature operation and maintenance model, and customer structure. Furthermore, with the development of AI, demand for intelligent computing center construction has surged, and companies that rely on technical advantages such as IB network optimization and liquid cooling are expected to benefit the core of this AI wave. We expect the company's net profit to be 21.91, 33.03, and 4.161 billion yuan respectively in 24-26, with year-on-year growth rates of 24.4%, 50.7%, and 26.0%, respectively. The company was given 30 times PE in 2024, corresponding to a target price of 38.21 yuan. For the first time, coverage was given a “recommended” rating.

Risk warning: Risk of changes in national energy efficiency policies, declining demand for IDC, delivery of new data centers falling short of expectations, and AIDC development falling short of expectations.