The computing power leasing sector heats up as listed companies accelerate their deployment

证券日报 记者 许林艳

Currently, the commercialization of AI intelligent agents is accelerating, demand for downstream application scenarios is being concentrated and released, and it directly drives explosive growth in compute power demand, significantly boosting the heat of compute power leasing.

On March 12, the compute power leasing concept sector showed active performance. As of market close on that day, multiple concept stocks, including Macro-scenic Technology Co., Ltd. and China Energy Engineering Group Co., Ltd., hit their daily trading limits. Individual stocks such as Beijing Guolian Times Information Technology Co., Ltd. and Beijing Huicheng Zidao Information Co., Ltd. (hereinafter referred to as “Huicheng Shares”) also rose in tandem.

“Compute power leasing is expected to become an important solution for easing the current shortfall in AI compute power supply. It can rapidly match AI enterprises’ short-term compute power needs, and especially for small and medium-sized enterprises that lack sufficient funding and technical capability to build data centers themselves, it can substantially reduce upfront investment costs. At the same time, it can improve the utilization efficiency of overall compute power resources and avoid compute power waste.” Yuan Shuai, Deputy Secretary-General of the Zhongguancun IoT Industry Alliance, said to reporters of the Securities Daily.

Meanwhile, rising rental prices have become a key issue that the market needs to pay attention to in its subsequent development.

An open-source Securities research report shows that, according to industry monitoring, starting in early 2026, an upsurge in AI compute power demand driven the compute power leasing market to enter a price-increase cycle. By the end of February, rental prices for high-end GPUs such as Nvidia H200 and H100 rose by 15% to 30% month over month. The hourly rental rate for H200 increased to between 7.5 yuan per card-hour and 8.0 yuan per card-hour; monthly rental fees rose to between 60k yuan and 66k yuan, an increase of 25% to 30%. The monthly rental rate for H100 rose to between 55k yuan and 60k yuan, an increase of 15% to 20%. The delivery cycle extended to 2027 Q2 (H200) and Q1 (H100).

Against this backdrop, Bai Wenxi, Chairman of Join Kunlun (Beijing) Asset Management Co., Ltd., said in an interview with reporters of the Securities Daily that from an enterprise operations perspective, businesses can effectively hedge business uncertainty caused by short-term price fluctuations by locking in long-term contracts and signing fixed-term leasing agreements with durations of 3 to 5 years. From the long-term development perspective of the industry, it is necessary to accelerate the improvement of an industry standards framework and clarify core metrics such as compute power performance and security levels, thereby guiding market participants to compete in an orderly manner. Relevant policy subsidies should be targeted precisely toward small and medium-sized enterprises in the industry and research institutions, avoiding excessive concentration of resources in leading companies.

Under the development trend of rising industry prices and intensifying competition, compute power leasing companies should avoid vicious price competition. Instead, they should build differentiated barriers by improving service quality and technical capabilities. “At the same time, compute power leasing companies need to balance the cost pressure of the real-industry AI transformation by rolling out exclusive discounted compute power packages for the real economy, and by jointly building compute power application scenario development with real-economy enterprises, thereby promoting a coordinated development pattern between the compute power leasing industry and the real economy—not merely obtaining short-term benefits through price hikes,” Yuan Shuai said.

The high level of optimism in the compute power leasing market has also attracted listed companies to accelerate their related business deployment through multiple approaches. Recently, on investor interaction platforms, the progress of listed companies’ compute power leasing business has become a hot topic of investor attention.

Relying on its own compute power application, Huicheng Shares has already released a one-stop product “Integrated Compute Power Management Service Platform” for smart-compute resource operations management. The platform supports integrated management of heterogeneous compute power resources, provides compute power resource transactions, service operations, and more, and can better meet scenario demands for different types of AI compute power leasing.

“Currently, the platform has already been deployed with applications within the company as well as in relevant data centers. In terms of compute power hardware, only a small amount of compute power resources have been leased to relevant customers for use,” a relevant person in charge from Huicheng Shares told reporters of the Securities Daily.

A notice from Hunan Aibalu Environmental Protection Technology Co., Ltd. states that the starting point of the company’s compute power center compute power leasing business was October 2024, and the overall construction and investment in the compute power center were completed in December 2024. Since the investment and construction of the compute power center began, the company has gradually strengthened its compute power leasing business capabilities, improved team building and technical reserves, and further enhanced its market-oriented capabilities. Therefore, in 2025, the company’s sales revenue and profit from its compute power leasing business increased significantly compared with 2024.

Coorstar Data Technology Co., Ltd. stated on an investor interaction platform that compute power services are already an important direction of the company’s business. To date, it has signed cooperation or service agreements with multiple customers.

Looking ahead, Everbright Securities believes that enterprises with compute-and-electric coordination technologies, green power compute power deployment, and customer resource advantages will occupy a leading position in competition in the compute power leasing market, becoming core beneficiaries of the development of the AI industry.

“From a long-term perspective, the development of compute power leasing still needs to address bottlenecks such as technological dependence and safety and compliance. In the future, the industry will evolve toward greening, service tiering, and ecosystem coordination. Enterprises need to achieve sustainable development through hybrid leasing models, cost reductions via technology, and policy coordination,” Bai Wenxi said.