Computing Power Rental Track Heats Up as Listed Companies Accelerate Layout

Securities Daily Reporter Xu Linyan

Currently, the commercialization and implementation of AI intelligent agents are accelerating, with downstream application scenarios releasing concentrated demand. This directly drives explosive growth in computing power demand, and the popularity of computing power leasing has significantly increased.

On March 12, the concept sector of computing power leasing was active. By the close of trading that day, several concept stocks such as Hongjing Technology Co., Ltd., China Energy Construction Corporation, and others hit the daily limit. Stocks like Beijing Guolian Video Information Technology Co., Ltd. and Beijing Huichen Zidao Information Co., Ltd. (hereinafter referred to as “Huichen Shares”) also saw their prices rise.

“Computing power leasing is expected to become an important solution to alleviate the current AI computing power supply gap. It can quickly match the short-term computing power needs of AI companies, especially for small and medium-sized enterprises lacking sufficient funds and technical capabilities to build data centers themselves, significantly reducing initial investment costs. At the same time, it can improve the overall utilization efficiency of computing resources and prevent waste,” said Yuan Shuai, Deputy Secretary-General of the Zhongguancun Internet of Things Industry Alliance, in an interview with Securities Daily.

Meanwhile, rising leasing prices have become a key issue to watch for future market development.

According to a research report from Kaiyuan Securities, industry monitoring shows that, driven by a surge in AI computing power demand at the beginning of 2026, the computing power leasing market has entered a price increase cycle. By the end of February, rental prices for high-end GPUs such as Nvidia H200 and H100 increased by 15% to 30% month-on-month. The hourly rent for H200 reached 7.5 yuan per card to 8.0 yuan per card, with monthly rents between 60,000 and 66,000 yuan, an increase of 25% to 30%. The monthly rent for H100 rose to 55,000 to 60,000 yuan, up 15% to 20%. Delivery cycles have been extended to Q2 2027 for H200 and Q1 2027 for H100.

In response to this trend, Bai Wenxi, Chairman of Zhonghe Kunlun (Beijing) Asset Management Co., Ltd., told Securities Daily that from an operational perspective, companies can hedge against short-term price fluctuations by locking in long-term contracts and signing fixed-term leasing agreements for 3 to 5 years. From a long-term industry development perspective, it is necessary to accelerate the improvement of industry standards, clarify core indicators such as computing power performance and security levels, and guide market participants to compete in an orderly manner. Relevant policy subsidies should be targeted at small and medium-sized enterprises and research institutions in the industry to prevent resources from overly concentrating in leading companies.

Under the trend of rising prices and intensified competition, computing power leasing companies should avoid vicious price wars by improving service quality and technological capabilities to build differentiated barriers. “At the same time, leasing companies need to consider the cost pressures of AI transformation in the physical industry by launching exclusive discount packages for physical industries and collaborating with physical enterprises to co-develop computing power application scenarios, promoting a coordinated development pattern between the computing power leasing industry and the physical industry, rather than merely seeking short-term profits through price hikes,” Yuan Shuai said.

The high prosperity of the computing power leasing market has also attracted listed companies to accelerate their related business layouts through various means. Recently, on investor interaction platforms, the progress of listed companies’ computing power leasing businesses has become a hot topic among investors.

Huichen Shares, relying on its own computing power applications, has launched a one-stop product called “Integrated Computing Power Management Service Platform” for intelligent computing resource operation and management. The platform supports heterogeneous computing resource integration management, providing transaction and service operation functions, better meeting the needs of different AI computing power rental scenarios.

“Currently, the platform has been deployed internally within the company and in related data centers. In terms of hardware, a small amount of computing resources are also leased to relevant clients for use,” a relevant person from Huichen Shares told Securities Daily.

Hunan Aibulu Environmental Protection Technology Co., Ltd. announced that the company’s computing power center leasing business is scheduled to start in October 2024 and be fully operational by December 2024. Since the construction of the computing power center, the company has gradually strengthened its leasing capabilities, enhanced team building and technical reserves, and further improved its market competitiveness. As a result, the company’s sales and profits from computing power leasing in 2025 are expected to grow significantly compared to 2024.

Zhaochuang Data Technology Co., Ltd. stated on the investor interaction platform that computing power services are already a key business direction for the company, and it has signed cooperation or service agreements with multiple clients.

Looking ahead, Everbright Securities believes that companies with integrated computing and electricity, green power computing layouts, and customer resource advantages will dominate the competition in the computing power leasing market and become core beneficiaries of AI industry development.

“Long-term, the development of computing power leasing still needs to address technical dependence, security, and compliance bottlenecks. In the future, the industry will evolve toward greening, tiered services, and ecological collaboration. Companies need to achieve sustainable development through hybrid leasing models, technological cost reduction, and policy coordination,” said Bai Wenxi.