CMSC: The new cycle of AI-driven computing power and electricity integration. The industry enters a stage where policies are implemented in parallel with pilot projects.

CMSC released a research report stating that the training of large AI models is driving an exponential increase in computing power demand and energy consumption, leading to a sharp rise in intelligent computing power usage. The inclusion of "computing power and electricity coordination" in a government work report for the first time highlights the importance of green electricity as a key factor in the low-carbon development of computing power. Not only can it effectively supplement the traditional electricity supply gap, but it also provides a feasible path for low-carbon and sustainable development of computing facilities. The industry has entered a stage where policies are being implemented alongside pilot projects, with a focus on the collaborative application of multiple entities in "optimizing computing with electricity and supporting electricity with computing" scenarios. It is recommended to pay attention to key beneficiaries in this field. The main points of the CMSC report are as follows: "Computing power and electricity coordination" included in a government work report for the first time On March 5, 2026, during the 14th National People's Congress session, the Premier of the State Council explicitly stated in the government work report the implementation of new infrastructures such as ultra-large-scale intelligent computing clusters and computing power and electricity coordination, strengthening national integrated computing power monitoring and dispatching, and supporting the development of public clouds. Computing power and electricity coordination, along with ultra-large-scale intelligent computing clusters, have become national-level new infrastructure projects. Liu Liehong, Director of the National Bureau of Statistics, provided an authoritative definition of computing power and electricity coordination at the 2026 China Development High-Level Forum on March 23: it refers to the deep integration of computing power infrastructure with the electricity system through digital technology, intelligent algorithms, and information networks, promoting resource dynamic matching and optimization configuration as a new infrastructure project to achieve a virtuous cycle of "strengthening computing with electricity and promoting electricity with computing." The main contents include promoting direct supply of green electricity, agglomerating green electricity supply, improving the support capacity of green electricity for computing power, promoting waste heat recovery, enhancing the benefits of green low-carbon circulation, and more. The continuous expansion of renewable energy capacity makes green electricity a key support for solving the "power shortage" dilemma in computing power centers The exponential growth in computing power demand and energy consumption driven by large-scale model training has led to a sharp increase in the use of intelligent computing power. Against the backdrop of explosive growth in computing power demand and increasingly prominent structural contradictions in electricity supply, green electricity can effectively supplement the traditional electricity supply gap and provide a feasible path for low-carbon and sustainable development of computing facilities. Currently, China's computing power and electricity coordination has entered a key stage of policy implementation and pilot exploration. Various policy documents have been successively released to coordinate the layout of computing power and electricity. By the end of 2023, the National Development and Reform Commission and other departments proposed that by 2025, green electricity should account for over 80% of the total energy supply at national hub nodes for new data centers, and several companies have also promised to achieve 100% renewable energy use by 2030. The development of computing power and electricity coordination includes two key elements: the computing power system and the electricity system The main actors in the computing power system include computing power suppliers and computing network operators, while the main actors in the electricity system include power generation entities, grid operators, and energy storage entities. Among them, computing power suppliers provide adjustable loads to the power system to support peak load management, while computing network operators facilitate the integration of computing power with the electricity system to achieve overall optimal scheduling. Power generation entities provide clean and low-carbon energy for computing power to promote decarbonization, grid operators ensure stable power supply for computing power, and promote electricity market trading, and energy storage entities stabilize fluctuations in new energy sources to enhance the reliability of computing power supply. Specific measures for computing power and electricity coordination include source-load interaction, storage-load interaction, source-grid-load-storage integration, and more. As a pioneer in the construction of a new power system, the State Grid Corporation of South China has begun early deployment in the field of "electricity-carbon-computing coordination" During the "14th Five-Year Plan" period, the State Grid Corporation of South China has established a cluster of "3+1+X" digital infrastructure, developed the "Electricity-Hong IoT" operating system and high-efficiency servers, and released the first domestic electricity-carbon-computing coordination operation system. Looking towards the "15th Five-Year Plan," the company adheres to a direction of intelligence, ecology, and integration, comprehensively deploying around infrastructure, technological innovation, scenario expansion, and ecological upgrading, establishing a benchmark for the development of computing power and electricity coordination in China with a fusion paradigm of "electricity as the foundation, carbon as the pulse, and computing as the wing." Investment recommendation: Focus on CSG Digital Power Grid Research Institute, China Southern Power Grid Technology, State Power Rixin Technology; other recommendations include GCL Energy Technology, Henan Yuneng Holdings, NYOCOR, China Energy Engineering Corporation (not covered by the computing sector). Risk warning: Risks of policy and mechanism implementation falling short of expectations; Risks of supply-demand mismatch and economic fluctuations.