Introduction

Recently, the National Development and Reform Commission, the National Energy Administration, the Ministry of Industry and Information Technology, and the National Data Bureau issued the “Action Plan for Promoting Mutual Empowerment between Artificial Intelligence and Energy” (hereinafter referred to as the “Action Plan”).

The “Action Plan” focuses on supporting the development of artificial intelligence through energy and empowering energy transformation with artificial intelligence. It deploys 29 key tasks aimed at ensuring a safe and reliable energy supply for computing facilities, promoting the green and low-carbon transformation of these facilities, and facilitating efficient economic collaboration between computing power and electricity.

Industry experts indicate that the introduction of the “Action Plan” signifies that energy is no longer just a passive supplier for the development of artificial intelligence, and artificial intelligence is not merely a technical tool for energy transformation. The deep coupling of the two will inject strong momentum into the high-quality development of China’s digital economy and the construction of a new energy system.

Energy Supporting Artificial Intelligence

The explosive growth of artificial intelligence has created unprecedented demands for energy supply. Currently, the electricity consumption of computing facilities is surging, characterized by high density, strong continuity, and sensitivity to power quality. There is an urgent need for more stable, green, and economical energy security. The “Action Plan” emphasizes the need to ensure a safe and reliable energy supply for computing facilities, stating that energy resource allocation should be coordinated with computing facility construction to strengthen energy support for computing development.

In terms of energy resource and computing layout, the “Action Plan” proposes to coordinate large-scale renewable energy bases with national computing hubs, promoting the orderly aggregation of computing facilities and internet backbone points in areas rich in renewable energy. It also suggests exploring the collaborative construction of million-kilowatt-level AI computing facilities and supporting energy systems in suitable regions.

A practical example of this policy is the recent launch of China’s first large-scale “computing-electricity synergy” green electricity supply project, a 500,000-kilowatt photovoltaic power station in Ningxia Zhongwei. This project employs a dual-track supply system combining physical direct supply and bilateral trading, effectively reducing operational costs while significantly decreasing carbon emissions.

This marks a typical case of scaling up the concept of computing-electricity synergy in China. From the first grid connection of the Ulanqab Zhongjin data project in July 2025 to the production of the first batch of units in the Gansu Qingyang gigawatt-level green electricity aggregation project in April 2026, numerous computing-electricity synergy projects have been established across Inner Mongolia and Xinjiang, with a cumulative installed capacity nearing ten million kilowatts.

To enhance the diverse electricity supply capabilities of computing facilities, the “Action Plan” proposes to establish and improve planning and construction standards for energy supply based on the actual conditions of the computing facilities, such as system scale and energy quality requirements. It encourages the direct connection of nuclear power and hydrogen energy to supply computing facilities and supports the configuration of grid-type energy storage to enhance supply stability.

The plan also emphasizes the importance of green and low-carbon energy in supporting computing development. It outlines the need for statistical work on the proportion of green electricity consumption and carbon emissions accounting for computing facilities, as well as improving energy efficiency and carbon effectiveness.

Artificial Intelligence Empowering Energy

Kang Yanbing, deputy director of the Energy Research Institute of the National Development and Reform Commission, stated at the AI + Energy Development Conference earlier this year that there are two core pathways for the energy transition revolution, both of which AI can significantly empower. The first is energy conservation and efficiency improvement, where AI can support efficient production and convenient living. The second is clean substitution, which involves reducing coal and oil use and promoting the decarbonization of the energy structure through the development of non-fossil energy sources. However, the instability of renewable energy necessitates the optimization of energy storage and peak regulation using AI and other digital technologies.

Artificial intelligence is not only an energy consumer but also a crucial force in driving the energy revolution. The “Action Plan” focuses on opening high-value application scenarios for AI in the energy sector and strengthening model innovation, systematically deploying pathways for AI to empower energy transformation.

In terms of application scenarios, the “Action Plan” emphasizes the need for AI technology innovation driven by scenario demands, accelerating the deep integration and large-scale development of AI technology across the entire energy supply chain.

Industry experts believe that AI’s application scenarios in energy encompass areas such as “AI + coal,” “AI + oil and gas,” and “AI + electricity” on the supply side, as well as intelligent manufacturing, smart energy consumption in urban and rural areas, and smart transportation on the consumption side, achieving precise energy supply and energy conservation through AI.

For instance, in the field of grid operation, AI is moving towards large-scale application. The State Grid Shandong Electric Power Company has collaborated with research institutions to establish an AI load forecasting model, achieving a peak load forecasting accuracy of 99.7% during the summer peak in 2025, effectively supporting the grid in handling a historical peak load of 130 million kilowatts.

In terms of model innovation, the “Action Plan” emphasizes the need for professional model innovation and the deep application of controllable hardware in the energy sector, achieving a profound coupling of AI technology with the energy industry.

Experts believe that the rich application scenarios combined with high-quality data resources will fully unleash the multiplier effect of AI empowering the transformation of energy, fostering a batch of replicable, scalable, and practical intelligent energy solutions.

Data is the fuel source for artificial intelligence. The “Action Plan” clarifies the establishment of a high-quality energy data development model that integrates governance, security, and circulation, fully leveraging the value of data elements and promoting the transformation of energy data from resources to assets.

Computing-Electricity Synergy as a Key Link

Computing-electricity synergy is the key link connecting energy supporting artificial intelligence and AI empowering energy. In the future, the competition in artificial intelligence will not only depend on models, chips, and computing power scale but also on the reliability of energy systems. Those who can achieve efficient synergy between intelligent computing power and green electricity first will gain the upper hand in AI industry competition.

To this end, the “Action Plan” specifically includes a chapter on “Promoting Efficient Economic Synergy between Computing Power and Electricity,” proposing to leverage the scale effect of computing-electricity synergy and explore the flexible adjustment potential of computing facilities.

The “Action Plan” clarifies the need to establish an interactive mechanism between computing power and electricity, using electricity market price signals to guide computing facilities in optimizing energy management and various forms of computing scheduling across networks and regions, enhancing the economic efficiency of computing facilities. It encourages computing facilities to participate as flexible adjustable resources in grid operations, improving the regulation capability of the power system and achieving mutual efficiency enhancement between computing facilities and the power system.

Computing-electricity synergy can alleviate the pressure of energy consumption growth and promote the deep integration of energy and industrial structures. Experts state that it breaks down the barriers between computing power and electricity, essentially constructing a dynamic matching mechanism between energy supply and computing demand.

In practical implementation, experts suggest that data centers participate in grid demand response, such as low valley cooling and peak load reduction, and promote the nearby consumption of wind and solar energy to achieve integrated source-grid-load-storage systems.

This concept has been explored in various regions. For example, in Anhui Province, the first “computing-electricity synergy” valley filling scheduling test was completed, where China Telecom Anhui Company smoothly transferred real-time computing tasks from the Hefei High-tech Zone Intelligent Computing Center to the Huaibei Telecom Cloud Computing Center, with the Huaibei center fully taking over the tasks three minutes later, increasing the single machine’s electricity load by 50% and effectively enhancing local green electricity consumption capacity.

In terms of market mechanism construction, the “Action Plan” encourages new computing facilities to sign long-term green electricity trading contracts with renewable energy generation enterprises, enhancing the proportion and stability of green electricity consumption and constructing an economically efficient green energy supply system for computing facilities. It supports various forms of participation in electricity trading, auxiliary services, and demand response.

Yuan Jun, deputy director of the National Data Development Research Institute, recently acknowledged the challenges of achieving the threefold goals of security, greenness, and economy in computing-electricity synergy, referred to as the “impossible triangle” challenge. He proposed that a unified national strategy is needed to incorporate computing power and grid planning into a cohesive spatial system, dynamically adapting to AI development. Additionally, strengthening technological breakthroughs in load forecasting and cross-domain collaborative scheduling is essential to enhance flexible interaction capabilities.

As computing power scales continue to expand, the electricity consumption of national data centers is expected to maintain rapid growth, with both computing demand and renewable energy installation growing quickly. China is accelerating the construction of a national-level computing infrastructure system that is interconnected, accessible, green, and safe, with future computing-electricity synergy transitioning from “electricity supporting computing” to “computing optimizing electricity,” deeply integrating clean green electricity with computing networks.

Conclusion

The shift from “AI +” to “mutual empowerment” signifies a profound adjustment in policy logic. Energy is no longer just an application scenario for artificial intelligence but has been elevated to a strategic infrastructure supporting AI development. The relationship between the two has evolved from one-way empowerment to deep integration.

The “Action Plan” not only requires computing facilities to be stably supplied with electricity but also positions them as a new type of load with high adjustment capabilities within the power system. This encourages market mechanisms to guide computing facilities in participating in grid peak regulation and demand response, transforming them from mere energy consumers to active participants in balancing the power system. The mutual empowerment of AI and energy creates a cycle where AI empowers green electricity production, and green electricity supports AI development, fostering a collaborative and efficient energy development landscape.