Implementation Opinions of the National Development and Reform Commission and the National Energy Administration on Promoting High-Quality Development of "Artificial Intelligence +" Energy

Implementation Opinions of the National Development and Reform Commission and the National Energy Administration on Promoting High-Quality Development of "Artificial Intelligence +" Energy

Development and Reform Commissions and Energy Bureaus of all provinces, autonomous regions, municipalities directly under the Central Government, cities specifically designated in the state plan, and the Xinjiang Production and Construction Corps, relevant central enterprises, and relevant industry associations:
To thoroughly implement the decisions and deployments of the Party Central Committee and the State Council on the development of artificial intelligence, to implement the relevant work requirements of the "Opinions of the State Council on Deeply Implementing the 'Artificial Intelligence +' Action" (Guo Fa [2025] No. 11), seize the major strategic opportunities of artificial intelligence development, highlight application orientation, accelerate the deep integration of artificial intelligence and the energy industry, and support high-quality development and high-level safety of energy, the following opinions are hereby proposed.
1. Overall Requirements

Adhere to Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era as guidance, thoroughly implement the spirit of the 20th National Congress of the Communist Party of China and the second and third plenary sessions of the 20th Central Committee, fully implement General Secretary Xi Jinping's important instructions on promoting the deep integration of artificial intelligence with the real economy and cultivating and expanding the intelligent industry. Relying on expanding deep integration application scenarios of artificial intelligence and energy fields as an important foundation, focusing on improving the level of innovative application technology of artificial intelligence in the energy field as the main direction, promoting the coordinated development of intelligent computing power and electricity as necessary support, and improving the innovation system for energy intelligence development as a key guarantee, efforts will be made to enhance the energy system's safety, reliability, flexibility, and efficient operation capabilities, ensure energy security, stable supply, and green low-carbon transition, accelerate the cultivation of new quality productivity, and provide strong support for the construction of a new energy system.

By 2027, the integration and innovation system of energy and artificial intelligence will be initially established, the foundation for coordinated development of computing power and electricity will be continuously consolidated, significant breakthroughs will be made in core energy technologies empowered by artificial intelligence, and applications will be more extensive and in-depth. Promote the deep application of more than five professional large models in industries such as power grids, power generation, coal, and oil and gas; explore more than ten replicable, easily promotable, and competitive key demonstration projects; explore empowerment paths for one hundred typical application scenarios; cultivate a batch of energy industry artificial intelligence technology application research and development innovation platforms; formulate and improve one hundred technical standards; cultivate a batch of compound talents in energy and artificial intelligence; explore the establishment of a financial support system for artificial intelligence technology research, development, and application in the energy field; form an innovation and development model of artificial intelligence technology in the energy field that fits China's national conditions; and initially show the effectiveness of intelligence in the energy field.

By 2030, the specialized technologies and applications of artificial intelligence in the energy field will reach a world-leading level overall. The coordinated mechanism of computing power and electricity will be further improved, establishing a green, economical, safe, and efficient computing power energy consumption model. Theoretical and technological innovations in the integration of energy and artificial intelligence will achieve significant results. Artificial intelligence technology in the energy field will realize cross-domain, cross-industry, and cross-business scenario empowerment, achieving breakthroughs in directions such as intelligent power regulation, intelligent energy resource exploration, and intelligent renewable energy forecasting. Embodied intelligence and scientific intelligence will be applied in key scenarios. A batch of globally leading "Artificial Intelligence +" energy-related research and development innovation platforms and compound talent training bases will be formed. A more complete policy system will be established to continuously guide efficient, healthy, and orderly innovation of "Artificial Intelligence +" energy, laying a solid foundation for high-quality energy development.
2. Accelerate Empowerment of Energy Application Scenarios

(1) Artificial Intelligence + Power Grid. Focusing on the requirements of power grid security, new energy consumption, and operational efficiency under the new power system, carry out applications such as power supply and demand forecasting, intelligent diagnosis and analysis of the power grid, and intelligent generation of planning schemes for power grid planning and design, and strengthen intelligent construction management of power grid projects; promote multi-scale intelligent simulation and analysis of the power grid, explore the application of artificial intelligence models in intelligent auxiliary decision-making and dispatch control of the power grid, and improve the safe, reliable, and low-carbon operation level of all elements of the power system including source, grid, load, and storage; steadily improve the intelligent level of key equipment development such as transmission and transformation; promote predictive maintenance of power equipment faults, build intelligent agents for power equipment health management with autonomous perception, decision-making, and execution capabilities, and improve lean management of equipment; promote integrated intelligent application of operation, distribution, and dispatch, build an intelligent support system for power grid operation services, and improve the intelligent service level for power customers throughout the process; promote the integration of artificial intelligence technology into the power emergency system and capacity building, and improve the intelligent level of disaster prevention, mitigation, and relief of the power system.
Column 1: Typical Application Scenarios of Artificial Intelligence + Power Grid

Intelligent Planning, Design, and Production Construction of Power Grid. Build applications such as intelligent prediction of power supply and demand, intelligent diagnosis and analysis of power grid operation, intelligent auxiliary decision-making for power grid planning, and intelligent design of transmission and transformation facilities. Apply artificial intelligence technology to carry out planning, design, and techno-economic analysis, promoting the transformation of power grid planning and design work modes towards intelligence. Focus on work perception and business monitoring during the construction phase, build applications such as artificial intelligence violation recognition, progress simulation, online monitoring, real-time analysis of control indicators, and intelligent management of work processes for power grid construction, promoting intelligent upgrading of power grid engineering construction.

Power Grid Dispatch and Operation. Under the background of the construction of a unified national electricity market, build intelligent applications in areas such as new energy power forecasting, load forecasting, offline simulation analysis, online safety analysis, extreme emergency handling, dispatch auxiliary decision-making, market clearing and operation optimization, and intelligent decision-making in the electricity market. Continuously improve the support system for new generation intelligent regulation technology to support the safe and stable operation of the new power system.

Power Equipment Status Evaluation and Intelligent Operation and Maintenance. Build applications such as intelligent perception and early warning of equipment status, intelligent localization and diagnosis of equipment faults, intelligent decision-making for equipment status maintenance, intelligent prediction of equipment disaster risks, and intelligent generation of maintenance work tickets, improving lean management of equipment.

Intelligent Operation and Management of Distribution Network. Build technical applications such as real-time perception, risk analysis, and intelligent decision-making of the distribution network, comprehensively improve the intelligent control capability and power supply reliability of the distribution network, and strengthen coordinated regulation of source, grid, load, and storage at the distribution network level.
Power Emergency Repair. Build auxiliary decision-making systems such as intelligent early warning of disaster risks in the power system, intelligent analysis of damage conditions, and intelligent decision-making of emergency plans. Promote intelligent application of power emergency repair technology and equipment, and improve the power system's disaster prevention, mitigation, and relief capabilities.

(2) Artificial Intelligence + New Energy Business Models. Focusing on energy supply security and green low-carbon transition needs, promote the application of artificial intelligence technology in flexible adjustment resources such as virtual power plants (including load aggregators), distributed energy storage, and electric vehicle grid interaction, improving load-side group control optimization and dynamic response capabilities; strengthen the application of artificial intelligence technology in new energy storage and coordinated optimal dispatch of power systems as well as full lifecycle safety, and promote intelligent optimization of renewable energy hydrogen production processes. Strengthen artificial intelligence technology empowerment in energy production processes for energy saving and carbon emission management, improve the comprehensive energy efficiency and carbon reduction level of multi-energy complementary integrated energy systems supplying electricity, heat, cooling, and gas. Promote the application of artificial intelligence in zero-carbon parks, intelligent microgrids, and computing-electricity coordination, improve the integrated intelligent operation level of source, grid, load, and storage, and promote local consumption of new energy.
Column 2: Typical Application Scenarios of Artificial Intelligence + New Energy Business Models

Precise Control and Intelligent Operation of Virtual Power Plants. The virtual power plant operator platform intelligently optimizes control strategies and generates control instructions based on grid regulation commands, market information, and dynamic changes in resource characteristics, achieving large-scale flexible resource aggregation optimization and control, and enabling smart trading decisions for virtual power plants participating in the electricity market.

Intelligent Optimization of Green Hydrogen Production Processes. Integrating multi-dimensional data such as wind and solar power fluctuation forecasts, hydrogen storage tank capacity, electrolyzer temperature, and catalyst status, based on artificial intelligence algorithms, intelligently drive dynamic optimization of electrolyzer current density, build an intelligent control system for the entire chain of electrolytic hydrogen production, hydrogen storage, and hydrogen use, and achieve millisecond-level matching of renewable energy power fluctuations and flexible loads of electrolytic devices.

Intelligent Carbon Reduction in Parks. Based on operational data from photovoltaic, energy storage, and other equipment, the park's intelligent carbon reduction collaborative control system dynamically optimizes energy scheduling strategies in real time. It automatically adjusts air conditioning temperatures, charging pile power, and equipment start-stop sequences by combining electricity prices and carbon emission factors. Through augmented reality visualization interfaces and voice assistants, personalized energy-saving suggestions are pushed to users, forming an intelligent collaborative model of "carbon-energy-cost."

Intelligent Operation of New Energy Storage. Targeting dynamic adaptation of new energy storage to power system dispatch, wide-area collaborative interaction, weak grid support, battery equipment safety monitoring, equipment evaluation, and maintenance, artificial intelligence technology is used to enhance multi-type energy storage coordination and control capabilities for weak grids. It builds an application system including wide-area collaborative optimization control of new energy and newly built energy storage, intelligent evaluation of energy storage power stations, smart operation and maintenance decision support, and full lifecycle safety, improving the power supply assurance capability of system-friendly new energy power stations.

Intelligent Marketing Services. For customer-facing service scenarios such as oil, gas, and electricity, intelligent applications are constructed including intelligent assistance for agent business acceptance, intelligent customer service, intelligent power supply plan generation, comprehensive energy use plan generation, intelligent dispatch of operation and maintenance work orders, and user energy anomaly diagnosis. This creates an interactive and accompanying new customer service model, enhancing the intelligent service level throughout the customer process.

(3) Artificial Intelligence + New Energy. Addressing the volatility and intermittency of new energy output, accelerate AI applications in high-precision power forecasting, electricity markets, smart operation of stations, new energy planning, and post-project evaluation. Continuously promote iteration and innovation of key new energy materials and products. Advance the development of large power forecasting models for complex scenarios and transitional weather toward smaller scales and higher accuracy, support wide-area new energy resource collaborative optimization, promote intelligent operation and maintenance of remote new energy stations, and create an integrated new energy intelligent production model of "weather forecasting + power forecasting + smart trading + intelligent operation and maintenance," fully supporting stable new energy supply.
Column 3: Typical Application Scenarios of Artificial Intelligence + New Energy

Meteorological Forecasting and Accurate New Energy Power Prediction. Build a meteorological service system centered on multi-temporal and spatial scale weather forecasting, establish multi-scenario multi-period algorithm large models for precise mining and analysis of nonlinear meteorological-power relationships, achieving accurate new energy power prediction.

Intelligent Operation and Maintenance of Remote Area Stations. Using technologies and equipment such as large models, voiceprint detection, remote sensing, robots, and smart wearable devices to monitor surrounding environments and equipment operation status in real time. Achieve multi-system intelligent linkage including drones, unmanned vehicles, unmanned boats, and intelligent control, improving equipment inspection efficiency and enhancing comprehensive operation efficiency of stations.

New Energy Planning and Design. Considering factors such as power generation efficiency and investment return rate, build an intelligent recommendation engine to provide optimal model matching solutions. Integrate large models with design software to quickly generate multiple design versions and evaluate key parameters, improving design efficiency and quality.

Smart Construction Site Development. Promote deep integration of artificial intelligence technology into the entire process of power construction project management including engineering scheme selection, personnel management, risk warning, and schedule control. Develop drone inspection systems, automatic risk judgment and warning systems, etc., to capture construction personnel violations in real time, building a "smart construction site" management platform throughout the construction process, helping to improve overall safety and quality of power construction projects.

(4) Artificial Intelligence + Hydropower. Focus on intelligent construction of hydropower projects in high-altitude and cold regions and smart dispatch operation of watershed hydropower station groups. Promote AI applications in hydropower project construction to enhance intelligent design and construction management levels. Advance integration of AI with traditional hydrological models, meteorological models, and large-scale reservoir dispatch technologies to improve accuracy of meteorological and hydrological bidirectional coupled forecasts. Develop intelligent applications for dispatch decision optimization. Promote integration of knowledge graphs, large models, intelligent agents, and other technologies into the new generation hydropower smart operation brain, forming intelligent solutions in key areas such as smart operation and maintenance of hydropower stations, lean maintenance, intelligent dam situational awareness, and smart management.
Column 4: Typical Application Scenarios of Artificial Intelligence + Hydropower

Intelligent Hydropower Project Construction. Based on multi-source remote sensing data fusion and AI technologies such as intelligent robots, establish an intelligent geological survey and design system for hydropower projects. Achieve digital and intelligent installation and commissioning of unit equipment, improving intelligent construction management levels of hydropower projects.

Joint Meteorological and Hydrological Forecasting. Based on watershed meteorological and hydrological bidirectional coupled forecast large models, build risk quantification tools for extreme flood and drought events. Fully integrate meteorological knowledge, hydrological knowledge, and watershed geographic information to improve accuracy and lead time of meteorological and hydrological forecasts.

Comprehensive Watershed Dispatch. Based on key technologies such as joint smart optimization dispatch, risk control, and simulation of watershed station groups, build precise dispatch decision optimization intelligent applications. Achieve real-time monitoring, analysis, and evaluation of water resource dispatch plan execution, optimize water resource allocation in time and space, improve water energy utilization, and increase power generation benefits.

Intelligent Equipment Operation and Inspection. Based on multi-source data including physical fields, acoustics, vision, intelligent sensors, and technologies such as knowledge graphs and large models, promote full-process intelligent upgrades in key hydropower equipment for status holographic monitoring, full lifecycle health management, intelligent operation and maintenance, and condition-based maintenance. Realize structured management of operation and maintenance knowledge and intelligent auxiliary decision-making systems based on large model-intelligent agents.

High-Quality Dam Operation. Build a database and knowledge graph of typical dam disease features. Combine intelligent dam sensing, fusion, diagnosis, and prevention-control theoretical methods to achieve multi-driven early identification, self-diagnosis, adaptive warning, and intelligent feedback control of dam safety status. Ensure safe operation of hydropower station dams and support high-quality reservoir dam operation management.

(5) Artificial Intelligence + Thermal Power. Focusing on clean carbon reduction, safety and reliability, efficient regulation, and intelligent operation of thermal power, collaboratively carry out AI empowerment and technological innovation in fuel control, production operation optimization and intelligent control, and full lifecycle equipment management business scenarios. Accelerate digital design, construction, and intelligent upgrades of thermal power, promote intelligent development and application of thermal power operation control systems, enhance intelligent monitoring and health management capabilities of key thermal power equipment throughout their lifecycle, and help further improve thermal power support and assurance capabilities.
Column 5: Typical Application Scenarios of Artificial Intelligence + Thermal Power

Intelligent Fuel Control. Based on multi-dimensional and multi-type data such as fuel market price fluctuations, inventory, coal consumption, 3D structure of coal piles, and coal quality analysis, use advanced sensing, image recognition, rule understanding, and intelligent agent technologies to achieve intelligent detection and control of fuel quantity and quality.

Production Operation Optimization. Based on large models and production operation related system data, realize intelligent upgrades of core business scenarios such as fuel blending, operation optimization, intelligent flexible peak regulation, and safe intelligent control during production operation, improving the intelligence level and efficiency of production operation.

Full Lifecycle Equipment Management. Based on large models and AI technologies such as robots, conduct real-time status monitoring of key equipment including turbines (including gas turbines), generators, and boiler heating surfaces through multi-type data. Achieve panoramic equipment status monitoring, health quantitative assessment, hazard identification and fault warning, remaining life prediction, operation plan adjustment, anomaly analysis and judgment, and closed-loop hazard management.

Intelligent technology supervision and evaluation. Relying on massive operational data of key equipment such as boilers, steam turbines (including gas turbines), and generators, as well as relevant materials for thermal power technology supervision, based on the multimodal analysis capabilities of the thermal power large model, deeply integrating thermal power characteristic scenarios to enhance the intelligence of technology supervision and the professional capabilities of personnel.

(6) Artificial Intelligence + Nuclear Power. Focusing on the safe development of nuclear power, build intelligent auxiliary support systems for nuclear power safety early warning, intelligent traceability analysis of power plant operation events, and emergency response. Conduct key research on special operation and maintenance robots for the nuclear industry, continuously promote technological upgrades such as automatic start-stop of nuclear power systems, explore AI-assisted plasma prediction and control, controlled nuclear fusion, and other technical paths, and steadily promote the nuclear power industry’s transformation towards a new model driven by data, guided by models, and intelligent management and control.
Column 6: Typical Application Scenarios of Artificial Intelligence + Nuclear Power

Nuclear Power Intelligent Safety Management and Control. Utilizing data governance and artificial intelligence technologies, focusing on operation event traceability, technical specifications, and operational parameter boundary conditions, intelligently identifying unsafe states of personnel, equipment, and environment, advancing technical research and application in safety early warning and intelligent emergency response scenarios.

Nuclear Power Intelligent Operation and Maintenance. Using data from structures, systems, and equipment/components at various stages, establish data-driven nuclear power plant models, promote the development of small AI models and professional large models for nuclear power, advance the application of AI technology in intelligent monitoring, early warning, diagnosis, and prediction of nuclear power systems, improve intelligent diagnosis and optimization capabilities of units, enhance one-click start-stop capabilities of key equipment, systems, and units, and expand the scope and depth of robotic operations in high-risk scenarios such as high radioactivity, underwater, and confined spaces.

Intelligent Control of Controlled Nuclear Fusion. Combining the multi-physical field coupling characteristics of controlled nuclear fusion devices, conduct research on intelligent control systems for controlled nuclear fusion based on AI technology, develop intelligent models for real-time plasma shape prediction and adaptive regulation of magnetic confinement parameters, achieving intelligent control of steady-state operation of tokamak plasma.

(7) Artificial Intelligence + Coal. Focusing on typical scenarios such as geological exploration, coal mining (stripping), coal washing and selection, production scheduling, safety management, and equipment management, stably acquire various operational data under complex geological, multi-working condition, and multi-spatiotemporal coordinated conditions, integrate and apply intelligent models to realize intelligent control and autonomous decision-making in the production process, assist in normalizing low- or no-person operation, steadily promote reduction of personnel, increase safety, and improve efficiency, further consolidating coal’s role as a bottom-line guarantee in energy security.
Column 7: Typical Application Scenarios of Artificial Intelligence + Coal

Digital Intelligence Empowerment of Coal Mine Geological Exploration. Based on professional large models for coal mines, integrating new technologies of high-precision surface exploration and dynamic intelligent underground detection, build a coal mine geological database under complex geological conditions, realize full-process dynamic collaborative management and early warning of mine geological information, ensuring efficient, rapid, green, and intelligent mine production.

Optimization and Intelligent Control of Underground Coal Mining Processes. Through multimodal perception, fusion of large and small models, collaborative control of equipment groups, and dynamic process optimization, extract coal-rock feature information, drive intelligent cutting, autonomous decision-making, and collaborative control of coal mining and tunneling face equipment groups, achieve autonomous operation of coal mining face production systems, efficient collaboration of exploration-tunneling-support-anchoring-transport at tunneling faces, and normalized low- or no-person operation, significantly improving mining efficiency and safety levels.
Autonomous Mining, Loading, and Transportation Unmanned Operation in Open-pit Coal Mines. Promote the integration of large model simulated blasting parameters with blasting operations, apply AI technology to quickly analyze mining and stripping progress, realize normalized remote control or autonomous operation of excavators, spoil bulldozers, and other auxiliary equipment within the mining-transportation-disposal production system, as well as large-scale unmanned operation of mining trucks, enhance the intelligence and accuracy of blasting, greatly reduce the number of personnel working underground, and improve production efficiency and safety levels of open-pit coal mines.

Rapid Coal Quality Detection and Intelligent Washing and Selection. Collect and build a coal quality characteristic database, dynamically predict key indicators such as coal ash, sulfur, volatile matter, moisture, and elemental content in real time, achieve intelligent recognition of coal quality characteristics, greatly improve the accuracy of online coal quality detection, provide real-time feedback of online coal quality detection data, optimize and adjust coal washing production process parameters, improve the qualification rate and stability of coal products. Develop professional models for coal washing and selection, establish industrial digital twins, realize dynamic information monitoring, trend prediction, and collaborative management throughout the coal washing and selection process.

Major Equipment Status Monitoring and Intelligent Operation and Maintenance in Coal Mines. Establish large models integrating real-time operational status and lubrication, temperature, and vibration detection data of major equipment, achieve fault diagnosis and intelligent early warning, promote preventive maintenance of coal mine equipment, significantly reduce production downtime caused by faults, and effectively lower maintenance costs.
(8) Artificial Intelligence + Oil and Gas. Focusing on cross-disciplinary collaborative research, on-site operation control, and production operation management, promote intelligent evaluation of exploration geological targets, intelligent optimization of development plans, intelligent adjustment of drilling and fracturing operation parameters, intelligent operation of refining units, real-time simulation of pipeline network operation, accelerate the research and application of intelligent production technology equipment such as intelligent drilling rigs, robots, drones, and intelligent sensing systems, promote intelligent linkage and automatic optimization throughout the production site, and drive the intelligent upgrading of the oil and gas industry chain.
Column 8: Typical Application Scenarios of Artificial Intelligence + Oil and Gas

Intelligent Empowerment of Oil and Gas Exploration. Improve the software intelligence level for exploration professional fields such as seismic, logging, and core outcrop, build professional large models for seismic logging processing and interpretation, create intelligent application systems for comprehensive evaluation of favorable geological targets, and realize demonstration applications of robots such as intelligent assisted driving of controlled seismic sources and seismic geophone placement.
Intelligent Management and Control of Oil and Gas Reservoir Development and Production. Develop intelligent technologies for oil and gas development data and knowledge, intelligent development optimization software, and professional large models, build a large model-driven collaborative research and production management decision-making platform, and establish a new model for smart oil and gas field development and production management and control.

Prediction and Maintenance of Marine Oil and Gas Production Environment. Focusing on environmental protection and major risk prevention and control needs in marine oil and gas production processes, through intelligent monitoring and anomaly early warning of the production environment, intelligent management and control of solid waste treatment, intelligent identification and emergency prediction of oil spills, form an integrated capability covering risk prediction, situational awareness, early accident detection, and cognitive decision-making for the entire ecological environment of oil and gas fields.
Intelligent Optimization of Engineering Technology. Promote intelligent design of surface engineering, intelligent optimization of drilling parameters, real-time intelligent lithology identification during logging, reservoir modification, intelligent fault diagnosis, and risk assessment, realize demonstration applications of well control robots, ensuring safe and efficient construction in complex geological environments.

Pipeline Network Simulation and Intelligent Regulation. Promote market insight and forecasting, real-time simulation and dynamic optimization of pipeline networks, efficient intelligent operation of stations and storage, integrated management of air, ground, and space lines, and monitoring and early warning of key equipment, achieve "black screen" intelligent regulation, and enhance the safety production, oil and gas supply security, and fair service capabilities of oil and gas pipeline networks.

Integrated optimization of refinery production and operation. Focusing on full-process plan optimization, intelligent safety production identification, preventive equipment maintenance, and other aspects, tackling scientific computing large models for new material research and development. Through collaboration between large and small models and hybrid modeling techniques, reduce process fluctuations, lower the probability of safety incidents, and enhance the intelligence level of production operations.
3. Increase the supply of key technologies

Focusing on technical bottlenecks such as data silos in the energy sector, fragmented computing power, algorithm black boxes, and high energy consumption of computing power, promote the development of common key technologies applicable to data, computing power, and algorithms in the energy field.

(1) Strengthen the data foundation. In response to the construction of high-quality data sets and data security requirements in the energy sector, promote the application of technologies such as intelligent data annotation, intelligent enhancement, and data synthesis. Advance the development of energy data classification and grading technologies, privacy computing technologies, intelligent dynamic data encryption, and cross-domain trusted traceability technologies. Optimize data sharing mechanisms, accelerate the formation of high-quality data sets in the energy sector, and ensure the full-process security and reliability of energy data.

(2) Strengthen computing power support. Addressing the needs for multi-source heterogeneous computing power integration and utilization under the combined rental and construction model in the energy sector, carry out key technology research on unified scheduling of multi-source heterogeneous computing power, intelligent task orchestration, integrated storage-computing-network fusion, and computing power pooling to improve intelligent computing service levels. Continuously monitor energy computing power demand, coordinate planning of computing power, electricity, and communication network resources, build a deeply integrated computing power and electricity collaborative development mechanism, and continuously increase the proportion of green electricity in computing power centers.

(3) Enhance foundational model capabilities. In response to the energy sector's needs for model security and interpretability, promote the construction of security capabilities for model algorithms and application systems. Increase research on multi-agent collaboration, interpretability, and lightweight model inference technologies. Deepen the application research of key artificial intelligence technologies such as machine vision, multimodal, and time series prediction in the energy sector. Promote deep integration of artificial intelligence with energy sector software. To address the energy consumption issue of AI computing, accelerate breakthroughs in green and low-carbon AI technologies, research energy supply technologies such as flexible DC power supply and modular small reactors, and encourage the application of efficient energy comprehensive utilization technologies such as data center liquid cooling, waste heat recovery, and intensive backup power.
4. Safeguard Measures

(1) Strengthen organizational implementation. Local energy authorities and related central enterprises should establish and improve working mechanisms according to the requirements, coordinate relevant plans, and accelerate the development of "AI+" energy in their regions and units based on actual conditions. Ensure all necessary elements are secured, explore the construction of a safety governance system, form a work pattern of coordinated top-down implementation and safe development, and accelerate the research, demonstration, and promotion of AI integration applications in the energy sector.

(2) Promote collaborative innovation. Focus on key common technologies and supporting specialized technologies for AI integration and innovation applications in the energy sector, and promote the construction of a number of industry R&D innovation platforms. Encourage enterprises to lead in collaboration with research institutions, universities, and social service organizations to build cross-field and interdisciplinary "AI+" energy innovation alliances aimed at technological innovation and integration applications. Deepen industry-university-research-application cooperation and build an open, collaborative, co-creative, and shared intelligent energy innovation ecosystem.

(3) Strengthen standards and norms construction. Based on in-depth summary of application demonstration practices, accelerate the formulation of a batch of technical standards and norms such as energy data governance, multi-source heterogeneous computing power integration, and typical scenario design. Promote the construction of AI standards system in the energy sector, explore the establishment of AI application evaluation indicator systems and industry-level AI application standard testing platforms, and improve the safe application level of AI technology in the energy sector. Encourage energy enterprises to lead the formulation of international standards, using technical standards to "go global" and drive the promotion and application of AI technologies and products in overseas energy markets.

(4) Conduct pilot demonstrations. Organize pilot demonstrations of AI applications in the energy sector, select a batch of replicable and easily promotable scenarios and enterprise benchmark applications. Encourage demonstrations of typical cross-field and cross-industry scenarios such as energy and transportation integration, oil and gas and new energy integration. AI-related technical equipment in the energy sector should be given priority inclusion in the scope of support for the first set of major technical equipment in the energy sector. Support qualified regions and enterprises to carry out various AI application pilots in the energy sector tailored to local conditions, deeply explore and pilot in aspects such as technological innovation, business models, development formats, and institutional mechanisms.

(5) Increase support efforts. Fully leverage the role of central government financial funds, rely on national major science and technology projects and key R&D plans in the energy and AI fields, and orderly promote AI technology application innovation in the energy sector. Utilize the multi-level capital market to support key hubs of technological innovation, guide social capital participation in AI technology project implementation and achievement transformation and application.

(6) Improve talent cultivation ecosystem. Encourage energy enterprises to jointly build "AI+" energy talent training bases with higher education institutions and research institutes. Design interdisciplinary curricula oriented by industry needs, focus on cultivating compound talents with knowledge of energy systems and AI algorithm application capabilities, and increase the supply of compound talents through industry-education collaboration.

Source: National Development and Reform Commission National Energy Administration