The essence of the synergy between computing power and power systems is the bidirectional deep coupling of computing power infrastructure and new power systems, realizing the closed loop of "empowering computing with power and promoting power with computing". In 2026, this synergy was written into the Government Work Report for the first time and included in the "15th Five-Year Plan". It is a top-level strategic deployment to coordinate the development of the digital economy, ensure energy security, implement the "dual carbon" goals, and build a unified national market.
Driving Forces for the Synergy Between Computing Power and Power Systems
From the power supply perspective, given the close integration of AI with social production and daily life, power consumption for AI computing power is growing exponentially. According to public information from the National Energy Administration, the power consumption of national computing centers reached 170 billion kWh in 2025, with a 3-year growth rate of 39.5%, far exceeding the growth rate of total societal power consumption. It is projected that by 2030, computing power consumption will exceed 700 billion kWh, accounting for over 5% of total societal power consumption. The power consumption of a single large-scale intelligent computing cluster is equivalent to that of a medium-sized town, and high-load services such as AI large model training and video generation continue to drive up power loads.
Despite the surging power demand for AI computing power, the consumption of renewable energy power faces difficulties. The installed capacity of wind and solar power in China accounts for over 56% of the national total installed capacity. However, wind and solar power output fluctuates with the weather, and local industrial loads in western renewable energy bases are insufficient. Coupled with the volatility of renewable energy generation, a large amount of clean power cannot be consumed locally, resulting in energy waste.
From the computing power perspective, the practical demands for the synergy between computing power and power systems mainly stem from the following four aspects.
01. Computing power business demands continue to grow, power consumption of computing facilities keeps rising, and the bearing pressure on the energy system continues to increase;
02. Computing power infrastructure has increasingly stringent requirements for power quality, power supply stability, and system resilience, which is particularly prominent in data centers hosting critical digital services and public welfare applications;
03. The proportion of volatile renewable energy connected to the grid continues to rise, creating an urgent need for a large amount of flexibly adjustable power loads on the grid. Computing power loads possess the regulatory potential for dynamic dispatch and collaborative optimization;
04. Constraints on energy efficiency improvement, carbon emission reduction, and sustainable development are constantly tightening, forcing the deep linkage between digital infrastructure and energy infrastructure.
The synergy between computing power and power systems is not simply "building data centers next to power plants," but a systematic strategic deployment by the state to coordinate digital security, energy security, "dual carbon" goals, regional balance, and industrial competitiveness. In the short term, it aims to solve the bottlenecks of AI computing power consumption and the pain points of renewable energy consumption. In the medium and long term, it seeks to build a modern industrial system with bidirectional empowerment between digital and energy sectors, green and low-carbon development, and safe and efficient operations, seizing the commanding heights in the global development of AI and new power systems.
Main Modes of the Synergy Between Computing Power and Power Systems
Based on various factors such as resource endowment, existing/newly added capacity, and multi-entity collaboration, the synergy modes can be divided into physical infrastructure, flexible dispatch, market-oriented trading, and cross-regional strategic linkage.
Physical Infrastructure
Physical infrastructure synergy focuses on heavy-asset physical construction, relying on hardware carriers such as dedicated green power supply lines and integrated "source-grid-load-storage" microgrids. It focuses on western computing hubs to achieve physical binding, closed-loop energy supply, and local consumption of renewable power and computing clusters, representing the fundamental implementation form of deep coupling between computing and power.
Direct Green Power Supply Mode
This mode requires the construction of dedicated transmission lines, adopting point-to-point direct connections (i.e., without passing through the public distribution network) to transmit power generated by wind and solar stations directly to computing clusters. It realizes "power following computing power" and is mainly applied in western computing hubs.
Integrated "Source-Grid-Load-Storage" Microgrid Mode
The integrated "source-grid-load-storage" refers to the unified planning and integrated operation of renewable energy sources such as wind and solar power (source), regional internal distribution networks (grid), computing center power loads (load), and regional energy storage (storage). This forms a self-sufficient regional microgrid, namely a closed-loop "wind-solar-storage-computing" microgrid, with the main grid serving only as a backup.
According to regional scale, this mode can be divided into three implementation forms. First is the park level (most prevalent), where a single computing center is equipped with solar power and energy storage to uniformly allocate computing center power consumption, solar power generation, and energy storage within the park. Second is the city and county regional level, mainly involving the construction of large-scale wind and solar power bases to support computing new cities within city and county regions. Under this form, unified planning and integrated operation are required for the power loads of the computing new city, power generation sources, internal distribution networks of the city and county regions, and regional energy storage. Third is the cross-provincial mega-base level, mainly targeting provinces in the northwest rich in renewable energy, such as the 10-million-kilowatt-level wind and solar power supporting computing clusters in West Inner Mongolia and East Gansu.
Flexible Dispatch
Flexible dispatch synergy is an asset-light and rapidly deployable mode. It relies on virtual power plant (VPP) aggregated dispatch and cross-regional computing power migration scheduling to dynamically match the rhythm of power supply and demand, realizing "computing following power." It is the core implementation approach for computing-power and power synergy in eastern regions, grid peak shaving and valley filling, and optimization of resource mismatch. Under this mode, it can be divided into two implementation forms.
VPP Computing Power Aggregation and Peak Shaving Mode
A VPP is a digital dispatch and operation system that relies on IoT, AI, and communication technologies to aggregate dispersed power resources in buildings, factories, and parks into a single entity. It uniformly accepts grid dispatch and participates in power market transactions like a physical power plant, representing a new type of virtual power business entity. Relying on the unified VPP dispatch platform, it aggregates the power loads of multiple dispersed computing centers in a city, integrating regional power resources into a holistically adjustable resource to participate in grid peak shaving and valley filling, as well as the ancillary services market. This realizes computing power coordinating with the power rhythm, i.e., "computing following power."
Cross-Regional Computing Power Migration Dispatch
Cross-regional computing power migration dispatch mainly refers to the "East Data and West Computing" project. This project transmits data generated in the east but not requiring real-time response to large data centers in the west for storage and computation via high-speed communication networks. The east retains only low-latency real-time computing power, achieving resource complementarity between the east and west. The national integrated computing power network and ultra-high voltage (UHV) west-to-east power transmission facilities are crucial foundations for the "East Data and West Computing" project.
Market-Oriented Trading
Market-oriented trading synergy is the preferred mode for asset-light stock transformation. It does not require the construction of new physical transmission facilities. Relying on the public grid and green power market-oriented trading mechanisms, it locks in green power supply through long-term power purchase agreements (PPA) and green certificate confirmation, realizing the low-carbon energy consumption and precise market matching of computing centers.
Cross-Regional Strategic Linkage
Cross-regional strategic linkage adopts a cross-provincial joint computing-power and power dispatch platform to connect the grid dispatch system with the national computing power dispatch brain, realizing bidirectional real-time linkage and matching of power and computing power. On the power side, wind, solar, thermal, and energy storage outputs are optimized according to the computing power load curve; energy storage is charged during computing power valleys, and power is fully transmitted during computing power peaks. On the computing power side, clusters participate in cross-provincial peak shaving and frequency regulation as flexible loads, earning revenue from power ancillary services and further reducing computing power electricity costs.
Key Technologies for the Synergy Between Computing Power and Power Systems
The realization of the synergy between computing power and power systems relies on multi-faceted technological breakthroughs, involving multiple disciplines such as computing, power, and communications.
Computing-Power and Power Mapping Characterization
This is the foundational theory and technical support for the synergy. Specifically, it requires accurately constructing a dynamic mapping characterization model for computing power loads, clarifying the quantitative relationship between the operating status of computing tasks and power consumption and supply quality. This direction is currently in the stage of theoretical verification and preliminary pilot. Research institutes have completed basic model construction and simulation tests, enabling the simulation of computing-power and power coupling relationships in simple scenarios. However, there is still significant room for improvement in the accuracy, real-time performance, and fit with actual scenarios of the mapping characterization.
Joint Dispatch Technology for Computing Power and Power
The core goal is to realize "computing moving with power, and power used by computing." Relying on intelligent monitoring and decision-making dispatch systems, it captures the dynamics of computing power loads and fluctuations in renewable power supply in real time, achieving rapid decision-making and precise dispatch. This technology already possesses basic dispatch functions, but the overall system has not yet reached full maturity.
Reliable Power Supply Technology with High Proportion of Renewable Energy for Computing Centers
Computing centers have extremely high requirements for power supply reliability and stability, while renewable energy is inherently intermittent and volatile. How to achieve stable power supply for computing centers with a high proportion of renewable energy is a key pain point in the implementation of the synergy. Currently, relevant technical solutions have been applied in some data centers, capable of supplying a certain proportion of renewable energy, but a mature application mode for scalable promotion has not yet been formed.
Interactive Control and Green Power Aggregation Supply Technology
This technology is the core driver for promoting the efficient consumption of green power and realizing the green transformation of computing centers. It is also the key support for connecting green power trading and consumption accounting. Its core is to build a bidirectional interactive control mechanism between computing power clusters and renewable energy station clusters to achieve dynamic matching between the two. At the same time, it integrates dispersed renewable energy resources through green power aggregation technology to form a stable green power supply capacity.
Computing Power Network Technologies Supporting the Synergy
The synergy between computing power and power systems is a crucial driving force for the development of computing power networks, while computing power networks are the key support for realizing this synergy. As a new national digital infrastructure, the national integrated computing power network uses high-speed communication networks as physical links. Relying on unified computing power identification and unified dispatch standards, it connects dispersed general computing power, intelligent computing power, and supercomputing resources nationwide. It realizes unified perception, intelligent dispatch, and on-demand usage of heterogeneous computing power across regions, data centers, and vendors, solving problems such as "isolation, supply-demand mismatch, and idle waste" of computing power in various regions.
Key computing power network technologies supporting the synergy include computing-power and power perception, unified orchestration of heterogeneous computing power, cross-regional computing power hot migration dispatch, computing-power and power synergy dispatch brain, and computing-power and power synergy digital twin and trusted metering.
Computing-Power and Power Perception
Traditional communication networks only monitor physical parameters such as bandwidth and latency. In the synergy scenario, the computing power network must perceive computing power status and power status. Based on fused data, it constructs a joint computing-power and power evaluation model to quantitatively output scores for green power abundance and computing power availability at each node. This provides a unified data foundation for cross-regional computing power migration, flexible computing power load regulation, and source-load-storage collaborative dispatch, solving the pain points of information fragmentation between computing power networks and power systems and the lack of collaborative basis for dispatch, providing basic support for realizing "computing power allocated with green power."
Unified Orchestration of Heterogeneous Computing Power
Abstracting diverse computing power such as CPU (central processing unit), GPU (graphics processing unit), and FPGA (field programmable gate array) into standardized adjustable load units realizes the quantification, adjustability, and dispatchability of computing power loads, which is the core foundation for computing power participating in grid demand response. By binding computing power loads to fixed energy consumption curves, the grid can issue adjustment instructions just like regulating energy storage and industrial loads, increasing the load of computing power to increase power consumption or decreasing the load to reduce power consumption. The dynamic scaling of computing power clusters is linked with real-time power signals; computing power nodes are automatically expanded when green power is sufficient, and idle computing power is contracted when wind and solar output declines, achieving grid load peak shaving and valley filling from the computing power side.
Cross-Regional Computing Power Hot Migration Dispatch
This technology realizes the uninterrupted and imperceptible migration of computing power tasks across east and west computing power nodes, completing the spatial synergy of "computing power following green power," and solving the spatiotemporal mismatch between east-west computing power demand and green power resources. The implementation method is to build a layered architecture of "eastern edge real-time computing power + western central green power computing power": low-latency rigid demand businesses remain at eastern edge computing power nodes; massive offline batch computing is entirely diverted to western computing power hubs, achieving optimal matching of computing-power and power resources at the spatial level.
Computing-Power and Power Synergy Dispatch Brain
The dispatch brain is the unified intelligent hub of the computing power network and power system. It connects the three major systems of computing power dispatch, "source-grid-load-storage" regulation, and power market trading, realizing minute-level computing-power and power linkage dispatch. It mainly includes the following modules.
Computing-power and power multi-source data fusion foundation: It integrates four types of global real-time data, including computing power cluster operation and energy consumption data, wind and solar renewable energy output forecast data, grid load and real-time electricity price data, and energy storage charge/discharge status data. It builds a unified time-series data middle platform to support second-level global data retrieval and joint modeling.
Multi-time-scale computing-power and power joint prediction: Integrating meteorological large models and computing power business time-series models to achieve multi-scale precise prediction, predicting power supply-demand gaps in advance, and carrying out computing power migration and flexible load regulation in advance to avoid short-term power impact on the grid.
Computing power load participating in VPP aggregated regulation: Uniformly aggregating dispersed computing power clusters nationwide into a VPP adjustable resource pool. The computing power network platform receives grid dispatch and demand response invitation signals, quickly adjusts flexible computing power loads, and realizes grid peak shaving and valley filling.
Computing-Power and Power Synergy Digital Twin and Trusted Metering
Digital twin and trusted metering ensure that the entire process of computing-power and power dispatch is traceable, measurable, and carbon-accountable, adapting to the needs of green power traceability, ESG (Environmental, Social, and Governance), and renewable energy consumption assessment. It mainly includes the following capabilities.
High-real-time global digital twin for computing-power and power: Constructing virtual mirrors of all elements including computing power clusters, wind and solar stations, energy storage, and grid lines to map the operating status of the computing-power and power system in real time. It supports simulation pre-rehearsal, simulating in advance the dispatch effects of computing power migration and flexible load regulation under scenarios of wind and solar fluctuations and grid peak scenarios, avoiding the risks of computing power outages and power supply-demand imbalance.
Computing-power and power joint trusted metering and blockchain evidence storage: Connecting computing power energy consumption metering, wind and solar power generation metering, and green certificate transfer ledgers. All metering data is uploaded to the blockchain in real time and is untamperable, forming a complete traceability chain. Every kilowatt-hour of computing power electricity consumption can be traced back to the corresponding wind and solar power station generation, meeting the legal evidence requirements for zero-carbon computing power certification, corporate ESG disclosure, carbon footprint verification, and consumption weight assessment.
Challenges and Countermeasures
Currently, China has formed four types of implementation modes for the synergy: physical infrastructure, flexible dispatch, market-oriented trading, and cross-regional linkage. Related technologies such as computing power networks supporting the synergy are developing rapidly. However, the overall progress is still in the pilot stage, and there are multiple bottlenecks for large-scale implementation. The implementation challenges are concentrated in the following five aspects.
01 Prominent Barriers to Mode Promotion
In the west, heavy-asset modes such as direct green power supply and "source-grid-load-storage" involve large investments, long construction cycles, and mismatched computing-power and power infrastructure. In the east, asset-light modes such as computing power peak shaving and cross-regional migration are constrained by rigid computing power tasks, transmission latency, and electricity price differences, resulting in limited adjustable resources. The virtual green power mode faces issues such as the difficulty of integrating fragmented green power and an imperfect green power traceability and confirmation system. The cross-provincial high-level linkage mode faces industry system barriers, with imperfect dispatch and revenue mechanisms.
Insufficient Adaptation of Core Technologies02
Computing-power and power data standards are not unified, and perception accuracy is limited. There is a lack of standardized modeling and flexible regulation capabilities for computing power. Cross-regional migration has many constraints, and the intelligence level of integrated dispatch is insufficient. Digital twin and trusted metering technologies involve high implementation costs and narrow coverage.
03 Lack of Cross-Regional Governance Mechanisms
National unified coordination standards are imperfect, the rights, responsibilities, and revenue distribution mechanisms among multiple entities are unclear, and the industry regulatory system lags behind.
Prominent Multi-Dimensional Development Contradictions04
There is a conflict between the stable demand for computing power and the flexible dispatch of power. The cost of green transformation for computing centers is high, resulting in insufficient motivation for enterprises, and cross-regional data security risks are prominent.
05 Poor Connection of Market Systems
The four major markets for computing power, power, carbon, and green certificates are fragmented. Cross-provincial trading processes are cumbersome, and market-oriented incentive and constraint mechanisms are not sound.
In response to the above challenges, the following countermeasures can be considered.
First, classify and optimize implementation modes. In the west, improve the synchronous planning of computing power and power and multi-element investment and financing mechanisms. In the east, revitalize flexible computing power resources, improve the green power trading confirmation system, and establish a cross-provincial joint dispatch mechanism.
Second, iteratively upgrade core technologies. Unify computing-power and power data interoperability and computing power modeling standards, upgrade cross-regional migration and intelligent dispatch technologies, and popularize digital twin and blockchain trusted metering systems in a lightweight manner.
Third, improve the global governance system. Strengthen top-level coordination, unify industry standards, establish multi-party collaborative consultation and benefit-sharing mechanisms, and regulate industry supervision.
Fourth, coordinate security and benefits. Guarantee computing power stability in a classified manner, expand green revenue channels, and build a solid cross-regional data security defense line.
Fifth, improve market mechanisms. Break down the barriers of the four major markets for computing power, power, carbon, and green certificates, simplify cross-provincial trading processes, and improve market-oriented incentive and constraint policies.
Outlook
Looking ahead, the synergy between computing power and power systems will present four major development trends: First, various modes will be standardized and implemented at scale globally, forming a normalized operational ecosystem with east-west complementarity. Second, technologies such as AI and digital twins will be deeply integrated to realize autonomous and imperceptible synergy of computing-power and power systems, solving the problem of spatiotemporal mismatch of resources. Third, an integrated trading system for computing power, power, and carbon will be built to form a sustainable market-oriented business closed loop. Fourth, the global governance system will be improved to deeply empower the construction of new power systems and the digital economy, becoming a core new business format that assists the "dual carbon" goals and realizes coordinated regional development.
Authors: Tang Xiongyan, Jia Xueqin, Zhang Yan, Cao Chang, China United Network Communications Co., Ltd. Research Institute
This article was published in Issue 8, 2026 of "Communications World", with the original title "Main Modes and Key Technologies of Computing-Power and Power Synergy".
Managing Editor/Layout: Zhu Wenfeng
Proofreader: Mei Yaxin
Supervisor: Liu Qicheng