Policy aligns clean power, computing infrastructure and AI innovation, targeting global leadership by 2030 while reshaping energy demand, digital infrastructure and industrial competitiveness.
On May 8, China’s central government officially released the Action Plan for Promoting Bidirectional Empowerment between Artificial Intelligence and Energy (NEA Dispatch on Science & Technology [2026] No. 34, dated April 8, 2026).
Jointly issued by NDRC, NEA, MIIT, and the National Data Administration, the plan sets out a comprehensive framework to integrate AI development with the energy system. It represents a decisive policy effort to align China’s rapidly expanding computing demand with its clean energy transition.
The plan outlines a roadmap to build a “safe, green and cost-efficient” energy system capable of supporting AI at scale by 2027, with global leadership targeted by 2030. At its core, it addresses a structural challenge facing major economies: the exponential growth in AI computing power—particularly for large language models and industrial applications—requires vast, reliable electricity supply, often in tension with decarbonisation and energy security objectives.
China’s response is to tightly couple energy infrastructure planning with computing deployment, effectively positioning data centres and AI clusters as integral components of the national energy system.
A coordinated policy linking power and compute
The action plan establishes a dual objective: deploying AI to optimize energy systems while simultaneously adapting energy systems to support AI’s accelerating demand. It emphasizes coordination across five key elements—energy, computing power, data, application scenarios, and AI models—to drive system-wide efficiency gains.
By 2027, China aims to establish a foundational system in which clean energy and computing infrastructure interact dynamically, supported by improved utilization of computing resources and mechanisms for sharing high-value energy datasets. By 2030, the country targets global leadership in both clean energy supply for AI computing and AI-enabled energy applications.
A defining feature is the co-location strategy: large-scale data centres and national computing hubs will increasingly be deployed in regions rich in renewable resources, particularly wind and solar bases in western China. While similar approaches exist in Nordic countries, China’s scale and level of central coordination are likely to be unmatched.
China’s 2026 Government Work Report includes “computing power–electricity coordination” as a core task of new infrastructure development, requiring that new computing facilities at national hub nodes use at least 80% clean electricity.
From grid-forming storage to hydrogen-enabled power supply
The plan outlines a suite of technical measures to stabilize and decarbonize power supply for AI infrastructure.
A central pillar is the diversification of energy sources for data centres, including not only grid electricity but also direct connections to nuclear and hydrogen-based systems. Hydrogen is positioned as a potential zero-carbon alternative to diesel backup generation, offering longer-duration storage and improved resilience, although commercial deployment remains at an early stage.
The policy also advances “grid-forming” energy storage technologies, which can actively regulate voltage and frequency rather than passively respond to grid signals—an important capability for hyperscale data centres with stringent power quality requirements.
In parallel, the plan calls for enhanced power quality management through advanced monitoring systems, real-time risk detection and specialized power conditioning equipment tailored to AI workloads, which are particularly sensitive to voltage fluctuations and transient disturbances.
Efficiency improvements form a second major pillar, with priorities including:
- High-efficiency cooling systems, notably liquid cooling for high-density AI chips
- AI-optimized server architectures and power supply systems
- Waste heat recovery for district heating and industrial reuse
- Exploration of next-generation low-power computing technologies, including photonic and quantum systems
These targets align with global best practice. Leading operators such as Google and Microsoft have already achieved power usage effectiveness (PUE) levels near 1.1; China’s policy aims to narrow this gap across its broader data centre base.
Early progress is evident. On May 6, construction began on the Wuxi Computing Equipment Industrial Park, described as China’s first mass-produced modular AI data centre (AIDC) system for large-scale deployment with a PUE below 1.2, an important step towards standardized, energy-efficient infrastructure.
Scaling green power for exponential AI demand
The commercial implications are significant. China already hosts the world’s largest data centre capacity, and AI-driven demand is expected to accelerate sharply. By linking data centre expansion with renewable energy deployment, the policy creates a structural demand anchor for clean power. GW-scale AI clusters could function as flexible loads, absorbing excess renewable generation and improving utilization rates for wind and solar assets—particularly in regions facing curtailment challenges.
The plan also introduces mechanisms to increase green electricity consumption, including participation in green power trading and renewable certificate markets, as well as “direct green power connections” that allow data centres to procure renewable energy more efficiently.
From a cost perspective, this integration could enhance competitiveness. Renewable generation costs in China are now among the lowest globally, and co-location reduces transmission losses and grid congestion, lowering total operating costs for compute infrastructure.
A blueprint with global resonance
China’s approach reflects a broader shift in how governments and industries conceptualize AI infrastructure—not merely as a digital asset, but as a core component of the energy system. By embedding computing demand into energy planning, the policy addresses both the risks and opportunities of AI-driven electricity growth.
For the energy sector, the policy also unlocks new application pathways. High-value AI use cases, ranging from grid optimization and renewable forecasting to predictive maintenance, are expected to scale through structured scenario development and data-sharing mechanisms.
For global cleantech and hydrogen stakeholders, the implications are clear: the next phase of the energy transition will be shaped not only by electrification and renewables, but also by the demands of computation itself. China’s strategy offers an early, large-scale test case of how these forces can be aligned and where new value pools may emerge across the energy–digital value chain.