On October 28, Beijing issued the notice on the "Planning for a New Energy System During the '15th Five-Year' Period." The plan explicitly states that it will coordinate energy resources with computing power layout to improve the quality of energy supply for computing facilities. New data centersmust use 100% green electricity. It promotes energy-saving and carbon-reduction retrofits for infrastructure such as computing facilities, communication base stations, and equipment rooms. Existing data centers must achieve a Power Usage Effectiveness (PUE) below 1.35, with energy efficiency and carbon efficiency per unit of computing power reaching international advanced levels.
The ultimate constraint on computing power is electricity. By embedding energy constraints for computing infrastructure into the energy system, the policy mandates that new data centers use 100% green electricity and requires existing data centers to keep PUE below 1.35. For the telecommunications and computing industries, this means bundling computing power layout, green electricity supply, trading mechanisms, and energy efficiency retrofits together. The project logic for data centers is shifting from "land acquisition, rack installation, and scale expansion" to "how computing power and electricity can collaborate."
Green Electricity Shifts from an "Add-on" to an "Entry Requirement"
In the past, when data centers discussed sustainability, they focused on PUE optimization, renewable energy ratios, and Green Electricity Certificates (GECs). Green initiatives were largely seen as operational optimizations or brand enhancements. Beijing's current plan lists the requirement for new data centers to use 100% green electricity as part of the energy system, making green electricity supply a prerequisite for projects. Without stable sources of green electricity and established trading arrangements, computing power projects are unlikely to pass the review for coordinated energy resource planning.
For operators and IDC enterprises, site selection no longer depends solely on land, network latency, and electricity prices; it now hinges on green electricity channels, trading contracts, and peak-shaving capabilities. As a load center with limited local green electricity, Beijing's plan simultaneously proposes expanding green electricity transmission into the city, building bases outside the municipal area, and promoting large-scale inter-provincial and cross-regional trading. This effectively places "external electricity entering Beijing" and "computing power entering Beijing" on the same strategic map.
Newly built data centers must address questions such as: Where does the green electricity come from? How is delivery settled? How can supply match the load curve?
Computing-Electricity Synergy Is Not Simply "Buying Green Electricity"
The challenge of achieving 100% green electricity lies not in signing a procurement contract, but in the spatiotemporal matching behind the "100%." Wind and solar power fluctuate, while computing loads require real-time stability. Inter-provincial trading involves transmission channels and time slots, and green electricity certification has specific criteria. The plan mentions orderly promotion of staggered green electricity trading and time-of-use trading at a large scale across provinces and regions, providing a mechanism for matching computing loads with green electricity curves.
Spatially, green electricity bases outside the municipal area connect to Beijing's data centers through cross-regional trading;
Temporally, time-of-use trading and staggered scheduling allow shiftable computing tasks to be concentrated during periods of green electricity surplus;
In terms of load, intelligent scheduling, energy storage, and demand response transform data centers from rigid loads into adjustable loads. AI training and other batch-processing tasks offer high scheduling flexibility, whereas inference and real-time services have limited flexibility and require more precise capacity planning. Communication base stations and equipment rooms are widely distributed and individually small, presenting different challenges for retrofitting and scheduling. However, the plan lists them alongside data centers, indicating that energy efficiency management for communication infrastructure is evolving from single-point operations to systematic energy management.
Short-term Cost Increases Trade Off for Long-term "Electricity + Carbon Efficiency" Competitiveness
Enterprises are most concerned about costs. Mandating 100% green electricity introduces several types of additional expenses: premiums for green electricity trading and GECs, service costs; organizational costs for inter-provincial and cross-regional trading; investments in energy storage, peak shaving, or load scheduling systems to match green electricity curves; and retrofitting costs to reduce existing data center PUE below 1.35, including cooling systems, airflow organization, power distribution, and intelligent O&M. For older server rooms, retrofitting costs may exceed those of new construction and could impact continuous operation.
A reduction in PUE directly decreases non-IT power consumption, freeing up usable racks and IT capacity. Procuring green electricity locks in long-term prices, reducing exposure to electricity price volatility. Against the backdrop of stricter energy consumption quotas and carbon efficiency reviews, the cost model for data centers should shift from "construction cost per rack" to "total lifecycle cost per effective unit of computing power," incorporating electricity fees, green electricity premiums, depreciation of retrofits, and carbon costs into a unified framework. Beijing's goal of achieving international advanced levels in energy and carbon efficiency per unit of computing power effectively shifts the competitive dimension from scale to efficiency.
For the telecommunications industry, higher entry barriers for new data centers favor operators, third-party IDCs, and computing service providers with capabilities in green electricity procurement, energy management, and intelligent scheduling. Leading enterprises can dilute costs through centralized green electricity procurement, participation in inter-provincial trading, and deployment of liquid cooling and intelligent O&M. Smaller projects may exit due to pressure on green electricity access and retrofit funding, potentially increasing industry concentration.
Conclusion
By mandating 100% green electricity usage for new data centers in its energy plan, Beijing presents enterprises with short-term challenges regarding green electricity premiums and retrofitting costs. In the long term, capabilities in green electricity acquisition, PUE optimization, and computing-electricity synergy will determine the economic viability and market entry eligibility of computing power projects. The shift of the computing industry from "scale expansion" to "green electricity synergy" represents a full-chain adjustment involving project approval, site selection, trading, O&M, and business models.
Editor/Layout: Sun Tian
Proofreader: Mei Yaxin
Supervisor: Liu Qicheng