China’s AI boom is consuming electricity at a pace that is beginning to strain the country’s power grid and carbon targets. Electricity consumption by internet data services and computing facilities rose 44 percent year on year in the first quarter of 2026, reaching 22.9 billion kilowatt-hours, according to the China Electricity Council. The China Academy of Information and Communications Technology estimates that consumption by computing centers could exceed 400 billion kilowatt-hours annually by 2030, and power costs already account for between 50 and 70 percent of operating expenses at some data centers.
The South China Morning Post reports that four national bodies, the National Energy Administration, the National Development and Reform Commission, the Ministry of Industry and Information Technology, and the National Data Administration, jointly released an action plan to address this challenge. The document proposes 29 measures to improve the integration of green power and AI infrastructure, and sets a goal to “build a new development pattern of mutual empowerment and deep integration between AI and energy” by 2030.
The 29 Measures: What the Plan Actually Says
The action plan encourages data center operators to continuously raise the share of green electricity in computing facilities through so-called “green certificate” and “green power trading markets.” It also promotes cleaner backup power systems to replace diesel generators, a significant source of emissions at facilities that require uninterrupted uptime.
On the hardware side, the plan promotes the adoption of domestically developed AI software and hardware in the energy sector, including the optimization of domestic AI chips for energy applications. This aligns with Beijing’s broader push for chip self-reliance and its call for Chinese tech companies to favor domestic semiconductors over foreign alternatives.
The plan is part of a wider effort to coordinate the expansion of computing capacity with renewable energy development while maintaining China’s carbon reduction commitments. China’s 2026 government work report, presented by Premier Li Qiang at the “two sessions” in March, pledged for the first time to better align power supply with the expansion of AI computing under the country’s new infrastructure agenda.
Why the Timing Matters
The surging demand for AI compute has accelerated efforts to shift data-center expansion toward China’s northern and western provinces, where wind and solar resources are more abundant and electricity prices are lower. Beijing’s action plan formalizes and accelerates this geographic rebalancing by making green energy use a key metric for approving and operating new data center projects.
Similar concerns over AI-driven electricity demand have emerged in the United States and Europe, where utilities and regulators have warned that rapid data-center expansion could strain power grids. China’s coordinated four-agency response represents one of the most comprehensive government-level attempts to address this challenge at scale.
For AI infrastructure companies, the plan signals that green energy compliance will increasingly be a prerequisite for new data center approvals, not an optional commitment. This is likely to reshape capital allocation decisions across China’s cloud computing sector, which includes Alibaba Cloud, Tencent Cloud, Baidu, and Huawei Cloud.
Implications for China’s AI Ecosystem
The emphasis on renewable energy dovetails with ongoing efforts in regions like Inner Mongolia and Qinghai, where vast solar and wind farms are being developed to supply clean power to hyperscale data centers. This trend echoes the strategic shift explored in our article on China turning to subsea data centers and offshore wind to ease AI computing bottlenecks.
The plan’s promotion of domestic AI chips in the energy sector also connects to the broader chip self-reliance agenda. As EastFrontier has reported, Chinese tech companies have been rapidly shifting procurement toward Huawei Ascend and Cambricon chips as US export controls restrict access to Nvidia hardware.
Adapting existing AI infrastructure to meet green mandates will require significant upfront investment and technical upgrades, potentially increasing short-term operating costs. This may prompt cloud providers to revise pricing models, as suggested by recent reports on China’s AI compute price wave.
Challenges Ahead
Despite the clear policy intent, the plan faces real implementation challenges. Renewable energy sources are intermittent, posing difficulties for data centers that require guaranteed uptime. Creating reliable green power supply chains requires investment in energy storage and smart grid technologies that are still maturing.
There is also the question of enforcement. The plan involves four separate agencies and touches on the operations of hundreds of data centers across multiple provinces. Effective coordination and compliance monitoring will be essential if the 29 measures are to translate into measurable emissions reductions.
For China, the stakes are high on both sides of the equation. Its AI ambitions require ever-larger computing facilities, while its climate commitments demand that those facilities run on clean energy. The May 9 action plan is the clearest signal yet that Beijing intends to pursue both goals simultaneously and expects the industry to follow suit.
For more on China’s evolving AI policies and infrastructure trends, see our recent coverage of Beijing’s AI chatbot regulations and Tencent’s AI investment commitments.