China’s AI Boom Strains Power Grid as Data Centers Surge
China’s rapid artificial intelligence expansion is placing unprecedented strain on the country’s power grid, as data centers and AI computing facilities demand massive and increasingly volatile amounts of electricity, experts warned at a Wednesday industry forum in Beijing.
The country is still in the early stages of coordinating AI computing demand with electricity supply, and fast-growing data centers are beginning to pose new challenges for grid planning and operations, according to Caixin Global. The issue is becoming more urgent as China’s AI buildout shifts grid concerns from total power consumption to real-time power delivery, flexibility, and local system stability.
The Scale of the Challenge
China’s data centers consumed approximately 140 billion kilowatt-hours (kWh) of electricity in 2024, representing about 1.4% of total national electricity usage. Bloomberg analyst Lv Jinghong projected at the BNEF Summit in March 2025 that by 2035, data centers will consume 400 billion kWh annually — roughly 3.2% of national power consumption, comparable to the entire electricity usage of Sichuan province.
China’s total electricity use topped 10 trillion kWh in 2025, a new record driven significantly by AI computing demand, CGTN reported.
A Fundamental Timing Mismatch
One of the most critical challenges is a construction pace mismatch. AI data centers can be built in 8 months to 2 years, while grid expansion and upgrades take 3 to 5 years, creating significant uncertainty for grid planners.
Wang Zesen, deputy director of the Power System Research Institute at State Grid Jibei Electric Power Co. Ltd., highlighted that data center utilization rates fluctuate between 10% and 80%, with extreme load fluctuations occurring every 0.5 to 1 millisecond. This risks grid frequency issues and potential chain-reaction shutdowns when combined with renewable energy sources, which are themselves intermittent by nature.
Policy Responses and Green Energy Solutions
China has been taking steps to address the challenge. In May 2026, the country brought online its first large-scale renewable energy project designed to supply power directly to data centers, located in Ningxia. The project has 500 megawatts of solar capacity with 1.5 gigawatts of wind power planned by the end of the year, generating 4.3 billion kWh annually, as reported by the South China Morning Post.
Beijing has also expanded its green electricity direct connection policy, allowing renewables to supply multiple users directly. China’s average Power Usage Effectiveness (PUE) — a measure of how efficiently a data center uses electricity — improved from 1.7 in 2016 to 1.48 in 2023, and new data centers must now meet a PUE of 1.3 or below.
The Coal-Renewable Paradox
China faces a genuine tension between its climate commitments — peak carbon by 2030, net zero by 2060 — and the energy demands of AI. While China leads the world in renewable deployment, coal still generates approximately 57% of the country’s electricity. China approved a record volume of new coal power plant construction in 2023 and 2024, partly to support AI-driven demand growth, according to EnergyPrices.net.
Major tech firms including Alibaba, Tencent, Baidu, and Huawei have committed to 100% renewable energy for data centers, but grid connectivity means many still draw from fossil-heavy sources. Alibaba and Tencent have also made early investments in nuclear fusion and small-reactor technologies as potential future energy sources.
Grid-Scale Battery Storage as a Buffer
China has deployed approximately 50 gigawatts of battery energy storage systems (BESS) by the end of 2025 — more than the rest of the world combined. This storage capacity is critical for smoothing out the volatility of both renewable generation and AI computing loads, providing a buffer that can absorb rapid fluctuations in demand.
Geopolitical Dimensions
US export controls on advanced AI chips — primarily Nvidia’s H100 and H200 GPUs — have forced Chinese operators to use less efficient domestic alternatives such as Huawei’s Ascend 910C processor. These chips require more power for equivalent workloads, further amplifying electricity demand and adding a geopolitical dimension to China’s energy challenge.
What’s Next
At the annual Two Sessions in March 2026, a deputy of the National People’s Congress called for stronger “power–computing coordination” at the national level, proposing coordinated planning, scaling up green energy and storage projects in western China, and training more talent skilled in both power systems and digital technology.
The 400 billion kWh projection for 2035 represents a tripling of current data center consumption. Whether China’s grid expansion can keep pace with AI data center construction — and whether efficiency improvements can outpace demand growth — remains one of the most consequential questions for the country’s energy future and its climate ambitions.