Something remarkable is happening beneath the rolling hills of Guizhou province, in the vast server halls of Beijing’s Yanqi Lake technology park, and in dozens of newly constructed hyperscale facilities across China’s interior. The country that once led the world in solar panel manufacturing and electric vehicle adoption is now racing to satisfy an entirely different kind of energy hunger: the insatiable electricity appetite of artificial intelligence.
China’s AI sector is consuming power at a pace that is reshaping the country’s electricity grid, accelerating investment in both renewable energy and coal, and creating new geopolitical dimensions to the global technology race. According to the International Energy Agency, data centres in China consumed approximately 200 terawatt-hours (TWh) of electricity in 2024 — roughly equivalent to the entire electricity consumption of Australia — and that figure is projected to more than double by 2030 as AI workloads intensify. You can explore the IEA’s full data centre energy analysis at iea.org.
The DeepSeek Effect: Efficiency Meets Scale
The release of DeepSeek’s R1 and V3 models in late 2024 sent shockwaves through the global technology industry — not because they were less capable than leading Western models, but because they appeared to achieve comparable performance at a fraction of the training cost. DeepSeek claimed to have trained its V3 model for approximately $5.6 million, a figure that made the hundreds of millions spent by OpenAI, Google, and Anthropic look extravagant by comparison.
The “DeepSeek moment” triggered a complex debate about AI energy consumption. On one hand, more efficient models could reduce the energy required per unit of AI output. On the other, cheaper AI tends to stimulate dramatically more demand — what economists call the Jevons paradox, where efficiency gains are more than offset by increased usage. The evidence from China so far suggests the latter effect is dominant: more efficient models have catalysed a wave of AI deployment across Chinese industry, government, and consumer applications, driving total compute demand — and thus electricity consumption — higher, not lower.
Chinese tech giants including Alibaba, Tencent, Baidu, Huawei, and ByteDance have all accelerated their data centre buildout programmes in 2025 and 2026. Huawei’s Cloud division alone announced plans to invest more than 100 billion yuan (approximately $14 billion) in AI infrastructure over three years, with a significant portion directed at new data centre construction in provinces with access to cheap renewable power.
Where China Is Building Its AI Infrastructure
China’s approach to data centre location is strategically calculated, driven by three factors: electricity cost, water availability for cooling, and political considerations around data sovereignty. The result is a geographic dispersal of AI infrastructure that is rapidly transforming several previously underdeveloped regions.
Guizhou Province has become China’s de facto data centre capital, hosting major facilities for Apple’s iCloud China operations, Huawei, and the government’s national data centre cluster. The province offers cool temperatures (reducing cooling energy costs), abundant hydroelectric power from its mountainous terrain, and significant political motivation to attract investment to one of China’s poorer regions.
Inner Mongolia benefits from flat terrain ideal for large-scale solar and wind development, low land costs, and proximity to Beijing via high-voltage transmission lines. Several of China’s largest hyperscale data centres — some exceeding 1 gigawatt (GW) of IT load capacity — are under construction or operational in the Hohhot area.
Xinjiang, in China’s far northwest, offers extraordinary solar resources — some of the best in the world — alongside coal reserves that provide firm backup power. Significant data centre investment has flowed into the region, though Western technology companies face reputational and supply chain pressures that make direct involvement complicated.
Sichuan and Yunnan provinces are targeting AI data centre investment on the back of their massive hydroelectric resources. Sichuan alone has more than 90 GW of installed hydroelectric capacity, making it one of the world’s leading sources of low-carbon firm power — ideal for AI operators seeking to demonstrate environmental credentials.
The Coal Paradox: Going Green While Keeping the Lights On
China’s AI energy surge is creating a genuine paradox at the heart of the country’s climate commitments. President Xi Jinping has pledged that China’s carbon emissions will peak before 2030 and reach net zero by 2060. China is also the world’s undisputed leader in renewable energy deployment, adding more solar and wind capacity in 2023 alone than the rest of the world combined.
Yet the sheer scale of new electricity demand from AI and other industrial loads is so vast that coal — which still generates around 57% of China’s electricity — remains indispensable as a baseload source. China approved a record volume of new coal power plant construction in 2023 and 2024, with much of the new capacity intended to provide grid stability as intermittent renewables expand.
The tension between China’s renewable ambitions and its coal dependency is nowhere more visible than in the data centre sector. Major Chinese cloud operators have made high-profile commitments to power their facilities with 100% renewable energy, but the physical reality of grid connectivity means many facilities are drawing power from a grid that remains heavily fossil-fuel dependent, particularly during periods of low wind and solar output or high seasonal demand.
China is investing heavily in solutions to this problem. The country has deployed more grid-scale battery storage than the rest of the world combined, with approximately 50 GW of battery energy storage systems (BESS) installed by end-2025. Ultra-high-voltage (UHV) transmission lines are being constructed to move renewable power from resource-rich western provinces to demand centres in the east — a grid of extraordinary scale and ambition that has no real parallel anywhere else in the world.
Water: The Hidden Resource Crisis
Electricity is not the only resource under pressure from China’s AI buildout. Data centres require enormous volumes of water for cooling — a typical large-scale hyperscale facility can consume millions of litres of water per day. In a country where water scarcity already affects hundreds of millions of people and entire river systems are under stress, the water footprint of AI is attracting growing concern.
The preference for locating data centres in Guizhou, Sichuan, and Yunnan — all water-rich provinces by Chinese standards — reflects in part a deliberate policy to avoid compounding water stress in the arid north. But as investment pressure intensifies, there are concerns that water availability constraints could become a binding limit on AI infrastructure growth in key regions.
Geopolitical Dimensions: The Chip War Meets the Power Grid
China’s AI energy surge is inseparable from the broader geopolitical contest over semiconductor technology. US export controls have restricted China’s access to the most advanced AI chips — primarily Nvidia’s H100 and H200 GPUs — forcing Chinese operators to either use less advanced chips, stockpile controlled chips before restrictions took full effect, or accelerate domestic chip development through companies like Cambricon, Biren Technology, and Huawei’s HiSilicon division.
The chip constraints have a direct energy implication: less efficient chips require more power to perform equivalent AI workloads, amplifying electricity demand further. Huawei’s Ascend 910C processor — China’s most capable domestically produced AI chip — is widely reported to be less energy-efficient than Nvidia’s latest offerings, though the gap is narrowing as Chinese chip design capabilities mature.
For global energy markets, the geopolitical dimension matters because Chinese AI data centre demand is now a meaningful driver of global electricity equipment markets — transformers, cables, cooling systems, and power electronics — creating supply chain pressures that affect grid development timelines everywhere. Our energy news section tracks how AI infrastructure investment is reshaping global commodity and equipment markets.
Impact on China’s Electricity Prices and Grid
For Chinese electricity consumers and industry, the surge in data centre demand is exerting upward pressure on electricity prices in affected provinces. China’s electricity market is undergoing gradual liberalisation, with more generation and consumption shifting to market-based pricing mechanisms. In provinces with heavy data centre concentration, spot electricity prices have risen during peak demand periods as AI facilities compete with industrial users for available power.
China’s National Development and Reform Commission (NDRC) has responded by tightening regulations on data centre power usage effectiveness (PUE) — a measure of how efficiently a facility uses electricity. New data centres in China must meet a PUE of 1.3 or below (meaning no more than 30% of electricity is used for overhead like cooling, compared to useful computing), with stricter requirements in regions with limited renewable energy access. The policy is driving rapid adoption of advanced cooling technologies including liquid cooling, immersion cooling, and AI-optimised airflow management.
What This Means for Global Energy Markets
China’s AI-driven electricity demand surge has implications that extend well beyond its borders. It is accelerating Chinese investment in solar panels, wind turbines, batteries, and grid infrastructure at a scale that is driving down global equipment costs — benefiting energy transition efforts worldwide. It is also intensifying competition for critical minerals including lithium, cobalt, copper, and rare earth elements used in both AI hardware and clean energy technology.
For energy investors and observers tracking the intersection of technology and commodity markets, China’s AI energy story is one of the most important trends of the decade. The decisions being made today about how to power China’s AI ambitions will shape not just the country’s electricity grid, but the trajectory of global energy demand, carbon emissions, and the competitive landscape of the most consequential technology race in history.
For the latest on renewable energy developments in Asia and globally, and how AI is transforming energy demand patterns, follow our dedicated AI & Energy coverage.
This article is for informational purposes only and does not constitute financial or investment advice. Always consult a qualified financial adviser before making investment decisions.
