China’s AI-driven data center boom, powered by coal and prioritized over climate goals, epitomizes the energy paradox of generative AI.

By Lydia Powell | Observer Research Foundation

Data centres (DCs) critical to AI applications consume rather than make available abundant quantities of energy.

In Brief by Probe International

Sam Altman’s 2025 vision of abundant energy and AI-driven prosperity by the 2030s obscures a critical contradiction: AI’s voracious energy demands.

Data centers (DCs), vital for AI, consume staggering power—generative AI queries use 30 times more energy than standard web searches.

China, a global leader in DC capacity, faces a dual crisis as its rapid DC expansion fuels surging electricity consumption (15% annual growth since 2015) and carbon emissions. Despite efforts to promote energy-efficient DCs since 2011, over 70% of China’s DC power still comes from coal-dependent eastern regions. While initiatives like the East–West Computing Project aim to harness western renewables, projections suggest fossil fuels will dominate DC energy until 2030, prioritizing data economy dominance over climate goals.

Read the original report in full at the publisher’s website here.