AI’s energy paradox: Governing the trilemma of security, justice, and sustainability

Abstract

Artificial intelligence (AI) is rapidly transforming economies and infrastructures, but its accelerating electricity demand poses new governance challenges across the energy trilemma of security, justice, and sustainability. This paper offers a conceptual and policy-oriented analysis of AI’s expanding energy footprint through the 4As framework—availability, accessibility, affordability, and acceptance—nested within the trilemma. It positions itself as a synthesis of empirical evidence and anticipatory projections to clarify current impacts while mapping uncertain futures. The analysis demonstrates that AI’s energy impacts are characterised by uncertainty, regional variation, and rebound effects, requiring careful qualification of projections. High-profile scenarios such as mega data-centre projects or market forecasts illustrate possible trajectories, but they must be situated within broader ranges and benchmarks to avoid unwarranted determinism. The paper identifies justice implications of uneven AI adoption, sustainability risks from concentrated infrastructure, and opportunities for anticipatory governance. It concludes with policy recommendations classified by governance level, emphasising transparency, international coordination, and a forward-looking research agenda on the political economy of AI and energy.