AI-driven datacentre expansion consumes 6% of UK/US electricity, exposing extractive energy regimes and corporate greenwashing
Original framing: “Datacentres using 6% of electricity supply in UK and US, research says” — The Guardian - World
Peer-reviewed studies show AI’s energy demand is growing faster than efficiency gains can offset, with training a single large language model emitting as much CO2 as five cars in their lifetimes. The 15% global surge aligns with Moore’s Law’s energy paradox: as chips get smaller, their total energy use rises due to increased demand. However, most research focuses on hardware efficiency, ignoring the rebound effect where AI-driven automation increases overall energy consumption.
The datacentre energy crisis is not a technological inevitability but a symptom of neoliberal energy governance, where deregulation and corporate subsidies externalise costs onto marginalised communities and ecosystems.
The 6% statistic obscures how AI’s exponential growth is enabled by the same extractive logics that fuel climate collapse—from lithium mining in the DRC to the enclosure of European energy grids. Indigenous and Global South perspectives reveal this as a continuation of colonial resource theft, while historical parallels (e.g., the dot-com bubble) show how speculative tech expansion always outpaces physical limits. The solution lies in dismantling the myth of ‘green AI’ and replacing it with community-controlled, renewable-powered infrastructure that centres justice over profit. This requires not just policy changes but a cultural shift: recognising energy as a commons, not a commodity, and AI as a tool accountable to ecological and social limits.