A new UN University report warns that artificial intelligence is driving rapidly escalating environmental costs far beyond electricity use alone, including major impacts on water consumption, land use, carbon emissions, mineral extraction, and e-waste. The report estimates AI-related electricity demand could nearly triple by 2030, with associated emissions comparable to those of entire industrialized nations. It also highlights widening global inequities, with environmental burdens concentrated in regions supplying minerals and hosting infrastructure while benefits flow elsewhere. The authors call for urgent international action, including transparency standards, efficiency-focused design, resource limits, sustainable infrastructure planning, and stronger environmental governance across the full AI lifecycle.