After billions in development, Microsoft quietly slashed expectations for Copilot, its AI wonderchild that can’t reliably perform basic tasks. The pattern is grimly familiar: massive investment chasing promised returns that never appear, followed by market correction when reality intrudes. But this bubble comes with a uniquely toxic twist—AI’s environmental cost accelerates the very climate crisis it claims it will solve.

Data centers now gulp enough water annually to supply hundreds of families while generating marketing copy riddled with errors. Training GPT-3 alone evaporated 700,000 liters of clean freshwater. Meanwhile, Australia’s Albanese government subsidizes data center expansion while manufacturing collapses, emissions targets slip through accounting tricks, and workers face automation without security.

The millenarian faith in AI’s salvation delays the unglamorous work we actually need: rapid decarbonization, public investment in resilience, and democratic reform to counter corporate capture. Time to stop waiting for miracles.