The hidden physical costs behind every AI prompt

A video from Imagination in Action features researcher Vijay Gadepally discussing the physical resources required to run artificial intelligence systems. He explains that while AI tools can feel intangible, they depend on real infrastructure including computer chips, data networks, cooling systems, electricity and water.

Gadepally argues that making AI more sustainable will require progress on two fronts at once: building more efficient AI models that use fewer resources, and improving the physical infrastructure that supports them, such as data centers and power grids.