Google engineered proprietary chips to slash the energy cost of machine intelligence.

Evidence

In 2016 it announced the development of a new chip for “deep learning inference” called the tensor processing unit (TPU). The TPU would dramatically expand Google’s machine intelligence capabilities, consume only a fraction of the power required by existing processors, and reduce both capital expenditure and the operational budget, all while learning more and faster. — Shoshana Zuboff, The Age of Surveillance Capitalism, 592

Categorization

• XREF: connects to: Digital technologies replace human attentional energy with electrical energy. — Yves Citton, The Ecology of Attention, 195

Broader Connections

• Digital technologies replace human attentional energy with electrical energy. #expands (0.78) — Yves Citton, The Ecology of Attention, 195