Photonic computing studies point to faster, greener AI
Two new studies from Koç University, published Saturday, suggest a way out of AI's energy crisis: do the computing with light.
The papers, in Communications Engineering and Communications Physics, demonstrate offloading parts of AI computation onto photonic hardware — a self-optimising multichannel optical system, and multimode fibre-laser cavities acting as nonlinear processors.
Moving computation from electronics to photonics could make AI far faster and dramatically more energy-efficient — a timely result as data-centre power demand soars worldwide.
Originally reported by Tech Xplore. Article compiled by the Deccan Age desk.
