Physicists create hybrid light-matter particles to replace electrons in AI cmpt (www.sciencedaily.com)

Researchers at the University of Pennsylvania have developed a groundbreaking hybrid light-matter particle called an exciton-polariton, which could revolutionize AI computing by drastically reducing energy consumption and speeding up processing times. Traditionally, computers rely on electrons for data processing, but the limitations of electron-based systems—such as heat generation and resistance—are becoming increasingly problematic as AI demands grow. Exciton-polaritons, which arise from the strong coupling of photons and electrons in a semiconductor, allow for efficient signal switching needed for computations while harnessing the speed of light to move data with minimal loss. This innovation has significant implications for the AI/ML community: it promises the potential for ultra-efficient photonic chips that can perform complex nonlinear operations without the need for slow conversions between light and electrical signals. In experiments, the Penn team achieved all-light switching using a mere 4 quadrillionths of a joule of energy—far below what traditional electronic systems require. If successful at scale, this technology could not only enhance the performance of AI systems but also pave the way for future quantum computing applications, addressing the power limitations posed by conventional electronic computing.