Eighty years after the launch of the ENIAC computer, researchers at the University of Pennsylvania are developing a novel approach to enhance computing efficiency. Physicists led by Bo Zhen are investigating the use of photons, which have no mass and carry information quickly, to address the limitations faced by traditional electronic computing.
According to Li He, a co-first author of the research published in *Physical Review Letters*, electrons show several shortcomings due to their charge, including energy loss as heat and resistance, which complicates the handling of large amounts of data. This limitation is increasingly critical as artificial intelligence (AI) demands more from computing hardware in terms of processing and cooling capabilities.
Zhen’s team has created exciton-polaritons, quasiparticles formed by coupling photons with electrons in a thin semiconductor layer. These exciton-polaritons enable strong interactions necessary for effective signal switching in computation, a process that photons alone struggle to perform due to their charge-neutral nature.
This advancement is particularly significant for AI applications. Zhen noted that while many photonic AI chips can perform basic calculations using light, they still require conversion of light signals into electronic ones for complex operations, which hinders speed and efficiency.
The team achieved all-light switching using exciton-polaritons at 4 quadrillionths of a joule—much less energy than is needed to power a tiny LED. If this technology can be scaled, it may allow photonic chips to directly process light from cameras, reducing energy demands in large AI systems.
Moreover, the research could also facilitate the development of basic quantum computing capabilities on semiconductors. Bo Zhen serves as the Jin K. Lee Presidential Associate Professor in the Department of Physics and Astronomy at the University of Pennsylvania. Li He, formerly a postdoctoral researcher in the Zhen Lab, is now an assistant professor at Montana State University.
The research included contributions from authors Zhi Wang and Bumho Kim from the University of Pennsylvania and received backing from the US Office of Naval Research and the Sloan Foundation.
