Researchers Propose Light‑Driven Memory Technology to Cut AI Energy Consumption

Scientists have introduced a light‑driven memory architecture for computing. The technology uses photons instead of electrons to store and retrieve data. Early experiments suggest it could reduce AI energy use by a substantial margin. Lower power consumption would address growing sustainability concerns in AI training. The approach may enable faster data access with minimal heat generation. Researchers plan further testing to validate performance at scale. If successful, the memory could be integrated into future AI accelerators. Industry watchers will watch for commercial prototypes and adoption timelines.