Revolutionary AI Computing Breakthrough: Single-Shot Tensor Computing at Light Speed
Researchers at Aalto University in Finland have achieved a revolutionary breakthrough in artificial intelligence computing. They have demonstrated single-shot tensor computing at the speed of light. This groundbreaking research was published in Nature Photonics on November 14, 2025.
The study shows the potential of performing complex AI calculations using just one pass of light through an optical system. This is a significant leap in the field of AI computing.
Dr. Yufeng Zhang from Aalto’s Photonics Group led the international collaboration. The team encoded digital data directly into light waves. This innovative approach allows computations to occur naturally and simultaneously. “Our method performs the same kinds of operations that today’s GPUs handle, like convolutions and attention layers, but does them all at the speed of light,” Zhang explained.
The breakthrough addresses the growing energy crisis in AI computing. Modern systems like GPT training consume massive amounts of electricity. Unlike traditional electronic circuits that process operations step-by-step, this optical approach enables calculations to happen instantaneously and passively. There is no need for active electronic switching.
Professor Zhipei Sun, leader of Aalto’s Photonics Group, noted that the technology could be integrated into photonic chips within 3-5 years. This promises dramatically faster and more energy-efficient AI systems. These advancements could transform everything from data centers to mobile AI applications.
Source: Aalto University