Chinese researchers have unveiled a revolutionary computing architecture that boosts processing speeds nearly fourfold, marking a significant leap for AI development and industrial applications. The breakthrough, led by Peking University scientists and published in Nature Electronics on January 9, 2026, integrates multi-physics-domain hardware to optimize frequency conversion processes critical to modern computing.
By combining two novel devices for frequency generation and modulation, the system performs Fourier Transform operations at 500 billion operations per second – a 285% increase from previous benchmarks. This advancement could accelerate progress in foundational AI models, brain-computer interfaces, and autonomous vehicle systems while reducing power consumption.
"Our architecture allows different computing paradigms to function in their ideal physical domains, whether electrical, optical, or charge-based," explained Tao Yaoyu, a lead researcher at Peking University's Institute for Artificial Intelligence. The technology's versatility makes it particularly valuable for edge computing applications requiring real-time processing of complex sensor data.
Industry analysts predict this innovation could reshape Asia's tech landscape, with potential applications ranging from smart city infrastructure to advanced communication networks. The development aligns with the Chinese mainland's strategic investments in next-generation computing hardware, positioning it as a key player in global AI infrastructure development.
