China is stepping up its chip game by starting mass production of non‑binary AI chips that blend traditional binary computing with probability‑based logic. Developed by Professor Li Hongge’s team at Beihang University, these chips are already set to enhance applications in aviation and heavy industry by improving fault tolerance and efficiency.
Professor Li highlighted two main challenges facing current technology: the power wall and the architecture wall. While binary systems efficiently transmit information, they often demand high energy consumption. Meanwhile, newer non‑silicon chips struggle to sync with conventional CMOS systems. To tackle these issues, his team created the Hybrid Stochastic Number (HSN) system, which smartly combines standard binary methods with stochastic computing. This approach not only optimises performance but also helps navigate around US-imposed chip restrictions.
