China has launched the first large-scale application of non-binary AI chips. This initiative integrates hybrid computing technology into sectors such as aviation and industrial systems.

It is led by Professor Li Hongge’s team at Beihang University in Beijing.

The new technology combines binary and stochastic logic. This combination improves fault tolerance and energy efficiency.

It also addresses issues in traditional computing, such as high-power consumption and challenges in integrating new chip architectures.

Li’s team developed a numerical system called Hybrid Stochastic Number (HSN). This system merges binary and probability-based numbers.

This method optimizes computational efficiency by reducing hardware needs while maintaining speed. It has been applied to design intelligent chips for touch displays and flight systems.

🔗 Source: South China Morning Post

🧠 Food for thought

1️⃣ China’s semiconductor self-sufficiency efforts represent a pivotal strategic response to export controls

This non-binary chip breakthrough comes amid China’s broader push to reduce dependence on foreign semiconductor technology, a vulnerability exposed by US export restrictions.

China historically relied on imports for 84% of its semiconductor needs, making technological independence a national security priority 1.

The development aligns with China’s ambitious “Made in China 2025” initiative, which aims to increase domestic semiconductor production from 16% to 70% by 2025, supported by massive government investments approaching $180 billion 2.

Professor Li’s hybrid computing approach represents a technological workaround to U.S. export controls that have specifically targeted advanced AI chips from companies like Nvidia, whose China market share has reportedly dropped from 95% to 50% due to these restrictions 3.

This alternative architecture development demonstrates how export controls can sometimes accelerate indigenous innovation rather than simply containing technological advancement.

2️⃣ The shift to hybrid computing architecture addresses fundamental limitations in traditional AI chip design

The hybrid stochastic computing approach tackles what Professor Li calls the “power wall” and “architecture wall” that have constrained traditional binary computing systems.