China’s Darwin Monkey brain-like computer brings AI closer to human thinking

China’s ‘Darwin Monkey’ Sets a New Benchmark in Brain-Inspired Computing

Zhejiang University has developed the Darwin Monkey, a brain-inspired computer system featuring over 2 billion artificial neurons—comparable to the brain of a macaque. Powered by 960 Darwin 3 chips, this system simulates complex cognitive tasks such as reasoning and content generation. Its design emphasizes energy efficiency and advanced neuromorphic computing, offering promising directions for AI development and brain science research.

World’s First Brain-Like Computer of This Scale

The Darwin Monkey operates on 960 Darwin 3 neuromorphic chips, creating a network with over 100 billion synapses. Zhejiang University has described it as “the world’s first brain-like computer based on a dedicated neuromorphic chip with more than 2 billion neurons.” The system is already applied in complex tasks including mathematics, logical reasoning, and content generation. It runs on a brain-like large model developed in partnership with Chinese AI company DeepSeek.

Bridging Artificial Intelligence and Neuroscience

With its vast number of neurons and synapses, the Darwin Monkey can simulate the neural structures of animals like macaques, mice, and zebrafish. This capability offers a new tool for brain science research, enabling more accurate study of neural processes. Such simulations can help researchers better understand biological brain functions and dynamics.

Neuromorphic Computing Explained

Unlike traditional AI systems that process data with continuous numerical values, the Darwin Monkey uses spiking neural networks. These networks send information in discrete pulses, more closely mimicking how biological neurons communicate. The Darwin 3 chip, developed in 2023 by Zhejiang University in collaboration with Zhejiang Lab, supports over 2.35 million pulsed neurons and hundreds of millions of synapses. It also incorporates specialized brain-inspired computing instruction sets and neuromorphic online learning mechanisms.

High Performance Meets Energy Efficiency

Despite its complexity, the Darwin Monkey is energy-efficient, consuming approximately 2,000 watts during normal operation. Pan Gang, director of the State Key Laboratory of Brain-Computer Intelligence at Zhejiang University, notes that the system’s large scale, high parallelism, and low power consumption offer a new computing paradigm for a variety of scenarios.

Building on Prior Success: From Mouse to Monkey

The Darwin Monkey is the successor to the “Darwin Mouse,” introduced in 2020. The Darwin Mouse featured 120 million artificial neurons. The new system reflects advances in neural system integration, interconnection technology, and brain-inspired operating systems, marking a significant step forward in neuromorphic computing.

For professionals interested in neuromorphic AI and brain-inspired computing, exploring courses on advanced AI topics can offer valuable insights into these emerging technologies.