Chinese engineers have unveiled Darwin Monkey, a groundbreaking brain-inspired computer system that incorporates over two billion artificial neurons, a scale comparable to the neural capacity of a macaque monkey.
The system integrates 960 custom-designed Darwin 3 neuromorphic chips, developed collaboratively by Zhejiang University, state-funded Zhejiang Lab, and Alibaba Group. Each individual Darwin 3 chip supports over 2.35 million neurons. Together, they create a massive neural network comprising more than 100 billion artificial synapses.
According to Zhejiang University, Darwin Monkey has successfully performed diverse cognitive tasks, including content generation, logical reasoning, and mathematical calculations, leveraging a large AI model created by the Chinese firm DeepSeek.
Beyond specific applications, developers highlight Darwin Monkey’s core strength: its ability to simulate entire animal brains for neuroscience research. The system is capable of simulating brain structures ranging from zebrafish and mice up to macaques, offering potential breakthroughs in understanding biological neural processes.
This Chinese achievement represents a significant leap in the global neuromorphic computing race, surpassing the scale of Intel’s previous industry-leading system, Hala Point, announced in April 2024 and deployed at Sandia National Laboratories, which features 1.15 billion neurons. The rapid progression from Intel’s 2023 large-scale system announcement to China’s larger implementation underscores the intense pace of development and competition in this field.
Neuromorphic computing fundamentally differs from traditional architectures, aiming to mimic the human brain’s structure and efficiency by using spiking neural networks and integrating memory and processing functions, helping overcome bottlenecks inherent in conventional systems.
Energy efficiency is a critical advantage highlighted by neuromorphic systems. Darwin Monkey operates with a power consumption of approximately 2,000 watts while performing its complex simulations. While significantly higher than the human brain’s estimated 20-watt operation, this power level represents major progress towards biological efficiency, especially when compared to the demands of traditional computing systems executing similar complex AI workloads. This efficiency is increasingly vital as global concerns mount over the surging energy demands of artificial intelligence, with projections suggesting data center energy consumption for AI could double by 2026.
The debut of Darwin Monkey signals substantial Chinese investment and technical prowess in this advanced computing domain, positioning the country competitively as the global neuromorphic market continues to grow, projected to reach $6.48 billion by 2024.
