Chinese researchers have pulled off a world-first with the unveiling of Darwin Monkey (“Wukong”), a supercomputer designed to mimic the workings of a monkey’s brain.
Developed by researchers at Zhejiang University, this neuromorphic system packs over 2 billion artificial neurons and more than 100 billion synapses, matching the complexity of a macaque brain and pushing the boundaries of AI and brain simulation.
The project is the result of years of work in neuromorphic computing, a field that designs computer systems to behave more like real brains.
The Darwin Monkey system is built on 960 Darwin 3 chips, neuromorphic processors developed in 2023 through a collaboration between Zhejiang University and Zhejiang Lab, a research institute funded by the Zhejiang provincial government and Alibaba Group.
Each chip supports over 2.35 million “spiking” neurons — units that communicate using pulses similar to how real neurons in animals transmit information. Altogether, the system operates with over 100 billion synapses, the connections that allow neurons to pass signals to each other.
“This is the world’s first brain-like computer based on a dedicated neuromorphic chip with more than 2 billion neurons,” Zhejiang University said on its official social media page, as reported by the South China Morning Post.
Beyond traditional AI
Unlike traditional computers that rely on GPUs and CPUs, neuromorphic systems like Darwin Monkey are designed to mimic the brain’s cognitive functions, such as memory, learning, and decision-making, while consuming far less power. Under regular use, the Darwin Monkey uses about 2,000 watts of electricity.
This energy-efficient architecture is particularly attractive as AI models become larger and more demanding. Traditional systems face growing challenges with heat, power usage, and latency; experts say brain-inspired designs could offer a solution.
The new machine is a significant upgrade from the university’s previous project, Darwin Mouse, released in 2020. That earlier version featured 120 million neurons; the new system increases that by more than 16 times, marking a major technological milestone in chip interconnection, neural integration, and operating system design.
“Wukong’s large-scale, highly parallel, and low-power features will provide a new computing paradigm for existing computing scenarios,” said Professor Gang Pan, the team’s leader and director of the State Key Laboratory of Brain-Machine Intelligence at Zhejiang University, in an interview with Science and Technology Daily.
What can Darwin Monkey do?
So far, the Darwin Monkey has demonstrated its capabilities in complex cognitive tasks, such as content generation, logical reasoning, and mathematics, using a large-scale AI model developed by Chinese startup DeepSeek.
But the potential goes far beyond everyday AI. The supercomputer’s neuron-level detail allows it to simulate the brains of animals, such as zebrafish, mice, and macaques. This opens new doors for neuroscience research, giving scientists tools to study brain function without relying on living test subjects.
A new milestone in the neuromorphic race
Until now, the most advanced neuromorphic system was Intel’s Hala Point, introduced in April last year, which featured 1.15 billion artificial neurons. That system was installed at the Sandia National Laboratories in the US. The Darwin Monkey has nearly doubled that figure, placing China at the forefront of neuromorphic research.
Neuromorphic computing is attracting global attention for its ability to combine low energy consumption, parallel processing, and intelligence, qualities that traditional GPU-based systems often struggle to achieve.
Experts say systems like Darwin Monkey may be critical in the next generation of AI, especially as traditional architectures face increasing limits in speed, memory, and energy use.