Chinese researchers have unveiled the “Darwin Monkey,” a significant advancement in neuromorphic computing, an AI version of a monkey’s brain. This project was developed by Zhejiang University in collaboration with the Zhejiang Laboratory.
The Darwin Monkey utilizes 960 chips, with each chip supporting over 2 billion spiking neurons and over 100 billion synapses, a capacity that reportedly approaches the number of neurons found in a macaque brain. It is touted as the world’s largest brain-like, or neuromorphic, computer and the first based on neuromorphic-specific chips. According to researchers, this development is considered a “step toward more advanced brain-like intelligence.” While macaques lack certain human capabilities, such as the brain circuitry for human speech, this technology could pave the way for more sophisticated robot animals.
The Darwin Monkey, with its 2 billion artificial neurons, surpasses Intel’s neuromorphic computer prototype, Hala Point, announced in April 2024. Hala Point features 1.15 billion neurons, roughly equivalent to an owl’s brain, and is being utilized by Sandia National Laboratory for “advanced brain-scale computing research” across various sectors, including commercial, defense, and basic science.
This latest achievement builds upon previous work by Zhejiang University researchers, who first developed the Darwin Mouse in 2020, featuring 120 million artificial neurons. The “brain-inspired” Darwin 3 chips were then created in early 2023, leading to the birth of the Darwin Monkey two years later. The system integrates China’s DeepSeek AI model to perform intelligent tasks such as logical reasoning, content generation, and mathematical problem-solving.
To enhance efficiency, the research team also developed a new operating system for the Darwin Monkey. This system aims to improve performance by achieving “concurrent scheduling of neuromorphic tasks and dynamic optimization of system resources, taking into account communication bandwidth and task characteristics.” The focus on efficiency in neuromorphic systems is also shared by Intel, which suggests Hala Point could help mitigate the “unsustainable rates” of current computing costs. Zhejiang University emphasizes that “brain-inspired computing systems can address the high energy consumption and computational complexity of existing deep networks and large models,” and that the Darwin Monkey’s “unsupervised online learning mechanism can bring revolutionary advances to AI.”
