Monkey brain, explained: Chinese researchers build AI system with 2B neurons to mimic macaque neurology

Chinese researchers at Zhejiang University have created the Darwin Monkey, an AI system that models a macaque’s brain using 960 specialized chips with over 2 billion artificial neurons and 100 billion synapses. The project represents the world’s largest neuromorphic computer and marks a significant step toward brain-inspired computing that could revolutionize AI efficiency and capabilities.

What you should know: The Darwin Monkey uses China’s DeepSeek AI model to perform complex cognitive tasks including logical reasoning, content generation, and mathematical problem solving.

• The system approaches the neural complexity of an actual macaque brain, making it the first neuromorphic computer based on specialized neuromorphic chips rather than traditional computing architecture.
• Researchers developed custom “brain-inspired” Darwin 3 chips in collaboration with Zhejiang Laboratory, a Chinese research institute, along with a new operating system designed for concurrent neuromorphic task scheduling and dynamic resource optimization.

The big picture: This achievement positions China at the forefront of neuromorphic computing, surpassing Intel’s previous milestone in brain-scale computing.

• Intel’s Hala Point system, announced in April 2024, contains 1.15 billion neurons—roughly equivalent to an owl’s brain—making the Darwin Monkey nearly twice as complex.
• The progression from the team’s 2020 Darwin Mouse (120 million neurons) to the current Darwin Monkey demonstrates rapid advancement in brain-inspired computing over just five years.

In plain English: Neuromorphic computing mimics how biological brains process information, using artificial neurons that communicate through electrical signals just like real brain cells. Traditional computers process information sequentially, but neuromorphic systems can handle multiple tasks simultaneously, much like how your brain can walk, talk, and think at the same time.

Why this matters: Neuromorphic systems could address the unsustainable energy consumption and computational complexity plaguing current AI models.

• “Brain-inspired computing systems can address the high energy consumption and computational complexity of existing deep networks and large models,” according to Zhejiang University researchers.
• The system’s “unsupervised online learning mechanism can bring revolutionary advances to AI,” potentially reducing reliance on massive datasets and computational resources.

What’s next: While human-level brain modeling remains distant, the technology could enable more sophisticated robotic applications.

• Macaque brains lack the neural circuitry for human speech and other advanced cognitive capabilities, according to Princeton University research.
• However, matching such brain models with robotic bodies could create more lifelike and capable artificial beings in the near term.

Competitive landscape: The achievement highlights China’s growing influence in advanced AI hardware development.

• Sandia National Laboratory, a U.S. research facility, currently uses Intel’s Hala Point for “advanced brain-scale computing research” spanning commercial, defense, and basic science applications.
• The focus on efficiency and energy reduction reflects broader industry concerns about the sustainability of current AI computing demands.