China has introduced what it describes as the world’s largest neuromorphic supercomputer, called “Darwin Monkey” or “Wukong.” Modeled on the neural structure of the macaque brain, the system contains over 2 billion artificial neurons and 100 billion synapses.
Highlights
- World’s Largest Neuromorphic System: “Darwin Monkey” (Wukong) simulates 2 billion neurons and 100 billion synapses, modeled after the macaque brain.
- Brain-Like Processing: Uses spiking neural networks (SNNs) to mimic biological neuron activity, enabling more realistic and parallel computation.
- Energy Efficiency: Runs on just 2,000 watts — similar to a household appliance — making it far more efficient than traditional supercomputers.
- Hardware Backbone: Powered by 960 Darwin III neuromorphic chips, each simulating millions of spiking neurons.
- Evolution of the Darwin Series: Builds on China’s earlier systems: Darwin Mouse (2020) with 120M neurons → Darwin Monkey (2025) with billions.
- Demonstrated Capabilities: Early applications include logical reasoning, math problem-solving, content generation, and multi-species brain simulations.
- Global Leadership: Surpasses Intel’s Hala Point (1.15B neurons), highlighting China’s ambitions in the neuromorphic computing race.
- Why It Matters: Offers a sustainable path toward AI and AGI by combining brain-inspired computation with low energy usage.
By mimicking how the brain processes information, the machine aims to advance research in artificial intelligence and accelerate efforts toward artificial general intelligence (AGI).
How Darwin Monkey Works
Unlike traditional supercomputers that rely on sequential binary processing, Darwin Monkey uses spiking neural networks (SNNs), which imitate how biological neurons communicate through electrical spikes.
- Hardware foundation: Built with 960 Darwin III neuromorphic chips, each capable of simulating millions of spiking neurons.
- Energy efficiency: Operates on just 2,000 watts, significantly lower than conventional high-performance computing systems.
- Parallel advantage: Processes data simultaneously, allowing more biologically realistic and efficient computation.
From Rodents to Primates
The system builds on earlier neuromorphic efforts in China,
- 2020 – Darwin Mouse: Modeled a rodent brain with 120 million artificial neurons.
- 2025 – Darwin Monkey: Advances to primate-level modeling with billions of neurons and synapses.
Developed by Zhejiang University and Zhejiang Lab with support from Alibaba Group, the project has already demonstrated applications in logical reasoning, mathematical problem-solving, content generation, and brain simulations ranging from zebrafish to macaques.
Position in the Global Neuromorphic Race
Darwin Monkey surpasses Intel’s Hala Point, which previously held the record with 1.15 billion artificial neurons. While direct comparisons are complex—since neuromorphic architectures vary—the system reflects China’s ambition to compete with global leaders in advanced AI hardware.
Neuromorphic computing takes a different path from traditional AI: instead of scaling raw computational power, it emulates the brain’s architecture, potentially offering both performance and energy efficiency gains.
Why It Matters
Researchers suggest that Darwin Monkey could serve as a vital platform for both neuroscience and artificial intelligence. Its energy efficiency—drawing power comparable to a household appliance—could pave the way for more sustainable AI infrastructure.
By modeling brain-like processes, the system also offers new opportunities to study cognition, potentially narrowing the gap between today’s narrow AI systems and future AGI.
Demonstrated Capabilities
So far, Darwin Monkey has shown early abilities in –
- Logical reasoning
- Mathematical problem-solving
- Content generation
- Neural simulations across species
Global Implications
China’s progress in neuromorphic computing is drawing international attention. Analysts note that the Darwin Monkey not only advances technical performance but also strengthens China’s position in the global AI race.
Its development underscores the growing strategic importance of sovereign innovation in brain-inspired computing.
