Darwin Monkey, the world's largest biological computer.

A traditional data center thinks in blocks of numbers, but the ambition of Darwin Monkey, Zhejiang University's new neuromorphic computer, thinks using impulses, electrical spikes and plastic connections, exactly as a biological brain does.

Stacking 2 billion pulsating neurons and more than 100 billion synapses to emulate the brain dynamics of a primate at scale, instead of devouring mega Watts, the system works in the 2000 Watt range in typical operation and already runs complete vision, hearing, language and learning applications, even running a Deepseek-like model for logical reasoning, content generation and mathematical problem solving.

It is a bold step to move away from more of the same of traditional chips and program with biology and mind, the Darwin Monkey brings together 960 Darwin Tree neuromorphic chips distributed in 15 Blade-type servers, each chip implements around 2.35 million spiking neurons and hundreds of millions of synapses, in addition to a set of instructions inspired by the brain and an online learning mechanism designed for continuous updating.

The secret is in the SN paradigm, Spiky Neural Networks, which processes events instead of samples at a fixed rate. When there is no stimulus, there is almost no consumption, when there is, the system responds with temporal spikes, encoding information in the when and the quantum of the shot. This electronic autonomy favors low latency, energy efficiency and plasticity, three pillars that classical computing has difficulty imitating when trying to model brains realistically.

This combination becomes particularly interesting when the application involves continuous interaction with the environment, highly responsive robotics, context-aware embedded devices and agents that need to monitor, decide and act without connection to the cloud.

Add to that the possibility of native sensory fusion, vision, audio and language, it is a platform made for AI that not only predicts, but perceives. And what is this technology for? In computational neuroscience, the scale of 2 billion neurons opens up space to simulate entire animal brains from fish to monkey with testable hypotheses about memory, stress and navigation

In automation and robotics, the Even Driven feature favors reactive control, now powering and maintaining response to disturbances. In Edge AI and medical devices, online plasticity promises continuous adaptation without costly offline training.

And for generative models, the combination between spiking networks and symbolic modules can reduce hallucinations and improve step-by-step reasoning, that is, it is a bet on how AI should evolve when it leaves the screen and enters the physical world, efficient, continuous, time-sensitive and capable of learning in the field. If you could reserve a piece of this electronic brain for a week, what real experiment would you run first? A robotic agent that learns by itself in a factory, a language model with online learning, or a memory simulation at the synaptic level.

Thank you for visiting my blog. If you like posts about #science, #planet, #politics, #rights #crypto, #traveling and discovering secrets and beauties of the #universe, feel free to Follow me as these are the topics I write about the most. Have a wonderful day and stay on this great platform :) :)

! The truth will set us free and science is the one that is closest to the truth!

Le immagini senza riferimento sono state create con l'intelligenza artificiale

---