China unveiled a groundbreaking step in brain-like computing with the launch of the INN inside computing body in Hengqin, Guangdong-Macao on Friday. The compact supercomputing system marks a major milestone in merging artificial intelligence with advanced hardware design.
Moreover, the INN inside computing body delivers supercomputer-level performance in a unit roughly the size of a household refrigerator. It uses an original intuitive neural network and a brain-inspired algorithm to process massive data with remarkable energy efficiency. Therefore, this new approach offers a model for future intelligent systems built on brain-like computing.
In addition, developers from the Guangdong Institute of Intelligent Science and Technology designed the system to combine power, compactness, and quiet operation. It features 1,152 CPU cores, 4.8 terabytes of DDR5 memory, and 204 terabytes of storage. Furthermore, the machine operates at noise levels below 45 decibels and consumes one-tenth the energy of standard supercomputers. Even under full load, the internal temperature stays below 70 degrees Celsius.
Researchers compared its impact to placing the capacity of a large library into a single home bookshelf. They noted that its efficient design makes high-level computing more accessible for businesses and research centers.
The system’s key innovation lies in its unified network structure that merges symbolic and numerical reasoning. As a result, this allows the machine to imitate how the human brain learns and makes decisions. The algorithm supports continuous learning, enabling the system to acquire new skills without losing old ones.
Cai Jiang, Director of the Joint Lab of Cognitive Neural Networks and Degeneration, said the technology gives artificial intelligence the power to reason. He explained that traditional AI only produces results, while this model can explain how it reached its conclusions.
In terms of testing results, the system shows impressive performance. The system completed a large-scale training task involving tens of billions of tokens in only 30 hours. Additionally, it achieved training and inference speeds of up to 100,000 and 500,000 tokens per second, respectively. These results match the output of multi-GPU clusters but with far less energy use.
Experts say the launch demonstrates China’s growing lead in high-performance and energy-efficient AI systems. Consequently, the success of the INN inside computing body could reshape the future of supercomputing, data processing, and AI development.
Finally, as global demand for smarter, greener computing rises, brain-like computing could become a core driver of next-generation technology. Researchers in Guangdong plan to scale production and integrate the model into intelligent industry applications. The project also signals China’s broader ambition to lead the world in innovative AI infrastructure.
