Researchers used the Fugaku supercomputer created a realistic digital model that advances brain research, disease studies, and AI-driven neuroscience.
Fugaku simulates a mouse cortex with 10M neurons and 26B synapses, advancing brain disease research more
Photo Credit: Allen Institute
One of the most detailed simulations of the mouse cortex has been developed by the researchers. In this case, the team managed to develop the model of the digital brain using the Fugaku supercomputer of Japan with approximately 10 million neurons and 26 billion synapses that cover the mouse cortex. This realistic model recreates finer details of the structure and activity of the neurons, and it opens up novel horizons of research on how the brain functions. Scientists at the Allen Institute (USA) and at the University of Electro-Communications in Japan were the leaders of the effort.
According to the Allen Institute, researchers combined biological data with supercomputing to recreate a virtual mouse cortex. The Allen Institute provided detailed cell-type and connectivity datasets, which were processed using the institute's Brain Modelling Toolkit and run on Japan's Fugaku supercomputer. Each of the ~10 million simulated neurons is modelled as a tree of compartments, capturing sub-cellular dynamics. Using Fugaku's full capacity, the team ran full-cortex simulations at roughly 32 seconds of compute per 1 second of real brain activity.
According to the researchers, the virtual cortex can reveal how disorders like Alzheimer's or epilepsy unfold across brain networks and allow testing of potential therapies in silico. Similar efforts are already underway: Stanford teams built an AI-based “digital twin” of the mouse visual cortex that predicts thousands of neurons' responses to novel visual stimuli. At EPFL, scientists generated synthetic wiring maps of the entire mouse brain (digital connectomes) that closely match experimental data.
Catch the latest from the Consumer Electronics Show on Gadgets 360, at our CES 2026 hub.