Stanford University has launched the Digital Brain project, bringing together researchers from multiple departments and schools. Over the next five years, the project aims to leverage large-scale, high-precision human brain imaging data and artificial intelligence technology to train brain foundation models, overcoming the limitations of traditional neuroscience research by shifting from predicting the responses of a single brain system to predicting whole-brain activity in complex situations—covering multiple types of cognitive tasks such as perception, memory, and language. The project adopts a data collection scheme involving high-frequency, long-term repeated measurements of a small number of participants, combining two complementary technologies—7T ultra-high-field functional magnetic resonance imaging and optically pumped magnetometer magnetoencephalography—to simultaneously acquire brain activity data with high spatial and temporal resolution. The project will open its data and models to the world. Its long-term goal is to help decipher abnormal brain networks in neurological diseases such as Parkinson's disease and Alzheimer's disease, providing new tools for clinical applications such as individualized neurological interventions; in the short term, it will serve as a new tool to assist neuroscience research, driving neuroscience to shift from local functional analysis to systematic prediction of the brain's dynamic behavior.