Building Silicon Brain Cube for Green and Trustworthy AI

A new hardware design called the Silicon Brain Cube stacks computing layers in three dimensions to cut the energy consumed by artificial intelligence while also making its decisions more trustworthy. Today’s AI systems, especially large language models, guzzle electricity because they constantly shuffle data between memory and processors. This project tackles that inefficiency head-on by building a wired-logic fabric that processes data directly where it sits, drastically reducing memory access. The UK and Japanese teams will jointly develop an AI model that lets users dial in their own trade-offs between speed, accuracy, energy use, and uncertainty—a feature essential for high-stakes decisions. The hardware aims to improve energy efficiency by more than an order of magnitude. If successful, the Silicon Brain Cube could reshape how AI runs in data centres, medical diagnostics, autonomous vehicles, and energy grids—anywhere that currently requires both heavy computation and reliable outputs. The project also plans to release an open-source repository of designs, tools, and tutorials, so the architecture can be adopted and improved beyond the grant period. This is applied engineering with a clear practical target: make AI both greener and more accountable.