Researchers have developed a nanoprocessor that uses light instead of electricity, enabling a significant reduction in energy consumption when performing complex calculations in trillionths of a second.

Challenges of Traditional Computing

As artificial intelligence models advance, conventional electronic devices struggle to deliver the necessary computational speed and energy efficiency. Standard computer chips process data by moving charged particles, which leads to considerable heat loss and high energy costs.

The New Photonic Chip

To overcome these limitations, an ultra-compact photonic chip was created that performs mathematical computations using light. The processor was designed with advanced computer simulations that accurately model the interaction of light waves in three-dimensional space. This approach allows the use of miniature physical elements—each smaller than the wavelength of light—as customizable data points. As a result, the chip achieves a computational density of about 400 million parameters per square millimeter. The size of the resulting nanostructures is just tens of micrometers, comparable to the thickness of a human hair.

Operating Principle

When light passes through these intricate nanostructures, the physical geometry of the chip automatically performs the mathematical operations required for machine learning. Since the system operates using photons, calculations are completed in trillionths of a second.

Testing and Future Prospects

To test the prototype, the photonic neural network was used to classify over 10,000 biomedical images, including scans of the chest, breast, and abdomen. In physical experiments, the system achieved a classification accuracy of around 90%, and up to 99% in simulations. Integrating artificial intelligence capabilities directly into nanostructures has enabled the creation of a scalable and energy-efficient platform that could significantly reduce the environmental impact of future computing infrastructure.