The chip performs convolution operations by converting data into laser light, passing it through miniature Fresnel lenses, and then converting it back to digital signals, enabling near-zero energy consumption for pattern recognition tasks.

Industry experts and researchers believe that optical AI computing will become a core component of future AI hardware, marking a significant step toward more sustainable and powerful AI systems.

Testing shows the chip can classify handwritten digits with approximately 98% accuracy, comparable to traditional electronic chips, while simultaneously processing multiple data streams through wavelength multiplexing.

A new photonic chip leverages wavelength multiplexing with different colored lasers, enabling it to handle multiple operations simultaneously, which significantly enhances scalability and speed in AI processing.

This chip also employs spectral multiplexing to process multiple wavelengths at once, reducing power consumption and increasing efficiency for AI tasks.

Developed through collaboration with institutions like UCLA, George Washington University, and the Florida Semiconductor Institute, this research indicates optical elements are already being integrated into commercial AI chips, such as those from NVIDIA.

The chip, called pJTC, uses optical components like Fresnel lenses and integrated lasers to perform complex mathematical operations at high speeds, operating at gigahertz frequencies.

Researchers at the University of Florida have created a light-powered photonic chip that drastically improves AI computation efficiency by replacing electrical components with optical ones, which reduces energy consumption and boosts processing speed.

This innovative chip, the pJTC, performs AI calculations using light, enabling faster processing with significantly lower energy use.

The design incorporates photonic components from optical transceivers and features photonic wire bonding, resulting in a compact and integrated chip.

The pJTC achieves up to 305 trillion operations per second per watt and 40.2 trillion per square millimeter, outperforming traditional electronic chips in efficiency.

This technological breakthrough could transform AI hardware across data centers, edge devices, and cloud services, supporting applications like autonomous vehicles and medical imaging with faster, more energy-efficient processing.