Optical and quantum computing research targets AI processing limits

Researchers and technology developers are advancing optical and quantum computing methods to address the power and bandwidth limitations of traditional electronic processors and GPUs.

At the University of Queensland, researchers are exploring Quantum Optical Reservoir Computing (QORC). This hybrid approach uses photonic quantum processors and boson sampling to create enhanced data representations, or ‘quantum fingerprints’. By using the quantum processor as a fixed transformation layer to enrich data before it reaches a classical machine learning model, the method aims to improve image classification performance, particularly in environments with limited training data.

In the United States, Q/C Technologies is collaborating with the Center for Integrated Nanotechnologies (CINT) at Sandia National Laboratories. The partnership aims to develop an optical processing unit (OPU) designed for AI inference workloads. By using the interference of light to perform mathematical operations, the technology seeks to overcome the scaling and energy constraints of conventional electronic circuits. The research focuses on identifying the nanophotonic components necessary to build a scalable architecture while addressing challenges such as nonlinear operations, memory integration, and computational precision.