Introduction

Optical computing, also known as photonic computing, is a type of computing that uses photons (light) instead of electrons (electricity) to perform operations (Anon., 2023). Computers traditionally rely on the flow of electrical currents through circuits and complex wiring. However, photonic computers may operate more compactly and efficiently through using light as a medium. We use optic technologies in everyday applications, ranging from fiber optic communication and data storage to imaging and sensing. However, optical computing, which involves using light to perform calculations and operations, is still an emerging technology. It was mostly neural networks that led to the renewed excitement in optical computing (McMahon, 2023) Neural networks have become a dominant part of machine learning and require a significant amount of computing resources, which optical computing can provide  (McMahon, 2023).

The embedded video below uses visuals to discuss and explain the concepts of photonic/optical computing:

Advantages

The outlook for optical computing looks promising, with photonic computing’s potential to provide a significant advantage over electronic computing in terms of bandwidth, speed, and energy efficiency, which is especially important for neural network processing. Bandwidth, in this sense, relates to the maximum amount of information transmitted through a medium. Photonics has a bandwidth around 100,000 times bigger than electronics, which means photonics can transfer greater amounts of data than electronics in a given time frame (McMahon, 2023).

Optical computers are energy efficient due to the use of light instead of electrical currents since photons (particles of light) can travel through optical systems with minimal energy loss. In contrast, electrical resistance leads to the generating of heat, resulting in energy wastage (Anon., 2023).

Computing at the speed of light and saving energy are not the only advantages of optics for computing:
Optical computing may also provide a high degree of parallelism: Photonic computing can perform multiple operations simultaneously, leading to faster computations, decreased size, weight, and flexibility making it an attractive option for computing (McMahon, 2023). The decrease in the size and weight of computing systems is a known trend of modern technology and makes the idea of photonic computers even more attractive.

How we started with electronics

Electronic telecommunication systems all started with telegraphy. Simple metal wires were used for baseband operations (channels transmitting signals in their original and unmodulated forms) before advancing into telephony (Athale & Psaltis, 2016). The increased demand for transmission bandwidth led to complex structures such as coaxial cables and, ultimately, fiber optic technology. Similarly, the starting point of computer technology mirrored the early stages of communication systems: simple electrical circuits operating at baseband frequencies.

The difference

As mentioned, our traditional computers use electrical currents that flow through circuits, and photonic computers use light as a medium, which means the fundamental difference lies in their means of representing and processing data.

The computers we use work with numbers because numbers can be written in binary form (1s and 0s). Computer components use electricity to represent ON or OFF states, where the ON state is the binary digit 1, and the OFF state is 0. Computers use binary to perform operations, and the processor is the component that is responsible for performing the operations. Optical computers use light to represent data and execute operations. The intensity and wavelengths of light may encode data, and through the use of optical components, we might manipulate the light and perform calculations. Some are working on optical architectures that use analog rather than digital computing.

Analog optical computing:

Analog optical computing is an approach to optical computing that uses analog signals rather than digital signals. These analog signals are continuous signals, while digital signals are discrete signals. Analog optical computing has the potential to be even faster and more energy-efficient than digital optical computing. We see advancements in this area with optical AI. Optical AI refers to the application of optical computing techniques in the field of artificial intelligence and, as mentioned previously, advancements in neural networks have sparked renewed interest in photonic computing due to its computing power possibilities. Researchers at NTT Corporation and the University of Tokyo have been working on optical AI and have made significant progress.

Sources and further reading

Anon., 2023. Artificial intelligence and the rise of optical computing: Information technology. Global Business Review, 3 January.

Athale, R. & Psaltis, D., 2016. Optical Computing: Past and Future. Optics and Photonics News, 27(6), pp. 32-39.

Denz, C., 1998. Optical Neural Networks. 1 ed. s.l.:Vieweg+Teubner Verlag.

McMahon, P. L., 2023. The physics of optical computing. Nature Reviews Physics.

Minzioni, P. & al, e., 2019. Roadmap on all-optical processing. Journal of Optics, Volume 21, p. 55.