Optical computing, an early approach later abandoned in favor of binary electronic circuits, is also advancing. I’m fascinated by the possibility of building computers that use light as the ‘working fluid’, sending photons around like our current chips do with electrons.
This is already happening: photonic silicon chips offer high energy efficiency and help overcome the delay problems in traditional GPU architectures. They can reduce the time it takes to train deep learning models, enabling the next generation of advanced AI. There are opportunities to integrate photonics with new low-power chip designs such as that of TR35 winner Hongjie Liu of Reexen Technology.
In the long run, such photonic circuits can help us approach or perhaps even exceed generally accepted limits in computer science. Theoretical work on photonic information processing suggests that light can be converted to heat and vice versa, opening up some notable opportunities for all-optical energy storage — essentially batteries made of photons — and alternative computer architectures.
Many of these projects are still primarily in academia, but we are slowly moving towards building larger, more fully integrated systems. If we can continue to think about how to integrate these ideas into entire computing systems, further progress should be made in the coming years away from traditional chips and towards a range of different forms of computing.
Prineha Narang is the Howard Reiss Chair Professor of Physical Sciences at the University of California, Los Angeles (and was a 35 Innovators Honoree in 2018).