Could future computers use light instead of transistors?

A major engineering problem is that photons do not naturally interact with one another very strongly. Electronic transistors can easily use interactions between charges to create logic gates, while optical systems need special materials or structures to make light signals affect each other. Converting electrical data to light and then back again can also introduce energy costs and delays that erase much of the advantage.

Yes, optical computing is an active research area. Light can move through photonic circuits at very high frequencies, and multiple wavelengths can sometimes share the same path, allowing a lot of data to be processed in parallel. These systems are being explored for specialized tasks such as artificial-intelligence workloads, data-center connections, and optical circuit boards. They are not yet general-purpose replacements for ordinary CPUs.

The headline speed advantage can be misleading. Signals inside modern electronic processors already travel at a significant fraction of light speed, so replacing electrons with photons does not automatically make every operation dramatically faster. The real benefits are usually bandwidth, parallelism, and reduced energy for moving data rather than simply faster clock speeds.

Physical design is another limitation. Light generally needs waveguides, mirrors, or other structures to be redirected, and those components can take more space than tiny electronic wires. Optical delay lines also tend to be much larger than electronic memory, so replacing caches and storage completely would be difficult. A more realistic future is hybrid hardware, with electronics handling control and memory while photonic circuits accelerate particular calculations or high-bandwidth connections.