Could future computers use photons instead of transistors?

The biggest issue is that photons don’t naturally interact with one another very strongly. Ordinary digital logic needs reliable switches and nonlinear operations, so creating compact all-optical versions of gates, memory, and amplification is difficult. Converting electrical data to light and then back again can also erase much of the speed or energy advantage.

A realistic near-term direction is hybrid hardware: electronics handle storage, control, and complicated logic, while photonic circuits handle communication or highly parallel mathematical operations. Optical links can offer enormous bandwidth, but that doesn’t automatically mean the entire processor runs 100 times faster, since memory access, synchronization, conversion, and heat still matter.

Yes—this field is generally called optical or photonic computing, and researchers are actively developing photonic processors. Light can move data efficiently and multiple wavelengths can share the same optical path, which is useful for high-bandwidth tasks such as neural-network calculations and signal processing.

Optical components are often physically larger than transistor circuits because light has to be routed through waveguides, bends, resonators, and other structures. That makes dense general-purpose optical CPUs difficult to build. Electronic transistors are slower in clock frequency, but they are tiny, cheap, robust, and already excellent at logic and memory.