Skip to main content

Technology

Photonic Computing

Light-based processors using photons instead of electrons for ultra-fast, energy-efficient computation and AI inference.

91 companies

Explainer

Photonic Computing explained

Photonic computing uses light instead of electricity to carry or process information. Photons have no charge, so light in a waveguide does not cause the resistive heating that current in a wire does, though lasers and modulators still use power, and many wavelengths can share one waveguide. The industry has two quite different parts: moving data with light, and computing with light.

Moving data with light is established. Silicon photonics puts lasers, modulators and detectors on chips and is used in optical transceivers. Computing with light is earlier. Some designs use interference in a network of waveguides to do matrix multiplication, the core operation in neural networks, using very little energy per operation.

How it differs from the usual approach

Electronic chips compute with transistors that switch on and off. Optical computing is mostly analog: the answer comes from how light waves combine, so precision is limited by noise and is usually in the range of a few bits to around eight. Light is also hard to store and hard to make interact, so there is no practical optical equivalent of a memory or a general-purpose logic gate. Practical photonic chips therefore convert between electrical and optical signals, and that conversion can use much of the energy saved.

Where it is used

  • Optical transceivers linking servers and switches in data centers.
  • Co-packaged optics that put optical engines in the same package as switch and processor chips.
  • Accelerating matrix maths for AI inference in research and early products.
  • LiDAR and optical sensing.
  • Specialised optical processors for signal processing and optimisation problems.

What to watch

  • The nearest commercial impact is in optical links between chips and racks, not in replacing processors.
  • Photonic interconnect for AI clusters is further along in reaching products than photonic computing.
  • Photonic matrix accelerators still have to show an advantage after counting conversion, control and packaging costs.
  • Whether optical computing can reach the precision and scale needed for large AI models is not settled.

Frequently asked questions

What is photonic computing?

It is computing that uses light in place of, or alongside, electrical signals. In practice the term covers optical data links, which are already in use, and optical processors for matrix maths, which are still early.

Are optical computers faster than electronic ones?

Light travels fast, but speed is not the main benefit. The appeal is low energy per operation and high bandwidth. Converting between electronic and optical signals, and limited precision, reduce the gain.

Is silicon photonics the same as photonic computing?

No. Silicon photonics is a manufacturing platform for optical chips. It is mostly used today for communication, and some companies use it to build photonic computing chips as well.

Explore Other Technologies