Photonic quantum computing
Photonic quantum computing encodes qubits in properties of individual photons, such as polarization, path, or time bin, and processes them with linear-optical elements (beam splitters, phase shifters) together with single-photon detectors. Photons interact weakly with their environment and with each other, so photonic machines can operate at or near room temperature but must overcome photon loss and the difficulty of making photons interact.
How it works
Because photons do not naturally interact, a naive optical circuit cannot perform a deterministic two-qubit gate. The 2001 KLM scheme showed that efficient universal quantum computation is nonetheless possible using only linear optics, single-photon sources, and detectors, by using measurement and feed-forward to induce effective nonlinearity (Knill, Laflamme, Milburn 2001). Most modern designs adopt the measurement-based, or one-way, model: a large entangled cluster state is prepared, and computation proceeds by measuring its qubits in chosen bases, with later measurements adapted to earlier outcomes (Raussendorf and Briegel 2001). This shifts the challenge from performing gates to generating and stitching together entangled resource states.
Strengths
- Photons have long coherence and are natural carriers for networking and communication, so photonic qubits integrate well with quantum links and distributed architectures.
- Many components can work at room temperature, avoiding the millikelvin cryogenics that superconducting circuits require, though photon detectors are often still cryogenic.
- Silicon photonics offers a path to manufacturing large numbers of identical optical components on chips.
Limitations
- Photon loss is the dominant error: every optical component and fiber has some probability of losing a photon, and lost photons cannot be recovered, so scaling requires very low loss and heavy error correction.
- Single-photon sources and detectors are imperfect; producing photons on demand with high purity remains difficult.
- Linear-optical gates are inherently probabilistic, so architectures rely on multiplexing and generating resource states offline to reach deterministic behavior.
Fault tolerance in this setting depends on Quantum error correction codes adapted to loss, which raises the resource overhead further.
Who is building it
- PsiQuantum pursues large-scale, fault-tolerant photonic computing using silicon photonics and a fusion-based measurement architecture, and describes plans for utility-scale machines rather than small demonstrators.
- Xanadu builds programmable photonic processors and released Borealis, which the company reported as a demonstration of a sampling task intractable for classical simulation (Xanadu 2022).
Independent groups have also used photonics for boson-sampling experiments claiming Quantum advantage on contrived tasks (Jiuzhang 2020). As with all such claims, the honest reading distinguishes a narrow, specially chosen benchmark from useful computation, and some results have been challenged by improved classical algorithms.
Encodings
Photonic qubits can be defined in several ways, and the choice shapes the architecture. Polarization encoding uses the horizontal and vertical states of a photon; path encoding uses which of two waveguides a photon travels; time-bin encoding uses early or late arrival. Some designs instead use continuous-variable states, encoding information in the quadratures of the light field, which is the basis of the squeezed-light approach used in several boson-sampling machines. Each encoding trades ease of manipulation against robustness to loss, and integrated silicon photonics is attractive because it can pack many such components onto a single chip with reproducible performance.
Status
As of early 2026 photonic systems have demonstrated advantage-style sampling and steady progress on integrated components, but no photonic machine performs error-corrected, cryptographically relevant computation. Like other modalities, photonics remains in the NISQ era and does not change the uncertain timeline to Q-Day or whether Bitcoin is quantum safe.
Sources
- A scheme for efficient quantum computation with linear optics (KLM) (Nature (Knill, Laflamme, Milburn), 2001)
- A One-Way Quantum Computer (arXiv (Phys. Rev. Lett.), 2001)
- Quantum computational advantage with a programmable photonic processor (Borealis) (Nature (Xanadu), 2022)
- Quantum computational advantage using photons (Jiuzhang) (Science, 2020)
Cite this entry
"Photonic quantum computing." postquantum.wiki. Updated July 11, 2026. https://postquantum.wiki/photonic-quantum-computing@misc{pqwiki-photonic-quantum-computing,
title = {Photonic quantum computing},
howpublished = {\url{https://postquantum.wiki/photonic-quantum-computing}},
year = {2026},
note = {postquantum.wiki, updated 2026-07-11}
}