computing platform profile
Photonic quantum computing
Photonic systems represent qubits with properties of individual light particles. They are naturally suited to communication and can operate with less cryogenic infrastructure at the qubit itself.
Technical deep dive
How it works
Sources prepare single photons or entangled optical states, interferometers transform their paths and phases, and detectors make measurements that implement or complete gates. Because photons interact weakly, architectures often use measurement, feed-forward, multiplexing, or matter interfaces. Loss in sources, switches, waveguides, couplers, and detectors compounds quickly as circuits grow.
Encoding and optical transformations
A photonic qubit can occupy one of two paths, time bins, polarizations, frequencies, or modes. Phase shifters and beam splitters implement precise single-qubit transformations by changing interference between modes. Detection measures photon number or arrival channel; superconducting nanowire detectors often provide the best efficiency and timing but require cryogenic operation.
Photons barely interact in ordinary materials. Two-photon gates therefore use interference between indistinguishable photons plus ancillary photons, measurement, and feed-forward, or they couple photons through an emitter, cavity, or nonlinear medium. Hong–Ou–Mandel interference is a key diagnostic: spectral, temporal, spatial, or polarization mismatch makes photons distinguishable and lowers entangling fidelity.
Measurement-based computation
Many architectures first build a large entangled cluster or graph state. Adaptive single-photon measurements consume that state while teleporting quantum information through it; the choice of later measurement bases depends on earlier outcomes. Fusion measurements stitch smaller entangled resources into a larger state, while multiplexing routes a successful probabilistic source into the needed place and time.
Quantum error correction can tolerate loss and operation errors only below architecture-specific thresholds. Every source inefficiency, coupling loss, propagation loss, switching loss, detector inefficiency, and failed fusion compounds. Resource estimates must therefore include photons attempted, not only photons that survived into the reported state.
Modularity and hardware stack
Silicon, silicon nitride, lithium niobate, and III–V platforms provide different combinations of low-loss routing, fast modulation, and on-chip emitters. Hybrid integration can combine the best components but adds alignment and packaging interfaces. Delay lines or quantum memories are needed while classical electronics calculate feed-forward decisions.
Room-temperature waveguides do not make the whole computer room-temperature. Deterministic emitters may need cryogenic stabilization, and leading photon counters are often superconducting. A useful machine must co-package optical loss, thermal control, phase stabilization, detector recovery, classical decoding, and manufacturable fiber or chip connections.
Reason to use it
What distinguishes the approach
Photons move naturally through networks, interact weakly with their environment, and can often be routed without cooling the entire optical circuit to millikelvin temperatures.
Unresolved engineering
Main scaling problem
Low-loss components and efficient, indistinguishable photon generation
Monthly evidence layer
Research watch
Named programs, dated developments, and primary or official sources. Company claims are identified separately from independent or peer-reviewed evidence.
Last reviewed 6 September 2026
Research groups & programs
- Xanadu and collaborators
Modular photonic architecture, interconnects, error-correction primitives, and system integration.
- PsiQuantum and collaborators
Manufacturable silicon-photonic sources, detectors, switches, chip-to-chip links, and fusion-based computing.
Current developments
2025
Modular subsystems were connected in a common architectureThe result was important because scaling depends on interfaces among sources, circuits, detectors, and networks, not one optical chip in isolation.
2025
Manufacturable photonic building blocks reached system-level benchmarksThe platform co-integrated generation, manipulation, detection, and networking, while reported operation remained conditional on photon detection and excluded loss.
Papers & official resources
- paperScaling and networking a modular photonic quantum computer
Nature · 2025
- paperA manufacturable platform for photonic quantum computing
Nature · 2025