A team at the University of Vienna led by Philip Walther has demonstrated long-term in-orbit operation of a programmable quantum photonic processor, according to the arXiv paper *In-orbit operation of a programmable quantum photonic processor* (arXiv:2609.25248v1, submitted 2026-09-21). The payload was launched aboard D-Orbit's ION SCV Passionate Paula on June 23, 2025, into a 510 km sun-synchronous orbit — reportedly the first long-duration in-orbit operation of such a processor.
Key hardware specs
- Size budget: 15 × 15 × 45.3 cm³, 9.8 kg, 10 W — roughly the weight of a portable projector
- Photon source: 405 nm pump + 1 cm ppKTP crystal generating 810 nm entangled photon pairs (waveguide writing by Roberto Osellame & Francesco Ceccarelli, Politecnico di Milano)
- Linear optics: 6-mode, 15 Mach-Zehnder interferometer silicon photonic circuit, each phase externally programmable
- Detection: single-photon detectors; hardware integration by D-Orbit
- Orbit conditions: 92-minute orbital periods; experiments run only in ~30-minute eclipse windows, since sunlight floods the 810 nm channel; the remaining ~62 minutes are spent cooling down.
- Headline result: In-orbit Hong-Ou-Mandel single-photon interference visibility of 0.908 ± 0.191, above the 0.5 classical limit by 2.14 standard deviations. Pre-launch ground measurement was 0.924 ± 0.003. The large in-orbit uncertainty reflects limited data collection time and low photon count rates.
- Radiation damage handling: physical shielding plus overbiasing degraded detectors from 7 V to 12 V to restore single-photon sensitivity — a standard but rarely disclosed technique for silicon avalanche photodiodes in space.
- Detectors C2/C3/C4 had already failed before launch; the lifetime of the remaining C1/C5/C6 is an open question, and dark counts accumulate over time.
- Thermal drift remains unsolved: the paper states that maintaining spectral degeneracy and autonomous thermal re-calibration of the circuit is still an open problem, currently requiring ground intervention.
- Boson sampling — feasible at this scale
- Quantum walk simulation — feasible
- Photonic neural network layers and quantum machine learning kernels — only at small scale
- Universal fault-tolerant algorithms — far out of reach at 6 modes
- [1] Steiner, S. et al., *In-orbit operation of a programmable quantum photonic processor*, arXiv:2609.25248v1, submitted 2026-09-21
- [2] Philip Walther group, University of Vienna
- [3] D-Orbit ION SCV Passionate Paula platform specs (Transporter-14 mission, launched 2025-06-23)
- [4] Roberto Osellame & Francesco Ceccarelli (Politecnico di Milano), ppKTP waveguide writing publications
- [5] Quantum Brief, *Two photons met on a glass chip 510 kilometres up and came out the same side*, 2026-09-24
- [6] Public roadmaps of PsiQuantum / Xanadu / Quandela
Key points
What can 6 modes actually do?
Compared with ground-based processors at 200+ modes (IBM, PsiQuantum, Xanadu, Quandela), the significance is not scale but proof of engineering feasibility: a lab-grade programmable quantum optical circuit can survive launch, radiation, vacuum, and thermal cycling and keep doing genuine quantum optics in orbit.
Stated limitations (from the paper)
1. Autonomous thermal re-calibration still requires ground involvement 2. 30 minutes per orbit and ~6 coincidences per second are far too little for any real classification task — the paper notes: *"A classification task needs far more than 30 minutes an orbit and six coincidences a second, and the payload loses a little efficiency every week it keeps orbiting."* 3. Dark counts accumulate; overbiasing is only a stopgap 4. Detector lifetime is finite
Bigger picture
The result sits at the intersection of three trends: rapid scaling of ground-based quantum photonic processors toward 100+ modes; commercialization of satellite quantum communication and QKD (Mynaric, SpeQtral); and low-Earth-orbit small satellites becoming the default platform for scientific payloads. Notably, the engineering barrier for "space quantum computing" is lower than for space-based classical HPC — no radiation-hardened GPUs needed, just stable entangled-photon sources and detectors.
Conclusion
The answer to "can a programmable quantum photonic processor work long-term in space?" is: yes, with conditions — a 30-minute-per-orbit window, imperfect but solid quantum visibility, ongoing ground calibration support, and a mission lifetime of a few years. Progress on longer eclipse windows, better thermal control, and longer-lived detectors would move in-orbit quantum photonic computing from engineering validation toward mission applications — toward space-based quantum nodes that *compute* with entangled photons rather than merely relay them.