Lucy: France’s Photonic Quantum Processor Connects to a National Supercomputer

At the CEA’s secure Bruyères-le-Châtel site, a photonic quantum processor known as “Lucy” has been installed and connected to a national supercomputer. Its appearance is understated, but its arrival marks an important change.

What is inside the black cabinet

Built by French start-up Quandela, Lucy encodes qubits in light. The machine currently contains 12 photonic qubits. Rather than being a laboratory demonstration only, it is a universal, programmable system. Its engineers created it as a modular product for deployment in a high-performance computing environment. At the TGCC campus, the CEA has linked it with the Joliot-Curie supercomputer. This connection is significant for practical workloads rather than simply for publicity images.

First universal photonic quantum processor on European soil, wired into a national supercomputer and prepared for remote users.

From circuits to light: why photons matter

Most quantum hardware is based on superconducting circuits, trapped ions or neutral atoms. Such approaches require ultra-cold refrigeration systems or sophisticated electromagnetic traps. Photonic machines follow another path, using individual particles of light within integrated optical circuits. Much of their hardware operates at room temperature. Compact cryogenic modules house the single-photon sources, which then send light into chips that direct, interfere with and measure it.

Photons offer several notable advantages. They connect naturally with telecommunications networks and can be routed over long distances. They are also less vulnerable to many of the environmental disruptions affecting solid-state qubits. The cabinet can be placed in a conventional data hall and connected by fibre. This creates possibilities for hybrid computing and secure networking that are harder for other platforms to achieve.

A European machine intended for shared use

The CEA runs Lucy for public research purposes. Its funding is provided through EuroHPC as part of the EuroQCS-France programme, while GENCI manages access for researchers. Developers can already familiarise themselves with Quandela’s software tools, including Perceval for photonic circuits and MerLin for quantum machine learning. Direct use of Lucy is due to expand throughout 2026.

Who gets access and when

• Academic groups submit applications through GENCI’s national allocation calls.
• Industrial R&D teams may seek pilot access to evaluate hybrid workflows.
• Training courses and webinars help users prepare before connecting to the physical hardware.
• Initial users can develop work on simulators before transferring it to Lucy when prepared.

Roughly 80% of Lucy’s components come from European suppliers, and the full machine reached the data hall after a twelve-month build.

Sovereignty by design

Quantum leadership is shaped by supply chains, and Lucy relies heavily on local production. Quandela manufactures its photonic circuits in Palaiseau, while the cryogenic modules are produced near Munich. Final assembly is carried out in Massy, south of Paris. The Franco-German collaboration involves attocube systems AG alongside funding streams from France 2030 and the National Quantum Strategy. Together, these arrangements limit dependence on export controls and allow European teams to manage maintenance and upgrades themselves.

A Franco-German supply chain

Photonic sources, laser units, optical stabilisation, packaging and control electronics are now sourced within a European ecosystem. The strategy is about more than displaying a national flag. It develops expertise in photonic chip design, cryo-optics, calibration, and quantum-classical control software. That expertise can also benefit sensors, telecommunications and secure communications.

HPC meets quantum in practice

At the site, Lucy communicates with Joliot-Curie through a managed software stack. Workloads are divided between classical computing nodes and the photonic processor. In practice, this involves classical pre-processing, followed by a quantum kernel and then classical post-processing. Engineers also intend to connect Lucy with the next-generation exascale system, Alice Recoque, once it enters service. Together, the three systems are intended to address problems involving more variables than classical machines can handle alone.

What hybrid jobs look like

• Grid optimisation: balancing European electricity networks amid variable renewable generation.
• Risk analytics: sampling financial worst-case scenarios while relying on fewer modelling shortcuts.
• Aerospace logistics: planning fleets and payloads subject to strict constraints.
• Materials R&D: examining molecular properties where classical approximations can skew results.
• Quantum machine learning: training compact models that make use of interference patterns.

How Lucy compares with rivals

Photonic quantum computing is a competitive field built around very different technical approaches. China’s Jiuzhang systems have delivered notable optical-sampling demonstrations, but remain analogue and designed for specific tasks. Canada’s Xanadu has advanced continuous-variable photonics. In the UK, ORCA Computing produces modular fibre-based machines that have already undergone Ministry of Defence testing. PsiQuantum is pursuing a large-scale silicon-photonic approach to fault-tolerant computers, supported by substantial industrial backing.

Machine	Platform	Scale	Home base	Status	Note
Lucy	Photonic, universal	12 qubits	France (EU)	Operational	Integrated with a national supercomputer
Jiuzhang 2.0	Photonic, analogue	≈100 modes	China	Experimental	Sampling showcase, not fully programmable
IBM Osprey	Superconducting	433 qubits	United States	Operational	Commercial access and clear roadmap
IonQ Forte	Trapped ions	32 qubits	United States	Commercial	High-fidelity gates, cloud access
ORCA demo systems	Photonic, fibre-based	Modular	United Kingdom	Pilots	Defence trials, modular architecture

What Lucy means for France and Europe

France now has an operational photonic quantum computer available for researchers to use. The message extends beyond national prestige. Europe requires hardware that it can buy, run and improve over time, and Lucy demonstrates that ambition. It also establishes an expectation: universal programmability, straightforward integration with classical clusters, and access for outside users.

The limitation, and why it still matters

Twelve physical qubits will not independently solve drug-discovery problems. Logical qubits created through error correction are still a medium-term objective. Noise control and calibration continue to determine which workloads perform effectively. Even so, hybrid quantum kernels can already be incorporated into classical workflows, reducing runtime or improving solution quality for certain cases. This is likely to be the first route to practical value.

How to prepare for the photonic era

Teams should begin with modest projects. They can simulate photonic circuits on classical nodes and move them to hardware when appropriate. They should learn how physical qubits differ from logical qubits, as well as understand error mitigation, batching and shot management. The quantum machine should be viewed as a specialist accelerator that works alongside GPUs and CPUs, rather than as a universal solution.

Choose one or two pilot use cases with measurable outcomes, such as a portfolio VaR calculation or route optimisation. First, create a classical solver as a baseline. Next, assess a quantum-assisted version. Record wall-clock time, solution quality and energy consumption. Retain all findings, including unsuccessful ones, because that record will inform the next grant application or budget request.

Why photons could transform networking first

The earliest benefits may emerge in secure communications and distributed computing. Photons can travel through existing fibre infrastructure and distribute quantum states between data centres. This could enable secure keys and, eventually, distributed quantum processing. Since Lucy is located at a supercomputing facility with robust networking and controls, it could serve as a hub for these experiments.

For the UK, the development is particularly relevant. Photonics already underpins major parts of British quantum activity. ORCA’s fibre-based approach, university laboratories in Bristol, Oxford and London, and industrial links to telecommunications provide a strong foundation. A French machine connected to a European supercomputer offers a natural counterpart for collaborative trials. Cross-Channel workflows could transfer photons, data and algorithms in both directions.