Bélénos, the photonic quantum computer developed by French startup Quandela, is now the world's most powerful machine of its kind, delivering a computational leap 4,000 times greater than its predecessor. Built in Massy, near Paris, it runs on photons rather than superconducting circuits, and it is quietly reshaping the global quantum computing landscape from a campus most people have never heard of.
France rarely tops global technology rankings without fanfare. But in June 2025, a company founded in 2017 on the Plateau de Saclay reached a milestone that neither IBM, NVIDIA, nor Google had achieved with photonic architecture: a 12-qubit photonic quantum processor with a performance gap so wide it belongs in a different category. The machine is called Bélénos, and almost nobody outside the quantum research community knows it exists.
Bélénos works with light, not cold
Most quantum computers on the market today rely on superconducting circuits or trapped ions. Both approaches require extreme cooling, typically close to absolute zero, which translates into enormous energy costs and deeply complex infrastructure. Quandela took a fundamentally different path from the start.
The photon as a qubit carrier
The company's entire architecture is built around photons, particles of light traveling at 299,792,458 meters per second. The key to making this work reliably is a photonic source called eDelight, developed by physicist Pascale Senellart-Mardon. This source produces single photons that are consistent, repeatable, and stable enough to serve as qubits — the basic unit of quantum information. Without that reliability, photonic quantum computing remains theoretical. With it, Quandela built a machine that scales.
A full-stack architecture that controls everything
What sets Quandela apart from many competitors is its full-stack approach. The company controls every layer of the system: the photonic source, the silicon optical circuits, the control electronics, the middleware, and the application-layer algorithms. This vertical integration gives engineers the ability to optimize each component in relation to the others, rather than assembling parts from different suppliers and hoping they work together. The result is a tighter, more coherent system, and a much faster path to performance improvements.
Bélénos is the direct product of that strategy. Between the previous generation and this machine, just two years passed. The computational power increased by a factor of 4,000. That is not incremental progress.
performance gain of Bélénos over the previous generation
Quandela's rise from lab spinoff to industrial supplier
Quandela was founded in 2017, growing out of research conducted at the CNRS ecosystem around the Plateau de Saclay. Its first significant funding round raised 15 million euros, backed by Quantonation and Bpifrance. By November 2023, a second round brought in 50 million euros, with participation from Serena, Crédit Mutuel Innovation, and the European Innovation Council. A third round, targeting more than 100 million euros, is currently in preparation.
From 6 qubits to industrial deployment
The commercial timeline is worth tracing carefully. At the end of 2022, Quandela opened the first European cloud access to a photonic quantum processor — a 6-qubit machine. That milestone, modest by today's standards, established the company as a genuine provider rather than a research project. By March 2024, OVHcloud had installed a private quantum computer. Scaleway began offering access through a "Quantum as a Service" model. Exaion, the EDF subsidiary, also became a client.
The machine called Lucy was delivered to the CEA, France's atomic and alternative energies commission. These are not pilot programs. These are operational deployments at major French institutions, and they represent a commercial foundation that most quantum startups in Europe cannot claim.
Quandela is a laureate of the French Tech 2030 program and participates in PROQCIMA, a French defense procurement initiative (DGA) targeting two universal quantum computer prototypes by 2032.
A team of over 100 people from 20 nationalities
Behind the hardware is a team of more than 100 people drawn from roughly 20 nationalities, based at the company's headquarters in Massy. The diversity is not incidental — quantum physics and photonic engineering draw talent from across Europe and beyond, and Quandela has positioned itself as a European hub for that expertise. Expansion into North America and Asia is already planned.
The roadmap points toward hundreds of qubits by 2030
Bélénos at 12 qubits is not the endpoint. Quandela's published roadmap is explicit: 50 physical qubits by 2028, and several hundred qubits by 2030. The next machine, Canopus, is scheduled for 2026 and will double the current qubit count.
By end of 2025, the company plans to deliver a fully integrated version of Bélénos coupled directly to a supercomputer, enabling hybrid classical-quantum computation. This architecture is significant because it allows quantum processors to handle the specific tasks they excel at — optimization, simulation, certain cryptographic operations — while classical hardware manages everything else. The hybrid model is increasingly seen as the realistic near-term path for quantum utility in sectors like finance, cybersecurity, energy, logistics optimization, and artificial intelligence.
Production capacity is also scaling. Quandela targets a cadence of four machines per year starting in 2025, and plans to open a second manufacturing facility in 2027.
Photonic quantum computers operate at or near room temperature, unlike superconducting systems that require cooling to near absolute zero. This makes them significantly more energy-efficient and easier to integrate into existing data center infrastructure.
Photonic quantum computing still faces one hard problem
The advantages of photonic architecture are real. Lower energy consumption compared to cryogenic approaches. Greater modularity, which makes scaling more straightforward. Better qubit stability, since photons are naturally isolated from environmental noise. And an architecture that integrates more naturally with fiber-optic communications infrastructure, which matters enormously for quantum networking.
But the challenge that no photonic quantum computer has fully solved is error correction. Quantum systems are inherently noisy, and building logical qubits from many physical qubits — the approach needed to run truly fault-tolerant algorithms — requires a very large number of physical qubits working in concert. The 2025 target for Quandela includes demonstrating the first error-free logical qubits, which would be a critical proof-of-concept. Reaching the millions of high-quality qubits that experts believe will be needed for transformative industrial applications remains a long-term challenge shared by every quantum computing approach, photonic or otherwise.
Just as researchers have found unexpected complexity beneath familiar surfaces, quantum engineers continue to uncover new layers of difficulty in building systems that are both powerful and reliable. And just as nations compete fiercely in advanced materials and propulsion technologies, the race for quantum supremacy involves strategic calculations that extend well beyond the laboratory.
What Quandela has demonstrated with Bélénos is that the photonic path is not a niche detour. It is a credible route to large-scale quantum computing, and France is currently leading it. The 4,000-fold performance jump between generations, achieved in just two years by a team of 100 people in a suburb south of Paris, suggests that the next few years will produce surprises that even the quantum research community has not fully priced in.
