France’s CEA research agency and quantum-computing company Alice & Bob have agreed to work on software that connects quantum processors with conventional supercomputers. The practical aim is not to replace today’s high-performance computing systems, but to make it possible to assign the small number of suitable tasks to quantum hardware when that hardware is ready.
The partnership matters because the question behind quantum computing is shifting from “who has the most qubits?” to how will quantum computers work with supercomputers? Software, error correction and workflow design are the pieces that determine whether a quantum processor can be useful in a real research or industrial setting.
At a glance
- CEA and Alice & Bob said on September 17 that they will work on quantum–HPC software integration.
- The project is intended to connect quantum processors with classical supercomputers, rather than treating them as separate systems.
- It is a long-term infrastructure effort: Alice & Bob’s hardware is planned for CEA in 2027, according to Reuters.
What the CEA–Alice & Bob partnership is trying to solve
Classical supercomputers remain far more capable for almost every workload used today. Quantum machines use qubits rather than ordinary bits, but current qubits are prone to errors and not all problems gain an advantage from quantum methods. A useful system therefore needs a way to divide work carefully: classical computing handles the large conventional workload, while a quantum processor is called on only for an algorithm that fits it.
According to the companies’ announcement, the collaboration focuses on accelerating quantum–HPC integration and industrial use cases. Reuters reported that it will build around Qaptiva, an open-source software project developed by Bull, and that CEA plans to add an Alice & Bob quantum computer in 2027.
That makes this an integration project, not a claim that a general-purpose quantum replacement for a supercomputer is already available. Alice & Bob says its approach centres on “cat qubits,” designed to reduce some error types at the hardware level; the company’s own glossary correctly stresses that fault tolerance means detecting and correcting errors while a computation is running.
Why software matters as much as the quantum hardware
Modern high-performance computing is already a mixed environment: CPUs, graphics processors, specialist accelerators, storage and networking software each handle different parts of a task. Adding a quantum processor creates another scheduling problem. Researchers need tools that decide when a quantum call is worthwhile, translate a problem into a quantum circuit, move results back to the classical system and measure whether the result is reliable.
Without that layer, a quantum device can be technically impressive but difficult for research teams to use. The partners’ stated focus on integration is therefore significant. It could help laboratories test hybrid workflows before quantum systems are mature enough for broad deployment.
Why it matters
Europe is trying to build capability across more of the computing stack, from research hardware to software and industrial use. The CEA is a major public research body, while Alice & Bob is a French quantum specialist. Their work will sit alongside existing CEA links with other quantum companies, Reuters reported.
For businesses, the near-term takeaway is modest: this does not make quantum computing a ready-made answer for everyday AI, cloud or analytics work. It is a step toward testing a future hybrid model. If the tools mature, quantum processors may eventually be used for narrow, high-value tasks in materials science, chemistry, optimisation or cryptography research while conventional systems continue doing most of the computation.
It also illustrates why the competition around advanced computing is not only about chips. Software standards and developer tools can create dependencies that last for years. The same pattern has been visible in the AI infrastructure buildout, where hardware matters but the software ecosystem determines how easily organisations can deploy it. See our analysis of Huawei’s Atlas 960 AI roadmap for another example of infrastructure strategy.
What happens next?
The first question is whether the partners can make hybrid workflows usable for scientists and engineers before Alice & Bob’s planned 2027 installation. The second is whether open tooling such as Qaptiva develops a broad enough community to support repeated, verifiable experiments.
Readers should be cautious about claims of imminent quantum advantage. The useful benchmark is not a laboratory demonstration alone, but whether a hybrid system can produce a result that is faster, cheaper or more accurate than the best available classical approach for a meaningful problem.
Sources
Alice & Bob newsroom announcement; Alice & Bob platform information; Reuters reporting. Read more in Technology.
