Summary
- Alice & Bob has doubled its 2030 target to 200 logical qubits, with a proposed Carbon system using 3,600 physical cat qubits.
- Its modular architecture would connect quantum processing units at cryogenic temperatures to address scaling limits associated with individual refrigeration systems.
- Further targets extend to 2,048 logical qubits in 2034, but the larger systems and their proposed applications remain unproven commercial capabilities.
French quantum computing company Alice & Bob has doubled its target for fault-tolerant quantum hardware, setting out plans for a 200-logical-qubit system by 2030 using modular processors designed to reduce the physical resources required for reliable computation.
The revised roadmap, published on 18 November, identifies a proposed machine called Carbon containing 3,600 physical cat qubits. The company previously targeted 100 logical qubits by 2030 and now plans further development stages leading to a system with 2,048 logical qubits by 2034.
Alice & Bob develops superconducting cat qubits, an approach intended to suppress certain quantum errors through the underlying hardware. The company argues that this could reduce the number of physical qubits required to create reliable logical qubits, helping address a major obstacle to building computers capable of performing long calculations.
The larger systems remain future targets. Although Alice & Bob has demonstrated individual technologies relevant to error correction and developed operational quantum hardware, it has not demonstrated a 200-logical-qubit computer or independently established the commercial performance anticipated in its 2030 roadmap.
Hardware design changes error correction demands
Quantum computers store and manipulate information using physical systems whose states can be disturbed by their environment and by imperfections in the equipment controlling them. These disturbances introduce errors that become increasingly important as calculations involve more operations.
Fault-tolerant quantum computing seeks to manage those errors so that reliable calculations can continue without the result being overwhelmed by accumulated faults. One approach combines several physical qubits into logical qubits, using additional operations to detect and correct errors.
The resulting hardware requirement can be considerable. A machine performing useful calculations may need far more physical qubits than the number of logical qubits available to execute an algorithm, particularly where the underlying devices have limited reliability.
Alice & Bob’s cat-qubit architecture is intended to change that balance. Its superconducting hardware suppresses bit-flip errors, one of the main error categories, reducing the burden placed on additional correction mechanisms.
The remaining errors still need to be addressed through suitable coding, measurement and control systems. Suppressing one error category does not make a physical qubit fully reliable or remove the need to demonstrate logical operations under realistic conditions.
The company’s Carbon target combines 3,600 physical cat qubits into a proposed system with 200 logical qubits. Those figures describe the intended architecture, not a measured conversion ratio that can automatically be applied to other quantum processors.
Performance also depends on the reliability of logical operations, the rate at which errors occur and how long calculations can continue before corrective mechanisms become insufficient. Counting qubits alone does not establish whether a machine can outperform conventional computing on a commercially useful task.
Modular processors introduce a different challenge
The updated roadmap places greater emphasis on connecting separate quantum processing units rather than continuously expanding a single processor within one refrigeration system.
Superconducting quantum hardware operates at extremely low temperatures, requiring specialised equipment to maintain stable conditions. As systems grow, the available space, wiring and control connections inside a cryogenic environment become increasingly difficult to manage.
A modular architecture could allow processors to be developed and tested separately, then combined to increase computing capacity. Alice & Bob intends to connect its modules at cryogenic temperatures, using the approach to address limits associated with a single cryostat.
Manufacturing repeatable processing units may also offer operational advantages. Designing smaller modules around established components can support more consistent development and testing than building every successive system as a unique, increasingly large installation.
However, connecting quantum processors introduces demanding engineering requirements. Quantum information must be transferred or shared between modules without introducing so many errors that the benefits of additional computing capacity disappear.
The control systems also need to coordinate operations across different processors, while error correction must account for faults affecting communication as well as those arising within individual modules.
Alice & Bob has not demonstrated the complete proposed modular architecture at its intended 2030 scale. Its development programme must establish that connections between modules can support reliable logical operations alongside the performance improvements expected from cat qubits.
Roadmap extends to 2034
The company’s near-term programme centres on Helium, a prototype system associated with its 2026 logical qubit milestone. Alice & Bob unveiled the Helium Quantum System in June, establishing an initial hardware platform, although that development should not be confused with demonstrating the later roadmap’s fault-tolerant capabilities.
The next named system, Beryllium, is targeted for 2028. It would provide six logical qubits and support a universal set of quantum operations, allowing the machine to perform the types of instructions required by general quantum algorithms.
Carbon would follow in 2030 with 200 logical qubits. Alice & Bob associates that stage with materials simulations, including calculations involving the Ising and Hubbard models used to study interactions in physical systems.
The company describes Carbon as its first intended utility-scale machine, capable of performing calculations beyond practical classical computing limits with economic value. That characterisation remains a development objective rather than a demonstrated advantage.
Oxygen is planned for 2032, targeting 1,024 logical qubits. The company identifies more complex chemistry calculations, including simulation involving the iron-molybdenum cofactor, alongside proposed cybersecurity applications using Shor’s algorithm.
Neon, targeted for 2034, would increase capacity to 2,048 logical qubits and support more demanding research in chemistry, materials and cryptography, including work intended to test the security of systems based on RSA-2048.
These proposed capabilities do not mean that existing encryption has already been compromised or that Alice & Bob has established a date when a particular cryptographic system will become insecure. Actual performance will depend on the complete machine, its error rates, available operations and the resources required to execute individual algorithms.
Research results support the revised plan
Alice & Bob points to experimental work conducted through France’s PROQCIMA quantum computing programme as evidence supporting the updated targets. The company reports building, assembling and operating three independent quantum processing units and demonstrating individual components required for logical error correction.
Its researchers also report progress towards the error performance needed for their proposed correction methods. These results provide evidence that elements of the architecture can operate on physical hardware, although assembling the components into a reliable logical qubit remains a separate technical milestone.
The company has expanded its chip design, fabrication and testing capabilities to support repeated hardware development. It also says AI-assisted research tools are helping accelerate aspects of device design, modelling and experimentation.
Those internal improvements may shorten development cycles, but they do not establish that the larger machines will meet their target dates or deliver the expected economic advantages.
Alice & Bob reports raising €180 million and employing more than 250 people, with operations in Paris and Boston. The resources support a programme requiring specialist equipment, manufacturing development and a highly trained engineering workforce.
Quantum computing also faces substantial operating costs. Cryogenic equipment, control electronics, maintenance and specialist facilities influence the economics of a system even after its technical capabilities have been demonstrated.
The financial value of a quantum application will therefore depend on whether the machine can complete useful work at an acceptable cost, compared with conventional computing and other available methods.
Alice & Bob has outlined a more ambitious route towards that objective, with modularity and reduced error correction overhead forming the basis of its strategy. The immediate evidence will come from its smaller systems, where researchers must demonstrate that the individual components already tested can produce reliable logical operations before the larger 2030 and 2034 targets become attainable.












