Fujitsu tests quantum stack on neutral-atom hardware

Fujitsu and Japanese quantum startup Yaqumo have begun testing Fujitsu’s STAR computing architecture and open-source quantum operations software on physical neutral-atom hardware, moving a collaboration that began with theoretical studies in April onto an actual machine from August.

The work tests whether architecture and software developed primarily around superconducting quantum systems can be adapted to another hardware modality. Yaqumo is developing computers based on neutral ytterbium atoms, while Fujitsu is providing its STAR architecture and Open Quantum Toolchain for OPerators and USers.

Neutral-atom machines use laser-cooled atoms held in optical traps as qubits. Their physical connectivity, control commands, operating procedures, and error behaviour differ from superconducting circuits, so architecture and software designed around one platform cannot simply be assumed to operate unchanged on another.

Fujitsu identifies qubit connectivity as a particular reason for testing the approach. Neutral-atom systems can support flexible interactions between qubits, potentially reducing some of the routing restrictions found in architectures where physical connections are limited to fixed neighbouring devices.

That characteristic could influence quantum error correction, although the practical benefit depends on the operation fidelity and control behaviour achieved by the physical hardware. Fujitsu is therefore moving beyond theoretical compatibility and testing the STAR architecture against the constraints of an actual neutral-atom system.

STAR — Space-Time efficient Analog Rotation quantum computing architecture — was developed for the Early-FTQC stage, where error-corrected computing remains constrained by the number of physical qubits available. Fujitsu says the architecture can reduce the qubit requirement for arbitrary-angle phase rotations, but also acknowledges that its effectiveness varies considerably between hardware types.

Testing it on neutral atoms should expose where those assumptions hold and where they need to change. Hardware connectivity, operation timing, physical error rates, and the available gate set can all alter the resources needed to execute the same algorithm.

The software work presents a parallel integration task. Fujitsu’s Open Quantum Toolchain covers functions required to make quantum machines available as remotely operated computing resources, from execution environments through job management and device operation.

Commands sent to neutral-atom hardware differ from those used with superconducting machines, requiring the companies to adapt command conversion, job management, device-status monitoring, and calibration functions. The aim is to retain a unified external interface while isolating more of the hardware-specific control below it.

That sort of abstraction is standard in conventional computing, but quantum platforms remain sufficiently different that a common interface cannot make the physics disappear. Software still has to understand which operations a machine supports, how they are scheduled, what error behaviour is present, and which calibration state applies when a job is executed.

Yaqumo was established in 2025 and is developing its hardware around ytterbium neutral atoms. The company aims to produce a working system with more than several hundred qubits and quantum-error-correction capability by fiscal 2027. That remains a development target rather than demonstrated system capacity.

The physical trials can therefore answer a narrower and more immediate question than whether neutral atoms will become the dominant quantum-computing modality. If STAR can exploit the connectivity available in Yaqumo hardware, it may reduce some architectural overhead; if the assumptions do not transfer, the design can be revised before larger machines are built around them.

The same is true for cloud operation. A useful quantum system needs continuous calibration, scheduling, status monitoring, fault handling, and translation from user workloads into hardware-level instructions. Making those functions work across different physical platforms is necessary if quantum computers are eventually to behave as shared computing infrastructure rather than individually operated laboratory systems.

Fujitsu continues to work across several quantum hardware approaches, including superconducting and diamond-spin systems. The Yaqumo programme adds neutral atoms to the environments against which its architecture and operations stack can be tested, providing physical evidence about what can be made portable and what remains tied to the underlying qubit technology.