Quantinuum updates Reimei and expands RIKEN Quantum-AI – Unite.AI partnership

Quantinuum and RIKEN announced an expanded partnership on November 10, 2026, after completing Reimei’s upgrade to the 56-qubit H2 system model. Plans include joint work toward a fault-tolerant “Made with Japan” quantum computer and a separate SoftBank deal to explore the possibility of bringing hybrid quantum AI services to Japanese AI data centers.

RIKEN celebrated the expansion with a ceremony at the Wako campus. Mitsuhisa Sato, director of the Hybrid Quantum-HPC Platform division, said the updated machine is expected to address larger and more complex scientific problems than the previous H1 system, while a newly signed memorandum of understanding is intended to expand research into quantum computing for artificial intelligence for science.

Hardware procurement was already public: an announcement on April 14, 2026 stated that the H2 had been delivered for assembly to replace the H1. The November development therefore represents the completed update and new partnership commitments, rather than the first disclosure of the H2 shift.

Three systems divide the IT work

Quantinuum’s technical brief describes Reimei as having fully connected high-fidelity qubits and divides the broader platform into three roles. Fugaku deals with large-scale classical modeling and consolidated scientific computing; ROQUO performs tensor network simulations, AI training, and massively parallel computations; Reimei uses select calculations involving quantum states and correlations that become prohibitively expensive to represent classically.

Quantinuum Nexus’ hybrid workflow manager coordinates these resources, allowing jobs on classical infrastructure to send quantum tasks to Reimei and incorporate the results into larger calculations. The arrangement gives the quantum processor a focused role within a larger scientific workflow rather than requiring it to handle an entire problem.

NVIDIA reported in July 2026 that ROQUO uses 540 Blackwell GPUs accessed via the GB200 NVL4 platform and connects to local quantum systems, including Reimei. The researchers were beginning to explore an evolutionary artificial intelligence framework, developed with NVIDIA and integrated with its CUDA-Q platform for quantum-classical computing, to generate Reimei circuits.

First research projects on quantum artificial intelligence

The materials project combines quantum simulation, classical tensor networks and generative artificial intelligence to build phase diagrams, which map how material properties change with temperature, pressure or magnetic field. ROQUO runs simulations of dense tensor networks in classically tractable regions, leaving Reimei to tackle deep temporal evolution circuits where long-range correlations and increasing entanglement increase classical simulation costs.

The team collected Reimei hardware data for a 6×6 two-dimensional spin model, trained a transformer for zero-noise extrapolation across parameter space, and applied active learning to prioritize additional quantum data collection. A generative model trained on the combined classical and quantum dataset is then run on ROQUO to sample conditions that have not yet been simulated.

Quantinuum reports preliminary results that suggest the model can make more accurate predictions than AI trained only on classical simulations and more precise predictions than Reimei alone, while cautioning that the experimental results may not be confirmed on a larger scale or in commercial use. A better understanding of magnetism at the atomic level could inform ultra-high-field magnetic resonance and smart power grids, although these are potential applications rather than implementations.

A second project involving RIKEN, NVIDIA, and Quantinuum is developing a scalable workflow spanning quantum processors, GPUs, and high-performance computing. His initial focus is on zinc porphyrin, a light-absorbing molecule, with an emphasis on the excited electronic states involved in light absorption rather than just ground-state searches.

Fugaku represents the broader molecular environment and the broader chemical workflow; ROQUO handles optimization, configuration interaction calculations, AI-assisted circuit design, and spectral reconstruction; Reimei searches for strongly correlated regions for important electronic configurations. Classical systems then use those candidate configurations to calculate energies and reconstruct experimentally meaningful properties, including the absorption spectrum of the molecule. This division avoids requiring the quantum processor to represent the entire biomolecular environment.

Both projects began before the public launch of Reimei’s services. After completion, Reimei is expected to start regular operations under the Trial User Program.

Commercialization plans beyond RIKEN

The “Made with Japan” research plan combines Quantinuum’s technology with Japanese university research and the manufacturing expertise of domestic component suppliers, drawing on lessons from integrating and managing Reimei. Under SoftBank’s separate agreement, the three organizations will evaluate infrastructure readiness and examine direct connections between quantum computers and artificial intelligence data centers, including whether such connections could support commercial hybrid services outside the research community.

Quantinuum warned that the memorandum and SoftBank deal may not produce definitive agreements, joint development or commercial services. Reimei continues to be part of the JHPC-quantum project funded by Japan’s New Energy and Industrial Technology Development Organization, placing the current work within a supported research program as the partners explore industrial use.

The upgrade establishes the hardware and workflow foundation, but the commercial test remains two-fold: reproducing early research results on a larger scale and transforming SoftBank’s exploration into services that extend beyond RIKEN’s research scope.