While quantum hardware remains small in scale, large-scale simulation is essential for validating algorithms and estimating performance. Fault-tolerant quantum computing (FTQC) in particular requires vastly larger qubit counts and circuit depths, making reliable resource estimates directly relevant to investment decisions. Selected for the ABCI-Q Grand Challenge, QunaSys Inc. ran simulations of circuits with more than 40 qubits using up to 1,024 GPUs. We measured optimal configurations and bottlenecks at different scales, addressing two challenges: building circuits correctly and executing them at scale. How can organizations access large-scale computing resources, and what makes a proposal successful? We share perspectives from both the provider and the user of these resources.

NMR measurements may not fully distinguish between candidates with very similar structures. Many-body echoes (OTOC), which probe the responses of multiple nuclear spins, are being explored as a way to obtain structural information that conventional NMR does not readily provide. As systems grow larger and more complex, however, calculating these responses places a heavy load on classical computers. Building on prior research that estimated interatomic distances and angles in small molecules in liquid crystals, QunaSys Inc. is exploring a method that compares responses calculated using quantum computing with measurements to narrow down candidate structures, with potential applications in the development, analysis and evaluation of materials such as pharmaceuticals and batteries. We discuss the approach of combining measurements and calculations, its current state and outlook, and future validation tasks such as determining the conditions under which it can be applied.

Many teams pursuing materials informatics (MI) encounter the same obstacle: not enough data to train their models. Data-driven approaches are powerful where sufficient data is available, but new materials and conditions often lack it. Reframing phenomena through mathematical models offers a potential way forward, helping teams target experiments before enough data has been collected and reducing detours in development. In this session, leaders who have put this approach into practice discuss what changed, how it changed, and what it took to establish the approach within their organizations. We also introduce the current state and next steps of PhysiLenz, QunaSys Inc.’s AI service supporting this approach.

Computational fluid dynamics (CFD) supports a wide range of industries, including energy, manufacturing and mobility. Yet higher accuracy and larger simulations bring significant computational costs. TOKYO GAS CO.,LTD. and QunaSys Inc. are conducting joint research to address this challenge with quantum computing and published their findings in a paper in June 2026. Based on the lattice Boltzmann method, the research uses Carleman linearization to incorporate nonlinear phenomena essential to fluid dynamics into quantum circuits, followed by implementation and validation. Drawing on their actual R&D, TOKYO GAS CO.,LTD. and QunaSys Inc. discuss why quantum computing is relevant to fluid simulation, the work and findings presented in the paper, and the outlook for industrial applications.

In polymerization, how many polymer chains of each length are produced is a key indicator that shapes a material’s properties. One way to predict it is to describe the reaction process with differential equations based on reaction kinetics. This approach, however, has to distinguish chemical species by chain length and composition, which means solving a very large number of differential equations. In joint research with the Matsumoto Laboratory at the Graduate School of Informatics, Nagoya University, QunaSys Inc. designed a concrete way to embed the reaction-rate matrix that appears in these equations into a quantum circuit (block encoding). Combining it with a QSVT-based solver, we estimated the numbers of qubits and gates required. The work extends from the polymerization of a single monomer to copolymerization, and the results have been published as a paper. The research brings quantum CAE methods to chemistry problems and points to new possibilities for quantum algorithm development across fields. This session presents the paper and our outlook going forward.

Researchers from several companies using PhysiLenz in their R&D join a discussion moderated by QunaSys Inc. about their day-to-day practice. What challenges led them to start using it? Where are they seeing promise? How might mathematical model-based development change their work? Whether you are considering adoption or already experimenting, this session offers perspectives you can relate to your own organization. Together with the earlier session, “Find Answers Before Running More Experiments,” this session traces the journey from the initial adoption decision to integration into everyday R&D.

As product design becomes more advanced, CAE must handle increasingly complex and large-scale computations. Through DQCC, DENSO CORPORATION continues to gather real business challenges across the company and evaluate quantum applications. A key finding is the importance of designing the entire workflow, combining quantum computing with classical computing, HPC and AI. In the previous fiscal year, DENSO CORPORATION and QunaSys Inc. implemented the LCHS quantum algorithm for heat-conduction simulation. By comparing it with classical computation and estimating resource requirements, they quantitatively examined the conditions under which quantum computing could offer an advantage in CAE. The focus is on how to use quantum computing effectively within design workflows. DENSO CORPORATION and QunaSys Inc. explore the current state of quantum CAE and the path toward practical use.

ADAPT-QSCI is a hybrid quantum–classical method that addresses a challenge in QSCI: efficient preparation of the input state. By repeatedly sampling electronic configurations on a quantum computer and performing diagonalization on a classical computer, it combines input-state preparation with ground-state energy calculations. Classical gradient calculations, however, become a new bottleneck. In joint research with JSR Corporation, QunaSys Inc. is accelerating this classical processing on supercomputers in the JHPC-Quantum quantum–HPC environment to enable demonstrations on real hardware. Moving from last year’s algorithm-design stage toward hardware demonstrations, how far can quantum computing and HPC bring NISQ quantum chemistry toward practical use? We discuss the current state and outlook from the front lines.

When applying quantum computing to your own challenges, the first question is where to run it. The speed of the classical processing before and after quantum circuit execution also determines how quickly you can iterate. Constructing and diagonalizing the subspace Hamiltonian in QSCI is a representative example.

This session approaches these challenges from both the infrastructure and software sides. The first half introduces computing environments increasingly available to industry, from the perspective of those building and operating infrastructure that connects quantum computers and supercomputers.

The second half focuses on QunaSys Inc.’s QURI SDK Enterprise. Focusing on execution performance on HPC, we examine both the speed achieved in comparisons with other implementations and the practical considerations for using the software in operational computing environments.

The session brings these two perspectives together: fast software and the infrastructure to run it, offering a view of the present and future of quantum–HPC integration.

A theoretical quantum speedup does not automatically translate into practical value for industry. SoftBank Corp. and QunaSys Inc. have quantified the logical qubit counts, gate counts, error rates and execution times required for practical-scale quantum chemistry and CAE, revealing the gap between current quantum hardware and the computational resources industrial applications need. For the use cases studied, the estimates include 100–1,000 logical qubits for quantum chemistry and minimum requirements of 15–30 logical qubits for CAE. In March 2026, these assessments informed a white paper presenting “10 ideas” to accelerate practical quantum computing. Those ideas are now moving into concrete validation and implementation. This session introduces what the numbers tell us about the current state of quantum computing and the work beginning to bring practical applications closer. How can we actively accelerate the path to practical quantum computing? We share the work now underway.

In analyzing interfaces and reactions, high-accuracy first-principles calculations (DFT) alone cannot always cover every condition you want to examine. Machine-learning interatomic potentials (MLIPs) use machine-learning models trained on the relationship between atomic configurations and their energies and forces, enabling rapid exploration of these conditions. QunaSys Inc. is evaluating how fine-tuning existing MLIPs on a small dataset can adapt them to specific applications. In a study of an adhesive interface between copper and resin, fine-tuning on 30 structures improved accuracy even for structures at angles not included in the training, while keeping the computation time at about one minute. Alongside practical approaches to using MLIPs, we will present our concept for incorporating high-accuracy reference data from quantum computers where accuracy tends to fall short, such as in reactions and dissociation.

Quantum hardware companies are publishing roadmaps for FTQC devices, but it remains unclear which applications will become feasible at each stage, at what scale, and with what practical value. Unlocking the potential of FTQC requires evaluating application performance in detail at the architecture level and optimizing both algorithms and architectures. This talk introduces QunaSys Inc.’s framework for application-driven FTQC research and its resource-estimation efforts. Through research on FTQC architectures designed to improve application performance, we also present an approach that connects applications, FTQC architectures and quantum hardware from end to end.