Toshiba’s “SQBM+” Delivers One of the World’s Largest-Scale Optimization Capabilities, Supporting Up to 1-Billion Variables

- Enhanced quantum-inspired optimization solution supports larger-scale combinatorial problems in finance, logistics, drug discovery, and other fields -

June 25, 2026

Kawasaki, Japan – Toshiba Corporation has announced SQBM+ Version 2.2, the latest advance in its quantum-inspired optimization solver. Available to partners worldwide from today, Version 2.2 takes the scale of addressable optimization problems to up to 1-billion variables, positioning it among the largest and most powerful solvers currently available*1. This expansion delivers advanced support for real-world decision-making on large-scale optimization challenges in fields as diverse as finance, drug discovery and logistics. Its enhanced solution accuracy, and improved operational efficiency through flexible GPU resource allocation further strengthen SQBM+’s overall capabilities.

SQBM+ is a quantum-inspired optimization solution based on the Simulated Bifurcation Machine, a combinatorial optimization solver built on Toshiba’s proprietary Simulated Bifurcation algorithm. Run on classical computers, SQBM+ quickly finds high-quality approximate solutions to complex, large-scale problems*2.

By tackling problems with up to 1-billion variables, SQBM+ Version 2.2 addresses optimization problems closer in scale to those found in real-world applications, expanding the practical use of high-speed combinatorial optimization. Potential applications include large-scale portfolio optimization in financial markets, real-time nationwide logistics optimization, and complex design challenges in drug investigation, such as mRNA vaccine design*3.

SQBM+ Version 2.2 also introduces flexible GPU resource allocation, a function developed in response to user demand. It improves computing efficiency by enabling multiple small-scale requests to run in parallel, while also allowing GPU resources to be concentrated on large-scale requests when needed.

To improve solution accuracy, SQBM+ Version 2.2 incorporates SchemaSBM, an algorithm that identifies patterns in candidate solutions and narrows the search space.

SQBM+ Version 2.2 is available to partners as software modules through Amazon Machine Images and as GPU software for on-premises environments.

Toshiba has applied SQBM+ across a diverse range of fields, including finance, drug discovery, logistics, energy management and materials development. Applications explored to date include financial market trading verification*4, new equity index development*5, computational drug discovery*6, and factory warehouse logistics optimization. Toshiba will continue to expand the scope of application to help address increasingly complex real-world challenges.

Key Features of SQBM+ Version 2.2

  1. Capacity Expanded to Optimization of One Billion Variables
    SQBM+ Version 2.2 increases the scale of quadratic unconstrained binary optimization problems that can be addressed to up to one billion variables, enabling the solution to handle larger problems involving vast numbers of possible combinations.
  2. Flexible GPU Allocation for More Efficient Operation
    SQBM+ Version 2.2 enables automatic backend allocation of GPU resources in response to multiple frontend requests, allowing simultaneous execution and efficient GPU utilization. Users can also specify the number of GPUs assigned to each request, which allows resources to be distributed across multiple small-scale problems or concentrated on a large-scale problem as needed.
  3. Enhanced Solution Quality with SchemaSBM Algorithm
    SchemaSBM identifies patterns in candidate solutions and narrows the search space, helping SQBM+ find higher-accuracy solutions more efficiently.
Figure: Flexible GPU resource allocation through front–end separation

Notes

  • “SQBM+” is a registered trademark or trademark of Toshiba Corporation in Japan and other countries.
  • All other company names and product names mentioned herein may be trademarks or registered trademarks of their respective companies.
  • Information in news releases and topics, including product specifications, availability and related links, is current as of the date of publication and may change without notice.