February 3, 2026

HanabiAI Co., Ltd.

HanabiAI Co., Ltd. attended Supercomputing Japan 2026, held at Tower Hall Funabori from February 2 to 3, 2026. Through this participation, we aimed to develop a structured understanding of how next-generation high-performance computing (HPC) technologies and applications are evolving toward 2030, and to clarify what is required for compute infrastructure to move from being technically feasible to being operationally usable, maintainable, and scalable in real environments. We also held extensive discussions with stakeholders across industry and research, and will incorporate these insights into the requirements definition, product planning, and validation strategy for our HanabiAI Embodied Comput Platform (the “Platform”).

Background

Supercomputing Japan 2026 is a professional event focused on next-generation HPC. Under the theme of “architecture and application outlook toward 2030,” the event explored structural shifts in HPC driven by advances in AI and quantum computing. It also highlighted the growing relevance of HPC to semiconductor technology, industrial competitiveness, and topics such as Sovereign AI, alongside changes in the pace and direction of policy, investment, and development support.

With keynotes, technical sessions, exhibits, and tutorials, the event provided a high-density forum for engineers, researchers, and industry practitioners to exchange views and build a multi-dimensional understanding of trends across architecture, software stacks, application deployment, and operational frameworks.

Key takeaways

During the event, HanabiAI conducted on-site observation and discussions through the lens of “platform capabilities required as HPC and AI convergence enters an engineering and production phase.” Our key takeaways are as follows.

1. Open-source-based stacks remain common, but adoption and operations costs are increasingly visible

In conversations with multiple exhibitors and practitioners, we found that many organizations still build and extend their compute platforms primarily on open-source software stacks. In addition, job submission and resource scheduling often remain centered on command-line tools and scripts. While open-source platforms offer flexibility and customization, real-world adoption frequently requires substantial engineering effort for integration, adaptation, and ongoing enhancements. The learning curve can be steep, and operations often depend heavily on scarce experienced engineering and SRE/operations talent.

As workloads scale, cross-team usage increases, and heterogeneous resources must be coordinated routinely, operational burdens can accumulate quickly—covering continuous maintenance, version upgrades, incident diagnosis, access control and isolation, and reliability assurance. As a result, total cost of ownership can remain high, reinforcing the need for converged computing platforms that provide platformization, visibility, and operational readiness for production environments.

2. Commercial platforms are largely led by overseas vendors

For commercial platforms designed for production use and continuous delivery, we observed that many deployed examples and productized offerings are provided primarily by overseas vendors. At the same time, locally optimized and controllable options tailored to the operational needs of industrial users in Japan can be limited in certain cases. This further reinforced the practical necessity of developing an HanabiAI Embodied Compute Platform that is both engineering-ready and operable in production.

3. We reconfirmed our platform direction and clarified the importance of AI for Science

Across discussions, we reconfirmed the validity of our platform direction: placing unified scheduling and orchestration at the core, and delivering system capabilities that support reproducibility, observability, and scalability for heterogeneous resources and complex workloads.

We also developed a clearer view that AI for Science—AI that accelerates scientific discovery—will be a key pull-through domain for HPC and AI convergence. For the Platform, meeting both HPC and AI computing requirements is a foundational capability to support the growth of AI for Science.

Summary

Through extensive exchanges with a broad set of stakeholders, we further validated that our HanabiAI Embodied Compute Platform is positioned to support general-purpose HPC and AI computing use cases, and we strengthened the key assumptions underlying this direction. We also refined the boundaries of platform requirements, particularly around engineering practicality for adoption and long-term operations.

Based on the latest feedback gathered at the event, HanabiAI plans to officially release the Platform in the near term and will offer free demonstrations and hands-on access to real systems. If you are interested, please contact us in advance to schedule a session.

Company Overview

Company Name:HanabiAI Co., Ltd.

Location:Tokyo, Japan

Business Areas:Centered on the development of robot AI brain technologies, we are building the development infrastructure and ecosystem for embodied AI robots

Website:https://www.hanabiai.jp