It was fantastic to be in Osaka, Japan, this past week for SCA/HPC Asia 2026. This year was particularly special as it marked the joint hosting of SupercomputingAsia (SCA) and the International Conference on High Performance Computing in the Asia-Pacific Region (HPC Asia). IBM Quantum and Storage helped make this event a success with our Platinum sponsorship.
The energy at the Osaka International Convention Center was high, with the conference theme "Everything with HPC – AI, Cloud, QC and the Future Society" setting the tone. Between the deep technical dives and the incredible Japanese hospitality, it was a memorable week. Here’s a look at what I got up to.
There was over 2000 attendees from over 40 countries with the largest exhibition floor holding over 90 booths. Complimenting this were 6 tutorials, 13 workshops and 19 Birds of a Feather (BOF)s.
Let’s start first with the IBM Storage Scale User Group
One of the main highlights for me was the IBM Storage Scale User Group. It’s always great to connect with the community in person. We had some excellent sessions diving into the latest updates and real-world use cases.
Held on Wednesday, January 28th, right in the heart of the Osaka International Convention Center (Grand Cube), this session was a reminder of why this community is so vital. It wasn't just about product roadmaps; it was about the people building the systems that drive discovery.
This highlights exactly what the User Group is about: it doesn't belong to IBM or any single vendor; it belongs to the users who manage petabytes of data daily. The goal is to develop a worldwide User Group Council that represents our needs and our challenges. It was a call to action for everyone in the room to get involved and shape the future of the platform.
The event kicked off with a deep dive into the IBM roadmap. The focus was clear: Eliminating data silos. With the introduction of the Storage Scale System 6000, IBM is tripling capacity and pushing the boundaries of what a global namespace can do. We discussed the new Data Acceleration Tier (DAT), which uses intelligent data placement to accelerate AI pipelines, delivering up to 28M IOPS for the most demanding training and inference tasks.
The core of the afternoon was dedicated to the technical breakthroughs that make Storage Scale the gold standard for AI and HPC:
- S3 Vectors and KV Cache: @Khanh Ngo from our IBM Storage CTO Office presented how Scale IBM Storage Scale is operating in in two pillars S3 Vectors on IBM Storage Scale and (2) KV Cache tiering.
- S3 over RDMA: @Anandhu Karattuparambil, from Scale CES-S3 development, explored how high-performance object storage is evolving. By using Remote Direct Memory Access (RDMA), IBM is bypassing traditional bottlenecks, allowing S3 data transfers to achieve the low latency and high throughput required for date-intensive workloads. This has already been possible with POSIX and protocols like GPU Direct Storage.
- AI Data Engine: Dr. @Frank Lee, Distinguished Engineer and Architect, again demonstrated where intelligence meets infrastructure with Content Aware Storage and Data Cataloging evolving into easier to use environments especially with the newly released Scale Model Context Protocol (MCP) component. A good walkthrough is available here.
- Multitenancy & Tiering: @Mathias Dietz, our Scale RAS Architect and STSM presented, how modern enterprises need to support multiple teams on a single platform without sacrificing security or performance. We looked at new strategies for granular multitenancy without slowing down the mission with the appropriate quality of service (QOS).
One of the highlights of the User Group was hearing from the users and ecosystem that makes this innovation possible.
- The University of Queensland shared their latest updates in how the IBM Storage Scale System 6000 accelerates their workload with TLC and QLC flash technology.
- NVIDIA joined to discuss the collaboration with IBM to show an example of Storage Scale becoming a backbone of NVIDIA-certified AI factories.
- Western Digital provided insights into the hardware evolution, specifically how high-density disaggregated flash can easily work with a software defined environment like IBM Storage Scale and bring performance with a look forward to tiering easily.
- Tuxera demonstrated how they can enhance the default IBM Storage Scale common protocols like SMB and NFS for highly performant and reliable in complex, hybrid environments.
If you weren't able to make it to Osaka, I highly recommend checking out the upcoming events on the Spectrum Scale User Group website! Here’s a view of where we will be this year!
- March 9 - 11 | Germany
- April 21 - 22 | Austin, US
- May 18 - 20 | London User Group
- June 22 | Germany Hamburg - ISC
- July 20 - 22 | San Jose, US - West Coast SVL
- September 14 - 15 | NYC Scale Day
- October 26 - 30 | Australia Melbourne - eResearch
- November 15 | Chicago, US - SC
Hope to see you at the next one!
Next, let’s talk about the Data Storage & Filesystems BoF
I also participated in the Birds of a Feather (BoF) session focused on Data Storage and Filesystems. These sessions are often where the "real" conversations happen, unscripted and direct.
That session, titled "Managing and sharing large scientific data sets," was the perfect chaser to the User Group. It reinforced the theme that while compute (GPUs) often gets the glory, data management is the gravity that holds everything together. We debated the maturity of current data management strategies and how we can better handle the explosion of instrument-generated data.
We discussed data management maturity and how the explosion of instrument and computationally generated data is forcing us to rethink our storage architectures. It’s not just about feeding the GPUs anymore; it’s about how we manage, move, and store that data lifecycle efficiently.
Looking Ahead
SCA/HPC Asia 2026 was a massive success. The convergence of AI and traditional HPC is accelerating, and the storage layer is right at the center of that evolution. It was great to compare notes with colleagues from across the Asia-Pacific region.
Leaving Osaka, I felt energized. The technology is moving fast—especially with the integration of AI capabilities into the storage layer itself—but the community is keeping pace.
Looking forward to the next SC ASIA and HPC ASIA events in 2027!