SC26 SovAISCC

As the HPC landscape shifts toward massive-scale AI infrastructures, the pursuit of a sovereign supercomputing ecosystem has become paramount. However, recent industry consolidations—most notably the acquisition of SchedMD (Slurm) by NVIDIA—have shattered the illusion that open-source software (OSS) is a guaranteed failsafe against vendor influence. While OSS and open standards remain the preferred foundation for reducing total cost of ownership (TCO) and avoiding traditional lock-in, the "open-core" and "corporate-stewardship" models introduce new risks of strategic divergence.

This workshop will explore the paradox of modern HPC sovereignty: how can sites build autonomous, transparent infrastructures when the most critical "open" components are increasingly controlled by hardware vendors? We will evaluate the current state of open-source tools within supercomputing environments and discuss whether true sovereignty requires a transition from passive consumption toward active community governance or strategic forking. Initiatives such as OpenCHAMI and the recently established High Performance Software Foundation (HPSF) represent a shift toward a "foundation-first" approach to software longevity. However, the persistent tension between community-led governance and the necessity of official vendor support remains a pivotal variable for large-scale production sites. Ultimately, this workshop seeks to move beyond an idealized view of open-source software toward a pragmatic, risk-aware strategy for the next generation of sovereign on-premises infrastructure.

Included within the scope are technical and governance aspects encompassing:

Open source and open standards for AI factories, supercomputing, and cloud operations.
Sovereign AI/HPC motivating use cases and state-of-the-practice.
Resource management and scheduling systems for AI, HPC, and cloud technologies.
Vendor alternatives for software-defined infrastructure (SDI).
Security benefits and limitations of using a sovereign stack.
Strategies for "sovereignty-by-design" to protect against license shifts.

Workshop Goals:

Community Building: Bridge the gap between interdisciplinary communities (HPC centers, cloud architects, and AI infrastructure providers) to foster a sovereign ecosystem.
Ideation & Strategy: Move beyond technical presentations to strategic discussions on governance, specifically addressing the "single point of contact" support model vs. community-led stacks.
Socialization of Early Works: Provide a platform for emerging open-governance projects (like OpenCHAMI) to present roadmaps and gather requirements from production sites.
Professional Development: Equip attendees with the knowledge to negotiate SLAs that permit deep-stack sovereignty without assuming full hardware-firmware incompatibility risks.

Important Dates

Submission deadline: Early September, 2026
Notification of Acceptance: Late September, 2026
Program Published: November, 2026

Workshop Committees

Organizing Committee

Organizing Committee (supercompcloud@googlegroups.com)

- David Y. Hancock, Indiana University
- Winona G. Snapp-Childs, Indiana University
- François Tessier, Inria
- Sadaf Alam, University of Bristol
- Maxime Martinasso, Swiss National Supercomputing Centre
- Alex Lovell-Troy, Los Alamos National Laboratory

Committee members are currently being solicited for future events. If you would like to participate in the committee please contact the workshop organizers at supercompcloud@googlegroups.com.

Google Sites

Report abuse