The SC26 supercomputing conference wrapped up its run at McCormick Place in Chicago this week, the event's first time in the city and, organizers said, its largest yet with more than 20,000 attendees. Argonne National Laboratory researchers were heavily represented throughout, presenting work spanning fusion energy modeling, cancer research and what the conference's own preview materials described as the accelerating convergence of artificial intelligence with traditional high-performance computing.
Kevin Hayden, a senior network engineer at Argonne, served as the conference's general chair, and Argonne-affiliated researchers ran hands-on tutorials through the Department of Energy's booth and the conference's Art of HPC exhibition, alongside a Regional Tech Hub showcasing Illinois institutions' work on quantum and next-generation computing. The conference's theme, "HPC Unites," reflected a broader shift previewed by Data Center Frontier ahead of the event, which described AI becoming central to scientific computing rather than remaining a bolt-on addition to conventional supercomputing workloads.
The scale of this year's gathering underscores how thoroughly AI workloads have reshaped the supercomputing field's own priorities, with machine-learning training and inference now accounting for a growing share of the massive computing clusters that national laboratories and universities operate for traditional scientific simulation. Organizers have not yet announced the location for next year's conference.