System Interconnect Architectures are Expected to Shift with Future HPC Procurements

Expertise is now a major concern for both HPC and AI centers, outranked only by budget concerns. When asked about barriers to furthering AI capabilities, AI-specific expertise was a significant concern. Popular responses included access to AI expertise (49%), skills in AI model development (47%), and skills in AI programming (36%). This development comes at a time when a third of HPC sites (33%) report a lack of knowledge or skilled staff and identified this issue among their top three barriers to expanding on-premises HPC. This data is from the eighth annual study of Hyperion Research's high-performance computing (HPC) end-user-based tracking of the HPC marketplace. It included 181 HPC end-user sites with 3,830 HPC systems.

Survey respondents cited GPUs as the most anticipated processing element within the next 12-18 months at a rate of 74%. When asked about which processing elements they expect to be incorporated into their HPC/AI/HPDA computing resources, the majority of respondents across all sectors expected GPUs to be first (74.0%) and TPUs (24.3%) as next most anticipated. Government and academia respondents reported the highest expectation at a rate of 84%. This data is from the eighth annual study of Hyperion Research's high-performance computing (HPC) end-user-based tracking of the HPC marketplace. It included 181 HPC end-user sites with 3,830 HPC systems.