Kisaco Research’s AI Infra Summit (formerly the AI Hardware & Edge AI Summit) has grown into one of the largest vendor-neutral AI infrastructure conferences in the United States. Three days at the San Jose McEnery Convention Center cover the full stack from accelerator silicon through systems, networking, and the data center facilities that host them, with several thousand attendees and a large exhibition.
The data center track has expanded each year as the facility became the bottleneck. Sessions from hyperscaler and neocloud operators on building and powering training clusters, on liquid-cooling deployment at scale, and on the networking fabrics that determine how large a single cluster can usefully be are the ones to plan around. Because the event is independent of any single silicon vendor, it is also one of the better places to compare the thermal and power requirements of competing accelerator platforms side by side.
Produced by: Kisaco Research
Format: Multi-track conference plus exhibition
Focus: AI accelerators, systems, networking, inference infrastructure, AI data center facilities
Relevance to data center engineering: Cross-vendor accelerator power and thermal comparison, cluster-scale cooling and power deployment, AI networking facility requirements










