You can now ask questions directly within our broadcasts using our new AskAI feature.

AI performance is shaped by how compute, storage, networking, memory, and software work together, not by hardware alone. As systems scale, infrastructure bottlenecks and software efficiency increasingly limit performance. With training and inference drive different patterns of resource usage, software-defined management is critical to maximize utilization. The question of where AI performs best, whether on-prem, in specialized environments, or in the cloud, now depends on how effectively the stack is orchestrated to eliminate friction and maximize throughput. Key discussion points include:

• How software-defined management improves efficiency across AI workloads
• How training and inference differ in resource usage
• Where bottlenecks arise across compute, storage, networking, and memory
• How on-prem and cloud environments affect AI performance at scale