The Network Enables AI at Scale: From Training to Agentic Workloads
Agentic AI is increasing the demands on enterprise infrastructure. A single request can trigger multiple model calls, tool interactions, and data retrieval steps, putting greater pressure on front-end network services. Behind those requests, scale-up and scale-out networks help GPUs work together efficiently. The network is no longer a passive layer. It plays a direct role in AI performance, utilization, and cost.
This session examines what AI networking optimization requires in practice: programmable DPUs that accelerate infrastructure services, resilient scale-up connectivity within the rack, and intelligent traffic management across racks. Drawing on AMD networking deployments in demanding training and inference environments, the discussion will connect real performance challenges to the networking capabilities that address them, and explore what changes as enterprises expand their use of agentic AI.
Shane Corban is a Senior Director of Technical Marketing at AMD’s Network Technology Solutions Group (NTSG), where he specializes in AMD’s data processing units (DPUs), AI Networking, and smart switch product lines. Before joining AMD, Shane spent 14 years at Cisco, focusing on product management and engineering for Nexus data center switching products. He also held engineering positions at Brocade and Sun/Oracle prior to his time at Cisco.
