This episode is available to stream on-demand.
Today’s network infrastructure is under pressure to deliver faster, more reliable, and more scalable connectivity. This session looks at how new network technologies and architectures are being deployed to meet the demands of large-scale training and real-time inference. It also explores how connectivity strategies are evolving to support both centralized and edge deployments. Key themes addressed:
- Meeting bandwidth and latency requirements for AI training and inference
- Upgrading to faster interconnects like 400G and 800G Ethernet to handle growing data volumes
- Using smart networking components to offload data tasks and reduce strain on CPUs and GPUs
- Strengthening edge-to-core connections to support distributed and real-time AI applications
