AI strategies are advancing quickly, but many data center fabrics were never designed for GPU dense, latency sensitive workloads. As organizations move from pilot clusters to production AI environments, network architecture is increasingly becoming the limiting factor.

This interactive roundtable will examine how to design and evolve a modern data center fabric capable of supporting high performance AI workloads across training and inference. Attendees will leave with a clear understanding of where bottlenecks typically emerge, how to modernize existing environments without disruption, and concrete next steps to prepare their infrastructure for AI at scale.

BWG Connect, Insight Enterprises & Cisco invite you to join an interactive event with industry peers to review & discuss practical insights across the digital space.

As always, there will be no sales pitch and all attendees are welcome to listen and learn.

• Many AI initiatives stall not at compute, but at the network fabric.
  • Where traditional spine leaf architectures struggle under GPU east west traffic
  • How modern fabric designs, including Cisco HyperFabric, are reducing congestion, improving job completion times, and increasing end to end throughput
• Moving AI from pilot clusters to production environments introduces operational risk.
  • Strategies for phased fabric evolution without disrupting mission critical workloads
  • How organizations are integrating architectures such as HyperFabric into existing Cisco environments while maintaining stability
• AI workloads create unpredictable traffic patterns across training, inference, and mixed environments.
  • Performance tuning techniques that reduce latency variability and improve consistency
  • Observability best practices that expose bottlenecks before they impact model performance
• Right sizing infrastructure for AI requires precision, not overprovisioning.
  • How leaders are balancing cost, performance, and scalability across GPU clusters
  • Where high performance networking architectures can eliminate hidden fabric constraints
• An AI optimized data center starts with an honest assessment.
  • Key indicators that your current fabric is not AI ready
  • A practical readiness checklist to evaluate fabric maturity and identify immediate next steps