Grant Shirk, Cisco | Networking for AI Summit
summary
Cisco argues campus and branch networks must scale for agentic AI not just datacenters, citing an on‑prem pharma workload that already outstrips a newly upgraded 400G core and a restaurant deploying small language models at the drive‑through for transcription and quality. Requirements include microsecond‑level latency, identity/MAC explosion control, end‑to‑end assurance and fabric‑based architectures. Recommended steps: consolidate management and assurance onto a single platform, plan for continuous high inference load (don’t bet against bandwidth), and adopt multiplayer workflows with human‑in‑the‑loop trust building.
AI-written summary · about
chapters (7)
- 0:00Intro
- 2:29Real World Examples of AI in Network Environments
- 5:31Integrating AI and Networking: Adapting to Dynamic Workloads and Evolving Requirements
- 10:45Building Trust and Speed in AI-Driven Networks
- 14:31Architectural Considerations for Future-Ready Networks
- 17:00Assessing Network Needs for AI Applications
- 21:13Future-Ready Network Strategies: Best Practices and Key Takeaways
