Learn about the critical role of networks in AI training clusters through this technical presentation that explores model partitioning techniques, GPU traffic patterns, and optimal network design principles for AI applications. Delve into detailed explanations of how networks support AI model training, various approaches to distributing AI models across systems, and the specific networking requirements for efficient AI operations. Discover the latest developments in UEC (Ultra-Ethernet Consortium) and their initiatives to enhance Ethernet capabilities specifically for AI networking applications. Presented by Sharada Yeluri, this 42-minute talk provides valuable insights for networking professionals and AI practitioners looking to understand the intersection of networking infrastructure and artificial intelligence systems.