The rapid adoption of AI and machine learning workloads, in combination with existing cloud-native applications, demands a network fabric at scale, with release cycles compressed from months to hours.