Building Sustainable AI Data Center Networks

Ethernet growth projections are surpassing all expectations, driven by the extreme demands AI workloads are placing on modern data center networks. In particular, 800G adoption and early deployments of 1.6T are moving forward at an unprecedented pace to meet the urgent need for a high-performance network fabric essential for AI training and inference. Facing rapidly increasing AI traffic loads, providers are, in many cases, massively overbuilding their data center Ethernet networks. While overbuilding is a pragmatic approach today, is it sustainable and for how long?

Hear data center networking experts Sameh Boujelbene, Vice President for Dell'Oro Group, and Aniket Khosla, Vice President of Wireline Product Management for Spirent, as they discuss the question of sustainability when it comes to building of AI data centers, explain the drivers behind the adoption of high-speed Ethernet in data centers, and examine why network efficiency will be the key to monetizing investments in HSE infrastructures.