Connect | Virginia // Debate: Are storage and networking AI densification's hidden bottlenecks?

Storage and networking are often considered “AI’s underserved core,” frequently overlooked despite being critical to high-performance workloads. Because storing data consumes far less power than processing it, storage is often deprioritised in the broader conversation. Similarly, networking is sometimes overshadowed by compute, even though it is a driver of extreme rack densification.

This panel will explore how to scale storage and networking sensibly while meeting the demands of AI workloads, by addressing:

• Managing latency, bandwidth, and rack density for high-performance compute clusters
• Aligning storage to the AI pipeline
• Can optical connectivity be a solution?