0
Deep & Emerging Tech·August 19, 2026·1 min read

Relativity Networks raises $22 million to bring a faster kind of fiber to data centers

Share

A 30% latency reduction at the fiber layer is not a rounding error for cross-cluster training and HFT-style workloads—it changes how you think about where to place latency-sensitive services. If you’re architecting AI infra, start modeling what hollow-core links between key racks or sites would do to your parallelism and synchronization costs.