India, May 6 -- The main change is the sharp increase in network density. As AI workloads become more distributed, the amount of data moving between points of presence and data centres rises significantly.

Today, networks are already moving from 800G deployments towards 1.6 terabits, 3.2 terabits and potentially higher in the future. That directly affects fibre requirements. In regional and metro networks, fibre counts have moved from 288 fibres to 864, 1,728, 3,456 and, in some cases, even beyond 13,000 fibres, depending on the architecture.

The current phase of AI adoption is therefore defined by much higher fibre density combined with relatively long-distance connectivity.

A major area of activity today is data centre interconnect, ...