An edge site is useful for inference when it reduces latency to the users or systems that matter, connects efficiently to core capacity, and can be operated consistently with the rest of your footprint.
Latency to the right place
Define latency targets to specific user populations or data sources, then evaluate sites against those targets rather than distance alone.
Local connectivity
Local peering, diverse metro fiber, and backhaul to core sites matter as much as the building.
Consistency across sites
Distributed inference multiplies the number of sites you operate. Common access, ticketing, and change procedures keep operations manageable.