Continuum of Compute: Akamai on What Enterprise AI Needs at the Edge
IoT News, Friday, May 15th, 2026
Akamai's Field CTO discusses how enterprises must architect distributed AI inference across edge locations based on workload demands rather than cloud-vs-edge preferences.
Akamai's Joseph Glover explains that enterprise AI is moving beyond experimentation into production deployment, requiring careful consideration of where inference should run based on business needs rather than simply choosing cloud or edge.
Unlike training, which belongs in centralized GPU clusters, inference must be distributed across locations where it delivers the most value-whether in retail, factories, or media workflows. This requires treating AI inference as a distributed systems problem encompassing routing, observability, resilience, security, and cost control.
Glover emphasizes that successful organizations design inference as part of application architecture, considering factors like latency, data sovereignty, and operational consequences unique to each deployment location.