Everyone hates massive data centers. This $18 billion CEO has a better way to get you the AI compute you need
TL;DR
Akamai CEO Tom Leighton lays out a contrarian vision for AI infrastructure: instead of ever-larger centralized data centers, he argues that distributed compute sitting closer to users delivers inference faster and more cheaply than central megaclusters. Nvidia is backing the approach. For Akamai, it is an attempt to turn its CDN legacy into an edge-AI story — and a counterpoint to the hyperscalers' data center buildout.
Nauti's Take
Leighton's angle has real potential: inference closer to the user cuts latency and cost, and Akamai already runs the global edge footprint to deliver it. The limit is training — large models still need centralized clusters, and distributed nodes do not change that.
Promising for teams with latency-critical AI features, far less so for anyone shopping for training capacity.
Sources