AKAM · Earnings call · 2026Q2T · Full report

Cloud Infrastructure Commitment

AKAMAI TECHNOLOGIES INC · 2026-08-06 · Importance 31 · Surprise 42

Akamai is positioning its distributed platform as a unified grid for AI inference spanning core and edge locations. Management said customers can train models centrally while running latency-sensitive inference near users, robots, vehicles, or factories across Akamai's footprint of more than 700 cities in 130 countries. The company is not targeting training of giant foundation models, instead focusing on model usage and training medium or smaller models on its platform. Management expects the distributed architecture to optimize AI performance, latency, scalability, and cost, including lower egress costs than competitors.

Key facts

• Akamai stated that customers are preordering GPU capacity months in advance, likened to a reserved instance model.