Amazon Web Services is directing its engineering teams to cut compute resource waste as demand strains server capacity. Underutilized EC2 instances reportedly account for roughly 65% of all running instances. Procurement timelines for new CPU capacity have stretched from hours to multi-day waits. Microsoft Azure and Google Cloud have both acknowledged capacity constraints during recent earnings calls. AWS has already raised GPU reserved capacity pricing by approximately 15% in early 2026.
Source: Read the original article
