CAST AI's recent report reveals significant waste in Kubernetes cloud infrastructure, with organizations using only 13% of the CPUs and 20% of memory on average for clusters with 50 to 1,000 processors. The analysis, which examined over 4,000 clusters across AWS, Google Cloud, and Microsoft Azure during 2023, highlights a trend of overprovisioning resources. While larger clusters (1,000 to 30,000 processors) show slightly better efficiency at 17% CPU usage, the overall waste remains alarming. The study points out that CPU utilization rates are similar between AWS and Azure at 11%, while GCP is slightly better at 17%. Memory usage across the platforms also shows little variation, with GCP at 18%, AWS at 20%, and Azure at 22%. As cloud service costs rise—spot instance prices increased by 23% from 2022 to 2023—this inefficiency becomes more pressing. Laurent Gil, CAST AI’s chief product officer, notes that many organizations lack visibility into their IT resource consumption. Complexities in managing these resources often lead to excessive provisioning, driving up unnecessary spending. Gil advocates for using machine learning and AI to automate the optimization of IT environments. Kubernetes is designed to scale dynamically with application needs, but developers often prioritize availability over cost, leading to further overprovisioning. As the number of Kubernetes clusters grows, IT teams are increasingly pressured to cut costs. Many are adopting platform engineering practices aimed at implementing efficient DevOps methodologies. However, Kubernetes remains challenging to manage, particularly for teams lacking programming skills. Increasing reliance on AI for routine management tasks, including cost control, could help bridge this expertise gap.