Case Study: Adobe cuts EKS compute costs 38% with ScaleOps

Adobe cuts Amazon EKS compute costs 38% with ScaleOps

Adobe, a major technology company with over 30,000 employees, faced a significant challenge managing resources across its more than 500 Amazon EKS clusters, which powered its AI development with over one million CPUs and thousands of GPUs. Developers were over-provisioning resources to ensure performance, leading to low utilization, waste, and high costs. Adobe needed a solution to automate resource management, eliminate waste, and accelerate AI development without compromising on performance.

The company implemented the ScaleOps Autonomous Cloud and AI Infrastructure Resource Management platform. ScaleOps provided continuous automatic optimization by rightsizing pods, intelligently bin packing nodes with Karpenter, and increasing GPU utilization. This solution automated resource management for 240 clusters during a proof of concept, which projected annual savings of $7 million. With full deployment, Adobe expects to save $30 million per year, representing a 38% reduction in compute costs, while freeing engineering teams to focus on higher-value work.

Adobe