August 2026 Summaries

Infracost CLI v2.15.0 adds native support for estimating costs from Azure ARM JSON templates without deployments, Azure credentials, or subscriptions, addressing cost changes that can depend on parameter files, conditional resource definitions, loops, regions, and VM-size expressions. The tool evaluates ARM templates using an interpreter designed to handle Azure’s extensive function set, copy loops, conditions, nested templates, custom functions, and parameter-resolution behavior, enabling separate estimates for discovered development and production parameter files. ARM estimates use the same Infracost pricing data, policies, tagging rules, local caching, CI integration, and pull-request cost reporting available for Terraform, CloudFormation, Kubernetes, and other supported infrastructure-as-code formats. Bicep files must first be compiled into ARM JSON, and limitations include no expansion of Template Specs, no support for sovereign clouds or Bicep metadata, and some Azure resource values reported as unknown rather than estimated.

Infracost Fix is a new Infracost Cloud capability that connects to AWS to identify, investigate, and remediate deployed cloud-cost waste, complementing Infracost Dev’s pre-commit cost analysis and Check’s pre-deployment policy enforcement. It uses specialized agents that scan AWS environments daily for idle or unused resources, tag-policy violations, and cost anomalies, validating findings against infrastructure configurations and prior feedback to prioritize high-confidence opportunities rather than generating raw alerts. Agents trace resources to Terraform or other infrastructure-as-code definitions where possible, explain the underlying cause of waste, attach supporting billing and code evidence, and recommend specific remediations with required validation steps. Users can approve agent-created pull requests, address changes through Infracost Dev, or assign work through Jira or GitHub, initially operating in a notification-based approval mode. The product cites a beta customer in which it identified $300,000 in potential savings and plans to add further agents, beginning with Kubernetes rightsizing.

Infracost proposes static cost estimation and label-policy enforcement for Kubernetes manifests, Helm charts, and Kustomize configurations directly from source repositories, editors, pull requests, and AI agent workflows without requiring cluster access, credentials, or billing APIs. It calculates estimated monthly costs from resource requests, replica counts, persistent storage, load balancers, and published OpenCost default rates, aiming to expose the financial implications of configuration changes before deployment; for example, increased CPU, memory, replicas, and storage can compound into substantial costs that are difficult to infer from a YAML diff alone. The tool also applies tagging policies to Kubernetes labels after Helm templating and Kustomize transformations, helping ensure workloads, volumes, and load balancers are labeled for later allocation or governance. Its estimates are intended for comparing changes rather than matching negotiated cloud bills, since they use standard rates and requested rather than actual resource usage. The text notes important limitations, including no runtime rightsizing, showback or chargeback, GPU pricing, or detailed handling of autoscaling, DaemonSets, and job runtimes, while stating that runtime usage comparison capabilities are under development.