See your cloud cost efficiency in one place

AWS Billing & Cost Management Dashboards now include a Cost Efficiency widget, which gives FinOps teams a quicker way to track optimization progress.

The widget shows efficiency scores over time with account and Region detail, and it links straight to the Cost Optimization Hub. That makes it easier to review how well your cloud spend is improving alongside other financial tracking.

Send email with simpler pricing choices

Amazon SES has introduced pricing plans, giving customers packaged options instead of only à-la-carte pricing.

The plans are Essentials, Pro, and Enterprise, and they bundle deliverability features with discounts. That simplifies procurement and may reduce total cost for high-volume senders.

Better cost visibility for your AI coding tools

CloudWatch just added Coding Agent Insights, so teams can track how coding agents are being used and what they’re costing.

This update ingests OpenTelemetry metrics from tools like Claude Code, Copilot, and Codex, then ties agent usage to delivery metrics and token spend. That gives engineering and FinOps teams a clearer way to measure whether AI coding tools are actually delivering value.

Get more room to standardize governance

AWS Organizations increased the resource control policy quota to 2,000, doubling the previous limit.

This gives large multi-account organizations more headroom for centralized controls. It should be especially useful for teams that want finer-grained guardrails across many accounts.

Get better reporting for Bedrock spend

AWS Data Exports now includes standardized Amazon Bedrock product metadata, which should make cost attribution much easier.

The export now includes fields like model provider, model name, inference type, pricing unit, and feature. That gives teams a more consistent way to identify and categorize Bedrock charges in cost reports.

Make S3 cold-data moves happen sooner

S3 removed the 30-day minimum for transitions to Standard-IA and One Zone-IA, which makes automated tiering more flexible.

Lifecycle transitions can now happen with a 0-day minimum instead of waiting 30 days. In practice, that means data that cools off quickly can be moved into lower-cost storage much sooner.

One-click migration for OpenSearch dashboards

Amazon OpenSearch Service now supports one-click migration from legacy OpenSearch Dashboards to OpenSearch UI, which should reduce upgrade effort.

The migration works across domains and serverless collections. That means teams can move off the older dashboard experience with less manual work and less operational friction.

Scale Prometheus much farther without custom work

Amazon Managed Service for Prometheus now supports up to 1.5 billion active metric series per workspace, which is a big step up for large monitoring environments.

AWS also raised the limit to 200k rules per workspace and allows many workspaces per account. That makes it easier to consolidate observability at scale instead of stitching together multiple custom monitoring systems.

Use new instance sizes for more flexible Redshift right-sizing

Amazon Redshift added rg.large and rg.12xlarge instance sizes, giving teams more options for analytical clusters.

These new sizes use Graviton CPUs and are positioned for better price-performance. That makes it easier to match cluster size more closely to workload needs.

Expand memory-heavy workloads with more EC2 coverage

Amazon EC2 R6in and R6idn instances are now available in more Regions, which broadens options for memory-optimized deployments.

These instances are built for memory-heavy and network-intensive workloads. With more Regional availability, teams have a better shot at choosing the right mix of performance and location.

Reduce storage costs with better log analysis queries

CloudWatch Logs Insights now includes 25 new query commands and functions, expanding what you can do directly in queries.

The additions cover statistical functions, time handling, null management, and lookups. That means more log analysis can happen in place instead of being exported for offline processing.