Keep historical billing data intact in Cost Explorer

AWS Cost Explorer now preserves historical billing data, even for accounts mapped into billing groups through Billing Conductor and Billing Transfer. AWS says the data remains priced at original billable rates.

That helps restore reporting continuity for customers who centralize billing. If your organization relies on grouped billing views, this is valuable because it keeps older data available in a way that still lines up with the original charges.

Make AWS Marketplace professional services cheaper to sell and buy

AWS Marketplace lowered the listing fee for professional services private offers from 2.5% to 0.5%. That’s a direct reduction in transaction cost for partners.

For buyers, this could also help lower procurement costs for services transacted through Marketplace. It’s a straightforward commercial change, but one that may improve how services are packaged and purchased.

Make CloudWatch troubleshooting faster without leaving the console

CloudWatch Application Signals just got a useful upgrade, and it’s aimed at helping teams triage issues more quickly. AWS added service health ranking on the application map, plus infrastructure, logs, and traces tabs in the service overview.

That means you can keep more of your debugging workflow inside CloudWatch instead of bouncing between tools. For FinOps teams, that can help cut down time spent on investigation and shorten the path to resolution, which can also reduce operational cost.

Query and visualize AWS metrics in one place

CloudWatch Query Studio is now generally available, giving teams a single workspace to query and visualize OpenTelemetry and AWS metrics. It supports PromQL and Metrics Insights.

This is handy when you want to compare performance data across accounts or regions without stitching together separate views. It can make cost and reliability analysis feel much more connected, especially when you’re tracking how infrastructure behavior affects spend.

Centralize metrics across accounts and regions more easily

CloudWatch Metrics Centralization is now generally available, and it automatically replicates metrics across accounts and regions into one destination account. AWS says this supports unified querying, alarming, and governance.

It also improves consistency for governance and alerting. If you manage many accounts, having a central view can simplify both oversight and troubleshooting.

Cut log analysis overhead with a single CloudWatch console

AWS has launched CloudWatch Log Analytics, a unified console that combines Logs Insights, Live Tail, and Contributor Insights. The goal is to make querying and visualizing logs simpler.

The experience is designed to reduce friction in day-to-day log work. In practice, that means less jumping around between tools and a clearer path from issue detection to root cause.

Native OpenTelemetry metrics now work directly in CloudWatch

CloudWatch now natively ingests OpenTelemetry metrics through OTLP, and it also supports PromQL queries. AWS says pricing is per GB ingested, with 15 months of included storage.

This is useful if you want to consolidate custom application metrics and AWS vended metrics in one place. For FinOps teams, that can make observability spending more transparent while simplifying how metrics are collected and queried.

The update also makes it easier to manage mixed telemetry sources without extra tooling. That can help reduce operational complexity while keeping costed observability in one service.

Expand object storage with rich metadata attached to files

Amazon S3 Annotations are now available, letting you attach large JSON, XML, or YAML metadata directly to objects. AWS says each object can hold up to 1 GB of this metadata, surfaced through S3 Metadata and Iceberg tables.

Save big on similarity search for large vector indexes

AWS has cut Amazon S3 Vectors query charges by up to 80% for vector indexes with more than 10 million vectors. The change applies to data-processed charges and doesn’t require application changes.

There is one cost note to keep in mind: the amount of data returned can affect fees, with the first 512 KB free. So while the price cut is significant, FinOps teams should still keep an eye on result sizes.

Return more results per vector search, but watch the data-returned cost

Amazon S3 Vectors now supports up to 10,000 similarity results per query, which is a big jump from previous limits. AWS says this is useful for multi-stage retrieval pipelines.

At the same time, AWS notes this can affect data-returned fees. That makes it a practical update for FinOps teams to track closely, since bigger result sets can change the economics of each query.

Give MySQL 5.7 more time on Aurora and RDS

AWS extended security patch support for MySQL 5.7 on Aurora and RDS through June 30, 2029. AWS also says there’s no price increase, and Year 3 pricing stays the same.

Cancel future-dated EC2 capacity reservations more flexibly

EC2 Future-dated Capacity Reservations now support cancellation, though charges may apply depending on when you cancel. AWS also says cancellation details will appear in the Cost and Usage Report.

That’s helpful if your planned capacity changes before the reservation starts. It gives teams more flexibility when forecasting demand and reduces the risk of being locked into a reservation that no longer fits.

Remove one big variable from GameLift server costs

Amazon GameLift Servers now includes network bandwidth at no extra charge for all generation 6+ instance types, excluding China. AWS says you now pay only for instance hours.

That removes a major unpredictable cost component for game studios. For teams trying to forecast hosting spend, this makes GameLift pricing easier to understand and manage.

The change is especially valuable when bandwidth use is hard to predict. It can help reduce surprises and make budgeting more stable.

Boost SQL Server memory while cutting license costs

Amazon RDS for SQL Server now offers X2m memory-optimized instances, with up to 4 TB of memory. AWS says these instances include RDS Optimize CPU, which can reduce SQL Server licensing costs by about 50% versus prior x2iedn instances.

This is a strong fit for memory-heavy workloads where licensing is a major cost driver. For FinOps teams, the licensing reduction can be just as important as the performance gain.

Improve price-performance for PostgreSQL, MySQL, and MariaDB workloads

AWS has launched Graviton5-based M9g DB instances for Amazon RDS in GA. AWS says they can deliver up to 23% better price-performance versus Graviton4 equivalents.

This is a practical upgrade path for supported database workloads. Better price-performance can translate into lower database TCO without changing the database engine family.