Every year after re:Invent, FinOps practitioners sift through hundreds of AWS announcements for the ones that affect their cloud costs. We pulled the top 10 highest-impact announcements to summarize.

1. AWS Database Savings Plans

As AWS has nudged people away from Reserved Instances, Database Savings Plans are a long-awaited addition for a more flexible alternative to Reserved Instances that can save up to 35% on RDS, Aurora, ElastiCache, MemoryDB, and Neptune. Unlike Reserved Instances, which lock you into a specific database engine and instance family, Database Savings Plans provide flexibility to change between database services and instance types while maintaining your discount. This makes them ideal for teams with evolving database architectures or those consolidating their commitment strategy across multiple database services. Full writeup here: https://www.vantage.sh/blog/aws-savings-plans-databases

2. Graviton 5

The newest generation of Graviton processors, powered by Amazon, is here for ARM workloads. AWS claims Graviton5 delivers the best price-performance for a broad range of workloads in EC2, offering steep savings compared to Intel and AMD processors in the same generation with comparable performance. For teams already running ARM-compatible workloads or willing to migrate, Graviton5 represents a significant opportunity to reduce compute costs without sacrificing performance.

3. AWS Lambda Managed Instances

Another highly sought out feature was AWS Lambda Managed Instances. You can now run functions on EC2 compute while AWS handles all infrastructure management, giving you access to EC2 pricing models like Compute Savings Plans and Reserved Instances for up to 72% savings on steady-state workloads. This bridges the gap between serverless convenience and EC2 cost optimization since you maintain the Lambda programming model and operational simplicity while unlocking commitment-based discounts previously only available to teams managing their own EC2 infrastructure.

4. S3 Intelligent-Tiering for S3 Tables

S3 Tables, announced just last year at re:Invent have been widely adopted as a fully managed way to store structured data using Apache Iceberg. Amazon now supports automatic cost optimization by moving data between access tiers based on usage patterns. When enabled, this feature monitors access patterns and automatically shifts infrequently accessed table data to lower-cost storage tiers without performance impact or retrieval fees, eliminating the need for manual lifecycle management. For organizations storing large analytical datasets in S3 Tables, Intelligent-Tiering can significantly reduce storage costs while maintaining instant access to all data when needed.

5. RDS SQL Server Developer Edition Support

AWS now supports SQL Server Developer Edition on RDS, offering a free SQL Server edition that includes all Enterprise Edition functionalities for non-production workloads. This eliminates SQL Server licensing costs in development and testing environments while maintaining feature parity with production configurations. Teams can now build, test, and validate applications with the full Enterprise feature set without incurring licensing expenses, making it significantly more cost-effective to run multiple non-production database instances for development workflows.

6. Lambda Durable Functions

Lambda Durable Functions address long-running workflows that require coordinating multiple steps over time. They enable you to build multi-step applications and workflows that run reliably over extended periods, from seconds to up to one year, without paying for idle compute time. This is particularly valuable for workflows that wait for external events, human approvals, or long-running processes where traditional Lambda functions would time out or require constant polling. Durable Functions pause execution during wait periods and only charge for active processing time, significantly reducing costs for orchestration workloads compared to keeping compute resources running or building complex state management systems.

7. Amazon S3 Vectors

S3 Vectors are described as the "first cloud object storage with native support to store and query vector data." This service targets teams currently using specialized vector databases or purpose-built solutions that can be expensive at scale. For those with RAG applications, semantic search workloads, or recommendation engines, S3 Vectors offers a more cost-effective alternative by leveraging S3's existing storage pricing while adding native vector querying capabilities.

8. RDS M7i/R7i Instances with Optimize CPU

Amazon RDS for SQL Server now supports M7i and R7i instances with the Optimize CPU feature, delivering substantial cost reductions for database workloads. These new instances offer lower costs compared to previous generation instances while providing the ability to customize vCPU counts on license-included instances. Since SQL Server licensing is per-vCPU, this is particularly valuable for workloads requiring high memory and IOPS but fewer vCPUs, you can reduce licensing costs while maintaining the performance your applications need. AWS also now bills instance costs and licensing fees separately, providing better visibility into where your database spend is going.

9. OpenSearch GPU Acceleration

Another win for teams with vector workloads, Amazon OpenSearch Service now offers GPU-accelerated vector search that delivers faster performance and lower costs compared to CPU-based implementations. The auto-optimization capability automatically balances search quality, speed, and resource usage, so you're not stuck manually tuning performance parameters.

10. Amazon EKS Capabilities for Workload Orchestration

Amazon EKS introduced new fully managed platform capabilities that handle workload orchestration and cloud resource management, eliminating the operational overhead of maintaining Kubernetes infrastructure. While not a direct cost reduction, this significantly reduces the engineering time and expertise required to manage Kubernetes environments, translating to lower indirect costs through reduced operational complexity. Teams can focus on application development rather than platform maintenance while maintaining enterprise-grade reliability and security.

Conclusion

This year's re:Invent announcements cover a wide range of services, but the common theme is giving teams more control over their cloud spending without sacrificing performance or operational simplicity. From Database Savings Plans providing commitment flexibility to Lambda Durable Functions eliminating idle compute charges, AWS continues to focus on cost optimization alongside new capabilities. As always, the key is identifying which of these features align with your specific workload patterns and cost optimization goals.