Amazon EMR 7.12 delivers 4.5x faster Apache Spark and Iceberg workload performance compared to open source Spark 3.5.6, with optimizations for metadata caching, parallel I/O, and adaptive query planning while maintaining Iceberg's ACID transactions and schema evolution features.

Amazon EMR 7.9.0 introduces significant performance optimizations for Apache Spark encryption, delivering up to 20% faster execution and cost savings for encrypted workloads without requiring application changes.

Medidata transformed its legacy batch ETL architecture into a modern real-time lakehouse using Apache Iceberg, Flink, and AWS services, reducing pipeline latency from days to minutes while centralizing data access and security.

Amazon EMR 7.12 runtime achieves 2x faster Apache Spark and Iceberg write performance compared to open source versions through optimized metadata operations, bloom filter joins, and parallel file writes.

CBRE partnered with AWS to build a unified AI-powered search and digital assistant using Amazon Bedrock that enables property managers to efficiently access and analyze structured and unstructured property data across multiple sources through natural language queries.

Myriad Genetics partnered with AWS to transform healthcare document processing using generative AI, achieving 98% classification accuracy, 77% cost reduction, and 80% faster processing while automating manual information extraction tasks.

Amazon SageMaker HyperPod now supports Managed Tiered KV Cache and Intelligent Routing to optimize LLM inference, reducing time-to-first-token by up to 40%, increasing throughput by up to 38%, and lowering compute costs by up to 28% for long-context applications.

Amazon Quick Suite User Group in Chicago brought together 90+ professionals to explore agentic AI capabilities, featuring hands-on workshops and demonstrations of how AI agents can unify research, business intelligence, and automation to transform workplace productivity.

Netflix consolidated its relational database infrastructure by migrating from self-managed distributed PostgreSQL to Amazon Aurora PostgreSQL, achieving up to 75% performance improvement and 28% cost savings while eliminating operational overhead.

Amazon EKS introduces Provisioned Control Plane, offering pre-allocated capacity tiers (XL, 2XL, 4XL) to deliver predictable, high-performance Kubernetes control planes for demanding AI/ML, SaaS, and mission-critical workloads at scale.

Amazon S3 Metadata is now available in 22 additional AWS Regions, enabling users to discover and understand S3 data through automated, near real-time queryable metadata across 28 total regions globally.

Amazon SageMaker AI now extends Flexible Training Plans to support inference endpoints, allowing customers to reserve guaranteed GPU capacity for model evaluation and production workloads without managing infrastructure.

AWS Compute Optimizer now detects unused NAT Gateways by analyzing CloudWatch metrics and route table associations, enabling customers to identify and remove idle network resources while protecting disaster recovery configurations for cost savings.

AWS now supports Apache Iceberg V3 deletion vectors and row lineage across EMR, Glue, SageMaker, and S3 Tables to optimize data lake performance and enable efficient tracking of data modifications at petabyte scale.

AWS Backup now offers a low-cost warm storage tier for Amazon S3 backups that automatically transitions data after 60 days, reducing long-term storage costs by up to 30% while maintaining all enterprise-grade protection capabilities.

Amazon Kinesis Video Streams now offers a cost-effective warm storage tier for extended media retention with sub-second access latency, complementing the existing hot tier for real-time access.

Amazon Bedrock introduces a Reserved service tier offering predictable performance with guaranteed tokens-per-minute capacity, flexible input/output allocation, and automatic overflow to pay-as-you-go Standard tier for mission-critical applications.

AWS Cloud Operations announces ten major innovations for 2025 including generative AI observability, AI-powered incident resolution, cross-account log centralization, and mobile app monitoring to help organizations operate efficiently in the AI era.

Amazon EKS Blueprints for CDK now supports EKS Auto Mode, enabling fully automated Kubernetes cluster management for compute, storage, and networking through infrastructure-as-code patterns.

Amazon Bedrock AgentCore Gateway now offers gateway interceptors that enable fine-grained access control, dynamic tool filtering, and secure identity propagation for AI agents accessing thousands of tools across enterprises.

Letta builds production-ready AI agents with persistent memory by integrating with Amazon Aurora PostgreSQL, which provides scalable storage, pgvector support for embeddings, and reliable state management across agent conversations and interactions.

Amazon Redshift delivers 2x faster Apache Iceberg query performance through optimizations including vectorized scans, JIT ANALYZE for automatic statistics collection, and distributed Bloom filters.

AWS Knowledge MCP Server now supports topic-based search across specialized AWS documentation domains like Troubleshooting, Amplify, CDK, and CloudFormation, enabling more precise and relevant search results for AI agents and developers.

AWS Glue now supports catalog federation for Apache Iceberg tables, enabling you to query remote Iceberg tables across multi-vendor catalogs like Snowflake Polaris and Databricks Unity Catalog without duplicating data or metadata.

AWS API MCP Server is now available on AWS Marketplace for deployment to Amazon Bedrock AgentCore, offering simplified container management with enterprise-grade security, scalability, and configurable authentication methods.

Apache Iceberg V3 introduces deletion vectors and row lineage capabilities across AWS services to optimize data lake operations by reducing write amplification, storage costs, and enabling efficient change tracking for compliance and audit requirements.

❜embed built a scalable AI recommendation engine on AWS that transforms Web3 applications by leveraging onchain behavioral data to deliver personalized content experiences, achieving 2-3x engagement uplift for early adopters.

AWS Step Functions orchestrates a scalable serverless pipeline combining Amazon Textract, Bedrock batch inference, and Knowledge Bases to automatically extract text and metadata from large document collections for intelligent, searchable repositories.

Condé Nast accelerated contract processing and rights analysis from weeks to hours using Amazon Bedrock and Claude to automate extraction, comparison, and categorization of complex licensing agreements across their global media portfolio.

Amazon SageMaker HyperPod now offers BatchRebootClusterNodes and BatchReplaceClusterNodes APIs for programmatic management of unresponsive or degraded cluster nodes across Slurm and EKS orchestrators.

Amazon Nova Sonic, a speech-to-speech AI model, enables natural voice interactions for telephony applications through multiple integration approaches including SIP infrastructure, cloud providers like Vonage and Twilio, and open-source frameworks like Pipecat and LiveKit.

UCLA's Office of Advanced Research Computing built a serverless AWS infrastructure using Amazon SageMaker AI and Amazon Bedrock to enable real-time AI-generated media for an immersive theatrical production of Xanadu, where audiences created digital scenery through mobile phone sketches.

AWS Private Certificate Authority now supports partitioned CRLs, enabling issuance and revocation of up to 100 million certificates per CA by distributing revocation information across multiple 1 MB partitions for improved scalability and application compatibility.

Amazon EKS Container Network Observability enables ML inference workload teams to monitor, visualize, and optimize network performance through service maps, flow tables, and performance metrics integrated with existing observability stacks.

Amazon Aurora PostgreSQL now supports versions 17.6, 16.10, 15.14, 14.19, and 13.22, featuring Dynamic Data Masking for sensitive data protection and improved performance enhancements.

Amazon EMR and AWS Glue now provide audit context support for Lake Formation APIs and Glue Data Catalog calls, automatically logging platform type and identifiers in CloudTrail to enhance compliance, security monitoring, and troubleshooting across data lake infrastructure.

Amazon EMR and AWS Glue now support fine-grained access controls for write operations on AWS Lake Formation registered tables, enabling secure read and write operations in single Spark jobs with consistent security policies.

Mobileye optimized its Road Experience Management (REM) change detection system by deploying deep learning inference on AWS Graviton-based EC2 instances with Triton Inference Server, achieving over 2x throughput improvement while reducing costs through better instance diversification and Spot...

IBM Instana Observability MCP Server enables conversational AI interactions with observability data through Kiro CLI and IDE, allowing developers to query metrics, traces, and alerts using natural language instead of dashboards or API calls.

Amazon SageMaker AI now supports advanced Amazon Nova model evaluation with custom metrics, LLM-as-a-judge comparisons, token-level log probabilities, metadata analysis, and multi-node scaling for comprehensive model assessment.

Octus successfully migrated from Elastic Cloud to Amazon OpenSearch Service using Migration Assistant, achieving 85% infrastructure cost reduction with zero downtime while improving search performance by 20% and reducing operational overhead by 30%.

Mem0 Open Source now integrates with Amazon ElastiCache for Valkey and Amazon Neptune Analytics to provide persistent memory capabilities for agentic AI applications, enabling agents to remember user preferences and maintain context across multiple sessions.

Amazon ElastiCache for Valkey enables semantic caching for AI applications by storing vector embeddings of queries and responses, reducing LLM inference costs by up to 86% and latency by up to 88% through intelligent cache hit matching with Amazon Bedrock.

AWS Glue 5.1 is now available with upgraded engines, expanded Apache Iceberg capabilities including format version 3.0, and AWS Lake Formation write access control for enhanced data integration and security.

Amazon Redshift now supports native Apache Iceberg write capabilities, enabling direct data writes to Iceberg tables in S3 and S3 Tables without intermediate ETL steps while maintaining ACID compliance and seamless integration with other analytics engines.

AWS Compute Optimizer now identifies unused NAT Gateways with no traffic over 32 days, enabling cost savings by recommending deletion of idle resources.

Amazon SES Tenants feature enables isolated reputation management for multiple customers or business units through step-by-step implementation of tenant creation, resource association, IAM policies, EventBridge monitoring, and automated reputation controls.

Amazon SES now offers tenant management capabilities that enable isolated email sending environments within a single AWS account, allowing organizations to segregate email reputation by customer, business unit, or workload while maintaining independent sending policies and monitoring.

Amazon Route 53 Accelerated Recovery enables DNS management continuity during regional disruptions by automatically failing over the control plane to Oregon Region, targeting a 60-minute recovery time objective for public hosted zones.

Amazon SageMaker HyperPod now supports Managed Tiered KV Cache and Intelligent Routing to optimize LLM inference performance, delivering up to 40% latency reduction and 25% cost savings for long-context prompts and multi-turn conversations.