AWS re:Invent 2025 and 2026
Feature Announcements by Service Category

1. Executive Summary

Key Findings:

AWS re:Invent 2025 and early 2026 announcements reveal a comprehensive platform evolution focused on AI-first development, autonomous operations, and enhanced enterprise connectivity. The most significant developments center on four strategic areas: revolutionary AI capabilities through the Nova model family, autonomous agent frameworks for operational excellence, modernized networking infrastructure, and unified serverless architectures.

Revolutionary AI Platform Capabilities:

The Nova 2 model family represents AWS’s most significant AI advancement, delivering specialized models for diverse enterprise needs. Nova 2 Lite provides cost-effective AI with million-token context windows, while Nova 2 Pro handles complex multistep tasks including video reasoning and advanced mathematics [1, 2]. Nova 2 Sonic introduces enhanced speech-to-speech AI with multilingual conversations, and Nova 2 Omni serves as an all-in-one multimodal model supporting text, images, video, and speech inputs [1, 3]. Most notably, Nova Forge enables organizations to build custom frontier models for $100,000 annually, democratizing advanced AI development [2].

Amazon Bedrock’s AgentCore platform provides enterprise-grade infrastructure for deploying AI agents at scale, featuring comprehensive services including Runtime for low-latency serverless environments, Memory for session management, Identity for secure access, and Gateway for API transformation [4]. Reinforcement fine-tuning delivers 66% accuracy gains over base models, while integration with leading providers including new Mistral Large 3 models expands model selection [5, 1].

Autonomous Operations and Development:

AWS introduces three frontier agents that fundamentally change operational paradigms. Kiro autonomous agent writes code independently, learning team workflows and operating for hours or days without intervention [6]. AWS Transform Custom delivers AI-powered code modernization, cutting execution time by up to 80% [7]. The AWS DevOps Agent acts as an autonomous on-call engineer, analyzing data across CloudWatch, GitHub, and ServiceNow for proactive incident response and root cause identification [7, 8].

These agents integrate with enhanced security capabilities through AWS Security Agent, which provides proactive application security from design to deployment with AI-powered design reviews and contextual penetration testing [7]. IAM Policy Autopilot automates policy generation from code analysis, supporting AI coding assistants like Claude, Cursor, and Cline through Model Context Protocol integration [9].

Developer Platform Modernization:

Lambda capabilities expand significantly with Lambda Managed Instances enabling functions to run on EC2 compute while maintaining serverless simplicity [10]. Lambda Durable Functions coordinate multiple steps reliably over extended periods from seconds to one year, featuring automatic checkpointing and recovery capabilities [11, 12]. Cross-account DynamoDB Streams access enables event-source mappings across AWS accounts [13].

AppSync Events introduces WebSocket-powered real-time PubSub APIs with automatic GraphQL code generation and built-in local resolvers for managing temporary pub/sub channels [14]. Step Functions gains Visual Studio Code extension with drag-and-drop capabilities and native integrations with over 200 AWS services plus external APIs [15, 16].

Infrastructure and Networking Enhancements:

Load balancing capabilities receive comprehensive upgrades. Application Load Balancer adds cross-zone support for zonal shift, header modification using regex patterns, URL and Host Header rewrite capabilities, and IPv6-only support for internet clients [17, 18, 19]. Network Load Balancer introduces UDP support for dual-stack configurations, weighted target groups for blue/green deployments, and stickiness support for TCP, UDP, and TCP_UDP listeners [17, 20].

PrivateLink expands with UDP support and cross-region connectivity, eliminating need for VPC peering or public internet access [21, 20]. VPC Lattice adds TLS listeners with mTLS, Service Network Endpoints, and custom domain names for FQDN access across VPCs and accounts [22, 23]. Route 53 Global Resolver provides secure anycast DNS resolution with built-in security controls and unified management for hybrid deployments [24].

Database and Analytics Transformation:

Aurora PostgreSQL Limitless delivers virtually unlimited scaling with distributed, leaderless architecture, while Aurora DSQL provides multi-region active deployment with fastest multi-region reads and writes [25, 26]. Database Savings Plans offer up to 35% cost reduction for Aurora serverless and up to 18% savings for DynamoDB on-demand throughput [27].

Amazon S3 increases maximum object size from 5TB to 50TB, introduces S3 Tables with Intelligent-Tiering, and delivers S3 Vectors general availability supporting up to 2 billion vectors per index with 100ms query latencies [28, 10]. QuickSight transforms into Quick Suite with AI agents for business insights, Quick Research for comprehensive cited insights, and Quick Flows for automated workflow creation [29].

Unified Observability and Serverless Architecture:

CloudWatch evolves into a unified platform with Apache Iceberg compatibility through S3 Tables, automatic data normalization supporting OCSF and Open Telemetry formats, and natural language queries alongside LogsQL, PPL, and SQL [30]. Comprehensive observability for generative AI applications provides insights into latency, token usage, and errors [31].

EventBridge achieves 94% latency reduction, improving average latency from 2,235ms to 129ms at P99, while adding cross-account targets and VPC access integration [32, 33, 34]. Step Functions and EventBridge gain access to VPC resources through PrivateLink and VPC Lattice, supporting private HTTPS endpoints, databases, and on-premises services [33, 35].

Business Impact:

These announcements position AWS as the comprehensive platform for AI-first enterprises, with autonomous agents reducing operational overhead while Nova models democratize advanced AI capabilities. The unified serverless architecture with enhanced networking eliminates traditional infrastructure complexity, while comprehensive observability ensures reliable operations at scale. Organizations can now build, deploy, and operate AI-powered applications with unprecedented simplicity and performance, supported by autonomous systems that proactively manage security, operations, and development workflows.

Strategic Implications:

The convergence of AI agents, enhanced networking, and unified serverless capabilities creates a new paradigm where infrastructure becomes truly invisible, operations become autonomous, and development becomes AI-assisted. This represents AWS’s most significant platform evolution since the introduction of serverless computing, fundamentally changing how enterprises approach cloud architecture and operations.

2.     Developer Services and API Enhancements

API Gateway: - Enhanced TLS configuration standardization for improved security posture [36]

AppSync:
- AppSync Events websockets for real-time PubSub APIs powered by WebSockets [14]
- Simplified development with automatic GraphQL code generation on backend and client sides [14]
- Built-in local resolvers for managing multiple temporary pub/sub channels [14]

Lambda:
- Lambda Managed Instances allowing functions to run on EC2 compute while maintaining serverless simplicity [10]
- Lambda Durable Functions for building applications that coordinate multiple steps reliably over extended periods from seconds to up to one year [11]
- Cross-account DynamoDB Streams access for event-source mappings [13]

Step Functions: - Visual Studio Code extension with drag-and-drop Workflow Studio capabilities [15]
- Native integrations with over 200 AWS services plus external third-party APIs [16]
- VPC resource access integration with EventBridge for private network connectivity [33]

AI Development Features:
- Multi-Agent Collaboration frameworks compatible with LangChain, LangGraph, and CrewAI [31]
- IAM Policy Autopilot for automated policy generation from code analysis [9]

3.     AI and Machine Learning Innovations

Nova Model Family:
- Nova 2 Lite offering fast, cost-effective AI for everyday tasks with extended thinking capabilities and million-token context window [1]
- Nova 2 Pro for highly complex, multistep tasks including video reasoning and advanced math equations [2]
- Nova 2 Sonic providing enhanced speech-to-speech AI with multilingual conversations and dynamic speech control [1]
- Nova 2 Omni as all-in-one multimodal model supporting text, images, video, and speech inputs while generating both text and image outputs [3]
- Nova Forge custom model training service enabling organizations to build frontier models for $100,000 annually [2]
- Nova Act for UI automation with over 90% reliability for enterprise deployments [1]

Amazon Bedrock Enhancements:
- 100+ Models now Available , New added : Gemma3 4B,12B,27B , Newmotron Nano2 9B , gpt-oss-safeguard-120b,20b m Qwen3- ..  [5]
- AgentCore Runtime providing low-latency serverless environments with session isolation [4]

- AgentCore Runtime supports bi-directional Streaming
- AgentCore Memory adds Episodic memory strategy  
- AgentCore Memory for managing session and long-term memory with relevant context [4]
- AgentCore Identity enabling secure access to AWS services and third-party tools [4]
- AgentCore Gateway transforming existing APIs into agent-ready tools [4]
- Reinforcement fine-tuning delivering 66% accuracy gains over base models [1] 

SageMaker AI Capabilities:
- Serverless MLflow integration for zero-infrastructure AI experimentation [1]
- Checkpointless and elastic training on HyperPod for instant recovery from failures [1]
- Unified Studio one-click onboarding eliminating weeks of setup time [37]
- SageMaker Data Agent for generating code and SQL statements from natural language prompts [38]

Frontier Agents:
- Kiro autonomous agent for code writing designed to learn team workflows and operate independently for hours or days [6]
- AWS Transform Custom for AI-powered code modernization cutting execution time by up to 80% [7]
- AWS DevOps Agent acting as autonomous on-call engineer for incident response [7]

4.     DevOps Infrastructure and Load Balancing

Application Load Balancer:
- Cross-zone enabled ALB support for zonal shift and zonal autoshift capabilities [17]
- Support for header modification using regex-based pattern matching (also benefits Security) [18]
- URL and Host Header rewrite capabilities eliminating need for custom application logic [19]
- Support for Load Balancer Capacity Unit Reservation for predictable performance [17]
- IPv6-only support for internet clients enabling modern networking architectures [17]
- Support for configurable HTTP client keepalive duration for optimized connections [17]

Network Load Balancer:
- UDP support for dual-stack Network Load Balancers enabling IPv4 and IPv6 protocols [20]
- Cross-zone enabled NLB support for zonal shift and zonal autoshift for high availability [17]
- Weighted target groups for blue/green deployments with numeric weights between 0 and 999 [17]
- One-click integration with Amazon CloudWatch Internet Monitor for enhanced observability [17]
- Support for configurable TCP idle timeout including gateway load balancer [17]
- Stickiness support for TCP, UDP and TCP_UDP listeners [17]

CloudFront:
- Private VPC origins enabling secure content delivery from private networks [39]
- Support for anycast distribution for improved global performance [40]
- Embedded POPs for enhanced edge computing capabilities [39]
- gRPC support for modern application protocols [39]
- Viewer mTLS authentication with bidirectional certificate verification [39]
- Trust Store integration supporting certificate chains up to 4 levels deep [39]

AWS DevOps Agent:
- Autonomous incident response acting as on-call engineer analyzing data across CloudWatch, GitHub, and ServiceNow [8]
- Proactive system reliability improvements across AWS, multicloud, and hybrid environments [8]
- Root cause identification and incident response coordination capabilities [8]

5.     Security and Governance Capabilities

IAM Declarative Policies:
- IAM Policy Autopilot open-source tool for automated policy generation from code analysis [9]
- Model Context Protocol (MCP) server integration with AI coding assistants like Claude, Cursor, and Cline [9]
- Deterministic code analysis generating identity-based IAM policies from AWS SDK calls [9]
- Up-to-date AWS service knowledge and reliable permission recommendations for AI assistants [9]

Security Incident Response:
- AWS Security Incident Response service with streamlined four-step process for comprehensive threat management [41]
- AI-powered investigation capabilities combining automated triage with agentic AI analysis [42]
- 24/7 access to AWS Customer Incident Response Team (CIRT) professionals [41]
- Automatic ingestion from GuardDuty and Security Hub integrations with preconfigured team management [43]
- Available in ten additional AWS Regions as of December 2025 [44]

GuardDuty Enhanced Detection:
- Extended Threat Detection expansion to container-based environments for comprehensive security coverage [45]
- Timeline views and attack sequence mapping for complex threat identification across applications and workloads [46]
- Unified visibility across virtual machine and container environments identifying multi-stage attacks [7]

AWS Security Hub:
- Near real-time analytics and risk prioritization with automatic correlation across multiple security services [47]
- Historical trends analysis with up to 1 year of data retention for period-over-period comparisons [47]
- Integration with Jira and ServiceNow for incident management workflows [47]
- Open Cybersecurity Schema Framework (OCSF) support enabling seamless partner integrations [47]

AWS Security Agent:
- Proactive application security from design to deployment with AI-powered design reviews [7]
- Static application security testing (SAST) integration in development lifecycle [7]
- Contextual penetration testing understanding unique security requirements and application architecture [7]

Governance (also benefits Security):
- RCP on AWS Organizations for enhanced resource control policies [7]
- Declarative policies general availability for simplified governance across AWS Organizations [7]

6.     Networking and Connectivity Services

PrivateLink:
- UDP Support for AWS PrivateLink enabling private connectivity for UDP-based applications including real-time gaming, VoIP, and media streaming [20]
- Cross-region PrivateLink connectivity to AWS services eliminating need for VPC peering or public internet access [21]
- Enhanced access control mechanisms with endpoint policies supporting organizational compliance requirements [21]
- Commercial partition support across all regions with simplified regional data controls [21]

VPC Lattice:
- TLS listeners with mTLS for enhanced security in service-to-service communication [23]
- Service Network Endpoints providing PrivateLink-similar functionality for application networking [23]
- Custom domain names for resource configurations enabling FQDN access across VPCs and accounts [22]
- IPv6 management endpoint support with dual-stack capabilities for modern networking [48]

Route 53 Global Resolver:
- Secure anycast DNS resolution through globally distributed IP addresses for hybrid deployments [24]
- Unified DNS management resolving both public internet domains and private Route 53 hosted zones [24]
- Built-in security controls including DNS traffic filtering, encrypted query support, and centralized logging [24]
- Operational overhead reduction while maintaining consistent security controls across distributed environments [8]

Network Observability:
- EC2 ETH Tools for enhanced network performance monitoring and troubleshooting [28]
- Infrastructure performance insights through comprehensive network analytics [28]
- CloudWatch flow monitor with Agent on EC2 for detailed network traffic analysis [30]

Enhanced Connectivity:
- Route 53 DNS service PrivateLink support enabling private API access without public internet [49]
- Cross-region Interface VPC endpoints for Route 53 connectivity eliminating inter-region networking complexity [49]
- Route 53 Resolver and Profiles PrivateLink integration for comprehensive private DNS management [50, 51]

7.     Database and Storage Modernization

Amazon Aurora:
- Aurora PostgreSQL Limitless providing virtually unlimited scaling with distributed, leaderless architecture [26]
- Aurora DSQL multi-region active deployment with fastest multi-region reads and writes [25]
- Aurora DSQL automatic infrastructure management with zero downtime maintenance [52]
- Database Savings Plans offering up to 35% cost reduction for serverless deployments [27]

Amazon DynamoDB:
- Database Savings Plans providing up to 18% savings for on-demand throughput workloads [27]
- Multi-Region Strong Consistency providing zero RPO for highly resilient global applications [53]
- Price reduction benefits through Database Savings Plans with up to 18% savings for on-demand throughput [27]

Amazon MemoryDB:
- MemoryDB Multi-region active-active deployment with up to 99.999% availability [54]
- Microsecond read and single-digit millisecond write latencies across multiple AWS Regions [54]

Amazon S3 Enhancements:
- Maximum object size increase from 5TB to 50TB enabling massive sensor datasets for automotive and manufacturing [28]
- S3 Tables with Intelligent-Tiering for automatic cost optimization and simplified replication across regions [10]
- S3 Vectors general availability supporting up to 2 billion vectors per index with 100ms query latencies [10]
- S3 Storage Lens performance metrics providing insights into application interaction patterns [55]

AWS Transfer Family: - Transfer Family for web apps integration with S3 enabling simplified file transfer workflows

Amazon FSx Integration:
- FSx for NetApp ONTAP S3 integration enabling seamless access to file system data through S3 APIs [10]
- Direct integration with AWS analytics, ML, and generative AI services without data movement [10]

8.     Analytics and Observability Enhancements

Amazon QuickSight Evolution:
- QuickSight transformation to Quick Suite with AI agents for business insights, research, and automation [29]
- Quick Research delivering comprehensive, cited insights from enterprise and public data sources in minutes [29]
- Quick Flows enabling automated workflow creation and business process management [29]
- Embedded Chat capability bringing agentic AI-powered conversations into existing tools [56]
- Enhanced SPICE capabilities with increased dataset limits up to 2TB and expanded string length support from 2K to 64K Unicode characters [57]

Amazon SageMaker Unified Studio:
- One-click onboarding eliminating weeks of setup time with automatic project creation [37]
- Direct console integration launching from SageMaker, Athena, Redshift, and S3 Tables pages [38]
- Serverless notebooks with built-in AI agent supporting SQL, Python, Spark, and natural language [38]
- SageMaker Data Agent generating code and SQL statements from natural language prompts [38]
- Cross-region subscriptions and IAM role-based subscriptions for flexible data access [57]

Data Lineage and Governance:
- Unified environment supporting data processing, SQL analytics, model development, training, inference, and generative AI development [58]
- Trusted identity propagation for Apache Spark interactive sessions with corporate identities [59]

AWS Glue Enhancements:
- Enhanced ETL capabilities with automatic scaling and serverless compute provisioning [38]
- AWS Glue now available in Asia Pacific (New Zealand) Region for closer data processing [57]

Amazon CloudWatch Unified Platform:
- CloudWatch unified data management with Apache Iceberg compatibility through S3 Tables [30]
- Automatic data normalization supporting OCSF and Open Telemetry formats [30]
- Natural language queries alongside LogsQL, PPL, and SQL through single interface [30]
- Comprehensive observability for generative AI applications and agents with insights into latency, token usage, and errors [31]
- Expanded log data management capabilities across operational, security, and compliance use cases [30]

9.     Serverless and Event-Driven Architecture Improvements

AWS Lambda Enhancements:
- Lambda Managed Instances enabling functions to run on EC2 compute while maintaining serverless simplicity and cost optimizations [10]
- Lambda Durable Functions for building applications coordinating multiple steps reliably over extended periods from seconds to up to one year [11]
- Automatic checkpointing, suspension capabilities, and recovery from failures without custom state management [12]
- Cross-account DynamoDB Streams access enabling event-source mappings across AWS accounts [13]

AWS Step Functions:
- Access to VPC resources through PrivateLink and VPC Lattice integration for secure cross-network connectivity [33]
- Enhanced workflow orchestration with native integrations to over 200 AWS services plus external APIs [16]
- Visual Studio Code extension providing drag-and-drop Workflow Studio capabilities for local development [15]
- HTTP Task support for private connections eliminating need for custom Lambda functions or SQS queues [35]

Amazon EventBridge:
- EventBridge latency 94% reduction with average latency improved from 2,235ms to 129ms at P99 [32]
- Cross-account targets for Event Buses enabling direct event delivery to SQS, Lambda, and SNS in other accounts [34]
- VPC access integration through PrivateLink and VPC Lattice for secure private API connectivity [33]
- Support for private HTTPS endpoints, databases, EC2 instances, and on-premises services via VPN or Direct Connect [35]

Enhanced Serverless Capabilities:
- Serverless compute automatic scaling and cost optimization with just-in-time resource provisioning [60]
- Unified serverless experience across Lambda, Step Functions, and EventBridge with consistent pricing models [35]
- Private network integration supporting legacy system modernization and hybrid cloud architectures [33]
- Comprehensive monitoring through CloudWatch metrics including IngestionToInvocationStartLatency and end-to-end latency tracking [32]

-- Generated by QuickSuite research