re:Invent 2025

Introducing frontier agents that revolutionize the software development lifecycle

Introducing frontier agents, a new class of autonomous, massively scalable agents that work for hours or days without requiring intervention. Today, we're announcing three that work across the SDLC. Discover each new agent and how they transform the way you build, secure, and operate software.
Unlock the freedom to invent. re:Invent 2025
Unlock the freedom to invent. Explore frontier agents
Product announcements
Frontier Agents

Kiro autonomous agent

Accelerate development with a frontier agent that handles your tasks, so you can focus on what matters most.
Get started with Kiro
Frontier Agents

AWS Security Agent

Proactively secure your applications with a frontier agent that scales security expertise from design to deployment.
Discover AWS Security Agent
Frontier Agents

AWS DevOps Agent

Drive operational excellence with a frontier agent that resolves and proactively prevents incidents.
Discover AWS DevOps Agent
Browse all announcements
re:Invent 2025

Operate AI agents securely and at scale

Run fleets of agents with confidence. Explore AWS agentic AI solutions that empower millions of agents to work together securely—transforming everything from daily operations to complex business processes.
Deploy production-ready agents with built-in security. re:Invent 2025
Deploy production-ready agents with built-in security. Explore Bedrock AgentCore
Product announcements
Agentic Platform

Amazon Bedrock AgentCore Evaluations

Deliver continuous, real-time monitoring to gain insights into agentic operations. Perform evaluations using built-in or custom evaluators, eliminating tedious infrastructure management.
Explore Bedrock AgentCore
Agentic Platform

Policy in Amazon Bedrock AgentCore

Deploy agents confidently across your organization. Trust your agents to act autonomously while knowing they'll always stay within your defined boundaries and compliance requirements.
Explore Bedrock AgentCore
Browse all announcements
re:Invent 2025

Build your own frontier models

Introducing the industry's first open-training models. Nova Forge allows customers to use proprietary data and preferences at every phase in model training.
Discover open training models. re:Invent 2025
Discover open training models. Get started with Nova Forge
Product announcements
Model Training

Nova Forge

A first-of-its-kind program that offers organizations the easiest and most cost-effective way to build their own frontier models using Nova.
Try Nova Forge
Browse all announcements
re:Invent 2025

More models, more freedom to invent

Match model behavior to workload demands. Amazon Bedrock now offers our most comprehensive model portfolio yet—featuring models like Nova 2 Lite, a fast, cost-effective reasoning model for agentic AI, and Nova 2 Omni, an all-in-one model unifying text, image, video, and audio understanding.
Powerful, cost-effective foundation models. re:Invent 2025
Powerful, cost-effective foundation models. Try Amazon Nova
Product announcements
Foundation Model

Nova 2 Sonic

Our second generation speech-to-speech model for real-time, human-like conversational AI applications.
Discover Nova 2 Sonic
Foundation Model

Nova 2 Lite

A fast, cost-effective reasoning model for most workloads.
Discover Nova 2 Lite
Foundation Model

Nova 2 Omni

All-in-one model for multimodal reasoning and image generation.
Try Nova 2 Omni
Foundation Model

Nova 2 Pro

Our most intelligent model for highly complex workloads.
Try Nova 2 Pro
Browse all announcements
re:Invent 2025

Store and query any data type with S3

Store and scale your data for AI and analytics workloads with Amazon S3. Work with objects, files, vectors, or tables while reducing cost, simplifying pipelines, and keeping your data ready for search, RAG, agents, and analytics.
Make your data AI-ready. re:Invent 2025
Make your data AI-ready. Explore S3 innovations
Product announcements
AI-Ready Storage

S3 Vectors

Store and query billions of vectors at up to 90% lower cost for search, RAG, and agent workflows. Now with 1B-vector indexes and scale to 10T vectors per bucket.
Discover S3 Vectors
Analytics-Ready Storage

S3 Tables

Manage Apache Iceberg tables with automatic cost optimization and global replication for DR, low-latency analytics, and global data sharing.
Get started with S3 Tables
FSx File Data Access

FSx for NetApp ONTAP with S3 Access

Access FSx for NetApp ONTAP file data through S3 APIs without moving or duplicating data, enabling AI and analytics on your existing file systems.
Explore FSx for NetApp ONTAP
Scale and Performance

50 TB objects and faster S3 Batch Operations

Store objects as large as 50 TB in S3 and run S3 Batch Operations 10x faster to transform, tag, and prepare datasets for AI/ML workflows.
Explore S3 Batch Operations
Browse all announcements
re:Invent 2025

High-performance AI training and inference

The most complete choice of AI compute from AWS-designed silicon to next-generation EC2 UltraServers. Whether you’re training trillion-parameter systems or deploying real-time agentic workloads, you can run any model at any scale with industry-leading price-performance.
Power AI innovation at scale. re:Invent 2025
Power AI innovation at scale. Explore AWS AI infrastructure
Product announcements
AI Infrastructure

Trn3

EC2 Trn3 UltraServer, powered by Trainium3, is the 4th generation AWS AI chip purpose-built to deliver the best token economics for the next generation of agentic, reasoning, and video-generation applications.
Explore Trn3
EC2 UltraServers

P6e-GB300

Amazon EC2 P6e-GB300 UltraServers accelerated by NVIDIA GB300 NVL72 offer the highest GPU performance for AI workloads in EC2. Ideal for the most compute and memory intensive AI models and use cases.
Discover new UltraServers
Browse all announcements
re:Invent 2025

End tech debt forever with AWS Transform

Smash technical debt and make your tech stack AI-ready with AWS Transform. AWS Transform is the first agentic AI service to accelerate enterprise transformation at scale. Modernize Windows, mainframe and VMware workloads or perform custom modernizations of Java, Node, Python, and more.
Smash your technical debt. re:Invent 2025
Smash your technical debt. Watch visual demonstration
Product announcements
Windows

AWS Transform for full-stack Windows modernization

Accelerate your Windows technology stack transformation by up to 5 times. Modernize .NET applications including UI frameworks, SQL Server databases, and deployment processes. Everything modernizes together, automatically.
Discover AWS Transform
Custom code modernization

AWS Transform custom

Eliminate tech debt at scale with AWS Transform. AI-powered custom transformation modernizes code, APIs, frameworks up to 5x faster. Features 6 built-in transformations (including Java, Node, and Python) plus any-to-any capabilities.
Discover AWS Transform
Mainframe

AWS Transform for mainframe

Reduce mainframe modernization timelines from years to months with agentic AI. AWS Transform analyzes legacy code, transforms applications to modern languages and architectures, and automates migration planning and testing.
Learn about AWS Transform
VMware

AWS Transform for VMware

Modernize VMware workloads at scale with agentic AI. AWS Transform for VMware automates your wave planning, dependency mapping, and network conversion to accelerate your migration to Amazon EC2 with confidence.
Learn about AWS Transform
Browse all announcements

Join the conversation on Builder Center

Participate in real-time discussions around keynotes, launches, and get insights from fellow builders.

Connect with builders now
Missing alt text value

Start using the new cloud services announced at re:Invent

Loading
Loading
Loading
Loading
Loading