Build, deploy, and scale AI with AWS
Find models, agents, tools, and infrastructure for every AI workload
Why teams choose AWS for artificial intelligence
Access ready-to-use AI solutions or build custom models, agents, and applications on a secure and scalable cloud foundation. Adapt easily as needs change, with hundreds of models and tools for every framework and workflow. Connect AI to business data, improve performance and cost with AWS-designed chips and optimization tools. Move into production with layered protection, from hardware-based isolation to controls that enforce what agents can access and do.
Benefits
Pick the best model, framework, or path to production for your use case. Amazon Bedrock gives you access to hundreds of leading models, including Anthropic and OpenAI, through a unified API.
Get more from every AI dollar you spend. AWS delivers the price-performance, cost optimization, and cost control to move AI to production faster and keep it running efficiently at any scale.
Scale AI on a foundation of trust. AWS delivers enterprise-proven security and reliability for every AI workload, backed by the same infrastructure trusted by millions of customers worldwide.
Find the right AI capabilities for your needs
AWS offers a complete portfolio of AI services, from managed foundation model access to purpose-built agents to AI-optimized compute. No matter where you are with your AI journey, there's a clear path forward with AWS.
Build AI agents
Build AI agents that answer questions and take action
Deploy AI agents that answer customer or employee questions, processes routine requests, and escalate complex issues to the right person. The agent draws on your organization's knowledge base to ground responses in your custom data, responds in the user's preferred language, and operates within compliance guardrails your team defines. AI agents handle routine queries end-to-end so your team can focus on higher-value work.
Learn more: Amazon Bedrock AgentCore | Amazon Connect
Customize models
Customize models on your own data
When your use case demands a model trained on proprietary data, Amazon SageMaker AI provides serverless customization across 20+ open-weight models using techniques that range from supervised fine-tuning to reinforcement learning. Host the finished model on Amazon Bedrock without managing infrastructure, and iterate on performance in days rather than months.
Learn more: Amazon SageMaker AI | Amazon Bedrock
Diagnose incidents and optimize costs
Diagnose incidents and optimize costs automatically
AWS DevOps Agent correlates alerts across regions, identifies root causes, suggests remediation paths, and routes fixes for approval. FinOps Agent flags cost anomalies and recommends right-sizing actions across AWS, multicloud, and on-premises environments. What previously required hours of manual triage resolves in minutes, keeping applications running and costs in check.
Learn more: AWS DevOps Agent | AWS FinOps Agent
Modernize legacy applications
Modernize legacy applications without a multi-year rewrite
AWS Transform analyzes, refactors, and validates legacy codebases across mainframe, Windows, and VMware environments. The agentic AI service converts business logic to modern languages, validates functional equivalence, and keeps the codebase current after migration. Projects that once took years now complete in months, so your engineers build new products instead of maintaining old ones.
Learn more: AWS Transform
Capabilities
Explore models, infrastructure, accuracy, governance, and security.
Amazon Bedrock gives you access to hundreds of models through a unified API. Open tools and standards help your AI work with the technology you already use.
AWS Trainium delivers up to 40% better price-performance for training and inference vs comparable instances. Graviton handles CPU-intensive agentic workloads, with the best price performance at AWS.
Automated Reasoning checks in Amazon Bedrock Guardrails use mathematical logic to validate AI responses against the policies you define, helping detect errors and explain the findings.
Amazon Bedrock AgentCore provides policy enforcement, evaluations, observability, and identity controls. AWS Agent Registry adds centralized discovery with approval workflows and semantic search.
The AWS Nitro System provides hardware-based isolation with no mechanism for AWS operators to access customer data on Nitro hosts. Its security design has been independently validated by NCC Group.
FAQs
AWS AI
Open allStart with Amazon Bedrock to build AI applications or Amazon Quick to put AI to work across your teams. AWS experts and partners can guide you to the right path and help you get started.
Pricing varies by service and usage, with options including pay-as-you-go and subscriptions. Visit each product’s pricing page to compare costs and choose the right plan.
AWS combines built-in security with controls for data access, encryption, and AI governance. Configure these controls to support your security and privacy requirements.