Amazon Bedrock — Glossary

AWS's managed service for using foundation models, including Claude, through the AWS account — AWS runs the inference and bills it as native AWS usage.

What is Amazon Bedrock?

Amazon Bedrock is AWS's managed service for building with foundation models. Claude is available on it as "Claude in Amazon Bedrock": you call the same Messages API shape you would use with the Claude API, but the request runs on AWS-managed infrastructure, with AWS as the data processor and no operator access for Anthropic. You authenticate with IAM, pick a region or an inference profile, and the usage is billed as a native AWS service.

That makes it the AWS-native way to get Claude, together with the AWS tooling around it: CloudWatch and CloudTrail for logging, IAM for access, AWS quotas for throughput. It is a different product from Claude Platform on AWS, which Anthropic operates and sells through AWS Marketplace.

When should you use Amazon Bedrock for Claude?

Choose Bedrock when AWS must be the sole data processor, when you need EU data residency (Bedrock offers EU inference profiles and in-region endpoints such as Stockholm and Ireland), or when spend has to retire against an existing AWS commitment. The trade-off is feature surface: Anthropic's server-side tools — web search, code execution, the Files and Batches APIs, Skills, Managed Agents — are not available on Bedrock, and new models follow AWS's release schedule. Plain inference, tool use, prompt caching and extended thinking all work. We build on Bedrock in AI & Automation and Cloud & DevOps engagements when a client's AWS estate calls for it. All seven ways to get Claude are compared in Seven paths to Claude.