Anthropic puts Claude on AWS
- Anthropic made Claude Platform on AWS generally available on May 11, giving AWS customers Anthropic’s native API and console through AWS accounts. - The key wrinkle is architectural: AWS handles IAM, billing, and CloudTrail, but Anthropic operates the stack and processes customer data outside AWS’s boundary. - That matters because it gives enterprises day-one Claude features on AWS, but forces harder choices around residency, compliance, and who actually runs AI.
Cloud AI usually comes in a familiar shape. You buy it from the hyperscaler, the hyperscaler runs it, and the model company sits behind the curtain. Anthropic just broke that pattern on AWS. As of May 11, Claude Platform on AWS is generally available, which means AWS customers can use Anthropic’s native Claude platform through their AWS accounts — but the service itself is operated by Anthropic, not AWS. ### So what actually launched? Anthropic put its own platform inside the AWS buying and identity layer. Customers get the native Claude API, the Claude Console, and a long list of platform features through AWS authentication and billing, without opening a separate Anthropic account flow for procurement and access. Anthropic’s release notes tie the launch date to May 11, 2026, and AWS calls it general availability, not a preview. (claude.com) ### How is that different from Bedrock? Amazon Bedrock is the old, cleaner cloud story — AWS operates the inference stack and acts as the data processor. Claude Platform on AWS flips that. Anthropic runs the platform, while AWS provides IAM, consolidated billing, commitment retirement, and CloudTrail logging. Basically, AWS becomes the front door and Anthropic becomes the operator behind it. (claude.com) ### Why is the security boundary the big deal? Because this is where the tradeoff stops being marketing and becomes architecture. AWS says customer data in Claude Platform on AWS is processed outside the AWS security boundary. That gives customers access to Anthropic’s “canonical” platform and same-day feature rollouts, but it also means this is not the answer for teams that need strict regional residency or want everything to stay inside a standard AWS-managed service perimeter. AWS says the offering is aimed at customers without those specific residency requirements. (platform.claude.com) ### What do customers get that they did not before? The short version is feature parity with Anthropic’s own platform. Anthropic says new features and betas ship the same day they hit the native Claude API. AWS lists Claude Managed Agents, advisor strategy, web search, web fetch, code execution, Files API, Skills, MCP connector, prompt caching, citations, batch processing, and the Claude Console as available through this path. That is a much fatter surface area than the older “just call the model” setup. (claude.com) ### Where do managed agents fit in? Managed Agents are the second half of the story. Anthropic has turned agent runtime into a product — a hosted harness for long-running, asynchronous work where Claude can read files, run commands, browse the web, execute code, and keep persistent event history. Instead of building your own loop, sandbox, and orchestration layer, you define an agent, define an environment, start a session, and stream events. Anthropic also notes that Managed Agents are available on Claude Platform on AWS, with some feature and session differences. (claude.com) ### Why would enterprises like this? Because it removes one of the biggest adoption frictions — separate vendor plumbing. Teams can keep AWS IAM, AWS invoices, and CloudTrail, while still getting Anthropic’s latest platform features. For companies with big AWS commitments, that matters a lot. Anthropic also says the service is available across most commercial AWS regions, with global and U.S. inference geographies. ### What is the catch? (platform.claude.com) Control is split. Procurement and access control look like AWS. Runtime behavior, feature timing, and data processing look like Anthropic. That can be great if your main problem is speed to the newest Claude capabilities. But if your main problem is compliance boundaries, data handling certainty, or minimizing cross-vendor complexity during audits, the design gets trickier fast. ### Bottom line? Anthropic did not just add another cloud endpoint. (claude.com) It proposed a different deal: let AWS handle the enterprise wrapper, let Anthropic keep the product. That gives customers more Claude, sooner — but it also makes the line between “inside my cloud” and “inside my vendor’s platform” a lot more important.
