S3 gets regional namespaces
AWS is scoping S3 bucket names to account+region (so names look like bucket-<account>-<region>), a change designed to stop global naming conflicts and make Infrastructure as Code—especially CloudFormation—less brittle devopscube tweet and RDarrylR thread. The rollout leaves existing buckets untouched, so IaC templates can be simplified without forcing mass renames devopscube tweet.
AWS published the change on March 12, 2026 (aws.amazon.com) and made the feature available in 37 AWS Regions, including AWS China and AWS GovCloud (US), at no additional cost what’s new. Bucket names created in the account-regional option must include a suffix in the form -{AWS account ID}-{region}-an (example: mybucket-123456789012-us-east-1-an), and the combined prefix plus suffix must be between 3 and 63 characters long blog example and naming rules. Creation requires sending the new request header (x-amz-bucket-namespace: account-regional) or using the corresponding Console/CLI/SDK options; the AWS CLI example appears directly in the announcement showing the header and --bucket-namespace flag how-to. Organizations can enforce account-regional-only bucket creation by using the s3:x-amz-bucket-namespace condition key in IAM policies and AWS Organizations service control policies (SCPs), as documented in AWS guidance for the feature security controls. AWS listed Console, S3 REST API, AWS CLI, AWS SDK, and AWS CloudFormation as supported surfaces for the new namespace option (aws.amazon.com), while the Terraform AWS Provider has an open issue tracking integration and notes the change depends on an AWS SDK for Go v2 release around March 12, 2026 Terraform issue. AWS documentation also clarifies that S3 bucket namespaces remain organized by partitions (aws, aws-cn, aws-us-gov, aws-eusc) and that customers can still choose between the shared global namespace and the account regional namespace when creating general purpose buckets namespaces doc.
