POLICY · MONEY · #3511
Track and cap Amazon Bedrock spend per IAM principal using CUR 2.0 and AWS Budgets
AWS describes a no-code approach to get per-identity Amazon Bedrock cost visibility and enforceable spend limits by combining Bedrock's automatic IAM principal attribution with cost allocation tags, CUR 2.0 or Cost Explorer for reporting, and AWS Budgets with budget actions that apply a deny IAM policy when thresholds are exceeded. The post outlines three steps—attribute (principal or session tags), report (activate tags and use CUR 2.0 or Cost Explorer; optionally query with Athena), and enforce (create scoped budgets with budget actions)—to automatically restrict Bedrock access per identity.
KEY POINTS
- AWS describes a no-code approach to get per-identity Amazon Bedrock cost visibility and enforceable spend limits by combining Bedrock's automatic IAM principal attribution with cost allocation tags, CUR 2.0 or Cost Explorer for reporting, and AWS Budgets with budget actions that apply a deny IAM policy when thresholds are exceeded.
- The post outlines three steps—attribute (principal or session tags), report (activate tags and use CUR 2.0 or Cost Explorer; optionally query with Athena), and enforce (create scoped budgets with budget actions)—to automatically restrict Bedrock access per identity.
- This makes per-user or per-role AI cost attribution and automatic enforcement feasible with native AWS tools, reducing engineering overhead and giving finance/engineering leaders clearer controls over Bedrock spend.
WHY IT MATTERS
This makes per-user or per-role AI cost attribution and automatic enforcement feasible with native AWS tools, reducing engineering overhead and giving finance/engineering leaders clearer controls over Bedrock spend.