Track and Cap Per-User Bedrock Spend with CUR 2.0 and AWS Budgets

On November 13, 2026, the AWS Machine Learning Blog published a solution describing how to achieve per-IAM-principal cost visibility and spend limits for Amazon Bedrock using only native AWS features, with no custom code required. The solution has three steps: first, Bedrock automatically records the IAM principal identity behind each inference request; second, cost allocation tags generate reports in Cost Explorer or CUR 2.0 exports, enabling cost tracking by identity or tag; finally, an AWS Budget is created for an individual identity or tag, and when spend exceeds the threshold, a budget action automatically applies a Deny IAM policy, preventing that identity from further Bedrock calls. Aimed at organizations managing multi-user Bedrock usage costs, the solution provides a complete workflow from cost aggregation and report analysis to limit enforcement, all built on native AWS capabilities.

AI overview of reporting · Updated 2 h前

Not enough continuous observations to show a trend yet.