AWS has released cost attribution features for Amazon Bedrock that automatically trace each inference request to the IAM principal (user or application) that made it.
The new line_item_iam_principal column in Cost and Usage Report 2.0 feeds Amazon Athena queries and CUDOS dashboards, letting teams see spending by user, service, team, or project without manual instrumentation.
This enables fine-grained cost governance and chargeback for multi-tenant and multi-service Bedrock deployments.
What happened
AWS has released tools to trace Amazon Bedrock inference costs back to individual IAM principals (users or applications) using Cost and Usage Report 2.0 data. A new line_item_iam_principal column automatically records which identity made each API call, enabling per-user and per-application cost visibility through Amazon Athena SQL queries or CUDOS dashboards.
Why it matters
Teams running Bedrock-powered services can now isolate spending by user, application, or team—critical for multi-tenant platforms, chargeback, and cost governance. Without this granularity, costs were pooled at the service level, making it impossible to attribute spend to individual services like a chatbot versus a document processor.
What to watch
Setup requires enabling IAM principal data in CUR 2.0 exports (which increases file sizes for high-volume workloads) and may take up to 24 hours for the first report. AWS provides optional automation via Claude Code or Kiro-CLI, or manual Athena configuration. CUDOS dashboards offer pre-built visuals alongside Athena's flexible SQL queries.
Ask the AI about this article →
This post is the second in a series on Bedrock cost attribution. Part 1 introduced the core feature—automatic tracing of inference requests to IAM principals—but this follow-up focuses on the practical analysis layer: how to visualize and aggregate that data for operational use. The body positions two complementary tools: Amazon Athena (for SQL flexibility, BI tool integration, and custom chargeback logic) and CUDOS dashboards (for pre-built, organization-specific visuals). The critical prerequisite is enabling the "Include caller identity (IAM principal) allocation data" checkbox in CUR 2.0 configuration, which populates both the line_item_iam_principal column and IAM principal tags.
The post then walks through three progressively sophisticated Athena query patterns. The first isolates cost by caller and model usage type. The second groups by known IAM principal tags (team, project, cost center) to answer aggregate questions like "How much did engineering spend?" The third uses UNNEST to dynamically discover tag schemas across an organization—useful when tag names are unknown or evolve over time. A real-world example then shows how a platform team running multiple services (document processor, chatbot) can isolate spend per service using role-based filtering. AWS acknowledges that enabling IAM principal data increases CUR file sizes, placing the burden of storage planning on the operator.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Anthropic is privately hoping to file for its initial public offering by the end of this month, targeting a ra…
Broadcom is reportedly seeking to borrow up to $100 billion in debt financing to support growth efforts at Ant…
As AI technology matures, the bottleneck in the industry is moving beyond semiconductor constraints like GPUs…
Elice Group, a South Korean AI infrastructure provider, announced the launch of the country's first AI data ce…
On August 12, AT&T's Chief Data and AI Officer said OpenAI models power about 25% of the telecom's total AI us…
On August 11, IBM announced a multi-year $240 million agreement with Together AI to deploy NVIDIA HGX B300 sys…
