AWS Clean Rooms: pricing, features, API and alternatives

A collaboration is a shared workspace. One member creates it and invites others. Each member adds "configured tables" that point at its own data in Amazon S3, AWS Glue, Athena, Snowflake or Databricks (AWS product page). The data stays where it is. Only the member allowed to query can run analyses, and only the members named as result receivers get output.

AWS sells three things on top of that core. Clean Rooms ML trains an AWS model for lookalike audiences or runs your own custom model on the joined data. It became fully available on 3 April 2024 and added custom models on 7 November 2024 (AWS document history). Entity resolution builds an ID mapping table so two members can join records that use different identifiers (glossary: entity resolution). Differential privacy adds calibrated noise to results, and it reached general availability on 10 April 2024 (AWS What's New, 10 April 2024).

The typical buyer is a media owner or a brand that already keeps data in AWS. Kinective Media by United Airlines, DIRECTV and Fox are on AWS's customer page (AWS customers). Amazon's own advertiser room, Amazon Marketing Cloud, runs on the same service.

Every configured table carries an analysis rule, set by its owner. There are three types (AWS user guide, analysis rules):

• Aggregation allows COUNT, SUM and AVG queries with an inner join. Each output row must contain a minimum number of distinct values, such as user IDs, and that minimum must be at least 2 (AWS user guide, aggregation rule).
• List returns row-level overlap between tables for enrichment or segment building.
• Custom allows any query or PySpark job, but only from analysis templates that the owner has reviewed and approved.

When a query touches several members' tables, the strictest rule wins. If one partner requires 100 distinct users per row and another requires 150, the query needs 150 (AWS user guide). Since 13 August 2026, owners can also set minimum aggregation thresholds and comparison controls inside custom rules (AWS document history).

Differential privacy works per collaboration. The owner turns it on in a custom rule, then sets a privacy budget and the noise added per query. AWS tracks the queries left, "similar to the gas gauge in a car" (AWS user guide, differential privacy). It uses SampCert, a sampler developed by AWS.

For partners that may only use encrypted data in the cloud, the C3R client encrypts columns before upload. Encrypted "fingerprint" columns can be joined and "sealed" columns can be selected, but most other SQL does not run on them (AWS user guide, C3R).

Everything in the console is also exposed through the AWS Clean Rooms API, so teams can script collaborations (AWS API reference).

AWS publishes prices. The figures below are AWS's US East (N. Virginia) examples, accessed 9 November 2026 (AWS Clean Rooms pricing). Prices vary by Region for compute.

A worked example from the same page: a 3-minute Spark SQL query on the default 32 CRPUs, run three times a day, costs $9.60 a day and $3,504 a year (AWS Clean Rooms pricing). Members who bring data also pay standard S3 and Glue request fees. The collaboration creator chooses which member pays for compute, and since 21 May 2026 that payer can be changed (AWS document history).

Strengths

• Price transparency. AWS is the only large clean room vendor that publishes unit prices and worked examples, so a buyer can model cost before a call with sales.
• Controls as configuration. Analysis rules, aggregation minimums, comparison controls, data access budgets (1 October 2025) and differential privacy are settings, not custom code (AWS document history).
• ML in the room. Lookalikes, custom models, distributed training (1 July 2025) and synthetic training data (30 November 2025) run without either side exporting rows.
• Wider data sources. Snowflake, Athena and now Databricks tables can join a collaboration without an extract job.

Limits

• Every member needs an AWS account, even when its data sits in Snowflake or Databricks. Collaborators are invited by AWS account ID, so partners that refuse AWS cannot join (AWS user guide, creating a collaboration).
• Some controls are SQL-only. A 5 November 2026 update lists aggregation thresholds, comparison controls and differential privacy as controls that do not apply the same way to PySpark jobs (AWS document history).
• Provider matching is US-only. Provider-based matching with LiveRamp runs only in three US Regions (AWS pricing).
• Platform change is constant. AWS retired its first SQL engine on 17 December 2025, so older templates needed migration (AWS document history).

• Whether AWS supports the IAB Tech Lab PAIR or ADMaP protocols. AWS documentation does not mention either.
• The custom ML price above 100 billion records a month. The pricing page renders that value dynamically, and it did not display on 9 November 2026.
• The number of active collaborations or customers. AWS does not publish it.
• Compute prices outside US East. AWS states they vary by Region but shows examples only for US East.

1. AWS, "AWS Clean Rooms pricing" (accessed 9 November 2026). https://aws.amazon.com/clean-rooms/pricing/
2. AWS, "Document history for the AWS Clean Rooms User Guide" (entries 12 January 2023 to 5 November 2026; accessed 9 November 2026). https://docs.aws.amazon.com/clean-rooms/latest/userguide/doc-history.html
3. AWS, "AWS Clean Rooms Differential Privacy is now generally available" (10 April 2024). https://aws.amazon.com/about-aws/whats-new/2024/04/aws-clean-room-differential-privacy
4. AWS, "Analysis rules in AWS Clean Rooms", user guide (accessed 9 November 2026). https://docs.aws.amazon.com/clean-rooms/latest/userguide/analysis-rules.html
5. AWS, "AWS Clean Rooms Differential Privacy", user guide (accessed 9 November 2026). https://docs.aws.amazon.com/clean-rooms/latest/userguide/differential-privacy.html
6. AWS, "Cryptographic Computing for Clean Rooms", user guide (accessed 9 November 2026). https://docs.aws.amazon.com/clean-rooms/latest/userguide/crypto-computing.html
7. AWS, "AWS Clean Rooms customers" (accessed 9 November 2026). https://aws.amazon.com/clean-rooms/customers/
8. AWS, "AWS Clean Rooms API Reference" (last published 9 November 2026). https://docs.aws.amazon.com/clean-rooms/latest/apireference/Welcome.html
9. AWS, "AWS Clean Rooms" product page (accessed 9 November 2026). https://aws.amazon.com/clean-rooms/
10. US Securities and Exchange Commission, EDGAR submissions record for Amazon.com, Inc. (accessed 9 November 2026). https://data.sec.gov/submissions/CIK0001018724.json