A few years ago, a CTV advertiser who wanted to match its customer file against a streaming audience usually did it inside someone else’s environment: a clean room vendor, a cloud provider, or a retail media network’s data platform. By 2026, the big streaming platforms each offer a clean room under their own name. Roku has one. Disney has one. Netflix put data clean room partnerships in its 2026 upfront pitch. The headline reads like a move away from renting Amazon’s or LiveRamp’s tools. The details show something more specific: the platforms now control the product, the data rules, and the commercial terms, while most of the technology underneath is still licensed. For buyers, that difference affects how comparable the results are across platforms.

What the Platforms Have Actually Built

Roku. Roku launched its clean room in April 2022 as a privacy-focused environment where advertisers and agencies match encrypted first-party data against Roku’s audience and viewing data for planning and measurement. Launch users included Omnicom Media Group, Dentsu, and Horizon Media. The clean room was built on Snowflake. In January 2025, Roku extended the idea into the Roku Data Cloud, which delivers streaming data drawn from its automatic content recognition (ACR) and first-party signals through agency and measurement partners, including Omnicom’s Omni, PMG’s Alli, Innovid, and iSpot, with a DSP integration planned. Roku said it would not charge SaaS fees. Its advertising leadership framed the company as being in the media business rather than the data business, meaning the data supports media sales instead of being sold on its own.

Disney. Disney’s clean room dates to 2021 and has been one of the most visible in streaming. Its self-service portal was built with Habu. In January 2024, Disney expanded access to more cloud service providers, naming AWS and Google Cloud, so advertisers could connect from the environment where their data already sits. That same month, LiveRamp completed its roughly $200 million acquisition of Habu. The clean room technology behind Disney’s portal therefore now belongs to LiveRamp.

Netflix. Netflix’s ad business moved quickly from launch to self-built ad tech. Its 2026 upfront announcement said it had integrated with Snowflake and Amazon Web Services for secure collaboration through data clean room technology and would add InfoSum by the end of 2026. It also listed planning and data clean room work with Dentsu, Horizon, Omnicom, PMG, and Tinuiti, alongside new Audience Insights and Reach Curve APIs.

Other TV platforms are following a similar pattern. Samsung Ads, for example, has partnered with Snowflake’s data clean rooms for outcome-based campaigns.

Owning the Product vs. Owning the Stack

Put these together and a consistent structure appears. The platform owns:

  • The data. First-party subscriber, viewing, and ACR data that no one else has.
  • The rules. Which queries are allowed, which outputs are returned, the aggregation thresholds, and how long advertiser data is kept.
  • The commercial terms. Whether clean room access is bundled with media spend, priced separately, or free.
  • The relationship. Advertisers work with the platform’s sales and data science teams, not a third-party vendor’s.

The technology layer, meaning secure compute, privacy-preserving joins, and cross-cloud connectivity, is still largely supplied by Snowflake, AWS, LiveRamp (through Habu), and InfoSum. Amazon participates in two roles at once: AWS Clean Rooms is general infrastructure that Netflix integrates with, while Amazon Marketing Cloud is Amazon Ads’ own clean room for its advertising and retail data.

So the accurate description is not that CTV platforms stopped renting clean room technology. It is that they stopped letting the clean room define the relationship. A streaming platform that controls its own clean room decides what advertisers can learn from its data, and it can tie that access to media buying on its own inventory.

Why the Platforms Want This

The motive is plain from how the offerings are sold. Clean room access supports media spend. Roku’s decision not to charge SaaS fees for its data offering makes that explicit. The data exists to make Roku inventory easier to plan, buy, and prove, not to be a separate data product. Disney’s long record of clean room adoption and Netflix’s push on planning APIs work the same way. If a streamer’s data can only be joined with an advertiser’s customer file inside the streamer’s environment, measurement of that streamer’s ads happens on its terms.

There is also a defensive reason. Retail media networks, with Amazon the largest, have trained advertisers to expect closed-loop measurement: exposure matched to purchase data in a controlled environment. Streaming platforms without their own collaboration layer would be selling reach while retail media networks sell outcomes. Building a branded clean room is how a streamer competes in the outcomes pitch.

What It Costs Buyers: Fragmentation

The downside for advertisers is the same one that shows up whenever each seller builds its own measurement environment. A brand buying across Roku, Disney, Netflix, and a few FAST channels may now be running separate matches in separate clean rooms, each with different privacy thresholds, match methods, output formats, and attribution rules. Even when two platforms run on the same underlying technology, their query rules and outputs are set separately.

That makes cross-platform comparison hard. A conversion lift reported by one platform’s clean room and a lift reported by another’s are not automatically comparable, because the join keys, holdout designs, and minimum audience sizes may differ. Deduplicated reach across platforms is harder again, since each clean room is built to answer questions about one platform’s audience, not about overlap with competitors.

How to Work With Platform Clean Rooms

Platform clean rooms are worth using. They give access to data that exists nowhere else. The practical approach is to use them without letting any one of them become the only measure of success.

  • Standardise your side of the join. Use the same customer file, the same identity keys or tokens, and the same conversion definitions in every platform’s clean room, so the differences you see come from the platforms, not your inputs.
  • Ask what technology each one runs on, and what the platform has changed. Two clean rooms on the same provider can still apply different rules. Get the aggregation thresholds and permitted outputs in writing.
  • Keep a cross-platform measurement layer. An outside measurement partner or your own environment should hold the deduplicated view of reach and outcomes across platforms. Platform clean rooms answer platform questions.
  • Clarify data retention and reuse. Know how long your uploaded data stays in each environment and whether the platform can use match results for anything beyond your campaign.

Streaming platforms have made the clean room part of the media offer, and that is unlikely to reverse. The platforms own the data and the rules, and the vendors supply the compute. Buyers who keep their inputs consistent, get each platform’s rules in writing, and hold their own cross-platform view will get the value of each walled environment without letting any one of them grade its own results.