What is a data union, and did any of them work?
A data union is a group of people pooling the data they individually produce, so that together they have something worth buying. One person's browsing history is worth almost nothing. Two hundred thousand people's, sold as a set with the members taking a share, is a product.
Pod has none of this. No pooling, no marketplace, no payment. This page explains a design that other people built and ran.
The mechanism
A smart contract holds the membership and the revenue split. People join, contribute a data stream, and the contract distributes proceeds when the pooled set is sold. A framework provides the contract and the plumbing so each union does not rebuild it.
The insight is straightforward and correct: the unit of value is the aggregate, not the individual record. Individual data pays almost nothing, which is the economics that ended an earlier marketplace. A union is the attempt to fix that by changing the unit rather than the price.
Did it work
Partly, and the honest answer needs both halves.
Six unions use the framework. The largest, a browser plugin that pays people for their web activity, reports over 200,000 members. A vehicle-data union pays drivers for streaming telemetry. A banking union does the same for transaction data. Three of them raised capital between them, which means somebody outside the crypto press believed the model.
That is a real result. Getting two hundred thousand people to install anything is hard, and getting them to install it in exchange for a share of a data sale is harder.
And the half that goes the other way
The layer underneath contracted, and the clearest evidence is the substrate's own reporting.
Its Q1 2026 transparency report records a major exchange delisting its token, and active staked node operators falling from 159 to 93 in a single quarter, attributed to operators becoming unprofitable after the price fell.
The unions depend on that network for transport and sales. When capacity is paid for in a token, the capacity is a function of a price nobody controls, and the engineering does not have to change for the arithmetic to.
What our own note claims, and why it is not on this page
The note says the framework's primary domain now redirects to an unrelated site, and reads that as operational decay.
The framework's site and its documentation both resolve and serve content. The lineage in that note covers several renamed entities, and we could not establish which one it refers to. So this page asserts decay about no named project, and reports instead the decline the substrate published about itself.
That is the seventh fact in our competitor notes to fail a check across two runs. Reporting a live project as dead is a worse error than missing that it died, and the asymmetry is why the page is written this way.
What it tells you about the pod model
Two things, and the second is the useful one.
A union is the opposite bet from custody. Its value comes from data being aggregated and sold; a pod's comes from data being minimised and withheld. You can want either. Wanting both in full is wanting aggregation and non-disclosure at once.
And it demonstrates that the payment problem is separable. The unions solved distribution and revenue sharing without solving storage, permissions or identity. Anybody claiming those four problems need one integrated answer should explain why, because the record here says otherwise.
Where we sit
Nowhere in this picture.
We avoid the substrate risk described above only because we have not built anything that could carry it. That is an absence, not a design decision.