DeepDAO is the industry's reference dataset for who the DAOs actually are — the analytics and discovery layer often described as the “CoinMarketCap for DAOs”. Where a governance interface like Tally / Cactus or Snapshot runs one DAO's votes, DeepDAO reads across the whole field: it aggregates treasury balances, proposal counts, voter participation and member overlap into one searchable place, so a claim like “the largest DAO treasuries” has a source you can check rather than a number someone asserted.

What it measures

For each organization DeepDAO indexes the numbers a DAO is usually judged by: treasury size and composition, cumulative and active proposal counts, unique voters and governance-token holders, and derived participation rates. It also tracks people across DAOs — the same wallet's activity in multiple organizations — which surfaces the delegate class and the voting-power concentration that a single DAO's dashboard hides. The result is one place to answer comparative questions: which DAOs are largest, which are actually active versus dormant, and how thin real participation tends to be.

Why it matters for the industry

DAO treasuries are on-chain and therefore public, but “public” and “legible” are not the same thing — raw addresses across a dozen chains are not an answer to “how big is this DAO and is it healthy?”. DeepDAO is the layer that turns that raw visibility into comparable figures, which is why its dashboards are routinely cited in governance research and by DAOs themselves. Its own tracker, readable again since the site returned, puts the aggregate treasury of every organisation it enriches at $13.6 billion across 2,461 organisations, of which $12.3 billion is liquid, out of 50,845 governances aggregated in total. Two treasuries clear a billion dollars — Mantle at $3.2 billion and Uniswap at $2.4 billion — followed by ENS ($819.2m), GnosisDAO ($714.2m), the Ethereum Foundation ($683.2m), Arbitrum ($622.7m), Optimism ($574.3m) and Sky ($375.7m). Every one of those figures is frozen at 24 November 2025 and none has moved since, which is the whole difficulty described below.

Limits of the data

The numbers are only as good as what's on-chain and labelled. Treasury figures are dominated by each DAO's own governance token, so a large headline treasury can be mostly an illiquid position that couldn't be sold without crashing its own price — a distinction covered in treasury management. Participation rates flatter or damn a DAO depending on whether you count token supply or holders. DeepDAO is the best available map, but a map of reported figures — reading it well means knowing what each number does and doesn't include.

Outage and return (2026)

DeepDAO's public site went dark in mid-2026 and came back. Both halves matter, because the site it came back as is not the site that went away.

The outage. Through July 2026 the hostnames deepdao.io, www.deepdao.io and app.deepdao.io carried no address record at either the Google or the Cloudflare public resolver (checked 31 July 2026) — a DNS-level absence rather than a server error or a bot block. The domain stayed registered through 5 July 2027 and api.deepdao.io kept answering 403 Forbidden to unkeyed requests, so it read as a withdrawal rather than a lapse. No shutdown was ever announced: DeepDAO's Substack has published nothing since April 2025.

The return, and what it actually is. Re-checked on 26 August 2026, deepdao.io resolves again — four CloudFront addresses at the Google resolver — and serves the full application: the organisations directory, treasury totals, the people graph, all of it. It also carries a banner the site did not have before, and the banner is the story:

“DeepDAO is not currently updating data. The last update was on 11.24.25. This site is a static snapshot meant to showcase the platform's breadth. Do not use it for research, analysis, or decision-making.”

So the dataset is live to read and disowned by the people who built it. It has been rolled back rather than resumed: everything on it is dated 24 November 2025, nine months before the site returned and six months older than the last Internet Archive capture taken while the site was still running. app.deepdao.io, the logged-in product, still does not resolve at all. What returned is a showcase, not a service.

For anyone citing it, this is the harder failure. A dead site announces itself: the link fails, a link checker flags it, the researcher goes elsewhere. This one answers HTTP 200, renders thousands of organisations and billions of dollars, and dates none of it anywhere a machine will look — the staleness is declared in on-page prose, which no link checker reads and no screenshot preserves. A figure lifted from it today looks freshly sourced and is nine months old. Treat every DeepDAO-sourced number, including the ones on this page, as dated 24 November 2025, and say so wherever it is quoted.

The sourcing gap it left is still open. Cross-DAO treasury and participation figures — in governance research, in press coverage, and on several pages of this wiki — traced back to DeepDAO as the one index reading across the whole field, and a frozen index does not restore that. The live substitutes each cover part of it: Tally / Cactus and Snapshot for one DAO's votes at a time, Boardroom for aggregated governance reads through its API only, its public app having broken in the same window (checked 22 August 2026), Dune and Token Terminal for treasury and revenue queries. None reproduces the people-and-membership graph across organisations that made DeepDAO distinct, and the snapshot reproduces it only as of last November.

How Caper approaches this

DeepDAO exists because most DAO activity is scattered across chains, forums and off-chain snapshots that need aggregating to be understood. Caper keeps the primary data in one place by construction: a caper's treasury, proposals, votes and executions all live in the same on-chain contract, so the figures an analytics tool would reconstruct are the native, canonical state rather than an after-the-fact estimate. The transparency DeepDAO works to surface for the wider industry is the default a caper starts from.

References

• DeepDAO, deepdao.io — read directly on 26 August 2026; every figure it serves is dated 24 November 2025 by the site's own banner. Note on the archive: the Internet Archive captures of 7 and 20 May 2026, cited here while the site was down, return only the application's JavaScript shell — “You need to enable JavaScript to run this app” — so they resolve at 200 without preserving any of the data they were cited for. A single-page application is not archivable by capture alone.
• DeepDAO, “Is This a Healthy DAO Treasury?” (methodology).
• Uniswap governance, DeepDAO Pro discussion.