Methodology
Updated 27 Oct 2026. 66 buyers, 83 list pages.
Why this exists
Every "how to sell your data to AI companies" page we could find is written by a broker or an affiliate, and each one points at its own offer. This site lists the buyers side by side with what they publish: the price anchor, the payout model and the exclusivity term, each with a source and a date. Nobody pays to appear here or to rank.
How we find buyers
The first list came from buyers' own pages for sellers, press coverage of AI data licensing deals, funding announcements and the public directories of data marketplaces. New buyers come from the same sources and from the submit form. A buyer is listed when it has a public page saying what it buys and at least one published term: a price, a payout model or an exclusivity term. Listing is free.
What we record
One record per buyer, with these fields. Every fact that is not derived carries a URL and the date we read it.
| Field | What it means | Where it comes from |
|---|---|---|
slug | The address of the profile, made from the name. | Derived |
name | The name the buyer trades under. | Buyer's site |
website | The page a seller starts from. | Buyer's site |
hq_country | Country of the head office, as a two-letter code; a buyer with no stated office is shown as worldwide. | Buyer's site or company register |
buyer_type | Broker, marketplace, licensing rail, collective, direct buyer, data collector and the like. | Buyer's site |
data_types | The kinds of data the buyer says it takes. | Buyer's site |
seller_types | Who the buyer says it buys from: creators, companies, studios, developers. | Buyer's site |
own_catalog | Whether the buyer licenses a catalog it owns itself. | Buyer's site |
buyers_named | AI labs or companies named as customers, marked "reported" when the source is press. | Buyer's site or press |
labs_named | Whether at least one lab customer is named. | Derived from buyers named |
payout_models | How the seller is paid: upfront, revenue share, per token, per hour, bounty and so on. | Buyer's site |
price_anchor | The published price, quoted verbatim with its date. | Buyer's site or press |
rate_published | Whether the anchor is a rate per unit of data. | Derived from the anchor |
exclusivity | Exclusive, non-exclusive or varies by deal. | Buyer's terms |
minimums | Minimum volume, size or quality the buyer asks for. | Buyer's site |
referral_program | Whether the buyer pays for introductions. | Buyer's site |
referral_terms | What the referral program pays, as published. | Buyer's site |
payouts_reported | Whether total payouts to sellers have been reported. | Buyer's site or press |
funding_total_usd | Total funding raised, in US dollars. | Press or company announcements |
founded_year | Year the company was founded. | Company site or register |
status | Active, acquired or withdrawn. | Press or company site |
editor_pick | Whether the editors chose the buyer as an editor's pick. | Editors |
editor_pick_hubs | ||
first_party | Whether the publisher owns the listing. False on every record. | Publisher |
summary | Two or three sentences on what the buyer takes and how it pays. | Written from the sources |
summary_edited | Whether a person has edited the generated summary. | Editors |
verdict | One line on why the buyer is listed. | Written from the sources |
notes | A short factual note shown on the card. | Written from the sources |
sources | Every URL a fact comes from, each with the date we read it. | Research |
price_anchor_source | The URL and read date of the price anchor. | Research |
last_verified | The last day every fact on the record was checked. | Research |
record_flags | Internal markers for facts waiting on a second source. | Research |
How lists are ordered
Every list on the site uses the same order:
- Editor's picks for the page's data type first.
- Then, on a data type list, buyers that name that data type explicitly.
- Then buyers with a published price anchor.
- Then name, A to Z.
That is the whole rule. Nothing in a record is a rank or a score, and nobody can pay to move up.
Editor's picks
The editors chose 26 of the 66 buyers as picks: buyers that publish their terms clearly, have a track record of paying sellers, or cover a data type few others do. The editors choose picks per data type, so a pick moves a buyer to the top of the lists for the data types it was chosen for and nothing else. The current picks:
- Calliope Networks
- Cloudflare Pay Per Crawl
- Copyright Clearance Center
- CSIMarket
- Defined.ai
- ElevenLabs Voice Library
- Freepik / EyeEm
- GoPro AI Training Program
- Human Archive
- micro1
- Mozilla Data Collective
- Photobucket
- ProRata.ai
- Sell My Codebase
- SimpleClosure Asset Hub
- SourceAudio + Musical AI
- Troveo AI
- Wirestock
- Created by Humans
- David AI
- Gradient Health
- Luel
- Mecka AI
- Protege
- Segmed
- TollBit
When a page exists
A list page exists only when it holds 3 or more buyers. A page whose list would be the same as its parent's is not built either; its address redirects to the parent. Price pages are the one exception to the duplicate rule, because they show the anchors as a table.
On this build 83 list pages are built. 792 candidate addresses redirect because they would hold fewer than 3 buyers, 2 because they would repeat their parent, and 38 are other phrasings of a built page, 832 redirects in all. Every redirect lands on a built page in one step.
Sources and dates
Every fact needs a URL and the date we read it, shown on the profile. Where a fact comes from press rather than the buyer, the profile says "reported". There are no ratings on this site: we do not collect reviews, we do not copy ratings from elsewhere and we do not score buyers. Each record carries the day it was last verified.
Prices
A price anchor is the figure a buyer publishes for what it pays: a rate per minute, per image, per token or per hour, a revenue share, or a deal size. It is quoted verbatim, with the date we read it and the page it is on; amounts are never converted or rounded. 26 of the 66 buyers publish one.
Each profile also shows seven "terms published" facts: price anchor, payout model, exclusivity, minimums, named lab customers, payouts in press and referral program. Each is shown as Yes or No with its source. They are never counted into a score and never order a list.
The questionnaire
Find your buyer asks three questions: what data you have, who you are, and what matters most (highest price, fastest payout, keeping rights non-exclusive, or no minimums).
Only buyers that take your data type are considered. Each gets 3 points for the data type, 2 if it buys from sellers like you, 2 if it meets your priority (a published rate, an upfront payout, non-exclusive terms or no minimums), and 1 if it is an editor's pick. Ties are broken by a number between 0 and 1 that changes once an hour, so everyone asking the same thing in the same hour sees the same three. The score is never shown.
Badges
Badges are worked out from the records on each build. Nobody applies for one and nobody pays for one.
- Editor's Pick: Chosen by the editors of Who Buys My Data from the buyers we list.
- Transparent Terms: Publishes its price anchor, payout model and exclusivity terms.
- Non-Exclusive Buyer: Licenses data on non-exclusive terms, so the owner can license it again.
- Publishes Rates: Publishes a rate per unit of data.
- Pays Referrals: Runs a referral program that pays for introductions.
- Top data buyer of the year, one badge per data type: the first three buyers on a built data type list in the order above.
Disclosure
No listing on this site is owned by or affiliated with the publisher. The publisher, 2V Automation, builds and runs this site and has no product in this directory. Where a referral link exists it is marked on the card with "Referral link, we may earn a fee".
Corrections
We write every listing from sources anyone can check. A company whose terms have changed updates them at the source and sends the links to [email protected]. The listing follows the record.
Changelog
| Date | Change |
|---|---|
| 26 Oct 2026 | Site launched with 66 buyers and 83 pages |