01 / 03
Describe the data, its size and who holds the rights to it.
Prices for data, read from the pages that publish them
21 pages on 19 data buyers, brokers and marketplaces are on the board, read up to 2026-10-11. The board reads 9 of them as publishing a price or a range, Google's as a winning bid in a bankruptcy court's records rather than a figure it advertised, and 6 of the 19 have a page that calls the figure it gives an estimate or illustration. So the first job in valuing a dataset is knowing which kind of number you are looking at. A second board, of public AI data licensing deals, finds a published price for 10 of the 22.
Source: What buyers and brokers publish as the price of data: every row, with its source
01 / 03
Describe the data, its size and who holds the rights to it.
02 / 03
We set it against what buyers and brokers have published for data of that kind, and say which of those figures are estimates and which are offers.
03 / 03
If it is worth taking to buyers, we explain how a licence would be run and what we would charge.
Our fee, agreed in writing before anything is shared: 12.5% of the deal when we act as the seller's exclusive agent, or about 10% when the data joins a bundle or pool. Any fee a buyer pays us is credited to the seller. No buyer, broker or marketplace pays to appear on this board.
| Buyer, broker or marketplace | Publishes a price or range? (the board's reading) | The figure, in the source's words where one is stated | Kind of figure (the board's reading) | Source |
|---|---|---|---|---|
| aitrainer.work | price only on a quote or offer | not stated | not stated | aitrainer.work sell to AI labs page |
| Created by Humans | price only on a quote or offer | not stated | not stated | Created by Humans home page (its own page, listed on this site's dataset page) |
| Defined.ai | publishes a figure | Elite Plan: 3,125 hours for €500,000 (55% discount) | a quote or guaranteed minimum | Defined.ai Data Access Plan press release |
| FileYield | price only on a quote or offer | not stated | not stated | FileYield data buyers guide |
| publishes a figure | Total Consideration (USD): $10,000,000 | a quote or guaranteed minimum | Spirit Airlines bankruptcy docket, Doc 1463, notice of auction results | |
| Grepped | publishes a figure | Your company’s data could be worth $20K–$5M. | an illustration or estimate | Grepped home page (its own page, listed on this site's dataset page) |
| License My Data | publishes a figure | Placements in this category have reached six and seven figures | an illustration or estimate | License My Data home page (its own page, listed on this site's dataset page) |
| Maryah | publishes a figure | LOWER RANGE $800K TO UPPER RANGE $1M | an illustration or estimate | Maryah business data licensing page |
| Mercor | price only on a quote or offer | not stated | not stated | Mercor Data page |
| micro1 | publishes a figure | $100K-$2M+ for approved data packages | not stated | micro1 company referral page |
| Monda (Gulp Data) | price only on a quote or offer | not stated | an illustration or estimate | Monda data valuation calculator page |
| Nyne | publishes a figure | ESTIMATED PAYOUT RANGE $383,083 - $896,969 | an illustration or estimate | Nyne data partnerships page |
| OpenAI | not stated | not stated | not stated | OpenAI Data Partnerships announcement |
| Opendatabay | publishes a figure | not stated | a quote or guaranteed minimum | Opendatabay marketplace home page (its own page, listed on this site's dataset page) |
| Protege | not stated | not stated | not stated | Protege for data providers page |
| Shutterstock | price only on a quote or offer | not stated | not stated | Shutterstock data licensing page |
| SimpleClosure AssetHub | price only on a quote or offer | not stated | not stated | AssetHub product page |
| Snowflake Marketplace | not stated | not stated | not stated | Snowflake Marketplace consumer page |
| Troveo | publishes a figure | AI labs and startups are actively paying six figures for company data exports. | an illustration or estimate | Troveo business data page |
Each buyer's own buyer, broker and marketplace pages were read through estate-fetch and stored whole (sha256 on every row). For every page, five questions were answered from its own words: publishes a price or range for data, the price or range it publishes, unit the figure is quoted in, what it says moves the price, kind of figure. Every stated answer carries the sentence it rests on, checked against the stored text; 'not stated' means the page read does not address the question. A buyer with several pages reads 'varies' where they disagree. 19 buyers, 21 pages.
Cite: “What data buyers and brokers publish as a price”, Dataset Valuation, https://datasetvaluation.com/ (updated 2026-10-11).
Every figure sourced and dated.
What data buyers and brokers publish as a price, which of those figures are estimates and which are quotes, and what that tells you about your own data.
What data buyers and brokers say moves the price of a dataset sold to AI labs, in the words of the pages we read, and the units their figures come in.
What a data valuation calculator or estimate form on a buyer's page returns, what those pages say their range is, and what decides the final figure.
How much AI companies pay for data where a figure is public: 10 of 22 licensing deals priced, and per-unit prices for video, audio, photos, books and code.
AI data licensing deals from the parties' own announcements, filings and court records: who bought what, the year, and the price where one was published, 10 of 22.
No published figure can answer that for a specific dataset. What the board shows is this: it reads 9 of the 19 buyers, brokers and marketplaces on it as publishing a price or range, Google's among them as a winning bid in a bankruptcy court's records rather than a range it advertised, and 6 of the 19 have a page that calls the figure it gives an estimate or illustration.(What buyers and brokers publish as the price of data: every row, with its source)
Mostly not. Of the 19 buyers on the board, read up to 2026-10-11, the board reads the figure for 3 as a quote or guaranteed minimum, 6 have a page calling theirs an estimate or illustration, and for the other 10 it records the kind as not stated. Some of those estimate pages say in their own words that the figure is not an offer.(What buyers and brokers publish as the price of data: every row, with its source)
For 14 of the 19 buyers on the board, a page on the board says what moves the price. The buyers' and brokers' own pages name factors such as volume, years of history, headcount, the systems connected, rights and quality. Google's is a bankruptcy court declaration instead, saying the auction ran until every bidder had a full chance to improve its offer. The board on our dataset page quotes the sentence each answer rests on.(What buyers and brokers publish as the price of data: every row, with its source)
Per company, for the largest group: 6 of the 8 figures on the board are quoted per company or data package. The rest are quoted per hour of speech data or come with no unit on the page the board read, and the unit column on the board says which.(What buyers and brokers publish as the price of data: every row, with its source)
Because pricing only on a quote is part of this market: the board reads 7 of the 19 buyers on it as giving a price only on a quote or offer, and for 3 the page on the board says nothing about price. Leaving them out would make a published price look more common than it is.(What buyers and brokers publish as the price of data: every row, with its source)
Where a deal's figure is public, it runs to millions of dollars. On our deal index, read up to 31 October 2026, 10 of 22 AI data licensing deals carry a published price: Reddit and Google at about $60 million a year by a press report, News Corp and OpenAI at $250 million over five years, and Google's $10,000,000 winning bid for Spirit Airlines' de-identified data at a bankruptcy auction. For 11 of the rest, nothing we read gives a price, and one has only a price per book.(AI data licensing deals on our index: every deal, with its source)
This site's own datasets, every figure quoted from its source and dated: