Who buys a codebase for AI training, and on what terms
8 of 15 buyers, brokers and marketplaces for a company's codebase publish a price or range on the pages we read, read 22 October 2026.
Publishes a price or range
What it takes with the code
Licence shape
The figure it publishes
Takes closing or shut-down companies
The full board, one buyer per row
One buyer to a row. Expand an answer for its source sentence, word for word, plus the page and the read date; "not stated" means that page does not cover the question.
| Buyer | What it takes with the code | Licence shape | Publishes a price or range | The figure it publishes | Takes closing or shut-down companies |
|---|---|---|---|---|---|
| aitrainer.work | The ones that do look like production software: years of commits from more than one… | not statedNot addressed in aitrainer.work sell to AI labs page Not addressed in aitrainer.work enterprise page | price only on a quote or offer Not addressed in aitrainer.work enterprise page | not statedNot addressed in aitrainer.work sell to AI labs page Not addressed in aitrainer.work enterprise page | takes closing or shut-down companies |
| DataVendor | Private repositories and software projects: source, history, and context that models can… | not statedNot addressed in DataVendor home page | publishes a figure | Mobile SDK suite $95k 420k LOC iOS + Android | not statedNot addressed in DataVendor home page |
| FileYield | structured records, audio and speech, medical imaging, legal documents, geospatial data… | exclusive or non-exclusive, by deal | price only on a quote or offer | not statedNot addressed in FileYield how it works page | not statedNot addressed in FileYield how it works page |
we engage in partnerships that include delivery of content in a few key areas: closed and… | not statedNot addressed in Google AI: Partnerships to improve our AI products | not statedNot addressed in Google AI: Partnerships to improve our AI products | not statedNot addressed in Google AI: Partnerships to improve our AI products | not statedNot addressed in Google AI: Partnerships to improve our AI products | |
| Grepped | Real repositories and the history behind them: commits, reviews, and issues that explain… | not statedNot addressed in Grepped home page | publishes a figure | Your company’s data could be worth $20K–$5M. | not statedNot addressed in Grepped home page |
| License My Data | The major AI companies are licensing ordinary workflow data: correspondence, customer and… | exclusive or non-exclusive, by deal | publishes a figure | Placements in this category have reached six and seven figures | not statedNot addressed in License My Data home page |
| Maryah | Maryah licenses the emails, sales records and support tickets you choose to share, with… | non-exclusive licence | publishes a figure | LOWER RANGE $800K TO UPPER RANGE $1M | not statedNot addressed in Maryah business data licensing page |
| Mercor | Repositories, commits, PRs, and issues | not statedNot addressed in Mercor Data page | price only on a quote or offer | not statedNot addressed in Mercor Data page | not statedNot addressed in Mercor Data page |
| micro1 | Source code documentation, engineering workflows, bug tracking, product planning, and… | not statedNot addressed in micro1 data estimate page | publishes a figure | $1M+ Highly unique, proprietary operational data with significant value for frontier AI… | not statedNot addressed in micro1 data estimate page |
| Monda (Gulp Data) | Calculate the value of your company's data | not statedNot addressed in Monda data valuation calculator page | price only on a quote or offer | not statedNot addressed in Monda data valuation calculator page | not statedNot addressed in Monda data valuation calculator page |
| Nyne | File contents plus their metadata, authorship, timestamps, and access permissions. | not statedNot addressed in Nyne data partnerships page | publishes a figure | ESTIMATED PAYOUT RANGE $383,083 - $896,969 | not statedNot addressed in Nyne data partnerships page |
| OpenAI | We’re interested in large-scale datasets that reflect human society and that are not… | not statedNot addressed in OpenAI Data Partnerships announcement | not statedNot addressed in OpenAI Data Partnerships announcement | not statedNot addressed in OpenAI Data Partnerships announcement | not statedNot addressed in OpenAI Data Partnerships announcement |
| SimpleClosure AssetHub | The proprietary software your team built, including codebases, repositories… | exclusive or non-exclusive, by deal Not addressed in Introducing the Asset Hub (SimpleClosure blog) | price only on a quote or offer Not addressed in Introducing the Asset Hub (SimpleClosure blog) | not statedNot addressed in AssetHub product page Not addressed in Introducing the Asset Hub (SimpleClosure blog) | takes closing or shut-down companies |
| Troveo | Warehouses, schemas, and repos… | non-exclusive licence | publishes a figure | AI labs and startups are actively paying six figures for company data exports. | not statedNot addressed in Troveo business data page |
| Turing (Project Lazarus) | Source code, reviews, tickets, and the iteration history behind what shipped. | not statedNot addressed in Project Lazarus programme page | publishes a figure | Lazarus turns that operating record into a single payout of up to $1M, and handles… | takes closing or shut-down companies |
The 85 sentences the answers come from
- aitrainer.work, aitrainer.work enterprise page, read 2026-10-08, What it takes with the code:
AI labs pay to license how companies work: support tickets, CRM history, chat, email and code.
- aitrainer.work, aitrainer.work enterprise page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 5,924 characters long.
- aitrainer.work, aitrainer.work enterprise page, read 2026-10-08, Publishes a price or range: The page is silent here; it was read in full on 2026-10-08, 5,924 characters long.
- aitrainer.work, aitrainer.work enterprise page, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 5,924 characters long.
- aitrainer.work, aitrainer.work enterprise page, read 2026-10-08, Takes closing or shut-down companies:
Choose "A company that has closed or is winding down" when you register.
- aitrainer.work, aitrainer.work sell to AI labs page, read 2026-10-08, What it takes with the code:
The ones that do look like production software: years of commits from more than one person, issues and tickets alongside the code, design docs, tests and deploy history.
- aitrainer.work, aitrainer.work sell to AI labs page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 6,280 characters long.
- aitrainer.work, aitrainer.work sell to AI labs page, read 2026-10-08, Publishes a price or range:
It depends on the buyer and the asset.
- aitrainer.work, aitrainer.work sell to AI labs page, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 6,280 characters long.
- aitrainer.work, aitrainer.work sell to AI labs page, read 2026-10-08, Takes closing or shut-down companies:
Sell private code, data or a closed company's records to AI labs
- DataVendor, DataVendor home page, read 2026-10-22, What it takes with the code:
Private repositories and software projects: source, history, and context that models can learn from.
- DataVendor, DataVendor home page, read 2026-10-22, Licence shape: The page is silent here; it was read in full on 2026-10-22, 2,805 characters long.
- DataVendor, DataVendor home page, read 2026-10-22, Publishes a price or range:
Mobile SDK suite $95k 420k LOC iOS + Android
- DataVendor, DataVendor home page, read 2026-10-22, The figure it publishes:
Mobile SDK suite $95k 420k LOC iOS + Android
- DataVendor, DataVendor home page, read 2026-10-22, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-22, 2,805 characters long.
- FileYield, FileYield how it works page, read 2026-10-11, What it takes with the code:
structured records, audio and speech, medical imaging, legal documents, geospatial data, code repositories, conversational logs, financial data, video footage, and more.
- FileYield, FileYield how it works page, read 2026-10-11, Licence shape:
Choose your model: one-time sale, recurring license, exclusive deal, or non-exclusive.
- FileYield, FileYield how it works page, read 2026-10-11, Publishes a price or range:
Discuss price, license type, exclusivity, and delivery format directly with the buyer.
- FileYield, FileYield how it works page, read 2026-10-11, The figure it publishes: The page is silent here; it was read in full on 2026-10-11, 8,811 characters long.
- FileYield, FileYield how it works page, read 2026-10-11, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-11, 8,811 characters long.
- Google, Google AI: Partnerships to improve our AI products, read 2026-10-11, What it takes with the code:
we engage in partnerships that include delivery of content in a few key areas: closed and offline datasets, enhanced metadata and signals, as well as real-time structured factual information for verification purposes.
- Google, Google AI: Partnerships to improve our AI products, read 2026-10-11, Licence shape: The page is silent here; it was read in full on 2026-10-11, 3,997 characters long.
- Google, Google AI: Partnerships to improve our AI products, read 2026-10-11, Publishes a price or range: The page is silent here; it was read in full on 2026-10-11, 3,997 characters long.
- Google, Google AI: Partnerships to improve our AI products, read 2026-10-11, The figure it publishes: The page is silent here; it was read in full on 2026-10-11, 3,997 characters long.
- Google, Google AI: Partnerships to improve our AI products, read 2026-10-11, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-11, 3,997 characters long.
- Grepped, Grepped home page, read 2026-10-08, What it takes with the code:
Real repositories and the history behind them: commits, reviews, and issues that explain why code looks the way it does.
- Grepped, Grepped home page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 4,047 characters long.
- Grepped, Grepped home page, read 2026-10-08, Publishes a price or range:
Your company’s data could be worth $20K–$5M.
- Grepped, Grepped home page, read 2026-10-08, The figure it publishes:
Your company’s data could be worth $20K–$5M.
- Grepped, Grepped home page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 4,047 characters long.
- License My Data, License My Data home page, read 2026-10-08, What it takes with the code:
The major AI companies are licensing ordinary workflow data: correspondence, customer and support history, project records, finance and operations, production code
- License My Data, License My Data home page, read 2026-10-08, Licence shape:
Depending on the buyer, that is a non-exclusive licence you can place again and again, or a one-time exclusive buyout.
- License My Data, License My Data home page, read 2026-10-08, Publishes a price or range:
Placements in this category have reached six and seven figures
- License My Data, License My Data home page, read 2026-10-08, The figure it publishes:
Placements in this category have reached six and seven figures
- License My Data, License My Data home page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 13,970 characters long.
- Maryah, Maryah business data licensing page, read 2026-10-08, What it takes with the code:
Maryah licenses the emails, sales records and support tickets you choose to share, with identifying details removed and terms you approve.
- Maryah, Maryah business data licensing page, read 2026-10-08, Licence shape:
4× Average sales per dataset
- Maryah, Maryah business data licensing page, read 2026-10-08, Publishes a price or range:
LOWER RANGE $800K TO UPPER RANGE $1M
- Maryah, Maryah business data licensing page, read 2026-10-08, The figure it publishes:
LOWER RANGE $800K TO UPPER RANGE $1M
- Maryah, Maryah business data licensing page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 7,326 characters long.
- Mercor, Mercor Data page, read 2026-10-08, What it takes with the code:
Repositories, commits, PRs, and issues
- Mercor, Mercor Data page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 4,081 characters long.
- Mercor, Mercor Data page, read 2026-10-08, Publishes a price or range:
Compensation is based on data volume, the tools you connect, and the depth of your data.
- Mercor, Mercor Data page, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 4,081 characters long.
- Mercor, Mercor Data page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 4,081 characters long.
- micro1, micro1 data estimate page, read 2026-10-08, What it takes with the code:
Source code documentation, engineering workflows, bug tracking, product planning, and technical knowledge bases
- micro1, micro1 data estimate page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 8,917 characters long.
- micro1, micro1 data estimate page, read 2026-10-08, Publishes a price or range:
$100k+ Qualified Enterprise Data Partnerships
- micro1, micro1 data estimate page, read 2026-10-08, The figure it publishes:
$1M+ Highly unique, proprietary operational data with significant value for frontier AI development
- micro1, micro1 data estimate page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 8,917 characters long.
- Monda (Gulp Data), Monda data valuation calculator page, read 2026-10-08, What it takes with the code:
Calculate the value of your company's data
- Monda (Gulp Data), Monda data valuation calculator page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 2,720 characters long.
- Monda (Gulp Data), Monda data valuation calculator page, read 2026-10-08, Publishes a price or range:
With the calculator, you can get an estimation of how much your company's data assets could be worth.
- Monda (Gulp Data), Monda data valuation calculator page, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 2,720 characters long.
- Monda (Gulp Data), Monda data valuation calculator page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 2,720 characters long.
- Nyne, Nyne data partnerships page, read 2026-10-08, What it takes with the code:
File contents plus their metadata, authorship, timestamps, and access permissions.
- Nyne, Nyne data partnerships page, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 3,973 characters long.
- Nyne, Nyne data partnerships page, read 2026-10-08, Publishes a price or range:
ESTIMATED PAYOUT RANGE $383,083 - $896,969
- Nyne, Nyne data partnerships page, read 2026-10-08, The figure it publishes:
ESTIMATED PAYOUT RANGE $383,083 - $896,969
- Nyne, Nyne data partnerships page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 3,973 characters long.
- OpenAI, OpenAI Data Partnerships announcement, read 2026-10-08, What it takes with the code:
We’re interested in large-scale datasets that reflect human society and that are not already easily accessible online to the public today.
- OpenAI, OpenAI Data Partnerships announcement, read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 6,041 characters long.
- OpenAI, OpenAI Data Partnerships announcement, read 2026-10-08, Publishes a price or range: The page is silent here; it was read in full on 2026-10-08, 6,041 characters long.
- OpenAI, OpenAI Data Partnerships announcement, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 6,041 characters long.
- OpenAI, OpenAI Data Partnerships announcement, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 6,041 characters long.
- SimpleClosure AssetHub, Introducing the Asset Hub (SimpleClosure blog), read 2026-10-08, What it takes with the code:
Through the Asset Hub, founders can license their source code to vetted buyers.
- SimpleClosure AssetHub, Introducing the Asset Hub (SimpleClosure blog), read 2026-10-08, Licence shape: The page is silent here; it was read in full on 2026-10-08, 5,475 characters long.
- SimpleClosure AssetHub, Introducing the Asset Hub (SimpleClosure blog), read 2026-10-08, Publishes a price or range: The page is silent here; it was read in full on 2026-10-08, 5,475 characters long.
- SimpleClosure AssetHub, Introducing the Asset Hub (SimpleClosure blog), read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 5,475 characters long.
- SimpleClosure AssetHub, Introducing the Asset Hub (SimpleClosure blog), read 2026-10-08, Takes closing or shut-down companies:
The Asset Hub is a platform where founders closing their companies can sell the assets they've built
- SimpleClosure AssetHub, AssetHub product page, read 2026-10-08, What it takes with the code:
The proprietary software your team built, including codebases, repositories, documentation, testing, and related technical assets.
- SimpleClosure AssetHub, AssetHub product page, read 2026-10-08, Licence shape:
AssetHub helps assess, package, and connect SourceCode and WorkspaceData with AI buyers so founders can generate value through sale or licensing opportunities.
- SimpleClosure AssetHub, AssetHub product page, read 2026-10-08, Publishes a price or range:
We review your assets to determine what can be purchased and what they're worth.
- SimpleClosure AssetHub, AssetHub product page, read 2026-10-08, The figure it publishes: The page is silent here; it was read in full on 2026-10-08, 6,730 characters long.
- SimpleClosure AssetHub, AssetHub product page, read 2026-10-08, Takes closing or shut-down companies:
Founders working with SimpleClosure can recover it through two distinct ways: the codebases they built, and the workspace data from how their teams actually worked.
- Troveo, Troveo business data page, read 2026-10-08, What it takes with the code:
Warehouses, schemas, and repos — the deepest, richest layer of the stack
- Troveo, Troveo business data page, read 2026-10-08, Licence shape:
Most data buyers pay once and disappear. A Troveo partnership is different: every new product is a new sale that you share in.
- Troveo, Troveo business data page, read 2026-10-08, Publishes a price or range:
AI labs and startups are actively paying six figures for company data exports.
- Troveo, Troveo business data page, read 2026-10-08, The figure it publishes:
AI labs and startups are actively paying six figures for company data exports.
- Troveo, Troveo business data page, read 2026-10-08, Takes closing or shut-down companies: The page is silent here; it was read in full on 2026-10-08, 4,973 characters long.
- Turing (Project Lazarus), Project Lazarus programme page, read 2026-10-22, What it takes with the code:
Source code, reviews, tickets, and the iteration history behind what shipped.
- Turing (Project Lazarus), Project Lazarus programme page, read 2026-10-22, Licence shape: The page is silent here; it was read in full on 2026-10-22, 4,212 characters long.
- Turing (Project Lazarus), Project Lazarus programme page, read 2026-10-22, Publishes a price or range:
Lazarus turns that operating record into a single payout of up to $1M, and handles everything else.
- Turing (Project Lazarus), Project Lazarus programme page, read 2026-10-22, The figure it publishes:
Lazarus turns that operating record into a single payout of up to $1M, and handles everything else.
- Turing (Project Lazarus), Project Lazarus programme page, read 2026-10-22, Takes closing or shut-down companies:
For operating and winding-down companies
Documents we read but did not count
We read 4 named documents that turned out not to be what this board asks about; why is given for each below. They count toward the reading, not the answers.
- HUD, best platforms for selling your codebase (2026): the charter admits it for: press and comparison pages: name buyers and report deals for discovery and dating only; never a buyer's price
- FileYield, data buyers guide: the charter admits it for: press and comparison pages: name buyers and report deals for discovery and dating only; never a buyer's price
- Let's Data Science: Turing buys failed-startup codebases: the charter admits it for: press and comparison pages: name buyers and report deals for discovery and dating only; never a buyer's price
- RuntimeWire: Google pilot buying Play Store developers' code: the charter admits it for: press and comparison pages: name buyers and report deals for discovery and dating only; never a buyer's price
The method in full
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: what it takes with the code, licence shape, publishes a price or range, the figure it publishes, takes closing or shut-down companies.
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.
15 buyers, 17 pages.
Not read yet, so not counted: 1 named document.
Where we looked (22)
- DataVendor (marketplace: codebases, tasksets, RL environments) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- SimpleClosure AssetHub (closing companies' codebases and workspace data) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- SimpleClosure, introducing the Asset Hub (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Grepped (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Troveo business data (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Mercor Data (connects GitHub: repositories, commits, PRs, issues) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Nyne data partnerships (payout estimator) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- aitrainer.work, sell code, data or RL environments (the publisher's buyer, broker and marketplace pages, cited for reporting)
- micro1 data estimate (tiers) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- License My Data (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Maryah business data licensing (the publisher's buyer, broker and marketplace pages, cited for reporting)
- aitrainer.work Enterprise (names Turing's up-to-$1M single payment) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- OpenAI Data Partnerships (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Monda / Gulp Data valuation calculator (the publisher's buyer, broker and marketplace pages, cited for reporting)
- HUD, best platforms for selling your codebase (2026) (read, set aside)
- FileYield, data buyers guide (read, set aside)
- Let's Data Science: Turing buys failed-startup codebases (read, set aside)
- RuntimeWire: Google pilot buying Play Store developers' code (read, set aside)
- FileYield, How it works (own terms and fees) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Turing Research, Project Lazarus (own programme page) (the publisher's buyer, broker and marketplace pages, cited for reporting)
- Turing, AGI Advance newsletter 44 (own launch note for Project Lazarus) (not yet read)
- Google AI, Partnerships to improve our AI products (own page) (the publisher's buyer, broker and marketplace pages, cited for reporting)
Cite it or embed it
8 of 15 buyers, brokers and marketplaces for a company's codebase publish a price or range on the pages we read, read 22 October 2026.
What it takes with the code, by buyer: stated 15 of 15 (15 buyers, read 22 October 2026).
Licence shape, by buyer: not stated 10 of 15, exclusive or non-exclusive, by deal 3 of 15, non-exclusive licence 2 of 15 (15 buyers, read 22 October 2026).
Publishes a price or range, by buyer: publishes a figure 8 of 15, price only on a quote or offer 5 of 15, not stated 2 of 15 (15 buyers, read 22 October 2026).
The figure it publishes, by buyer: stated 8 of 15, not stated 7 of 15 (15 buyers, read 22 October 2026).
Takes closing or shut-down companies, by buyer: not stated 12 of 15, takes closing or shut-down companies 3 of 15 (15 buyers, read 22 October 2026).
Credit as: "Who buys a codebase for AI training, and on what terms", updated 2026-10-22, https://sellcodebasetoai.com/datasets/codebase-buyer-terms-and-prices/.
Save the CSV (85 answers, compiled 2026-10-22)