See if your data qualifies

Guides · Published

Best AI training data licensing companies (2026)

Six AI training data licensing options for 2026, judged on cost, control, ownership and evidence. Marketplaces, media suppliers and a free matchmaker compared.

The best AI training data licensing option in 2026 depends on what you have to license. For internal business records such as support tickets, CRM history and project files, Smelted is a free matching service that finds buyers for your company. For licensed photos, video and media, Troveo, Depositphotos and Versos are the names to compare.

We make Smelted, so it's on this list; we judged it on the same criteria as the others.

How we picked

Every entry is judged on five criteria:

  • What data it licenses: internal business records, media, or collected web data.
  • Cost to the data owner: whether the owner pays anything to take part.
  • Control: whether the owner chooses what is shared and with whom.
  • Ownership terms: whether the owner keeps the data and what the buyer receives.
  • Evidence: whether the entry's claims can be checked against a page we reviewed.

We also looked at market context. The AI training dataset market was worth about $2.82 billion in 2024 and is projected to reach $9.58 billion by 2029, a compound annual growth rate of roughly 27.7%, according to Ficstar. Epoch AI has projected that the stock of high-quality public text data could be effectively exhausted for training between 2026 and 2032, one reason labs increasingly license non-public data, as Troveo notes.

Figures on this page come from the sources linked in each entry. Where an entry does not state a term, we say so rather than fill the gap.

The list

1. Smelted

Smelted is a free matching service that connects US companies holding internal business records with AI data buyers. It finds buyers that fit a company, gives a ballpark range and introduces the company to two or three matching buyers. The buyer pays Smelted's fee, so the company pays nothing (Smelted).

Best for: founders and C-level leaders at US companies with 50 to 500 employees that hold years of records in tools such as Zendesk, Jira/GitHub, Salesforce/HubSpot, Bullhorn, AutoCAD/Revit or Benchling.

Strengths: the company licenses a copy and keeps ownership. It chooses what is included and which buyers it meets. Names, contact details and IDs are removed before anything is shared. Nothing is paid up front, and the company does not give up equity.

What to watch: value depends on headcount, years of records and how connected the data is, so there is no price until a buyer reviews a sample. Payment usually arrives a few weeks to two months after signing, so plan on one to three months from first call to payment. Buyers who take smaller companies pay less, so companies under 30 employees should expect a smaller offer. Smelted covers business records, not media files.

2. Troveo

Troveo runs a licensing marketplace where rights holders grant AI training rights across video, audio, gaming, robotics and business data (Troveo).

Best for: rights holders with video, audio or media libraries who want a marketplace rather than individual contracts.

Strengths: the marketplace model addresses the problem of many small licensors. Troveo reports that it has paid more than 20 million dollars to over 7,000 rights holders. Its own deal data also gives a useful benchmark: reported deals range from a few million dollars a year for mid-size publishers to $250 million over five years for News Corp and OpenAI (Quartz).

What to watch: the payout figure comes from Troveo itself and is not independently verified. The pages we reviewed do not set out the seller's fee or the terms for each deal.

3. Opendatabay

Opendatabay publishes a 2026 guide that compares commercial and general AI training licences for both buyers and sellers (Opendatabay).

Best for: teams that need to decide which licence type fits their data before they list or buy.

Strengths: the guide sets out licence types side by side, which helps a seller understand what they are granting.

What to watch: this is a guide rather than a record of completed deals. Check the platform's current seller terms directly before you list anything.

4. Depositphotos

Depositphotos writes about rights-cleared AI data licensing, meaning datasets where the rights to each file are clear and documented (Depositphotos).

Best for: image and stock media buyers who need clear rights before training.

Strengths: the article focuses on reducing legal risk in AI projects, which matters for buyers who must show clear provenance.

What to watch: the material is about media, not business records. We did not review its licence terms, so confirm them before you sign.

5. Versos

Versos publishes a list of direct suppliers of licensed AI training data, naming media companies such as Getty Images and Shutterstock (Versos).

Best for: buyers who want to approach named media suppliers directly.

Strengths: the list is a clear starting point for finding established media sources.

What to watch: it is a list of suppliers, not a marketplace a new seller can join. The entries reflect Versos's own selection.

6. Ficstar

Ficstar is a data services firm that collects web and product data at scale and offers AI training data as one of its services (Ficstar).

Best for: buyers who need collected web or pricing data rather than licensed company records.

Strengths: Ficstar says it has run managed web data collection for more than 20 years and processes over a billion product prices a month. It names SOC 2, ISO 27001 and GDPR alignment as common requirements in regulated sectors. These are its own claims.

What to watch: Ficstar is a services provider, not a route for a company to license its own records. Its page also warns that a headline accuracy figure can mislead, so ask how a provider performs on hard cases.

Comparison

OptionData licensedCost to data ownerControl over what's sharedOwnership termsEvidence we could check
SmeltedInternal business recordsFree to the company; buyer pays the feeCompany chooses content and buyers; names and IDs removedCompany keeps ownership; licence to a copyCompany's own description of its service
TroveoVideo, audio, gaming, robotics, business dataNot stated on the page reviewedPer-deal terms not statedLicence for AI trainingSelf-reported payout figures
OpendatabayLicensed AI training dataNot statedNot statedCompares commercial and general licencesPublished guide
DepositphotosRights-cleared images and mediaNot statedNot statedRights-cleared datasetsPublished article
VersosMedia from named suppliersNot applicable (supplier list)Not applicableNot statedPublished list
FicstarCollected web and product dataServices contract, not statedNot statedNot statedSelf-reported client and volume claims

FAQ

What is AI data licensing?

AI data licensing is an agreement that lets a company use its data to train an AI model, usually for a fee. The licence can cover a copy of the data while the owner keeps ownership of the original. Terms vary, so read the licence type before you sign. Opendatabay compares the main commercial and general licence types.

Where can I get training data for AI?

Buyers usually source training data from marketplaces, direct media suppliers, or data services firms. The right source depends on the data type. Media buyers often start with direct suppliers such as those listed by Versos. Buyers who need web data can look at services like Ficstar.

How do I sell internal business data to AI companies?

Start with the 2-minute form at smelted.ai/get-started. Smelted then looks for buyers that fit your records and introduces you to two or three of them. Your company chooses what is included, and names, contact details and IDs are removed before sharing. Companies with 50 or more employees can also book a call at calendly.com/smelted/intro-call.

It can be, but whether a particular licence is lawful depends on your contracts, your customers' terms and your jurisdiction. Smelted does not give legal or compliance approval, so check with your own counsel before you grant any licence. Depositphotos discusses how rights-cleared datasets can reduce legal risk for buyers.

Sources