AI data buyers: profiles of the companies that buy company data

The six types of buyer

The buyers fall into six types. Each one buys differently, pays differently and carries a different risk. The same groups are used in our comparison of data monetization companies.

• AI data vendors

  22 companies

  Acquire or license company data and supply it onward, mostly to AI labs, on their own account.

• Data marketplaces

  12 companies

  List or license a seller's data to buyers and take a cut or revenue share, including licensing agents.

• Wind-down buyers

  3 companies

  Buy or broker the data, code and archives of companies that are closing or in restructuring.

• AI labs

  8 companies

  Model developers that license data directly, mostly from publishers and named data partners, and buy through vendors.

• Brokers

  9 companies

  Represent the seller in a sale for a fee and do not take the data onto their own books.

• Big tech infrastructure

  1 company

  Infrastructure that charges AI companies for access to web content and pays the site owner.

All companies, A to Z

• AfterQuery AI data vendor
• AI Training Data Broker Broker
• Amazon AI lab
• Anthropic AI lab
• Appen AI data vendor
• Apple AI lab
• Centific AI data vendor
• Cloudflare Big tech infrastructure
• Cognor Labs Broker
• Collinear AI AI data vendor
• Datarade Data marketplace
• DataVendor Data marketplace
• Defined.ai Data marketplace
• Endless Broker
• FileYield Data marketplace
• Google AI lab
• Grepped Data marketplace
• Halluminate AI data vendor
• Handshake AI AI data vendor
• Hilco Global Wind-down buyer
• Jottings Broker
• Labelbox AI data vendor
• LicenseData.io Broker
• Magma Data marketplace
• Mercor AI data vendor
• Meta AI lab
• micro1 AI data vendor
• Microsoft AI lab
• Miro Advisory Broker
• Mode AI data vendor
• NVIDIA AI lab
• Nyne AI data vendor
• Ooak Data AI data vendor
• OpenAI AI lab
• Opendatabay Data marketplace
• Pewter Data Broker
• Polyshares AI data vendor
• pre.dev AI data vendor
• Protege Data marketplace
• Replay AI data vendor
• Scale AI AI data vendor
• Sell My Company Data Broker
• Shutterstock Data marketplace
• SimpleClosure Wind-down buyer
• Snorkel AI AI data vendor
• Sunset Wind-down buyer
• SuperAnnotate AI data vendor
• SupersonIQ Broker
• Surge AI AI data vendor
• Telegraph Lab AI data vendor
• The General Data Company AI data vendor
• Troveo Data marketplace
• Turing AI data vendor
• Versos AI Data marketplace
• Wirestock Data marketplace

How these profiles are built

Each profile is built from the company's own pages, from filings and from press coverage. Every value cites a source that was opened and independently re-checked. Figures the company has not confirmed itself are marked "reported". Where a company's own pages give different figures, each one is listed with its source. "Not published" means we found no figure, not that none exists.

The full dataset behind these pages is public. Download the dataset (JSON). For the market ranges quoted elsewhere on this site, see sources and methodology.

Reviewed and not listed

We review companies that appear on public buyer lists and list only those with a confirmed program or named deal. These companies were reviewed and are not listed, for the reasons given.

• Human Native. Acquired by Cloudflare (announced January 15, 2026). Its site now shows only a note about joining Cloudflare and no seller program, so there is no standalone program to list.
• License My Company Data. Describes itself as a matching and referral directory that lists buyer programs and is paid referral fees by buyers. It says it does not buy, hold, license or broker records, so we treat it as a source of leads rather than a buyer.
• Oracle. We found no Oracle program, deal or disclosure of buying or licensing company data for AI. Its October 2026 statement that its AI Media Models were trained on rights-cleared data names no licensors or terms, and Oracle's own pages on the subject returned an error (HTTP 403).
• Reflection AI. Describes its training data only in general terms (data from commercial partner agreements, proprietary vendor data and datasets it purchases) and names no data partner, deal or seller program, so it does not meet the inclusion rule for AI labs.
• Safe Superintelligence Inc.. We found no data-buying or licensing program, deal or disclosure. Its site does not mention data, it has released no product, and the press coverage we found covers only compute and investment deals.
• SpaceXAI. Only informal talks have been reported (September 2026, by Crypto Briefing and Breitbart, citing people familiar with the matter), and those reports say the talks may not lead to a deal. We found no specific data-buying program or completed deal, so it does not meet the inclusion rule for AI labs.
• Thinking Machines Lab. Its Inkling training summary says it concluded commercial licensing agreements with rightsholders and acquired third-party content, but it names no licensor, dataset or deal, and we found no specific program or named deal, so it does not meet the inclusion rule for AI labs.
• Zendō Labs. We found no program to buy, license or broker company data. Its site says only that it builds RL environments, evaluations and data for frontier AI labs.