Skip to content
AI Data Payout

Search AI Data Payout

Jump to a company, a ranking or a tool

Updated October 2026

Best human-data vendors for AI teams

Independent and sourced. Some companies pay us a referral fee. It never changes our ranking. How we earn

Mercor is the largest vendor by reported run rate, above $2B[2], but confirmed a cyberattack in March 2026[4]. Surge AI reported $1B to $1.2B in 2024 revenue with no outside investors[14], and Scale AI is 49% owned by Meta[10]. Few vendors publish quality or security evidence, so most scores rest on two or three criteria, and we show exactly which.

  1. 01Top pick
    Hs

    Handshake AI

    Pays people from its career network as AI training fellows.

    Best forLabs that want to draw on a large career network

    Reported ARR
    About $1.15B total gross ARR, with the AI training unit near $1B1

    Watch out for: Little published on vetting or security1

    7.9/10How we rate
  2. 02
    Me

    Mercor

    Supplies expert contractors whose work labs buy as training data.

    Best forLabs buying expert data at the largest scale on the market

    Reported run rate
    Above $2B gross run rate (June 2026)2
    Valuation
    $10B after a $350M Series C3

    Watch out for: Confirmed a cyberattack in March 2026; several class actions followed.4,5

    7.4/10How we rate
  3. 03
    Ap

    Appen

    Data annotation and collection at scale.

    Best forBuyers who want a listed company with public financials

    FY26 revenue guidance
    US$270M-$300M (ASX: APX)6
    7.4/10How we rate
  4. 04
    m1

    micro1

    Recruits experts for AI training work and licenses company operating data.

    Best forLabs that also want licensed company data alongside expert work

    Reported run rate
    $500M gross run rate in August 2026, up from $100M eight months earlier7
    Latest round
    Raised $100M+ at a $4B valuation8

    Watch out for: Net revenue reported far below the gross run rate7

    7.4/10How we rate
  5. 05
    Pr

    Prolific

    Human participant platform for research and AI evaluation.

    Best forTeams that need human participants for research and evaluation

    Estimated revenue
    About $350M annualized, gross (April 2026, estimate)9

    Watch out for: Revenue figure is an outside estimate9

  6. 06
    Sc

    Scale AI

    Managed data labeling and human data for AI teams.

    Best forTeams with no competitive overlap with Meta

    Ownership
    Meta holds 49% after a deal reported at $14.3B; founder Alexandr Wang joined Meta10
    Valuation
    Above $29B with Meta's investment11

    Watch out for: Ownership conflict: Meta holds 49%, and competes with many of Scale's customers.10

  7. 08
    Sg

    Surge AI

    Human data for model training, sold to AI labs.

    Best forLabs that want a large vendor with no outside owners

    Reported revenue
    Over $1B in 2024, and profitable14
    Outside funding
    None; Surge says it has never raised outside funding15

    Watch out for: Little published on quality or security14

  8. 09
    Tu

    Turing

    Supplies coding data to OpenAI and other LLM builders.

    Best forLabs buying coding data

    Latest round
    $111M Series E at a $2.2B valuation; about $300M ARR16
  9. 11
    To

    Toloka

    Human data and evaluation for AI teams.

    Best forTeams comparing independent data vendors

    Ownership
    Nebius keeps a majority economic stake after a Bezos Expeditions-led investment, but gave up voting control18

Some companies pay us a referral fee. It never changes our ranking. How we earn

How we scoreHuman-data vendors for AI teams: five criteria, published weightsShow

Expert depth and vetting25%

  • 9-10: documented expert pools in named fields with published vetting
  • 7-8: documented expert pools, vetting described
  • Not scored: no public evidence

Quality systems20%

  • 9-10: published gold sets, agreement rates or third-party audits
  • 7-8: quality process described in detail
  • Not scored: no public evidence

Security and neutrality20%

  • 8: SOC 2 Type II documented and no reported incidents
  • 6: SOC 2 documented, but an incident reported in the last 12 months
  • 5: part-owned by a company that competes with its customers
  • Not scored: no public evidence

Scale and speed20%

  • 9.5: reported run rate of $2B or more
  • 9: $1B or more
  • 8: $500M or more
  • 7: $250M or more
  • 6: under $250M
  • Not scored: no reported figure

Transparency15%

  • 8-9: public terms, pricing or rate cards
  • 6-7: public program terms, no pricing
  • Not scored: little published

A score is published only when criteria worth at least 35% of the weight are evidenced. The overall score is the weighted mean of the evidenced criteria. Commissions are never an input. Full methodology

FAQ

Hiring a data vendor: questions, answered

  • A human data vendor supplies the people whose judgments, answers and corrections are used to train and test AI models. Some, like Mercor, sell expert work to AI labs at very large scale. Others, like micro1, also license operating data from companies. This site ranks both kinds for the reader each one serves.2,19

  • Mercor is the largest by reported figures, with a gross run rate above $2B in June 2026. Surge AI's 2024 revenue is reported at $1B to $1.2B, and Handshake's gross ARR at about $1.15B. micro1 reported a $500M gross run rate in August 2026, Snorkel $375M, and Prolific about $350M by an outside estimate.2,14,20,1,7,12,9

    Most of these are gross figures from unnamed sources, and gross and net revenue can differ widely. TechCrunch put micro1's net revenue at $150M-$200M against its $500M gross run rate.

  • Each audience has its own published rubric of five weighted criteria, and every sub-score cites the facts behind it. A score appears only when we can evidence at least 35% of the rubric's weight; otherwise the card says there is not enough public data. Commissions never enter the calculation.

    Read more

  • Ownership is the first thing to check. Meta has held 49% of Scale AI since June 2025, which matters to any lab that competes with Meta. Surge AI says it has never taken outside funding, so no lab owns part of it. Ask every vendor for its owners, its SOC 2 report and its incident history before a pilot.10,15

  • Mercor confirmed a cyberattack on 31 March 2026, tied to the compromise of the open-source LiteLLM project, and several class actions were filed in the Northern District of California on 1 April. Mercor later called the affected data a very limited subset. No reliable count of affected people has been published.4,5

  • Mercor states a SOC 2 Type II attestation on its data page. None of the other companies in our current sources publishes one where we could check it. That does not mean they lack one: ask each vendor for its report under NDA, and treat a refusal as a signal.21

Sources (21)

  1. 1.Dealroom, citing The Information, published 14 Apr 2026. Accessed 14 Oct 2026.
  2. 2.Dealroom, citing The Information, published 6 Jul 2026. Accessed 14 Oct 2026.
  3. 3.TechCrunch, published 27 Oct 2025. Accessed 14 Oct 2026.
  4. 4.TechCrunch, published 31 Mar 2026. Accessed 14 Oct 2026.
  5. 5.HR Dive, published 13 Apr 2026. Accessed 14 Oct 2026.
  6. 6.Appen ASX announcement, published 27 Aug 2026. Accessed 14 Oct 2026. Company's own claim.
  7. 7.TechCrunch, published 20 Aug 2026. Accessed 14 Oct 2026.
  8. 8.Forbes, published 22 Sep 2026. Accessed 14 Oct 2026.
  9. 9.Sacra (estimate), published 19 May 2026. Accessed 14 Oct 2026.
  10. 10.TechCrunch, published 13 Jun 2025. Accessed 14 Oct 2026.
  11. 11.Scale AI blog, published 12 Jun 2025. Accessed 14 Oct 2026. Company's own claim.
  12. 12.SiliconANGLE, published 22 Sep 2026. Accessed 14 Oct 2026. Company's own claim.
  13. 13.SiliconANGLE, published 22 Sep 2026. Accessed 14 Oct 2026.
  14. 14.Reuters (via AOL), published 1 Jul 2025. Accessed 14 Oct 2026.
  15. 15.Forbes. Accessed 14 Oct 2026. Company's own claim.
  16. 16.TechCrunch, published 6 Mar 2025. Accessed 14 Oct 2026.
  17. 17.Bloomberg, published 16 Sep 2025. Accessed 14 Oct 2026.
  18. 18.Nebius newsroom, published 7 May 2025. Accessed 14 Oct 2026. Company's own claim.
  19. 19.micro1 company referral page. Accessed 14 Oct 2026. Company's own claim.
  20. 20.Forbes. Accessed 14 Oct 2026.
  21. 21.Mercor Data page. Accessed 14 Oct 2026. Company's own claim.