Updated 17 Oct 2026
The AI training data market, in numbers.
Who is paying for data, how fast, and what the companies in the middle earn and raise. Every bar and row is built from records with a source.
Licensing deals per quarter
Deals made public between data owners and AI companies. Most are publishers; private company-data sales are rarely announced.
Lawsuits and rulings per quarter
The legal pressure that makes buyers pay for data with clear rights.
What the data vendors earn
The newest headline figure each vendor published. Run rates are often the best month times twelve, and gross revenue includes the 60 to 70 percent that passes through to the experts doing the work.
- Figure$2BMeasurerun-ratePeriodrun-rate Jun 2026
- Figure$1.2BMeasurerevenuePeriodFY2024
- Figure$1BMeasureannualized revenuePeriodrun-rate early 2026
- Figure$1BMeasurerun-ratePeriodrun-rate Oct 2026
- Figure$870MMeasurerevenuePeriodFY2024
- Figure$699MMeasurerevenuePeriodQ2 2025
- Figure$500MMeasurerun-ratePeriodannualised revenue May 2026
- Figure$375MMeasurerun-ratePeriodrun-rate Sep 2026
- Figure$309MMeasurerevenuePeriodQ2 2026
- Figure$300MMeasurerun-rate (projection)Periodprojected run-rate end of 2026
- Figure$300MMeasureARRPeriodARR Mar 2025
- Figure$200MMeasurerun-ratePeriodannualized revenue Sep 2026
Money raised by data companies
Disclosed funding rounds of companies that make, buy or broker training data, in US dollars per quarter.
The largest rounds
The ten largest disclosed rounds raised by companies that make, buy or broker training data.
- $1B
Series F, May 2024, valued at $13.8B
- $400M
SignSplit
Strategic seed, Oct 2026, valued at $1B
- $350M
Series E, Sep 2026, valued at $3.5B
- $350M
Series C, Oct 2025, valued at $10B
- $200M
Series F, Jan 2022, valued at $3.5B
- $130M
Series A, Jul 2026
- $111M
Series E, Mar 2025, valued at $2.2B
- $100M
Series E, Oct 2026, valued at $1.3B
- $100M
Series B, Sep 2026, valued at $4B
- $100M
Growth, Sep 2025, valued at $2B
Everything recorded, newest first
- 16 Oct 2026BrokerageDigiday
- 16 Oct 2026BrokerageOoak: buyer programOoak (partner page)
Earn up to $1M
- 16 Oct 2026BrokerageHandshake AI: price rangeHandshake (data partnerships page)
$100K to $4M
- 16 Oct 2026BrokerageScale AI: referral termsScale AI (data partnership page)
up to $50,000
- 16 Oct 2026BrokerageTuring (Project Lazarus page)
- 16 Oct 2026BrokerageTuring (Project Lazarus): referral termsTuring (Project Lazarus page)
up to $10k
- 16 Oct 2026BrokerageSuperAnnotate: price rangeSuperAnnotate (data partnerships page)
$100K to $1M+
- 16 Oct 2026BrokerageLabelbox (Prism): price rangeLabelbox Prism
around $50K to $1M+
- 16 Oct 2026BrokerageNyne: price rangeNyne (sell company data page)
$100K or more
- 16 Oct 2026BrokerageHUMXN Research Labs: price rangeHUMXN
typically start around $250,000 and run into the millions
- 16 Oct 2026BrokerageHUMXN Research Labs: referral termsHUMXN referral partners
10% of the deal, up to $100,000 per deal
- 16 Oct 2026BrokerageMode (BizBuyData): price rangeBizBuyData (Mode)
$100K to $5M
- 16 Oct 2026BrokeragePolyshares: price rangePolyshares
$100K to $2M and above
- 16 Oct 2026BrokeragePolyshares: transactionPolyshares case studies
$450,000
- 16 Oct 2026BrokerageTroveo: price rangeTroveo
six figures
- 16 Oct 2026BrokerageDatarade: broker economicsDatarade provider application
$6,000 a year and 20% commission
- 16 Oct 2026Brokeragepre.dev: price rangepre.dev
$6,000 for the example repository; source code $100 and git history $40 per million tokens
- 16 Oct 2026BrokerageDefined.ai: price rangeDefined.ai partnerships
$600K+ average partner revenue; $1M+ for 5+ partners
- 16 Oct 2026BrokerageChinese face-licensing platforms for AI short dramas (unnamed): price range36Kr
100 to 500 yuan per AI short video for ordinary people; several thousand up to 10,000 yuan for professional actors; highest single listing 6,000 yuan (elderly man)
- 16 Oct 2026BrokerageGrepped: price rangeGrepped, site configuration (config.js)
1.25 x industry x (employees x years) to the power 0.85; high end 3 times the low
- 16 Oct 2026Brokeragemicro1: referral termsmicro1, sub-affiliate programme
20% of their payout
- 16 Oct 2026BrokeragePewter (pewterdata.com): referral termsPewter, partner terms
10% of the licensing payment, up to $100,000 per business
- 15 Oct 2026DealAxios, The Financial Times, Forbes, Post and Courier, Texas Tribune, Shaw Media, Oil City (publishers named in Google's October 2026 pilot update) licensed data to GoogleGoogle (The Keyword blog)
Paid; no amounts or contract lengths stated
- 15 Oct 2026DealCNN licensed data to GoogleTheWrap
Financial terms not disclosed; multiyear
- 15 Oct 2026DealMassiveMusic (part of Songtradr) licensed data to GRAIPR Newswire
Terms not disclosed
- 15 Oct 2026LegalNL Times
- 15 Oct 2026BrokerageGrepped: intermediaryGrepped, marketplace page code
an open catalogue of lots for buyers
- 14 Oct 2026FundingLabelbox: Series ELabelbox (company blog)
more than $100 million in Series E financing, including equity and a credit facility (the split is not stated). Participants: Wipro Ventures, a16z, B Capital, Tau Ventures, Snowpoint Ventures and others., valued at $1.25 billion pre-money
- 14 Oct 2026FundingLabelbox: Credit facility (part of Series E financing)Labelbox (company blog)
Amount not stated separately; included in the 'more than $100 million' Series E financing (do not add to the equity figure)
- 14 Oct 2026BrokerageGrepped: price rangeGrepped, site configuration (config.js)
sum of the picked sources' points x team size x years x industry; high end 2.5 times the low
- 13 Oct 2026LegalMashable
- 13 Oct 2026FundingPrime Intellect acquired MoondreamPrime Intellect
price not disclosed
- 13 Oct 2026BrokerageHUMXN Research Labs: buyer programHUMXN journal (press release)
$50 million
- 13 Oct 2026BrokerageHUMXN Research Labs: price rangeHUMXN journal (press release)
$20 per hour of approved footage
- 13 Oct 2026BrokerageMode (BizBuyData): buyer programBusiness Wire (paid release), read on Yahoo Finance
$5 Billion
- 13 Oct 2026BrokerageMode (BizBuyData): referral termsBusiness Wire (paid release), read on Yahoo Finance
the greater of $50,000 or 6% of the organisation's payment
- 13 Oct 2026BrokerageSimpleClosure AssetHub: intermediaryMIT Technology Review
over 150 deals; expects to return over $10 million by end of 2026; 47% of its deals now include code or data sales
- 13 Oct 2026BrokerageSimpleClosure AssetHub: transactionMIT Technology Review
roughly $25,000 for the code base; a pending amount (likely around $15,000) for workspace data
- 13 Oct 2026BrokerageGrepped: referral termsGrepped, referral terms (effective 13 Oct 2026)
referred owners' data only through Grepped
- 13 Oct 2026BrokerageGrepped: referral termsGrepped, referral terms (effective 13 Oct 2026)
2% of each referral fee
Cite this
AIData.Broker, “The AI training data market, in numbers”, reviewed 17 Oct 2026. https://www.aidata.markets/market
Free to reuse under CC BY 4.0 if you credit AIData.Broker and link to this page. Each row names the publication that stated it, and words quoted from a source keep that source's own terms.