Key findings

1
The largest price recorded: micro1 for Spirit Airlines (debtor), $12,500,000 (stated, 3 Sep 2026).
Figure 2
2
Polyshares: $100K to $2M and above (typical deals) for company operational data (Slack, email, Drive, spreadsheets, CRM history, support tickets, call recordings, accounting and ERP exports) (29 Oct 2026).
Figure 2
3
The largest buyer program: micro1, $1 billion over the next 12 months (9 Oct 2026).
Figure 3
4
micro1 pays referrers $50,000 up to a $700,000 deal; 7.5% of deal value above (stated, 29 Oct 2026); the terms are in the figure.
Figure 4

Figure 1

Who is hiring to buy data today (live)

Which labs and vendors have data-acquisition, partnership or licensing roles open right now?

Role	Company	Side	Stated pay
Research Program Manager, Data AcquisitionPosting	OpenAI	AI lab	$239,000-$328,000 per year
Software Engineer - Data Aquisition (systems)Posting	OpenAI	AI lab	$295,000-$536,000 per year
Strategic Sourcing Lead, Professional Services – Human DataPosting	OpenAI	AI lab	$234,000-$260,000 per year
Strategic Data & Licensing PartnershipsPosting	Mistral AI	AI lab	n/a
Data Partnerships Coordinator (1-year contract)Posting	Cohere	AI lab	$110,000-$150,000 per year
Data PartnershipsPosting	Thinking Machines Lab	AI lab	$300,000 - $350,000
Member of Technical Staff - Engineering Lead, Data AcquisitionPosting	Reflection AI	AI lab	n/a
Data Acquisition Lead, Frontier EnvironmentsPosting	Scale AI	Data vendor	$182,400-$228,000 per year
Data Partnerships RepresentativePosting	Handshake	Data vendor	$100,000-$120,000 per year
Director, Data PartnershipsPosting	Snorkel AI	Data vendor	$215,000-$315,000 per year
Data Acquisition LeadPosting	AfterQuery	Data vendor	$200,000-$350,000 per year

Sample. Read today from the public job boards of 25 AI labs and 14 human-data vendors. Recomputed daily.

The desks a data owner or a broker sells to. Full detail in the report Who Is Buying AI Training Data.

Source. AITraining.jobs Research, Company Data Brokerage: Buyers, Prices and Terms, live data.

Figure 2

Deals and prices paid

What have buyers paid for companies' records and archives?

Who	What	Price or terms as published	Terms	Date	Basis
Polyshares (companies with an operating history (sellers); AI labs (buyers, unnamed))Polyshares	company operational data (Slack, email, Drive, spreadsheets, CRM history, support tickets, call recordings, accounting and ERP exports)	$100K to $2M and above (typical deals)	Buys as principal ('We are the buyer'); no fee or commission to the seller; offer within one to two weeks of intake; payment net 30 to 60 once the data is approved, anonymized and delivered; identifying fields stripped by a specialist PII partner; the company keeps its business and data and licenses defined material for a defined term. Does not disclose which labs take which data.	29 Oct 2026	Stated
Polyshares (four companies (roofing, general contractor, solar, global marketing agency; names confidential))Polyshares	operational records (perpetual exclusive licenses)	$110,000 to $450,000 (four selected licenses)	Case studies: home services (roofing company) $450,000; construction (general contractor) $300,000; energy (solar company) $275,000; professional services (global marketing agency) $110,000; each a perpetual exclusive license, anonymized before delivery. The homepage says the founders 'licensed data on their own companies before buying it from anyone else'; the page does not say which licenses those were.	29 Oct 2026	Self-reported
micro1micro1	company operational data (Enterprise Data Partnership)	$100K-500K (Qualified Enterprise Data Partnerships)	Replaces the '$100k+' floor published with the program on 9 Oct 2026. Same page: data licensed only to labs and enterprises in the U.S. and allied countries; micro1 has 30 days to review delivered data, then pays within 30 days of the company's invoice. Eligibility unchanged: 30+ full-time employees, primarily English, US first.	28 Oct 2026	Stated
micro1micro1	company operational data (large-scale or multi-team)	$500K-2M (large-scale operational datasets or ongoing participation across multiple teams)	Replaces the '$500k+' tier published with the program on 9 Oct 2026.	28 Oct 2026	Stated
micro1micro1	company operational data (highly unique, proprietary)	$2M+ (highly unique, proprietary operational data)	Replaces the '$1M+' tier published with the program on 9 Oct 2026.	28 Oct 2026	Stated
micro1micro1	company operational data (typical payment)	$150K to over $2M (typical payments)	Step 4 of the process on the same page; the page headline reads 'We're paying out $200k-2M+ for the data your business already generates'. The FAQ adds that micro1 has 30 days to review and approve delivered data and pays within 30 days of invoicing.	28 Oct 2026	Stated
micro1micro1	agent trajectories (tasks run by the company's AI agents in its own workflows)	$1-2M annually (agent trajectories)	New on the page: optional agent build-outs (months 1-2) and continuous agent evaluation from launch; trajectories licensed with the company's approval, recurring from launch.	28 Oct 2026	Stated
company-data buyers (unnamed, per Business Insider) (companies (sellers))Reclaim The Net	workplace records (Slack, email, documents, code)	six-figure offers, with a decision demanded within 48 hours	Business Insider names Polyshares and micro1 among the brokers; Polyshares cofounder Ryan Lock describes buyers offering 'double market' with a Friday deadline. Relayed by Reclaim The Net; the Business Insider article itself was not read.	14 Oct 2026	Reported
SimpleClosure (brokered sale) (a crypto startup winding down (seller); buyer not named)MIT Technology Review	codebase; workspace data (Google, Notion, Figma, Slack)	roughly $25,000 for the codebase; workspace data pending (likely around $15,000)	Founder's account of a sale brokered by SimpleClosure; the workspace data sale was still pending.	13 Oct 2026	Self-reported
micro1micro1	company operational data (Enterprise Data Partnership)	$100K-$2M+ for approved data packages	Company requirements listed: 30+ employees, US prioritized, primarily English. Read again 2026-10-29: the page still states $100K-$2M+ for approved data packages but lists the requirements as '30+ employees US and Western markets English All industries'.	11 Oct 2026	Stated
micro1micro1	company operational data (estimator example)	$50,284 - $92,776 (example output of micro1's payout estimator)	Indicative estimate only, not a quote or binding offer. Inputs: employees, years of data, business entities, share of English, HQ region, regulated or sensitive data. Final offer depends on data sources, size, quality, access terms and due diligence. Read again 2026-10-29: the estimator now asks only employees, years of data and headquarters region and returns its range from a remote service; no new example was read.	11 Oct 2026	Stated
GreppedGrepped	company records (messages, code, documents, CRM and finance records, tickets)	$20K–$5M; 'Up to 6 figures per workspace'	Buyers see redacted previews before a sale; 'When they purchase, the money comes to you'; 'Get paid in as little as 7 days.' No longer on the live homepage read 2026-10-29, which reads 'Premium licensing for company data worth millions.'; the range is still archived on 17 Oct 2026.	11 Oct 2026	Self-reported
GreppedGrepped	company operational records (estimator example)	$29K to $41.5K	Estimate from team size, founding year and revenue. Reply with scope and offer within a few business days; read-only access to approved sources; payment within 7 days of data passing review; 'two to three weeks' from form to payment; originals deleted after processing. On 2026-10-29 /business/ redirects to the homepage; its estimate form shows a range only after contact details are given.	11 Oct 2026	Stated
Ooak Data (operating, winding-down and closed companies)Ooak Data	company records (email, Slack, Teams, documents, tickets, code, CRM)	from $20K (smallest closed) to several hundred thousand dollars	Price depends on years of history, headcount, number of tools and how deeply they connect. Payment goes to the entity, not individuals; signer must hold signing authority. Raw archive deleted within 30 days (contract obligation).	11 Oct 2026	Self-reported
micro1micro1	company operational data (Enterprise Data Partnership)	$100k+ (Qualified Enterprise Data Partnerships)	Eligibility: 30+ employees, established documentation, US first then other Western markets, English. Companies retain ownership; PII removed when relevant; some datasets may get synthetic rewrites. Pay factors: dataset size, workflow complexity, quality, domain expertise, uniqueness. Superseded: by 28 Oct 2026 the page lists $100K-500K, $500K-2M and $2M+ (see those records); the Wayback Machine still shows these tiers on 14 Oct 2026.	9 Oct 2026	Stated
micro1micro1	company operational data (large-scale or multi-team)	$500k+ (large-scale operational datasets or ongoing participation across multiple teams)	Eligibility: 30+ employees, established documentation, US first then other Western markets, English. Companies retain ownership; PII removed when relevant; some datasets may get synthetic rewrites. Pay factors: dataset size, workflow complexity, quality, domain expertise, uniqueness. Superseded: by 28 Oct 2026 the page lists $100K-500K, $500K-2M and $2M+ (see those records); the Wayback Machine still shows these tiers on 14 Oct 2026.	9 Oct 2026	Stated
micro1micro1	company operational data (highly unique, proprietary)	$1M+ (highly unique, proprietary operational data)	Eligibility: 30+ employees, established documentation, US first then other Western markets, English. Companies retain ownership; PII removed when relevant; some datasets may get synthetic rewrites. Pay factors: dataset size, workflow complexity, quality, domain expertise, uniqueness. Superseded: by 28 Oct 2026 the page lists $100K-500K, $500K-2M and $2M+ (see those records); the Wayback Machine still shows these tiers on 14 Oct 2026.	9 Oct 2026	Stated
micro1 (unnamed sellers)Fortune	company data (unspecified)	more than 50 data transactions in the preceding 45 days	Stated in micro1's Spirit court filing; no sellers, prices or AI-lab customers named.	14 Sep 2026	Self-reported
Turing (five to ten failed startups (unnamed))Fortune	source code (failed-startup codebases)	average in the tens of thousands of dollars	CEO statement to The Information, relayed by Fortune.	14 Sep 2026	Self-reported
micro1 (Spirit Airlines (debtor))Research Suite by Stretto (chapter11cases.com), docket summary	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$12,500,000	All cash from cash on hand, no financing contingency; described as a $2.5 million (25 percent) premium over the Successful Bid. Offered covenants excluding disciplinary, medical, EEO and union material. The Debtors had not responded on the docket through 11 Sep 2026. The ombudsman's 5 Oct 2026 supplement asks for the same further review of micro1, which proposed to de-identify the data itself.	3 Sep 2026	Stated
SimpleClosureSimpleClosure	company assets (code, operational data)	tens of thousands to hundreds of thousands of dollars; potentially millions at enterprise scale	Based on 350+ startup datasets. About 37% of raw code disappeared after removing third-party or generated code and duplicates. Demand cited in manufacturing, energy, banking and insurance, professional services and logistics.	3 Sep 2026	Self-reported
unnamed codebase-buying platform (cited by Troveo) (code owners)Troveo	source code (private repositories)	5,000 dollar baseline per repository	Adjusted up or down after grading; platform not named.	3 Sep 2026	Reported
Troveo (compiling reported figures)Troveo	source code and company archives	a few thousand dollars per graded repository; 10,000 to 100,000 dollars per shutdown deal; hundreds of thousands for a complete company archive	Operating companies usually license and keep ownership; an outright sale is 'mainly for shutdowns'. Exclusive licences command a premium. Rights work makes deals take 'Weeks, not days'.	3 Sep 2026	Reported
Google (Spirit Airlines (In re Spirit Aviation Holdings, No. 25-11897, Bankr. S.D.N.Y.))Research Suite by Stretto (chapter11cases.com), docket summary	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$10,000,000	Google opened at $5 million cash, with a third-party de-identification process paid for by the buyer. Sale not approved as of the docket review through 11 Sep 2026 (later status: see the 5 Oct 2026 ombudsman record and the 9 Oct 2026 adjournment to 6 Nov 2026).	14 Aug 2026	Stated
Mercor (Spirit Airlines (debtor))Research Suite by Stretto (chapter11cases.com), docket summary	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$7.5 million alternate bid; also $5.2M, $7M and $10M offers	Mercor offered $7 million and later $10 million conditioned on using its own de-identification tools instead of a third party; neither was selected. If the Google deal is not consummated, the Debtors may accept the Alternate Bid. The ombudsman's 5 Oct 2026 supplement asks to investigate Mercor's early-2026 supply chain security incident before opining on it as a buyer.	14 Aug 2026	Stated
multiple unnamed buyers (Warmly (spelled 'Wormly' in this source), acquired by HubSpot)The Financial News (fnnews.com), summarizing The Information	source code and employee communications	as much as $300,000	CEO Maximus Greenwald said he received four such offers and turned them all down; HubSpot said no data was included in the acquisition. AI Weekly (15 Aug 2026), citing The Information, says Mercor approached Warmly.	14 Aug 2026	Reported
SimpleClosure (defunct startups (sellers); AI labs and data firms (buyers, unnamed))Forbes	defunct-company archive (code, Slack, email)	$10,000 to $100,000 per company; nearly 100 deals; over $1 million recovered	CEO Dori Yona's figures to Forbes. SimpleClosure removes personally identifiable information before sale.	16 Apr 2026	Self-reported
Sunset (defunct companies (sellers))Forbes	defunct-company archive	'similar prices' to SimpleClosure (whose typical range is $10,000 to $100,000 per company)	Pricing factors per CEO Brendan Mahony: company size, age and 'data richness' (internal traceability and cross-platform linkage). A Jira ticket tied to a specific code commit is worth more than a standalone document; healthcare and finance command a premium.	16 Apr 2026	Self-reported
SimpleClosure (wind-down and sale) (cielo24 (seller); buyer not named)Forbes	defunct-company archive (13 years of Slack, Jira, email, Google Drive)	hundreds of thousands of dollars	Former CEO Shanna Johnson's account; buyer and exact price not disclosed.	16 Apr 2026	Self-reported
Turing (founders and engineering leaders (sellers))Turing (AGI Advance newsletter)	legacy codebases (COBOL, .NET, Java)	$10K-$100K+ per legacy codebase	Project Lazarus as launched, for sharing private codebases and the operational artifacts around them (Jira tickets, PRDs, architecture docs, support threads). By 29 Oct 2026 the Lazarus page offers one payout of up to $1M for a company's operating record (see that record).	25 Mar 2026	Stated
Turing (founders and engineering leaders (sellers))Turing (AGI Advance newsletter)	modern codebases (pull requests)	$1,000 per PR, up to $10K per codebase	Project Lazarus launch terms.	25 Mar 2026	Stated
Turing (founders and engineering leaders (sellers))Turing (AGI Advance newsletter)	org artifacts (tickets, specs, docs, support threads)	$50 per artifact	Project Lazarus launch terms; paid only when bundled with a codebase.	25 Mar 2026	Stated
RL environment vendors (neolabs and frontier labs)Epoch AI	RL environment (contracts)	often six to seven figures per quarter; $300k-$500k seen by a neolab researcher	From interviews with RL-environment founders and lab researchers. Included as a brokerage input: Grepped's referral portal accepts RL environments as referred supply.	12 Jan 2026	Reported
RL environment vendors (AI labs)Epoch AI	RL environment (software replicas)	around $20k per website replica; around $300k for a Slack-grade replica	From interviews with RL-environment founders and lab researchers. Included as a brokerage input: Grepped's referral portal accepts RL environments as referred supply.	12 Jan 2026	Reported
RL environment vendors (AI labs)Epoch AI	RL task (prompt plus grader)	$200-$2,000 per task; $20k per task rare	From interviews with RL-environment founders and lab researchers. Included as a brokerage input: Grepped's referral portal accepts RL environments as referred supply.	12 Jan 2026	Reported
RL environment vendors (AI labs)Epoch AI	RL environment (exclusive vs non-exclusive)	roughly 4-5x	From interviews with RL-environment founders and lab researchers. Included as a brokerage input: Grepped's referral portal accepts RL environments as referred supply.	12 Jan 2026	Reported

Sample. 36 records, to 29 Oct 2026, each from the publication linked beside it.

Source. AITraining.jobs Research, Company Data Brokerage: Buyers, Prices and Terms, live data.

Figure 3

Who is buying, and on what terms

Which labs and vendors run programs to buy or license company data, and what do they ask for?

Who	What	Price or terms as published	Terms	Date	Basis
micro1 (companies that are no longer operating (sellers))micro1	defunct-company data (Defunct Company Data Partnership)	$100K-$2M+ for approved data packages (defunct companies)	New partnership type on the referral page. Company requirements listed: '30+ employees', '10+ Countries', 'English', 'All industries'. Referral reward $75K+ (see the referral record).	29 Oct 2026	Stated
Turing (operating, winding-down and closed companies (sellers))Turing	a company's operating record (communications, tickets, wikis, knowledge bases, databases, source code)	a single payout of up to $1M	One payment, no retainer or recurring commitment; valuation after sampling, 'usually within a week'; intake call within one business day, under NDA. The FAQ says code is 'never bought on its own', the company keeps ownership and Turing acquires 'defined rights to the prepared record'; strongest demand in the US, UK, Canada and the EU. The site's process page says 'Repos typically pay $10k–$100k+ depending on quality and complexity.'	29 Oct 2026	Stated
Ooak Data (companies (sellers))Ooak Data	company operational and business data	Earn up to $1M	Same page still says deals to date run from $20K, the smallest closed, to several hundred thousand dollars; '2 to 4 weeks'; raw archive deleted within 30 days of delivery (contract obligation).	16 Oct 2026	Stated
Surge AI (companies' data archives (sellers, unnamed))MIT Technology Review	company data archives (Slack, email, calendars) for reinforcement-learning gyms	no price disclosed	Nick Heiner, who leads Surge AI's environment work, says one company's Slack history with variety and depth can be worth more than another's Slack, email and calendars combined, and that buyers quote 'wildly different numbers' because valuations rest on 'research theses' about how many useful tasks come from, for example, 100,000 lines of code.	13 Oct 2026	Self-reported
Mercor (companies in eight target industries)Mercor	company records from connected tools (Slack, Teams, Gmail, Outlook, Salesforce, GitHub, Drive, Notion, Confluence; 100+ tools)	amount not published	Target industries: Legal; Healthcare; Finance & Insurance; Professional Services; Manufacturing; Retail & E-commerce; Real Estate; Logistics & Supply Chain. No fees: Mercor covers extraction, anonymization and transfer, with no deductions. Paid by wire on delivery of the anonymized dataset to the buyer, typically within 2 to 4 weeks of the signed agreement. Anonymization fully automated; no lab buyer sees raw data.	11 Oct 2026	Stated
Ooak Data (companies (sellers))Ooak Data	company operational and business data	Earn up to $300K	Advertises '2 to 4 weeks' and 'No engineering work required'; in-house anonymization, no third party touches the raw archive. Superseded: the headline reads 'Earn up to $1M' from 16 Oct 2026 (see that record).	11 Oct 2026	Stated
micro1 (operating companies (sellers); capital from Citi and Hercules Capital)Business Wire (via FinancialContent)	company operational data (SOPs, knowledge bases, internal documents, CRM data, project histories, QA processes)	$1 billion over the next 12 months	Buys and licenses data through the Company Data Partnerships program; de-identified data is turned into reinforcement learning environments. Financed through capital provided by Citi and Hercules Capital (structure not disclosed).	9 Oct 2026	Stated
SpaceX (troubled or defunct startups)Bloomberg	customer and operational information of failed startups	n/a	Internal discussions only, per people familiar with the matter; no deal or price reported.	17 Sep 2026	Reported
micro1 (50 partner companies (unnamed))micro1 forum	company workflow data	$100K-$2M+ each, 50 companies	Company Data Partnership Program; companies and amounts per company not disclosed.	1 Jul 2026	Self-reported

Sample. 9 records, to 29 Oct 2026, each from the publication linked beside it.

Source. AITraining.jobs Research, Company Data Brokerage: Buyers, Prices and Terms, live data.

Figure 4

What brokers and referrers are paid

How are brokers and referrers paid when a deal closes?

Who	What	Price or terms as published	Terms	Date	Basis
micro1 (referrers (anyone over 18))micro1	company operational data (Enterprise Data Partnership)	$50,000 up to a $700,000 deal; 7.5% of deal value above	Page headline now 'earn $50,000+' and the card '$50K+ Referral reward'. Referral enters payout approval once the company's data is transferred and approved; paid by wire. A deal for an opportunity not listed in the referral portal pays 7.5% of deal value. Referrers may not use micro1 email addresses or present themselves as micro1 staff.	29 Oct 2026	Stated
micro1 (referrers (anyone over 18))micro1	defunct-company data (Defunct Company Data Partnership)	$75,000 up to a $1 million deal; 7.5% of deal value above	Card reads '$75K+ Referral reward'. Same payout conditions as the Enterprise Data Partnership referral.	29 Oct 2026	Stated
Turing (referrers)Turing	a company's operating record	up to $10k	Referral form on the page; the amount per referral and the payment conditions are not stated.	29 Oct 2026	Stated
Grepped (Grepped Partners (relationship holders; membership by application))Grepped	company data deals brought by partners	50/50 split of the spread	Partners bring the relationships; Grepped supplies lab connections, de-identification, data assurance and closers. The page says the biggest brokers 'cap referrers' fees and pocket 80 to 90% margin on every transaction' (no broker named). Separate from the open referral portal (10% of the owner payout, up to $100K).	14 Oct 2026	Stated
SimpleClosure (startups shutting down (sellers); foundation AI labs, simulated-workspace vendors and tech companies building their own models (buyers))MIT Technology Review	code and workspace data (emails, chats, documents)	over 150 deals; 47% of SimpleClosure deals include code or data sales	Asset Hub added data sales in April 2026; the projection of more than $10 million returned to startups by the end of 2026 is restated (see the 3 Sep 2026 record). CEO Dori Yona's figures.	13 Oct 2026	Self-reported
micro1 (referrers (anyone over 18))micro1	company operational data (Enterprise Data Partnership)	$50,000 referral reward	Same page says 'No cap on payouts' and, in its FAQ, 'You can earn up to $50K per referral, depending on the size and value of the partnership.' Workflow partnerships pay after onboarding plus a minimum revenue threshold; Corpus partnerships pay once data is accepted by micro1. Credited to a micro1 wallet. Companies that applied before clicking the link are not attributed. Referring does not guarantee placement. Superseded by 29 Oct 2026: the card reads '$50K+ Referral reward' and the FAQ pays $50,000 up to a $700,000 deal and 7.5% above (see that record); the Wayback Machine still shows this version on 20 Oct 2026.	11 Oct 2026	Stated
databroker.studio (independent micro1 affiliate) (micro1 (buyer))databroker.studio (Tom Wang, independent micro1 referral partner)	operating-company data (email, Slack, tickets, docs, workflows)	$50,000	Affiliate page, not an official micro1 site. Targets 20+ employees (30 to 200 ideal), US first, then UK and Canada. Discloses that micro1 pays the affiliate a separate referral fee. Describes micro1's Enterprise Data Partnership as '$100K–$2M+' and 'A one-time payment'. Advises accountants and lawyers to check their rules on referral fees.	11 Oct 2026	Self-reported
databroker.studio (independent micro1 affiliate) (micro1 (buyer))databroker.studio (Tom Wang, independent micro1 referral partner)	defunct-company archive (Teams or Slack history, deliverables)	$75,000	Targets companies with 50+ employees at peak and Microsoft Teams or Slack history; priority US, UK, Canada, India.	11 Oct 2026	Self-reported
Mercor (referrers)Mercor	company workflow data	referral bonus, amount not stated	Paid when the referred company goes live and starts earning ('you receive your referral bonus').	11 Oct 2026	Stated
Grepped (referrers)Grepped	company records, datasets and RL environments	10% of what we pay the owner	Paid 'When the owner you referred is paid. If they don't sell, there's no fee.' First link the owner fills the full form through wins; no fee if the owner applied alone first. No account needed; referrer need not work at the company. The linked full referral terms were not read. No legal entity named on the page.	11 Oct 2026	Stated
Grepped (referrers)Grepped	company records, datasets and RL environments	up to $100K per referral; reached at an owner payout of $1M	Paid 'When the owner you referred is paid. If they don't sell, there's no fee.' First link the owner fills the full form through wins; no fee if the owner applied alone first. No account needed; referrer need not work at the company. The linked full referral terms were not read. No legal entity named on the page.	11 Oct 2026	Stated
Grepped (AI labs and research teams (buyers, unnamed))Grepped	de-identified business workflow data	no buyer price or spread published	Grepped licenses to buyers under its own agreements with data owners and pays owners itself (referral fee is defined on what Grepped pays the owner), so its margin between lab price and owner payout is not disclosed. Samples de-identified and under NDA.	11 Oct 2026	Stated
FileYield (data sellers)FileYield	company datasets (any)	8% for sellers	Calls itself a private data brokerage. Does not perform PII stripping or legal certification; claims no standing agreements with companies in its buyer directory; a transaction 'may take weeks or months, or may not close'.	11 Oct 2026	Stated
FileYield (data buyers)FileYield	company datasets (any)	6% for buyers	Calls itself a private data brokerage. Does not perform PII stripping or legal certification; claims no standing agreements with companies in its buyer directory; a transaction 'may take weeks or months, or may not close'.	11 Oct 2026	Stated
DataVendor (sellers of codebases, datasets and tasksets)aitrainer.work (third-party referral and job site)	codebases, datasets and tasksets	seller keeps 80% of each sale	Sales mostly non-exclusive; sign-up by invite. Third-party description, not DataVendor's own page.	11 Oct 2026	Reported
micro1 (referrers)micro1 Inc.	company data (license or purchase)	seven and one-half percent (7.5%)	Applies only if no Referral Payout is specified on the Site for that company. Payout is a gross amount inclusive of taxes and fees. The clause is unchanged in the version 'Last Updated: October 14, 2026' (read 2026-10-29), which adds a Sub-Affiliate Bonus and an Influencer Bonus, amounts set on the Site.	8 Oct 2026	Stated
micro1 (referrers)micro1 Inc.	company data (license or purchase)	no amount (discretion clauses)	Timing and manner at micro1's sole discretion; no payout earned 'unless and until micro1 confirms it as payable'; holdback or clearance period possible; clawback and set-off if the company deal is cancelled, refunded, reversed or unwound; micro1 decides 'the original lead source'; termination at any time without notice, after which 'all pending and unpaid Referral Payouts are forfeited, except as required by applicable law.' Unchanged in the version 'Last Updated: October 14, 2026' (read 2026-10-29).	8 Oct 2026	Stated
Grepped (Erik Newsham)ARR Club	brokered company data	$4M MRR ($48M ARR run rate)	Unaudited founder claim relayed by a tracker; no buyers, deals or payments disclosed.	6 Oct 2026	Self-reported
Troveo (company data owners (sellers); more than 40 active buyers)Troveo	business data (operating records) among other licensed data	no fees and no deductions from payouts; more than 40 active buyers	Seller signs one license; Troveo handles rights review, packaging and every buyer conversation; the company is never named publicly; paid on every sale. Restates over $20 million paid to rights holders.	6 Oct 2026	Self-reported
BuyMyDeadStartup (buyers of dead-company archives)BuyMyDeadStartup	defunct-company archive (code, documents, operational records)	$1,500 a month	Free for sellers, no cut of any sale; listing board only, no custody of data; says it is not a broker or escrow.	9 Sep 2026	Stated
SimpleClosure (venture-backed startups (sellers); foundation labs and RL-environment builders (buyers))SimpleClosure press release (syndicated via Webull)	source code and operational data (AssetHub)	more than $10 million	350+ companies supported; AI buyers up 9X versus 2025; AssetHub opened to operating companies; free for SimpleClosure dissolution customers (fee for others not disclosed). Multi-pass de-identification; sensitive employee and health records not sold.	3 Sep 2026	Self-reported
micro1 (multiple customers)TechCrunch	off-the-shelf datasets resold to multiple customers	as high as 80% to 90%	Figure from an unnamed person familiar with micro1's finances.	20 Aug 2026	Reported
ProtegeThe Financial News (fnnews.com), summarizing The Information	licensed training data (mixed; Protege's catalog is mainly clinical and media)	from $30 million last year to at least $100 million this year	CEO Bobby Samuels' figures, relayed from The Information.	14 Aug 2026	Self-reported
Ooak Data (multiple frontier labs (unnamed))Y Combinator (Launch YC post by Ooak Data)	company records turned into anonymized 'digital twins' for agent training	well above 7 figures in revenue; data from 20 companies	Seeks companies with 20+ FTEs, operating, winding down or recently closed.	14 Aug 2026	Self-reported
Verb (consumers (sellers); AI labs expected as biggest buyers)Axios	consumer personal data (shopping habits, social media use; adjacent to company data)	no price disclosed	Users set their own price; Verb provides an estimate; 'The more you share, the more you can charge.' Excludes passwords, text messages and health data. CEO Cyrus Beschloss expects AI labs to be the biggest buyers.	6 Aug 2026	Stated
Mercor (referrers)UsagePricing (third-party pricing tracker)	company workflow data	up to $100K per referral (shown 29 Jul to 22 Sep 2026)	Tracker records the figure on Mercor's Data page from 2026-07-29 and its removal in the 2026-09-22 rebuild. Not visible on the live page on 2026-10-11.	29 Jul 2026	Reported
Troveo (7,000+ licensors (rights holders))Troveo	mixed licensed data (mainly video and audio; business data program added)	over $20 million paid out; 95 percent of licensors exclusive	Commission not disclosed here; Troveo's 6 Oct 2026 guide says it takes no fees (see that record).	23 Jul 2026	Self-reported
micro1 (referrers)micro1 forum	company operational data	up to $25,000 per referred company; $5,000,000 committed to the program	Paid once the referred company enters a paid data partnership with micro1; company must identify the referrer during onboarding. Superseded by the $50,000 headline on the referral page read 11 Oct 2026 and by the tiered terms read 29 Oct 2026.	16 Jun 2026	Stated
LabelSets (dataset sellers)Toolradar (third-party tool directory)	labeled datasets (marketplace)	15% commission; sellers keep 85%	Free to list, per the directory. Not confirmed on LabelSets' own site.	26 May 2026	Reported
Datarade (Monda) (data providers)ZoomInfo Pipeline (competitor of Datarade)	datasets (general data marketplace)	30% commission	Page served in Dutch. Also: quote-based annual provider subscriptions, $0.30 per GB egress on Monda-hosted infrastructure, free for buyers. Competitor-published.	22 May 2026	Reported

Sample. 30 records, to 29 Oct 2026, each from the publication linked beside it.

Source. AITraining.jobs Research, Company Data Brokerage: Buyers, Prices and Terms, live data.

Figure 5

Law and policy that shape the trade

Which rulings, laws and policies change what company data can be sold?

Who	What	Price or terms as published	Terms	Date	Basis
Spirit Airlines debtors (U.S. Bankruptcy Court, S.D.N.Y.) (Google (successful bidder); Mercor (alternate bidder); micro1 (notice of competing bid))Chapter 11 Cases (Stretto)	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$10 million sale; hearing moved to 6 Nov 2026	Reply deadline 4 Nov 2026, 12:00 p.m.; hybrid hearing in White Plains. The ombudsman's 5 Oct supplement (Dkt. 1684) supports the sale to Google; if the court considers Mercor she asks to investigate its early-2026 supply chain security incident first, and she asks for the same review of micro1, which proposed to de-identify the data itself. The hearing falls after this review; no ruling to 29 Oct 2026.	9 Oct 2026	Stated
Members of Congress (led by Sen. Elizabeth Warren and Rep. Steven Horsford) (Google; Spirit Airlines)The Record (Recorded Future News)	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$10 million deal	Letter signed by 121 lawmakers per the article (Reuters: 'more than 120'). Asks to exclude as much employee information as possible, set use limits and run an independent employee confidentiality review.	8 Oct 2026	Stated
Frankfurt Kurnit Klein & Selz (Daniel M. Goldberg) (companies selling data)Legal 500 (Frankfurt Kurnit Klein & Selz, Daniel M. Goldberg)	company records (email, chat, code, HR)	n/a	Also warns that once data trains a model 'it cannot practically be pulled back out' and advises excluding privileged, HR, payroll, health, union and customer-confidential material rather than relying on de-identification.	6 Oct 2026	Stated
Consumer privacy ombudsman (Lucy L. Thomson) (Google; Spirit Airlines)Business Insider (via Yahoo Finance)	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	$10 million sale	Ombudsman recommended approval; approval is up to the court; hearing scheduled for 14 October (adjourned on 9 Oct 2026 to 6 Nov 2026; see that record). Any other bids (Mercor, micro1) would need additional consumer-privacy review.	5 Oct 2026	Reported
California Privacy Protection Agency (registered data brokers)Husch Blackwell (Byte Back)	personal data inside brokered datasets	$6,000 plus up to 2.99 percent processing fee	Since 1 Aug 2026 brokers must process DROP deletion requests at least every 45 days; $200 per day fine for failure to register; independent audits from 2028. Oregon $600, Texas $300.	5 Oct 2026	Reported
U.S. Court of Appeals for the Third Circuit (Thomson Reuters v. ROSS Intelligence)LawNext (LawSites)	proprietary third-party content (Westlaw headnotes) used for AI training	n/a	Ruled for Thomson Reuters on copyrightability and fair use (training use not fair use); ROSS says it will seek Supreme Court review.	30 Sep 2026	Stated
State data-broker registries (CA, OR, TX, VT; CT and NJ to follow) (data brokers selling or licensing personal data)Husch Blackwell (Byte Back)	personal data inside brokered datasets	n/a	Definitions differ by state; California covers selling personal information of consumers with whom the business has no direct relationship.	30 Sep 2026	Reported
U.S. Court of Appeals for the Ninth Circuit (Doe v. GitHub (GitHub, Microsoft, OpenAI))Haynes Boone	open-source code used for training	n/a	Open-source licence contract claims continue in the district court; the panel took no position on whether Copilot output infringes.	16 Sep 2026	Stated
Consumer privacy ombudsman (Lucy L. Thomson) (Google; Spirit Airlines)Research Suite by Stretto (chapter11cases.com), docket summary	bankruptcy-estate corporate archive (de-identified enterprise data: email, Teams, files, tickets, code)	n/a	Report conditions approval on recommended conditions (Attachment A, not yet filed as of 11 Sep). Sale hearing adjourned to 30 Sep 2026 [Dkt. 1601]. Google's 9 Sep response argued the ombudsman's role is limited to personally identifiable information.	8 Sep 2026	Stated
U.S. Department of Justice (OpenAI copyright MDL (S.D.N.Y.), addressing The New York Times' claims)IPWatchdog	copyrighted text used for LLM training	n/a	Non-binding; argues training is 'exceedingly transformative'. Bears on public content, not on contract and privacy limits on private company data.	1 Sep 2026	Stated

Sample. 10 records, to 29 Oct 2026, each from the publication linked beside it.

Source. AITraining.jobs Research, Company Data Brokerage: Buyers, Prices and Terms, live data.

For labs, data vendors and researchers

Buy the data behind this report

The figures above are a summary. Behind them are records we collect ourselves, every day:

0 roles tracked, each with its history
Every role on 0 job boards, read since n/a (every hour since 18 Oct 2026): its pay and place, and the first and last day it was open.
0 apply clicks in 28 days
Which roles people actually apply to, by field, pay and kind of work. It happens on this site, so nobody else can count it.
39 lab and vendor hiring boards
The job boards of 25 AI labs and 14 human-data vendors, read every day: who is hiring for human data, and who is hiring to buy it.

What you can buy

The records
The rows behind any figure, with their history, not only the totals.
A regular feed
The same figures every week or month, as files.
A custom cut or benchmark
Your countries, languages or fields, or your own platform against the market.

Building AI and need the experts behind this data? (you contract them yourself) or (we deliver the expert work).

Investing in AI-data companies? Hiring radar is these records role by role, every day, as files.

Method and definitions

Company announcements and terms pages, court filings and reporting, each linked; live roles from the labs' and vendors' public job boards. The live figures are computed every day from this site's own data, the same way for every reader; every sourced row links the publication it is taken from, with the figure as that publication states it. Each figure states its sample beneath it.

Company data: A company's operating records (email, chat, tickets, code, CRM, documents), licensed or sold, usually de-identified, to build RL environments and training sets.
Stated, reported, self-reported: Stated: a party's own announcement or a filing. Reported: press. Self-reported: a broker or founder describing its own business.

What this data cannot tell you

Each report sets its sourced records beside a live figure computed from this site's own data every day; the sourced records are added as they are published and reviewed.
Compiled from public sources: an event that was never announced or reported is missing, and private companies' figures are what they or the press stated, not audited.
Most deals are private: the published prices are the ones parties chose to announce or courts made public.

How the job feeds are read and roles are classified: site methodology. How fresh, complete and consistent the data is, checked every day: data quality and coverage.

Editions

The live report is the dataset as last reviewed, beside figures recomputed every day from this site's own data; new events are added as they are published. Each complete month is also published as an edition, frozen two days after the month ends, so a figure you cite from an edition never moves.

Live data (this page)Sources as last reviewed, live figures daily
October 2026 editionFreezes on first reading

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 (creativecommons.org/licenses/by/4.0) with a link to this page. Live figures change daily: cite a monthly edition for numbers that will read the same later, or keep the access date.

Citation

AITraining.jobs (2026). Company Data Brokerage: Buyers, Prices and Terms (live data, 12 Jan 2026 to 29 Oct 2026). https://cyberjobs.cc/research/company-data-brokerage (accessed 11 Nov 2026)

BibTeX

@misc{aitrainingjobs_company_data_brokerage_live,
  author = {{AITraining.jobs}},
  title = {Company Data Brokerage: Buyers, Prices and Terms (live data, 12 Jan 2026 to 29 Oct 2026)},
  year = {2026},
  howpublished = {\url{https://cyberjobs.cc/research/company-data-brokerage}},
  note = {Accessed 11 Nov 2026},
}

Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 81 research reports