Who Sells Data (Secludy)
Who's buying data
A sourced comparison of programs that pay companies for their data: what each pays, the data it wants and who qualifies, with quotes as of Oct 14, 2026. It is the best single page for comparing buyers.
Learn · 109 sources
Read what buyers, brokers, lawyers and critics are saying
Every card links to the original and names its author. Posts from buyers and our referral partners are labelled. We show the best-rated items first; switch on “everything” for the full set, critics included.
37 shown
A micro1 referral partner lays out the micro1 company-data program: one-time payments of roughly $200K to $2M+, who qualifies (US, 20+ desk staff, 2+ years, 5+ tools), excluded data types, the export and QA timeline, and a flat referral fee. The author is paid through referrals, so the piece is promotional.
if you sell AI training data, see how you compare: we found 159 companies selling data, RL envs and evals, every fact sourced plus 38 programs that pay companies for their data: whosellsdata.com 8 of the 14 that pay companies directly publish a dollar range. most of the Show more
Secludy's CEO launches Who Sells Data, a sourced directory of 159 AI-data vendors and 38 programs that pay companies for their data. It notes that only 8 of the 14 direct buyers publish a dollar range.
MIT Technology Review
From a buyerAI labs are buying data from dead startups to train their models
A reported feature on the wind-down industry's data-resale arm. One founder netted about $25K for code with more pending for workspace data, and SimpleClosure says 47% of its deals now include code or data. Sunset prices data on headcount, years, volume and tool usage, while lawyers warn the data can be 'legally radioactive'.
Business Insider
AI labs are offering companies big money to hand over their employees' private Slack DMs and emails
An investigation into the operating-company market. Owners get six-figure 'exploding offers' with 48-hour deadlines, and data is priced by volume and anonymized before resale. It quotes sellers, brokers and a law professor who calls employee consent an ethical gray area.
A broker reports the funnel behind referral outreach for data deals: about 7,800 contacts in three weeks produced 46 engaged leads and 13 next steps, mostly architects, law firms and accountants. It also compares referral terms across micro1, Polyshares and Pewter.
Pewterdata.com has a $30M mandate to purchase business data over the next 8 weeks. We're actively buying from companies with 20 or more employees and at least 5 years in business across six verticals: Service. Manufacturing. Logistics. Healthcare. Ecommerce. Show more
Pewter says it has a $30M mandate to buy business data within eight weeks from firms with 20+ employees and 5+ years in business, across six verticals. It offers referrers 10% of deal value, capped at $100K.
Corey Ganim
he made $1,000,000 in 30 days as an AI data broker
A full podcast interview with Polyshares co-founder Ryan Lock on how a data broker finds companies, does diligence, de-identifies records and sells licenses to labs. The description includes a Polyshares referral link, so the host has a commercial interest.
10K viewsWatch on YouTube ↗
beri.net
micro1 Offers $100K+ for CRM Records That May Not Be Yours to Sell
Argues that micro1's offer will likely reach a business-unit head before legal does. It notes that the program's public terms name no end buyer or exclusivity and that CRM records may hold customer data you cannot license, and recommends taking inventory and checking rights first.
Grepped's founder describes what labs are actually asking for (finance, legal, semiconductors, manufacturing, niche workflows, RL environments with graders) and warns that a generic email archive is worth less than owners hope. He gives typical deal sizes, exclusivity norms, the payment timeline and referral terms, and he flags misconceptions spreading among new brokers.
This guy did $1,000,000 in 30 days as an AI data broker. His model: frontier labs have used all the data on the internet and need private company data to keep improving. His firm finds businesses with the right data then de-identifies it, packages it, and sells it to the labs. Show more
A thread summarising a podcast interview with Polyshares co-founder Ryan Lock: labs buy full workflows including rejected fixes, small companies are often not worth the fixed legal cost, names are replaced with stable pseudonyms, and sellers should avoid perpetual licenses.
The Next Web
micro1 commits $1bn to company data for AI agents, financed by Citi
micro1 plans to spend $1B over 12 months licensing company data, financed by Citi and Hercules Capital. Payouts are tiered at $100K+, $500K+ and $1M+, US firms with 30+ employees are prioritised, and some datasets are synthetically rewritten.
micro1
From a referral partnermicro1 announces commitment to spend $1B on enterprise data acquisition
micro1's official announcement of $1B for enterprise data over 12 months. It explains that de-identified company data is used to build RL environments reflecting real business operations.
OfficialRead ↗
Polyshares' Ryan Lock explains what buyers want from brokers: arrive with a data inventory (each tool, years available, volume) and realistic price expectations. He lists the deal terms to negotiate, such as exclusivity, acceptance, payout timing, scope of use, sub-licensing, de-identification and liability, and notes that switching chat tools mid-history cuts value.
Frankfurt Kurnit Klein & Selz (Technology Law blog)
Selling Company Data for AI Training: What to Know Before You Say Yes (via Passle)
A law-firm client alert with a practical checklist: confirm you have the right to sell (third-party confidential data, platform API terms, privilege), honour old privacy promises, check employee notices and NLRA issues, and remember that CCPA treats such a transfer as a sale. It recommends excluding high-risk data rather than relying on scrubbing, and tight deal terms.
micro1
What every business owner should know abut AI data licensing
micro1 and a referral partner explain to business owners what data licensing involves, which kinds of operational data are valuable, and how anonymization protects what makes a business unique.
211K viewsWatch on YouTube ↗
Mercor has now partnered with > 300 enterprises to help them monetize their data, paying companies up to $10M per enterprise. Across these companies, we’ve encountered 90 different systems: -55% contained Slack -53% GitHub -53% Google Drive -50% Gmail -40% Figma -35% Jira A Show more
Mercor's CEO says the company has partnered with more than 300 enterprises, paying up to $10M each. He shares which systems those companies used (Slack 55%, GitHub 53%, Google Drive 53%, Gmail 50%, Figma 40%, Jira 35%) and notes that a 200-person firm runs on many of the same tools as a large enterprise.
Troveo
From a buyerExclusive vs Non-Exclusive Data License, Explained
A guide to exclusive versus non-exclusive data licenses. Exclusivity raises the price of one deal but blocks other buyers for the term, and field-of-use carve-outs are a middle path. It is written by a data marketplace.
TIME
Why Google Bid $10 Million for a Failed Airline’s Data
An explainer on why Google paid $10M for Spirit's corporate data: RL environments for agents that do white-collar work need real emails and chats. Surge's Nick Heiner frames the purchase as a bet that agents will generalise across the economy.
An investor essay describing real-world company data as labs' newest procurement category. It cites fast revenue growth at suppliers like Mercor and Fleet and warns that demand is lumpy and concentrated in a few labs.
Hacker News
Google has acquired the data of failed US airline Spirit
The largest Hacker News discussion of the trend. Commenters doubt that 100M emails and 30M call recordings can really be de-identified, question the legality, and speculate the data will be used to build simulated airline environments.
The Register
Google buys crashed airline Spirit’s data at auction, because AI
A detailed inventory of what $10M bought from Spirit: over 100M emails, 500M Teams items, 30M recorded calls, plus SharePoint, OneDrive, ServiceNow, Oracle and SAP records. It is a useful reference for what 'company data' means in practice.
Reuters
Google to buy Spirit Airlines business data for $10 million
A wire report that court filings show Google as the winning $10M bidder for Spirit Airlines' business data, with Mercor as backup bidder.
NewsRead ↗
Sequoia Capital
RL Environments Explained: How AI Agents Learn Real-World Work | Brendan Foody, Mercor
Mercor's CEO explains the three parts of an RL environment: worlds (a real project's messages, docs and files), apps (clones of tools like Salesforce) and tasks with rubric-based graders. It is the clearest explanation of why labs want internal company records.
55K viewsWatch on YouTube ↗
micro1
The hidden asset every business already owns
The founder of a marketing agency explains why his company joined micro1's data partnership and what he chose to share. It is one of the few on-record seller accounts, though it was published by the buyer.
1.2K viewsWatch on YouTube ↗
The Information
Why Data Labeling Startups Are Paying HVAC Companies $150,000 For Data
The Information reports how micro1 approached the CEO of a Troy, Michigan HVAC company in June to license his firm's data. It is the clearest example of buyers moving past tech startups into ordinary trades businesses at about $150K per deal.
RL environments are used to train AI models to operate in the real world. To build them, we use data that captures how companies across different industries make decisions, collaborate, and get work done. Through the micro1 Company Data Partnership Program, we're partnering with Show more
micro1 explains that it uses company operational data to build RL environments, and that its program pays $100K to $2M+ for anonymized data. It also says referral rewards rose to up to $50K.
micro1
How to earn up to $2M training AI models with anonymized company data
micro1's CEO walks through the Company Data Partnership Program: which business workflows qualify and how compensation of $100K to $2M+ is set by the scope and value of the data.
1.2K viewsWatch on YouTube ↗
Chris Barber
How to sell RL envs and data to AI labs: Interview with Sean Cai
An interview with Sean Cai on which data areas are under-supplied, what the sales process with a lab looks like, and why labs renew or expand purchases. He says good-quality data is rarely turned down for budget reasons.
3.4K viewsWatch on YouTube ↗
Forbes
AI's New Training Data: Your Old Work Slacks And Emails
The story that started the trend. cielo24's former CEO sold 13 years of Slack, email, Jira and Drive records through SimpleClosure for 'hundreds of thousands of dollars', and buyers like Surge AI use such data for RL gyms. Privacy experts raise concerns.
AI labs are paying hundreds of thousands of dollars to buy email, Slack and Jira threads from dead startups as feedstock for ‘reinforcement learning gyms,’ which specialize in using defunct company data to build simulated work environments forbes.com/sites/annatong…
One of the first widely shared posts on the trend: AI labs paying hundreds of thousands of dollars for dead startups' email, Slack and Jira data to build RL 'gyms', relaying Forbes' April 2026 report.
Epoch AI (Gradient Updates)
An FAQ on reinforcement learning environments
Based on interviews with 18 people across RL-environment startups and labs, this FAQ finds that contracts often run six to seven figures per quarter and individual tasks cost about $200 to $2,000. Enterprise workflows are named as a major growth area.
Surge AI
RL Environments and the Hierarchy of Agentic Capabilities
Surge AI put nine frontier models in a simulated workplace with 150 jobs and found most lacked basic common sense. It shows concretely why labs want realistic company environments.
BlogRead ↗
TechCrunch
Silicon Valley bets big on 'environments' to train AI agents
The late-2025 explainer on RL environments, including reports that Anthropic leaders discussed spending over $1B on them. This is the demand that later drove company-data purchases.
EPIC (Electronic Privacy Information Center)
EPIC Files Amicus Brief Challenging Spirit Airlines’ Bankruptcy Sale of Employee Data to Google
EPIC and Prof. Seema Patel's amicus brief arguing the Spirit sale agreement protects consumer data but not employee data, and that de-identification does not prevent inferences about workers.
micro1
From a referral partnerMonetize your company's operational data
micro1's program page lists eligibility (30+ employees, US first, English), payout tiers ($100K+, $500K+, $1M+), the data it wants (SOPs, CRM, project histories) and sample review before use.
OfficialRead ↗
Handshake AI
From a buyerData Partnerships: Get Paid for Your Data
Handshake AI buys operational data across industries, with partnerships from $100K to $4M+ depending on what is shared. Data is de-identified, and use is limited by contract.
OfficialRead ↗