How much is your company's data worth to AI companies?
Public deals for a company's own records run from about $25,000 to $10 million. See what sets the price, where a small company stands and how to get estimates.
Last checked November 20, 2026
Your company’s own records, such as Slack or Teams history, email, support tickets and documents, can be worth anywhere from $25,000 to several million dollars to an AI company. This is about a company licensing what it owns, not about the personal data that apps sell about you. Where your records land depends on how many people wrote them, over how many years, and how hard they would be to find anywhere else.
Public deals run from $25,000 to $10 million
The biggest price on record for one company’s working records came out of a bankruptcy. Spirit Airlines sold former employees’ email and chat to Google, an estimated 100 million emails and half a billion Microsoft Teams messages.
Google agreed to leave out health and medical data, drug and alcohol test results, safety and disciplinary records and identification documents. From The Record, November 18, 2026.
After the auction closed, micro1 offered $12.5 million for the same data, Fortune reported. The court approved Google’s purchase instead.
A company of 60 to 100 people sells for less, often in six figures. Telegraph lists three deals it paid for: $350,000 to a 60-person accounting firm, $450,000 to a 70-person engineering firm and $650,000 to a 100-person freight broker. At the small end, a closed crypto startup told MIT Technology Review it sold its code for about $25,000. Grepped’s founder says its average deal is now $1.1 million to $1.3 million, a figure nobody else has checked.
Reddit’s licensing deals came to $203 million over two to three years, according to its IPO filing. That’s a platform with millions of users, so don’t price a company against it. Every deal we can source is in our deals ledger.
Published price ranges start at $10,000 and pass $3 million
Several AI companies publish a range before they see your records:
- micro1 pays $100K to $500K for a qualified partnership, $500K to $3M for large-scale datasets or ongoing participation across teams, and $3M+ for “highly unique, proprietary operational data.”
- Telegraph lists data licensing payments of $100K to $4M, “subject to data review and agreed terms.”
- Scale gives “$10K–$1M+” as an “illustrative value per data partnership, scaling with cadence,” which means it pays more for records delivered on a schedule.
These are ranges, not quotes. Each company sets its price after it reviews a sample of your records.
People, years and location move the price most
More people over more years leave more examples of work, so both raise the price. micro1’s own estimator shows about $203,000 to $377,000 for a US company with 30 staff and five years of records, and $379,000 to $888,000 for 100 staff and ten years. micro1 and Telegraph both cut the figure for companies outside the US: Canada and Europe count at three quarters, everywhere else at 30%.
What the records hold matters too. Telegraph says it weighs “the data’s depth, quality, originality, and permitted uses.” SourceX says “the most valuable packages cover many systems,” and it values records that are hard to reproduce or that capture expert workflows. A decade of engineering calculations or legal matter files is worth more than the same years of lunch plans in a general channel.
Terms change the number as well. An exclusive license pays for a promise not to sell the records again, and records you can’t license, like a client’s confidential files, shrink the package. Exclusive or non-exclusive sets out what each company asks for.
See which AI companies would take your records
Answer up to eleven questions and get matched with several potential buyers at once. It's free, and we never email or call you first.
A company shows up here once your answers meet everything it asks for.
Your answers stay anonymous unless you ask for an introduction.
A small company’s records can still be worth something
Most published size lines sit between 30 and 50 people, and most of them bend:
| Company | Size it asks for |
|---|---|
| micro1 | 30+ full-time employees |
| Scale | “40+ employees is our preferred starting point”, assessed case by case |
| SourceX | Usually 50+ full-time employees at peak; “a smaller company with truly one-of-a-kind data can still be a strong fit” |
| Maryah | Suggests 30+ people and three years, but “fit depends on the data, not just company size” |
| Telegraph | “We work across industries and company sizes” |
| Mercor Data | Its law firm listing takes 5 to 300 attorneys and staff |
A 12-person company with five years of Slack history is under most of these lines. A 12-person law firm, medical practice or engineering shop with deep, specialist records may not be. Industry pages show which companies name your field.
Estimates come first, and the offer follows a sample
The AI companies that publish estimators call the figure a starting point. SourceX says “an estimate is indicative, and not every qualifying company receives an offer.” The worth estimator lets you compare estimates from several potential buyers side by side, and how AI data buyers estimate payouts shows the formulas behind them.
If an offer arrives before you’ve compared anything, run it through the offer check first.
Questions owners ask about price
Are a closed company's records worth anything?
Yes. micro1 has a program for companies that no longer operate, open to all industries, that pays $100K to $3M+ for approved data packages from companies with 30+ employees and records in English. Selling after shutdown covers what to keep before you cancel the tools.
Is the payout a one-time fee or ongoing?
Usually one fee. SourceX says “there are no royalties or usage fees.” Maryah offers a choice between one payment and a share of licensing revenue, and Scale pays more for records delivered on a schedule.
Does Dataconto take part of my payout?
No. What you’re paid is between you and the buyer. How we get paid explains our side.
Buyers in this guide
Sources
- The Record, Judge signs off on Google purchase of Spirit Airlines data (November 18, 2026), checked November 20, 2026
- Fortune, micro1 tries to beat Google's bid for Spirit Airlines' data (September 14, 2026), checked November 15, 2026
- MIT Technology Review, AI labs are buying data from dead startups (October 13, 2026), checked November 15, 2026
- FourWeekMBA, quoting Grepped's founder on average deal size (October 24, 2026), checked November 15, 2026
- Reddit, Form S-1, filed February 22, 2024, checked October 22, 2026
- Telegraph Lab, Data Partnerships (payments and past deals), checked November 20, 2026
- micro1, Company data (tiers and eligibility), checked November 20, 2026
- micro1, data payout estimator, checked November 4, 2026
- micro1, Defunct Company Data Partnership, checked November 20, 2026
- Scale AI, Data Partnerships, checked November 20, 2026
- SourceX, FAQ, checked November 20, 2026
- Maryah, business data (payment options), checked November 20, 2026
- Maryah, How much is business data worth, checked November 4, 2026
- Mercor Data, law firm listing, checked November 5, 2026
More guides
- Acqui-hired? What happens to the old company's records
After an acqui-hire the old company often keeps its Slack, tickets, docs and some code. Find out who owns them now, who can sign, and which buyers license them.
- Before you license company data to AI: a rights checklist
Seven checks before you sign: who signs, customer contracts, personal data, health records, your privacy policy, vendor terms and third-party files.
- Can a company in administration or bankruptcy sell its data?
Usually yes, if the administrator, trustee or assignee signs. Learn who has authority, why customer personal data stays out and who licenses the records.