Don't Do It Yourself

How to Sell Biotech and Life Sciences Data to AI

A biotech or life sciences company sells its data to AI by licensing its own lab notebooks, experiment run records and trial documentation to a buyer like Mercor.

Mercor pays biotech and life sciences labs $50,000 to $875,000. It wants several years of unpublished experiment archives from an organization of 10 to 1,000 people, anywhere.

Failed runs count. Mercor says they’re “as valuable as successful ones.”

For the full process across every industry, start with our guide to how to sell your company’s data to AI companies.

Key takeaways

  • Mercor’s biotech listing pays $50,000 to $875,000 and wants 1,000 or more runs with inputs and measured outcomes.
  • It rules out published data and data owned by a sponsor or partner.
  • Wound-down biotechs and labs with years of archives are what Mercor calls a strong fit.
  • Human-subjects data has to be de-identified, and the original consent decides what you can do with it.
  • Records describing an invention you haven’t filed on need patent counsel before they go anywhere.

What lab data do AI companies buy?

AI companies buy the record of how science gets done: lab notebooks, structured run records with measured results, analysis files and de-identified trial documentation. Mercor wants unpublished work the organization owns, including failed experiments, in systems like Benchling, LabArchives or a LIMS.

Mercor’s Biotech and Life Sciences Lab listing, read November 13, 2026, splits it up like this:

Data Mercor’s view Main risk
Lab notebooks and run records Core of the license Ownership, if partners or sponsors funded the work
Measured results and analysis files Core of the license Unfiled inventions
Trial documentation, de-identified Core of the license Consent scope and patient privacy
Clinical study reports with patient-level data Adds to what you earn De-identification and sponsor ownership
Genomic data collected with research consent Adds to what you earn Consent terms and controlled-access rules
In-house assay and analysis tools Adds to what you earn Third-party code inside the tools
Scientists’ write-ups of experiments Adds to what you earn Names and unpublished findings

Mercor names the companies it wants:

  • Biotech and biopharma companies
  • Contract research organizations
  • Genomics and diagnostics labs
  • Consumer genomics companies
  • Academic spin-outs

It works with Benchling, LabArchives, Dotmatics, LIMS systems and Medidata Rave.

Mercor also runs a broader R&D records listing. It covers chemistry, materials, batteries, preclinical biology and clinical research, and wants the reasoning: hypotheses, protocols, decision logs and failed work.

How much is biotech data worth to AI companies?

Mercor’s Biotech and Life Sciences Lab listing pays $50,000 to $875,000, depending on the volume and depth of records. Its separate chemistry and materials R&D records listing pays $10,000 to $300,000. Volume matters: Mercor asks for 1,000 or more runs.
Mercor listing Company earns Size bar
Biotech and life sciences lab $50,000 to $875,000 10 to 1,000 employees
Chemistry and materials R&D records $10,000 to $300,000 Not stated
Semiconductor and materials $100,000 to $750,000 10 to 2,000 employees
Medical practice or health system $80,000 to $820,000 10+ clinicians or 2,500+ patients

Depth drives the number. A run record with its protocol, raw outputs, analysis and the decision that followed is worth more than a results table on its own.

Mercor also shows a referral fee of up to $110,000 for biotech labs. That’s paid to someone who refers a lab, not to the lab, and the full amount needs 30 or more employees.

Our guide to how much company data is worth to AI compares payouts across industries.

Can you sell human-subjects and patient data?

Only in de-identified form, and only if the consent and your agreements allow it. Under the Common Rule, data counts as human-subjects research when the subject’s identity can readily be ascertained. HIPAA covered entities also need patients’ authorization to sell their health information.

The Common Rule, 45 CFR 46.102, covers research that uses “identifiable private information.” That’s information where the subject’s identity “is or may readily be ascertained” by the investigator.

45 CFR 46.104(d)(4) exempts secondary research where the data is recorded so subjects can’t readily be identified. The recipient also mustn’t contact or re-identify them.

HIPAA adds a payment rule for covered entities. 45 CFR 164.508(a)(4) requires an authorization that states the disclosure “will result in remuneration.”

Genomic data needs extra care. The NIH Genomic Data Sharing Policy applies to NIH-funded genomic research submitted from January 25, 2015, and controlled-access data comes with its own use terms.

Before anything goes in scope, pull the consent forms. Check whether they covered future research use, sharing with companies and commercial use.

Our guide to anonymizing data for AI training covers HIPAA’s 18 identifiers. The medical practice and digital health guides cover patient records in more depth.

Do sponsor, collaboration and grant agreements stop you selling lab data?

They often decide who owns it. Mercor rules out data owned by a sponsor or partner, which covers most CRO work done for a client. Collaboration agreements can split data rights, and federal grants give the government rights to the data they paid for.

Sort your archive by who paid for the work:

  • Your own programs: work you funded. Usually yours to license.
  • Sponsored studies: CRO and contract work done for a client. The client usually owns the data.
  • Collaborations: joint work with a university or company. The agreement decides who can license what.
  • Federally funded work: grants and contracts. The government has rights to the data.

For federal grants, 2 CFR 200.315(d) gives the government the right to “obtain, reproduce, publish, or otherwise use the data produced under a Federal award.”

That doesn’t stop you licensing grant-funded data. It does mean you can’t promise a buyer that nobody else will ever get it, so raise it before you agree to exclusivity.

Should you file patents before licensing lab data?

Yes, if the records describe an invention you want to protect. Mercor wants unpublished data, which is exactly where unfiled inventions sit. US law treats public disclosure before filing as prior art, with a one-year grace period for the inventor’s own disclosures.

35 U.S.C. 102(a)(1) bars a patent on an invention that was “otherwise available to the public” before filing. Section 102(b)(1)(A) gives inventors a one-year grace period for their own disclosures.

Whether a license under NDA counts as making something public is a question for patent counsel, not a buyer. The safe order is to file, then license.

The commercial risk is separate. A model trained on your unpublished results may help a competitor reach the same answer faster.

Older programs and abandoned projects carry the least risk. They still show how your scientists work.

What should you leave out of a lab data deal?

Leave out data owned by sponsors or partners, anything already published, identifiable patient or donor data, genomic data whose consent didn’t cover sharing, and records describing inventions you haven’t filed on. Mercor rules out the first two itself.
  • Sponsor- or partner-owned data without written permission
  • Published data, which Mercor doesn’t want
  • Identifiable patient or donor data, including sample IDs that link back to people
  • Genomic data collected under consent that didn’t allow commercial sharing
  • Unfiled inventions until counsel clears them
  • Current lead programs you’d rather keep as trade secrets

Who buys biotech and lab data?

Mercor is the buyer publishing ranges for lab data, through its biotech listing and its R&D records listing. micro1 buys operational data, such as SOPs and documentation workflows, from companies with 30 or more staff, which can fit a lab’s processes but not its results.
  • Mercor: lab notebooks, run records and trial documentation from labs of 10 to 1,000 people. Pays $50,000 to $875,000.
  • micro1: operational processes and documentation, listed on its data partnerships page. It wants 30 or more full-time staff and publishes tiers from $100,000 to over $3 million. You can apply to micro1 here.

Our list of companies that buy business data for AI has each buyer’s terms.

What are the steps for a biotech lab, start to finish?

A lab licenses its data in eight steps: count runs and years of records, sort by who paid for the work, pull the consent forms, flag unfiled inventions, apply to a buyer, de-identify and clean the export, have counsel review the license, then check a sample.
  1. Count your records. Runs with inputs and measured outcomes, failed ones included, and how many years they cover.
  2. Sort by who paid. Your programs, sponsored studies, collaborations and grants each have different rules.
  3. Pull the consent forms for any human-subjects or genomic data, and check what they allowed.
  4. Flag unfiled inventions. Patent counsel decides what stays out.
  5. Apply to a buyer. Mercor asks what you keep, in which system, whether you own it and under what consent.
  6. De-identify and clean. Remove scientist names, sample IDs that link to people, and partner or sponsor names.
  7. Have counsel review the license. Our AI data licensing agreement checklist lists the clauses to check.
  8. Check a sample before delivery, and confirm nothing published or sponsor-owned slipped in.

Common mistakes when selling biotech data to AI

The most common mistake is treating CRO and sponsored work as the lab’s own data. These are the others:

  • Throwing out failed runs, which Mercor values as much as successes
  • Assuming old consent forms cover licensing to a company
  • Leaving sample IDs that link back to donors or patients
  • Promising exclusivity on grant-funded data
  • Licensing records on an invention before the patent is filed
  • Including data that’s already published
  • Reading Mercor’s $110,000 referral fee as what the lab gets paid

Frequently asked questions

Can a biotech company sell its data to AI?

Yes, for records it owns outright. Mercor’s biotech listing wants lab notebooks, run records and trial documentation the company owns, and rules out data owned by a sponsor or partner and data that’s already published.

How much do AI companies pay for biotech lab data?

Mercor’s Biotech and Life Sciences Lab listing pays $50,000 to $875,000, depending on the volume and depth of records. Its separate chemistry and materials R&D listing pays $10,000 to $300,000.

Do failed experiments have value to AI companies?

Yes. Mercor says failed experiments are as valuable as successful ones, and asks for 1,000 or more runs with their inputs and measured outcomes, failed runs included.

Can you sell clinical trial or patient data to AI?

Only de-identified, and only if you own it and the consent allows it. If you’re a HIPAA covered entity, selling protected health information needs an authorization that says you’ll be paid.

Can a wound-down biotech sell its lab data?

Yes, and Mercor names wound-down biotechs as a strong fit. Check first who owns the records, what the consent forms allowed and whether any federal grant or sponsor agreement covers them.

Should you file patents before licensing lab data?

File first if the records describe an invention you want to protect. US law gives inventors a one-year grace period for their own disclosures, but patent counsel should decide what goes in scope.

Last updated: November 13, 2026

Disclosure: some links on this page are affiliate links. If you sign up through one, we earn a commission at no extra cost to you.

This isn’t legal advice. Have counsel review consent forms, agreements and patent questions before you license any lab data.

Don't Do It Yourself
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.