R&D organizations can license unpublished experiment records they own, and failed experiments are as valuable as successful ones. Published buyer ranges run $100K to $750K for semiconductor and materials companies with 1,000+ runs linking recipes to measured results, $50K to $875K for biotech and life-sciences labs with notebooks and run records, and $10K to $300K for chemistry and materials records of the reasoning behind research decisions. Semiconductor and biotech buyers accept organizations anywhere, and labs that have wound down qualify.
- Linking each run's settings to its measured result is what buyers pay for.
- Keep failed, null and repeated runs, with dates. Published results alone do not qualify.
- Data owned by a customer, foundry, sponsor or partner stays out.
Three kinds of R&D records and what buyers want
As read on 1 November 2026:
| Record type | Company earns | Core requirement |
|---|---|---|
| Semiconductor and materials | $100K to $750K | 1,000+ runs, builds, wafers or experiments with recipe or process settings linked to measured performance |
| Biotech and life-sciences lab | $50K to $875K | ELN or structured experiment records, 1,000+ runs with inputs and measured outcomes |
| Chemistry and materials R&D records | $10K to $300K | Linked research episodes: question, evidence, experiments, results and decisions |
What makes R&D records valuable
Buyers agree on these points:
- Every run kept, failed ones included, with dates.
- The same recipe built or run again on different days.
- Measurements under more than one condition.
- Notebooks that record why a run failed and what changed next.
- Records made at the time rather than written up afterwards.
Systems the records usually sit in
Buyers name MES, SPC and yield systems, LIMS, electronic lab notebooks, Cadence, Synopsys and OSIsoft PI for semiconductor work, and Benchling, LabArchives, Dotmatics, LIMS and Medidata Rave for life sciences.
Ownership and control checks
Check these before describing any dataset:
- Data owned by a customer, foundry partner, sponsor or collaborator stays out.
- Published papers, datasheets and already-published data do not qualify.
- Semiconductor and some materials technology can be export-controlled under the US EAR and the EU Dual-Use Regulation.
- Clinical trial data needs de-identification, and the original consent may limit secondary use.
- Licensed third-party IP, including design libraries, stays out.
Can companies outside the US license these records?
Semiconductor and biotech buyers accept organizations anywhere; no location rule is published for chemistry and materials records. Personal data is rare in lab records outside clinical work, so export controls and ownership are usually the main checks.
The privacy law that applies to any personal data left in the records:
| Country | Main rule to check | Country guide |
|---|---|---|
| United States | EAR export controls; ownership | US guide |
| United Kingdom | Export Control Order 2008 | UK guide |
| Ireland | EU Dual-Use Regulation | Ireland guide |
| Germany | EU Dual-Use Regulation | Germany guide |
| Canada | Export Control List | Canada guide |
| Singapore | Strategic Goods (Control) Act | Singapore guide |
Next step
Start with a company-level description: which systems hold the records, how many years they cover and who owns them. No files are needed. Request a free review.
Origin Data Partners is independent and is not affiliated with any AI data buyer. We help owners compare buyers and find the best fit for their records. See what your data is worth.
Common questions
Can a biotech sell its lab notebooks to AI companies?
Yes, for unpublished records the organization owns. Published buyer requirements want electronic lab notebooks or structured experiment records with 1,000+ runs and measured outcomes, failed runs included.
How much do AI companies pay for R&D data?
Published buyer ranges run $100K to $750K for semiconductor and materials companies, $50K to $875K for biotech and life-sciences labs, and $10K to $300K for chemistry and materials research-decision records.
Are failed experiments worth anything?
Yes. Buyers of all three kinds ask for failed, null or repeated work. Notes explaining why a build or experiment failed are named as especially useful.
Can a startup that shut down license its research?
Yes. Buyers say labs and startups that have wound down qualify, and biotechs that have wound down are a strong fit. See selling company data after closing.
Sources and further reading
Origin Data Partners is independent and not affiliated with any buyer named on this page.