# Should you license direct to an AI lab or through a broker?

> Deciding whether to sell direct to an AI lab or through a broker? Most licensing runs through a marketplace, since no lab publishes an intake for company data.

By Matt Burns, Vaultfish. Canonical: https://vaultfish.com/answers/sell-direct-to-ai-lab-or-through-a-broker  
Published: 2026-09-05  
Updated: 2026-10-14

## The short answer

License through a marketplace or direct program, since no lab publishes a company-data intake. micro1's partnership, Handshake AI and Troveo are visible routes. Google, the only named lab-side bidder, won Spirit Airlines' archive with a $10M bid, pending court approval. Marketplaces add matching and privacy review, often with unpublished fees.

Legally, a "data broker" resells consumer data and must register in some states. So this site calls the companies in between marketplaces and buyers.

## Do AI labs buy company data directly?

OpenAI announced a Data Partnerships effort in November 2023. It aimed to collect large-scale datasets from outside groups, with a focus on data that "expresses human intention" ([TechCrunch](https://techcrunch.com/2023/11/09/openai-wants-to-work-with-organizations-to-build-new-ai-training-data-sets)). It targets large datasets and does not describe company records. I found no equivalent intake page from other labs.

Epoch reported that labs are doing "substantially more in-housing" of environment building. It named [Mercor](/buyers/mercor), Surge, [Handshake](/buyers/handshake-ai) and Turing as vendors ([Epoch](https://epoch.ai/gradient-updates/state-of-rl-envs)). Public reporting I read names no enterprise AI vendor buying company email or support records, outside bankruptcy sales.

## What did the one public lab-side deal look like?

Spirit Airlines' archive went to a bankruptcy auction. Google won the auction with $10M, and the sale is pending court approval. Mercor bid $7.5M. [micro1](/buyers/micro1) offered $12.5M after the auction closed ([Time](https://time.com/article/2026/08/25/google-spirit-airlines-ai-data-RL/)).

Google said a third party would scrub personal information first. A flight attendants' union objected that the data would keep "referential integrity."

Figure: comparison of Three routes from company data to an AI lab: Direct to a lab (No intake found), Direct data program (emphasized: Publishes tiers you can check, Published tiers), Marketplace (Fee often unpublished). Direct data programs like micro1 are emphasized because they publish tiers a company can check.

## What does a marketplace add?

[Troveo](/buyers/troveo)'s own guide says licensing on your own means "finding buyers, negotiating terms, and handling packaging and privacy yourself, which is why most owners go through a marketplace" (a marketplace's view, [Troveo](https://www.troveo.ai/resources/sell-data-to-ai-companies)). Handshake AI says it de-identifies and cleans data on its side before delivery to labs ([Handshake AI](https://joinhandshake.com/ai/data-partnerships/)).

## What does a marketplace cost?

Troveo charges owners no fees and shares revenue on each sale. The split is unpublished. A success-fee advisor takes a percentage of what you receive, agreed in writing.

micro1 pays companies directly under data partnerships, so the question there is the amount it pays. It states tiers of $100K+, $500K+ and $1M+ (self-reported). Ask each one for the figure in writing.

A marketplace also becomes a second place your data sits. Ask who holds it and when it is deleted, using [how to choose an AI data broker](/guides/how-to-choose-an-ai-data-broker).

## What should you do first?

Confirm you hold the rights to the records with the [rights check](/rights-check). Then compare [who handles company data](/guides/best-ai-data-brokers). The [buyer directory](/buyers) lists public terms. Talk to a privacy or commercial lawyer in your state before you sign.

micro1 may pay us a referral fee if a company we introduce goes ahead with it. [How we are paid](https://vaultfish.com/disclosure)

[Apply to micro1's data partnership (paid link)](https://req.micro1.ai/post/74520e8b-4673-4da5-b855-53609aaf6710?referralCode=f82e519e-5d64-4e0e-9574-760c84371cf2&utm_source=referral&utm_medium=share&utm_campaign=data_referral) · [micro1's own data-partnerships page](https://www.micro1.ai/data-partnerships)

_Paid link. [How we are paid](https://vaultfish.com/disclosure)_

_Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one._

## Sources

- [TechCrunch](https://techcrunch.com/2023/11/09/openai-wants-to-work-with-organizations-to-build-new-ai-training-data-sets)
- [Epoch](https://epoch.ai/gradient-updates/state-of-rl-envs)
- [Time](https://time.com/article/2026/08/25/google-spirit-airlines-ai-data-RL/)
- [Troveo](https://www.troveo.ai/resources/sell-data-to-ai-companies)
- [Handshake AI](https://joinhandshake.com/ai/data-partnerships/)