Non-dilutive funding: your options, including licensing your data

What non-dilutive funding means

Dilutive funding means selling part of your company. You issue new shares to an investor, and every existing owner now holds a smaller slice.

Non-dilutive funding is cash you raise without issuing equity. Your cap table stays the same. You still give something up, though. It might be repayments and interest, a share of future revenue, a tax filing and the records to back it, or rights to use something you own.

That's why the useful question isn't "dilutive or not". It's "what does this option cost me, and can my business carry that cost?"

The main types of non-dilutive funding

Grants

Grants are awards you don't repay, usually tied to a research or public goal. They tend to be competitive and slow, and they come with reporting rules.

In the US, the largest research program for small businesses is SBIR and its sister program STTR. SBIR.gov describes it as "equity free funding through federal agencies to American small businesses", funded through 11 participating federal agencies. As of October 16, 2026, agencies may issue a Phase I award up to $335,937 and a Phase II award up to $2,239,578 without seeking SBA approval (SBIR.gov). The SBA itself notes that it "does not provide grants for starting and expanding a business" (SBA).

In the UK, start with Innovate UK. In the EU, see the European Innovation Council's funding opportunities.

Revenue-based financing

A provider advances cash and you repay it as a share of your revenue. The University of York's Enterprise Works puts it simply: "You pay back a percentage that fluctuates along with your revenue" (University of York).

It suits companies with steady, predictable sales. It's still repayment, so a slow quarter stretches it out rather than making it go away.

Venture debt and bank loans

Bank loans are the familiar version: a fixed amount repaid with interest, often against collateral or cash flow.

Venture debt is a loan for startups that already have venture investors. The British Business Bank describes it as "lending funds to early-stage, high-growth companies that are already supported by venture capital", usually used between equity rounds (British Business Bank). Read every term in the agreement, because lenders set conditions that differ from deal to deal.

IP-backed financing and royalty financing

Here your patents, designs or copyrights support the money you raise. WIPO explains that IP rights "can be used to secure financing, either by pledging them or transferring rights to cash flows derived from these assets" (WIPO). Royalty financing is the second version: an investor pays now in exchange for a share of the royalties your IP earns later.

WIPO also notes that when IP is used as collateral, ownership usually stays with the borrower, but "a lender may put certain conditions on how the IP can be used in the future." That can limit what else you do with it.

R&D tax credits

Most of the regions where Lunchbreak works have research tax incentives. In the US, see the IRS research credit page (Form 6765). In Canada, see the SR&ED tax incentives. In the UK, see HMRC's R&D tax relief guidance.

Eligibility, rates and filing rules are set by each tax authority, and they change. This is squarely a job for your accountant.

Licensing your data and IP

Licensing means you let someone use something you own, under written terms, in exchange for payment. Companies have long licensed patents and software. A newer version is licensing the records behind your work, such as design histories or lab notebooks, to teams that train AI models. The rest of this guide covers that option.

What each option costs you

This table is qualitative on purpose. Rates and amounts vary by provider, program and company, so get real quotes.

Licensing your company's data as non-dilutive funding

AI models learn from examples of how experts actually work. Public data runs thin in specialized fields, so the detailed history inside an engineering firm or a lab can be useful training material.

Lunchbreak's offer, in its own words: "License your past design and research records to the teams training frontier AI. You keep ownership."

On ownership, Lunchbreak's /businesses FAQ says: "We license specific usage rights defined in your agreement. You keep ownership of the records and can continue to use them however you like."

A few things follow from that:

• It's not a loan. You're granting usage rights under an agreement, so there's no repayment schedule. It also doesn't involve selling shares.
• You decide what's in scope. You "choose the systems, folders and date ranges" and "can exclude any project, client or record".
• The data stays private. Lunchbreak's FAQ says "Data is never published." Identifiers are removed before anything moves, and you review the anonymized dataset before it is approved.
• You can stop. You can stop any time before signing, and Lunchbreak deletes what you shared and confirms the deletion in writing.

Lunchbreak is not a lender, investor or funder. It licenses records. If you want this to sit alongside a raise, a loan or a wind-down, your advisers should look at how it fits.

The two most common reasons owners look at it, based on the options in Lunchbreak's estimator, are "Revenue from records we already have" and "Value before we wind down". If you're closing, see what to do with company records when winding down.

Which records have value

Not every archive is worth licensing. As Lunchbreak's page puts it, "Buyers pay for context, not file counts." The thinking between the files is what matters most.

Engineering and CAD records

Lunchbreak's CAD page lists "Native CAD files, PDM vaults, engineering change orders, drawings, FEA and CFD studies, and RFIs." Change orders and review threads are the strongest part, because they show why a design changed.

If that's your archive, you can estimate a CAD archive.

Life sciences research records

On the research side, Lunchbreak's life sciences page lists lab notebooks, negative experiments, programme decisions and validation work. Protocol revisions and failed experiments count, too.

How an archive is weighed

Lunchbreak scores an archive on four things before a sample review:

Source: lunchbreak.com/businesses.

Example estimates

Lunchbreak publishes three example estimates on its /businesses page. They're examples, not offers.

Indicative only. Offers depend on a sample review, rights and anonymization. No commitment.

Non-dilutive funding for biotech

Biotech burns cash for years before revenue, so founders tend to stack several non-dilutive sources.

Grants and partnerships

Research grants are the usual first stop: SBIR and STTR in the US (SBIR.gov), Innovate UK, and the European Innovation Council. Many biotechs also partner with larger companies on a program, trading some rights to an asset for payments. Each of these has its own terms, so involve your lawyer and your board early.

Licensing records from discontinued programmes

Shelved programmes still hold value. The lab notebooks, negative results and the decisions about why a programme stopped are the kind of reasoning that's hard to find in public data.

Lunchbreak's life sciences page lists "Companies with discontinued programmes" among the teams it works with, along with discovery CROs, chemistry R&D groups and materials discovery companies. It works with research records only: "discovery programmes, protocols, methods and experimental results."

If that describes you, estimate a research archive.

Who qualifies, and when data licensing is not a fit

Lunchbreak's /businesses FAQ says you generally need "at least two years of operations, 20 or more employees at some point, earned revenue, and clear rights to the records." It adds: "Firms that are operating, winding down, or already closed can all qualify." Lunchbreak works with companies in the US, Canada, the UK and the EU.

Data licensing is not a fit when:

• The records are patient-level. Lunchbreak does "not license patient-level data or records from hospitals, clinics, care delivery or routine diagnostics."
• The data is export controlled. It does "not accept ITAR or EAR controlled technical data."
• The company is very young. Under two years of operations doesn't meet the bar.
• The rights are unclear. If a client, sponsor or partner may own the work, sort that out with your lawyer first. You can exclude any project or client you're unsure about.

How it works

Lunchbreak's process has three steps:

1. Estimate. Answer ten questions on /businesses for an indicative range.
2. Sample review. Lunchbreak reviews a sample of your records. Identifiers such as names, emails, and client and project names are removed automatically, and specialists who know your field then catch what patterns miss.
3. Written offer. "A sample review turns your estimate into a written offer." You review the anonymized dataset before it's approved.

Lunchbreak doesn't publish a timeline, and it can't promise one. If you have a deadline, say so early and don't plan around a payment date until terms are signed.