A company with a working prototype, registered in the right country, building something that helps vulnerable children, can read the UNICEF Venture Fund record end to end and conclude it is a good fit. Up to $100,000, no equity taken, a year of mentoring attached. Then one clause decides the whole question, and it is not about the money or the mission. It is about who owns the result.
The record requires open source licensing. Not as an aspiration — as a term, with named licence families for each kind of output. A team whose plan depends on proprietary code is not a weak applicant to this fund. It is an ineligible one, and the sentence that establishes that sits well below the award figure.
This is a price paid in rights rather than in cash, and it behaves differently from every other kind of grant cost. A match is a number you can find in your own accounts. A licence condition is a decision about your product that reaches past the grant period and does not reverse when the money is spent.
The Short Answer
On funding listings, “open” occupies two different positions. Sometimes it is a condition you accept to receive money offered for another reason, enforced through named licences. Sometimes it is the deliverable itself, and openness is simply the work being bought. The first costs you rights; the second does not.
UNICEF Names the Licence Families, Not Just the Amount
Per the UNICEF Venture Fund’s general funding opportunity record, the fund offers up to $100,000 in equity-free funding plus one year of mentoring. Eligible applicants are private for-profit companies registered in a UNICEF programme country with an open source technology prototype that has potential to positively impact vulnerable children.
Then the conditions. The record requires real-time measurable public data and adherence to open source licensing, specified by output type: OSI-approved licences for software, CERN, MIT or TAPR for hardware, and CC-BY for design and content. The process the record describes runs through an eligibility check, an expression of interest, internal board review, an RFP, and an investee announcement. No application deadline appears on the record, so treat the timing as rolling and confirm it against the listing.
Read the precision asymmetrically, because that is how it is written. The award is a ceiling — “up to” $100,000, with no floor and no stated typical size. The licence requirement is exact, naming specific standards bodies for three distinct categories of output. The soft number is the money. The hard number is the licence.
“Equity-free” is doing real work in that sentence, and it is also narrower than it first sounds. The fund is not taking a share of the company. It is requiring that the thing the company builds be released under terms anyone can use, and requiring that the performance data be public and live. For some teams that is already the plan and the condition costs nothing. For others it is the entire business model, and the $100,000 is unreachable at any level of application quality.
This Is an Eligibility Test You Can Run Before You Write
Most eligibility screens require assembling documents — a determination letter, audited financials, a board roster, evidence of a designation someone else maintains. This one requires a decision.
Because the record names licence families by standard rather than describing them loosely, an applicant can resolve fit in an afternoon. Can the software ship under an OSI-approved licence? Can the hardware designs go out under CERN, MIT or TAPR? Can the design and content carry CC-BY? Is the team willing to publish real-time measurable data on how the thing performs? Four yes-or-no answers, none of which require a lawyer to begin, settle whether the rest of the process is worth entering.
That is unusually cheap diligence, and it is available only because the record is specific. A listing that said “we favour open approaches” would leave the same question unresolved through several rounds of effort. Teams working across startup funding meet the vaguer version constantly, and it is worth noticing when a funder has done the work of being precise.
At NSF, Open Is the Job Being Purchased
The same word on a different record obligates nothing of the kind.
Per the National Science Foundation’s record for Pathways to Enable Secure Open-Source Ecosystems (PESOSE), the program supports translating open-source science and engineering research products into safe and sustainable ecosystems addressing national and societal challenges. The record describes open-source tools — software, hardware, machine learning models, languages and data platforms — as designed to be shared because they are publicly accessible and modifiable. PESOSE supports creating managing organisations for these ecosystems, covering governance, distributed development and broad user communities across academia, industry and government, and it supports work on the safety, security and privacy of existing ecosystems by addressing technical and socio-technical vulnerabilities.
The solicitation seeks three types of proposal: scoping and planning the establishment of an ecosystem; establishing and expanding a sustainable one based on a “robust, promising open-source product” meeting an emergent societal or national need; and improving the safety, security and privacy of an existing ecosystem and its products. The record carries a deadline of March 2, 2027 and an award maximum of $40,000,000.
Nothing there is a condition on the applicant’s own intellectual property. Openness is the subject matter. An applicant is being paid to build the governance and security of something already public. The licence question that decides the UNICEF application does not arise, because there is no proprietary alternative being given up — the premise of the work is that the artefact is shared.
That difference is invisible in a search result. Both records match a query for open source funding. One of them will ask a company to change what it owns, and the other will ask a team to run an ecosystem.
FAIROS Makes You Choose a Track, and Suggests a Phone Call
Per the NSF record for Findable Accessible Interoperable Reusable Open Science (FAIROS), the program supports open science activities including research and cyberinfrastructure development advancing FAIR research data management; piloting new models of scientific communication and publication; developing data portals, research data commons and research data management as a national service; and lowering barriers to accessing, curating, integrating, linking, managing, sharing and storing data across disciplines irrespective of data size. The maximum award is $6,000,000 and the record’s deadline is April 14, 2027.
Two structural facts matter more than the dollar figure. First, proposals must select one of two tracks — Disciplinary Improvements aimed at targeted scientific communities, or Cross-Cutting Improvements applying across most disciplines — and the activities in the proposal must match the chosen track. Second, the record strongly recommends contacting a cognizant program officer before proposing: for disciplinary work, one in the directorate closest to the major disciplinary impact; for cross-cutting work, one in the Office of Advanced Cyberinfrastructure. Standard research proposals are the only type accepted.
The record states that FAIROS is undertaken in support of the NSF Public Access Initiative. That is the thread connecting this family: these are programs built to implement a public-access policy, which is why openness arrives as scope rather than as a concession. The applicant is on the same side of the requirement as the funder.
A Standing Door With No Deadline on It
Per the NSF record for Cyberinfrastructure for Public Access and Open Science (CI PAOS), the program sits within the Office of Advanced Cyberinfrastructure and aims to catalyse socio-technical partnerships supporting research data infrastructure, through early-stage collaborative activity among cyberinfrastructure researchers, scientists, research computing and data management experts, research labs, university libraries and other communities of practice. Awards run from $50,000 to $600,000.
The record states that NSF accepts proposals pursuant to this program description year-round, and that NSF may periodically issue Dear Colleague Letters encouraging proposals on particular themes. Proposers must target one or more of three pathways the record names: Competency Building, Capability Building, and Community Building. The record also notes that a primary feature of successful projects is a collaborative team spanning science and engineering, research data science and information science, and that proposals with an international component are welcome.
No deadline is not the same as no clock, and here it means something specific: the program is a standing door whose emphasis shifts when NSF publishes a letter saying so. The useful habit is watching for those letters rather than watching for an opening date. Programs that behave this way across federal funding reward a different kind of attention than a dated competition does.
NIH Funds Curation of Data Nobody Here Owns
One record completes the pattern by removing the applicant’s product from the question entirely.
Per the National Institutes of Health record for Computational Approaches to Curation at Scale for Biomedical Research Assets, the National Library of Medicine seeks to accelerate access to and availability of secure, complete datasets and computational models that can serve as the basis for biomedical discoveries. The record calls for innovative at-scale computational approaches that increase the speed and scope of curation, for data mining and knowledge discovery from growing quantities of biomedical data. It is an R01 with clinical trials not allowed, carries a maximum of $250,000, and shows a deadline of April 15, 2027.
The asset being improved already exists and belongs to the research community. Curation is the deliverable. An applicant gives up no rights, because the rights in question were never theirs. This is the furthest position from the UNICEF record while using recognisably the same vocabulary — and the two sit in the same search results.
What to Check Before Treating Open as Alignment
The index holds 43,000+ open opportunities across federal, state, local, foundation and corporate sources (OpenGrants data, verified September 11, 2026), refreshed daily (OpenGrants data, verified August 10, 2026). Amount, deadline, geography and applicant type are structured fields. Whether a licensing requirement attaches to your own work is not a field. On these five records it lives in prose, and on one of them it is the clause that decides eligibility.
Two questions separate the cases quickly.
Is openness described as something I must do to my output, or as the thing being built? UNICEF names licence families for software, hardware, design and content — that is a requirement pointed at the applicant’s product. PESOSE, FAIROS, CI PAOS and the NLM notice describe ecosystems, portals, governance and curation. Those are deliverables, and they cost no rights.
If it is a requirement, is it named precisely enough to test? Named standards are a gift. OSI, CERN, MIT, TAPR and CC-BY are checkable against a product plan today, which turns a long application into a short decision. Vague preference language does the opposite, and the right response is to ask the program contact before investing.
When the answer is that openness is the price, the follow-on work is commercial rather than editorial — what the licence means for the roadmap, and whether the figure on the record is worth it. Teams weighing that against non-dilutive alternatives often end up comparing it with SBIR and STTR programs, where the intellectual-property treatment is governed differently again, and the mechanics of award terms sit in our knowledge base.
Subscribe to Funding Friday, our weekly grant digest, and the figures arrive already checked against the listing.
Common Questions
Does equity-free mean there are no strings? No. It means the funder is not taking ownership of the company. On the UNICEF Venture Fund record the consideration is different in kind: open source licensing under named standards for software, hardware, design and content, plus real-time measurable public data. Those obligations attach to the product rather than the cap table, and they do not end when the grant period does.
Can I accept an open source condition for one product line and keep the rest proprietary? The records here do not answer that, and we will not guess at it. The UNICEF record states the licensing requirement and names the licence families; it does not define the boundary of what counts as the funded output. That boundary is exactly the question to put to the fund before the expression of interest, because the answer determines whether the application is viable at all.
Are the NSF open science programs only for universities? The records do not restrict eligibility to universities. CI PAOS describes teams drawn from cyberinfrastructure research, science, research computing, data management, research labs and university libraries, and PESOSE describes user communities across academia, industry and government. For the authoritative eligibility statement on either, check the listing — the indexed descriptions summarise scope rather than reproduce the eligibility section.
Why would a funder publish the licence requirement more precisely than the award size? Because they serve different purposes. An award ceiling is a budgeting fact that varies per project, so “up to $100,000” is the honest form. A licence requirement is a binary gate the funder needs every applicant to apply identically, so it gets named standards. The practical consequence is that the most load-bearing sentence on a record is often not the one with a dollar sign in it.