Where the need is, and where the grant money lands
From FY2020 to FY2023, the highest-need tenth of US counties received $102 per resident a year in grants; the lowest-need tenth, $349.
From FY2020 to FY2023, the highest-need tenth of US counties received $102 per resident a year in grants; the lowest-need tenth, $349.
Organizations that share funders gather into communities, lit by where the grant dollars land, with any funder's grants drawn on top.
32% of nonprofits that lose a big grant are gone or a quarter smaller two years on, against 11% when it continues.
54% of funder relationships are still funded 3 years on.
That is 9% of everything the funds granted.
Plinth is a US nonprofit and foundation intelligence database built from public IRS Form 990 and 990-PF filings: about 211,000 grantmakers — every US filer that itemized a grant — and 20.1 million grants, resolved into a multi-year funding network. It shows who funds any nonprofit, every foundation's grantees, which funders co-fund the same organizations and which portfolios resemble each other, and it is available through a public grants API.
Across the IRS filings tracked here (fiscal years 2017–2023), US grantmakers reported $1.18 trillion in grants to roughly 429,000 nonprofits. The figures come from Form 990 and 990-PF filings, which the IRS releases on a 12–24 month lag, so each is dated to its fiscal year.
About 211,000 organizations filed grantmaking returns in the period covered here: every electronically-filed Form 990, 990-EZ, and 990-PF, together making around 20.1 million grants.
It is read from public IRS Form 990 / 990-PF e-file filings and the IRS Business Master File: primary government data rather than a licensed directory. The data is free to search and free to query via the API. See the methodology for how each figure is built.
No. 15.2 million of the 20.1 million grant lines are resolved to a specific recipient EIN. Foundations file recipient names as free text, and an identification is attached only when the evidence clears a measured precision bar, so the remaining lines count in totals without appearing on any recipient's funding page. In practice a smaller nonprofit's page can be missing funders whose filings name it ambiguously: treat a funding page as a floor, not a census. Per-layer match rates, including the weak layers, are published on the data-quality page.
Yes. Alongside foundation grants, we track federal government funding (grants and contracts, from USASpending) and state government grants (from state transparency checkbooks), so a nonprofit's page shows its philanthropic and government funding together. How each is linked differs and it matters: state portals often publish an EIN, which is a direct join, but USASpending identifies recipients by federal UEI rather than EIN, so federal awards are matched by a confidence-scored name-and-location resolution. We favor precision over recall there: an award is left off rather than attached to the wrong organization, so federal coverage is partial by design, and state coverage is limited to the handful of states whose checkbooks publish usable recipient data today. Treat the government total on a nonprofit's page as a floor, not a full account; measured resolution rates for both layers are on the data-quality page.
Yes. The funding graph includes overseas NGOs, universities and government bodies that receive grants but have no US EIN, plus international grant flows. See the international funding and entity pages.