layout: true name: fullheader background-image: url(../apachecon/img/neonbrand-258972-unsplash-sm.jpg) background-size: cover

layout: true name: contributions background-image: url(../apachecon/img/rawpixel-600782-unsplash-sm.jpg) background-size: cover

layout: true name: thanks background-image: url(../client/img/stuart-guest-smith-150560-169.jpg) background-size: cover

layout: true name: fosdem background-image: url(../fossbackstage/img/KaraSowles-FOSDEM2024-Slide6.png) background-size: cover

layout: true name: logorb class: left background-image: url(img/ComCodeLogo.png) background-repeat: no-repeat background-position: bottom .8rem right 5rem background-size: 10%

template: fullheader

Coming up next

Foundation Funding: How Much?

Shane Curcuru

template: fullheader name: start

Foundation Funding: How Much?

Who funds FOSS foundations, and 10 years of history

@ShaneCurcuru

template: logorb

Resources

Resources and data are open source:

FOSSSustainability.com/ComCode

Foundation Sponsorship Modeling

US Foundation 990 tax form data

Actively seeking submissions!

.bottomnote[license: Apache-2.0]

??? Please follow along at fosssustainability.com/comcode for links to slides and all the data and code associated with this project!

template: logorb name: topics

Topics

  • Aspects of Sustainability
  • Foundation Sponsorship Modeling
  • Financial Data (US 990s)
  • NEW: Historical data & trends

??? Topics:

  • Fiscal sustainability - all the different ways projects might get funded
  • Modeling sponsorship models of major foundations
  • Comparative US Foundation revenue and asset data
  • NEW this year: IRS e-file data expanded and historical trends

Press SPACEBAR

–

  • Finding better questions to ask

??? I hope to leave everyone with enough information and ideas, so that you can start asking better questions when you’re working on sustainability. Note: this is US centric data; would love help collating non-US nonprofit data.

template: fullheader name: aspects

Aspects of Sustainability

??? Why is open source funding so hard? Because there are so many different aspects or ways that different people, organizations, and ecosystems think about funding and sustainable work.

template: logorb

Aspects of Sustainability

.left-column-equal[ Who

  • Users
  • Contributors
  • Maintainers
  • PMC / TSC
  • Foundation Boards ] .right-column-equal[ What
  • Software Companies
  • Non-Software Companies
  • Educational Institutions
  • Governments / Regulators
  • Standards Bodies
  • NGOs ]

.bottomnote[https://fosssustainability.com/aspects/]

??? Listening to financial sustainability conversations, I’m struck by how siloed they are. Each participant is coming in with their own needs and their own vocabulary, and it’s often hard to find efficient ways to translate one aspect’s needs to the vocabulary or taxonomy some other aspect needs. Note that we’re focusing on financial sustainability in particular, but the same concept applies to most other areas, like communities, education, or energy.

template: logorb

Aspects of Sustainability

.left-column-equal[ Who

  • Users
  • Contributors
  • Maintainers
  • PMC / TSC
  • Foundation Boards ] .right-column-equal[ What
  • Software Companies
  • Non-Software Companies
  • Educational Institutions
  • Governments
  • Standards Bodies ]

.bottomnote[https://fosssustainability.com/aspects/]

??? STORY: Maintainers keep seeing companies with OSPOs or FOSS Funds talking about funding open source projects - but few see the actual money, and certainly not enough to make a living. For companies without OSPOs, how do you show the value of FOSS to their leadership (i.e. those with budgets)? How do we help maintainers look for funding? And which maintainers need the funding? Many maintainers work for big software companies already; and who a maintainer works for is opaque to others.

template: logorb

Aspects of Sustainability

.left-column-equal[ Who

  • Users
  • Contributors
  • Maintainers
  • PMC / TSC
  • Foundation Boards ] .right-column-equal[ What
  • Software Companies
  • Non-Software Companies
  • Educational Institutions
  • Governments
  • Standards Bodies ]

.bottomnote[https://fosssustainability.com/aspects/]

??? Another example: while university curricula may cover technical mechanics of open source development, where are the classes helping new contributors understand how to seek funding or otherwise market their project as a whole, not just as a codebase? How do we expand the basic “community based open source” ethos in the US/Europe to the rest of the world?

template: logorb

Aspects of Sustainability

.left-column-equal[ Who

  • Users
  • Contributors
  • Maintainers
  • PMC / TSC
  • Foundation Boards ] .right-column-equal[ What
  • Software Companies
  • Non-Software Companies
  • Educational Institutions
  • Governments
  • Standards Bodies ]

.bottomnote[https://fosssustainability.com/aspects/]

??? And currently, with the CRA and PLD here in Europe, it’s likely that any major open source project will likely face new legal hurdles in Europe. How do we connect the Foundations that manage and provide support to many of these major projects with the Standards Bodies making up the new rules? Or the Governments, that are forcing complexity on FOSS projects who don’t have the resources to handle that complexity?

See also: Gregor Bransky talk, Vlad Miricioiu keynote, Dirk’s work at CRA

template: fullheader

Aspects of Sustainability

How do we translate between these groups?

.bottomnote[https://fosssustainability.com/aspects/]

??? Even with the best of intentions, many of the stories we tell don’t have the right impacts on different groups. Even as governmental policy makers are studying and planning for how to help the FOSS ecosystem, the needs of each group are often poorly understood, because they’re not getting translated into policy speak, or classroom theory.

template: fullheader name: finance

Who Funds FOSS Foundations?

??? OK, Shane, you’ve shown us all the complexity, how about answering the question?

template: fosdem

??? Just to give you an overview of the complexity of just the “how do we get money there” let’s just ask: how can you actually transfer funds? There are many, many different ways, making this a complex topic (so: ask better questions!).

This is an excellent holistic overview of how funding typically gets to open source contributors, foundations, and more. From Kara Deloss’ excellent talk at FOSDEM 2024.

template: fullheader name: finance

Who Funds FOSS Foundations?

Corporate Sponsors

??? The largest income source for common FOSS Foundations are corporate sponsorships.

But really: sponsorships are only part of the equation. Depending on the type of foundation and the ecosystem they’re in, the amount of sponsorships vs. individual donations vs. “program income” (event tickets, services fees, etc.) or event tickets varies widely.

template: fullheader name: sponsorship

Modeling Sponsorship Programs

Modeling actual foundation programs

??? So - let’s analyze actual foundation sponsorship programs and quantify them. What levels are offered, what benefits (if any) do you get, and how much likely income does sponsorhips bring to actual foundations?

So I built a model and analyzed 30+ foundations or sub-foundations to see the data - and new this year, we have historical data going back ~10 years for a select group of foundations (mainly: ones that make it easy to find this data).

template: logorb

Sponsorship Modeling Criteria

  • Only publicly posted programs
  • Not tracking event sponsorships
  • Level amounts are approximate
  • Tracking 39 foundations so far
  • NEW: Historical data for some foundations

.bottomnote[https://fossfoundation.info/sponsorships]

??? I’m focusing on quantifying how traditional foundation corporate sponsorships are modeled. All data is drawn from publicly posted sponsorship programs in the past year or so. This is not including event sponsorships, because those have a different typical structure, and aren’t expected to be continuing.

IMPORTANT: Numbers are approximate! A number of programs offer sliding scales by numbers of employees or revenue; we track the top end of each level’s cost. Also, we don’t know the actual details of sponsorships; some may pay more or less depending on discounts or other factors.

Model schema explainer: https://fossfoundation.info/sponsorships#sponsorship-model-structure

template: logorb

Sponsorship Model Schema

  • Program level map: first, second,…, community, etc.
  • Track funding amount per level
  • Model the benefits for levels
    • Governance /or/ Advisory roles
    • Services, Events discounts
    • Marketing partnerships
  • Scrape websites or landscape.yml for data

.bottomnote[https://github.com/Punderthings/fossfoundation/blob/main/_data/sponsorships-schema.json]

??? Most organizations include scraping definitions to automatically parse listings of sponsorship pages; some are manually drawn maps as of a specific date. We now have historical data for a select set of foundations; help wanted expanding our historical parsing of more foundations.

Model schema explainer: https://fossfoundation.info/sponsorships#sponsorship-model-structure

template: logorb

Sponsorship Program Dataset

.left-column-equal-vsm[

  • Adélie Linux
  • Arch Linux
  • ArduPilot
  • Apache Software Foundation
  • Academy Software Foundation
  • CD Foundation
  • Cloud Native Computing Foundation
  • Debian
  • Drupal
  • FINOS Foundation
  • FreeBSD
  • Free Software Foundation Europe
  • Gentoo
  • Gnome
  • GraphQL Foundation
  • Haskell
  • Hyperledger
  • JS Foundation ] .right-column-equal-vsm[
  • Linux Foundation
  • LF AI Foundation
  • LF Energy Foundation
  • LLVM Foundation
  • NumFOCUS
  • Open Mainframe Foundation
  • Open SSF Foundation
  • Open Stack Foundation
  • OSGEO
  • Open Source Initiative
  • OWASP
  • Plone
  • PostgreSQL
  • Python
  • Rails Foundation
  • Raspberry Pi NA
  • Software Freedom Conservancy
  • Software in the Public Interest ]

.bottomnote[https://fossfoundation.info/sponsorships]

??? IMPORTANT: this is only a subset of foundations, mostly depending on which foundations fit the criteria and were easiest to write scrapers for - or had a small enough sponsor list I could manually hardcode it. This does not represent the whole ecosystem!

template: logorb

Sponsorships - Apache Numbers

.left-column-equal[ Funding Sponsors

  • First: 8 Platinum
  • Second: 6 Gold
  • Third: 8 Silver
  • Fourth: 11 Bronze ] .right-column-equal[ In-Kind Sponsors
  • First: 11
  • Second: 5
  • Third: 3
  • Fourth: 9 ]

.bottomnote[book values (USD): Cash ~1.6M, In-Kind: ~1.75M · Oct 2026]

???

We can see a good mix of sponsor levels; in-kind (targeted) sponsorship is nearly as large as cash. These calculations of “book values” are done by multiplying the number of listed sponsors at each level, times the “price” of that sponsorship level in the prospectus; they are only approximations.

Data note: counts are from the Oct 1 2026 Internet Archive capture of apache.org/foundation/sponsors. A year ago (Sept 2025) it was 10/9/5/9 cash: the drop at Platinum and Gold is mostly companies moving down a level (Yahoo to Silver, IBM to Silver) plus Geico leaving. Book value at list prices: 8x125K + 6x50K + 8x25K + 11x6K = ~1.57M cash; in-kind ~1.75M. Cash is now below in-kind.

template: logorb

Sponsorships - Apache Who’s Who

.left-column-equal-sm[ First Level Cash Sponsors

  • Amazon + In-Kind
  • Apple
  • Google
  • Huawei
  • Meta
  • Microsoft + In-Kind
  • Snowflake
  • Visa ] .right-column-equal-sm[ First Level In-Kind Sponsors
  • Cloudbees
  • DLA Piper
  • Fastly
  • GitHub
  • Gradle
  • JetBrains
  • JFrog
  • Sonatype
  • Yahoo ]

??? Here’s who sponsors the ASF at the highest level. I have data files on who sponsors all the foundations tracked in the model online at FOSSFoundation.info, showing which companies like sponsoring only at the top level, for example.

template: logorb name: howweknow

NEW! Sponsor Model History

  • 1,120 dated sponsor lists for 38 sponsor programs
    • 610 from landscape git histories (LF, CNCF, OpenSSF…)
    • 414 from Internet Archive captures (Apache since Jan 2016)
    • 72 OpenStreetMap wiki revisions; 11 FSFE yearly pages
  • Every list keeps a link to its source capture or commit
  • Each sponsor becomes a span: level, first seen, last seen

.bottomnote[NEW: history/sponsorships/ at github.com/Punderthings/fossfoundation]

??? This is the big data change since last year: instead of a single snapshot, fossfoundation.info now records when each sponsor appeared, changed level, and left. 26 programs have 9 or more recorded lists; the rest have one or two so far. Short gaps of a month or less are bridged. Caveats: Internet Archive captures can miss short-lived changes; some sponsor pages were restructured (OWASP reset its list in Nov 2023, OpenStack and Eclipse changed page scope), and I exclude those from the trend numbers. PRs welcome: more historical sources, especially for European foundations.

template: logorb name: asfhistory

Apache Sponsors Over Time

.bottomnote[Cash sponsors: 53 (2023) to 32 (2026); in-kind steady at ~29]

??? Every bar is the sponsor list on apache.org on July 1 of that year, from Internet Archive captures (2026 is June 1). In-kind sponsors were first listed separately in 2018, which is also why cash drops that year: some sponsors moved to the in-kind list. Between July 2023 and July 2024, 22 sponsors left. Estimated sponsorship income (cash + in-kind, at list prices): 4.03M (2023) to 3.34M (2026). The cash drop is real money: the ASF 990 shows contributions falling from 2.38M (FY2024) to 2.08M (FY2025). NOTE TO SELF, do not show without checking first: 17 Bronze sponsors dropped off in 2023-24, and 14 of them vanished in one step between the Feb 6 and Mar 26 2024 captures (Bronze went from 27 to 13), which looks like a page cleanup or policy change rather than 14 separate decisions. Check with ASF Fundraising before mentioning; don’t name companies on screen. NOTE Dropped Feb-Mar 2024: aboutpaf.com, bestecasinobonussen.nl, cafe24.com, craftercms.org, goread.io, hotwaxsystems.com, leovegas.com, miro-kredit.ch, openlogic.com, praca.pl, purevpn.com, tii.ae, twitter.com, writersperhour.com. Later in 2024: dimealley.com, dollarhand.com (Sep), cerner.com (Dec), this is likely a recording error, not necessarily sponsors choosing to drop the ASF.

template: logorb name: loyalcore

Apache’s Longest Sponsors

21 Apache sponsors listed since 2018 or earlier are still here

.left-column-equal[ Cash

  • Google, Meta (Platinum)
  • Bloomberg, Union Investment (Gold)
  • Capital One, Red Hat (Silver)
  • RX-M (Bronze) ] .right-column-equal[ In-Kind
  • DLA Piper, Microsoft, Sonatype
  • Atlassian, Datadog, phoenixNAP
  • Google, PagerDuty, SURF, Virtru… ]

.bottomnote[Thank you! Long-term sponsors are the foundation of a stable budget]

??? Note: Google announcement on being one of the ASF’s first sponsors, and having 20 consecutive years of sponsorship!

A thank-you slide, and the tenure point: most sponsors that leave do so within two years, but there is a core that has stayed for 8+ years. “Since 2018 or earlier” because in-kind tiers were first listed separately in 2018; cash sponsors are tracked from January 2016. Full in-kind list: DLA Piper, Microsoft, Sonatype (first level); Atlassian, Datadog, phoenixNAP (second); Google, Hopsie, No-IP, PagerDuty, PCCC, Sonic, SURF, Virtru (fourth).

template: logorb

Validating Sponsorship Models

  • Estimate: Sponsor list × model level price
  • Accounting: IRS 990 contributions + program revenue
Foundation Model (cash) 990 (latest) Ratio
Apache (FY Apr 2025) 2.05M 2.08M 0.99
OpenInfra (2024) 4.95M 4.26M 1.16
Python (2024) 1.93M 4.03M 0.48
NumFOCUS (2024) 0.94M 8.13M 0.12
Linux Foundation (2024) 35.1M 213M 0.16

.bottomnote[Sponsor lists 2024-25; 990s via FOSS990Parser. Indicative only]

??? For a sponsor-funded charity like the ASF, simply multiplying listed sponsors by the prospectus price predicts the real 990 number within 1%. Where the ratio is far below 1, most money comes from somewhere else: fiscally sponsored project donations (NumFOCUS), PyCon (PSF), or events and training (LF). Reminder: these rounded figures have various caveats, but are indicative of the levels seen.

template: logorb

Who Sponsors Everyone?

.left-column-equal[ Foundations sponsored (out of 39)

  • Google: 18
  • Microsoft: 16
  • Red Hat: 14
  • Amazon: 12
  • Bloomberg: 10
  • Capital One: 9 ] .right-column-equal[ ASF top sponsors elsewhere
  • Microsoft: LF, FINOS, FreeBSD, GraphQL…
  • Bloomberg: CNCF, NumFOCUS, OSI, OWASP…
  • IBM: CNCF, Hyperledger, OpenSSF…
  • Apple: CD, CNCF, OpenSSF ]

.bottomnote[37 foundations, ~2,900 sponsor companies; 181 sponsor 3+ foundations]

??? Note: only a small set of companies show up as sponsors of multiple foundations we track. Counts are by sponsor company domain, with subdomains merged (aws.amazon.com counts as amazon.com, cloud.google.com as google.com) plus the fossfoundation.info host_aliases.json table (opensource.google, techatbloomberg.com). GitHub (9 more) is counted separately from Microsoft. Sponsor lists are from the Sept 2025 scrape of each foundation’s sponsor page; and there are some obvious foundations we are not tracking in these numbers.

template: logorb name: short

Sponsorships Are Short

Top three cash levels only, 10 US c3s + 9 LF programs, 10 year span:

  • 347 top-tier sponsors left; median stay 1.4 years
  • 62% left within 2 years
  • Current top-tier sponsors: 2.7 years so far
  • Apache: 24 left, median stay 2.4 years
  • Only 8 of 347 stepped down a level; the rest left entirely

Renewals matter as much as new logos

.bottomnote[First, second and third cash levels; excludes in-kind, community and lower tiers]

??? US c3s plus Debian: 89 top-tier sponsors left, median 1.8 years. LF programs: 258 left, median 1.4 years (there, “third” is usually the broad general-membership tier).

Sponsors already present when tracking began are left out of these medians, because we don’t know when they started. Comparison from general US fundraising (Fundraising Effectiveness Project 2023 data): donors giving under $100 renew at about 31%; major donors at about 66-68%. FOSS corporate sponsors renew at higher rates than that most years, but the ones who leave go quickly.

template: logorb name: sneezes

When a Sponsor Sneezes

Departures since 2023, across the foundations we track:

  • Intel: left 8 sponsor programs
  • VMware: left 5, including Apache (Feb 2024)
  • Twitter: left 4, including Apache
  • Didi: left 4, including Apache

A few corporate budgets underwrite many foundations at once

.bottomnote[Sponsor spans 2023-2026, fossfoundation.info history/sponsorships]

??? Example: when one company changes strategy, several foundations can lose a sponsor at the same time. Sponsor income is highly variable number, and can affect the ecosystem, not just individual foundations. Intel: CNCF, FINOS, FreeBSD, FSFE, LF AI & Data, LLVM, OpenSSF, OpenStack (Intel also newly joined FINOS and FreeBSD after 2023 before leaving). VMware: Apache, CNCF, LF, LF AI & Data, OpenSSF - Apache’s Platinum listing ended Feb 2024, shortly after Broadcom’s acquisition of VMware closed (Nov 2023). Twitter: Apache, CNCF, GraphQL, LF. Didi: Apache, CNCF, LF, LF AI & Data. Counts are by company web domain, so subsidiaries are counted separately; these are lower bounds.

template: logorb name: squeeze

The 2024 Sponsor Squeeze

.bottomnote[Standalone c3s lost sponsors in 2024-25; LF programs barely dipped]

??? Retention = of the sponsors listed on July 1 last year, what share is still listed at the same foundation this July 1. Disclaimer: this is a small random subset of foundations, but does show a trend.

Left: Apache, Debian (via SPI), FreeBSD, Gentoo, Haskell, LLVM, NumFOCUS, OSGeo, Python SF, Software Freedom Conservancy. Right: ASWF, CD Foundation, FINOS, GraphQL, Hyperledger, LF AI & Data, LF Energy, Open Mainframe, OpenSSF. US c3s: 90% (2023) to 66% (2024) and 69% (2025), recovering to 78% so far in 2026. Python SF’s list is volatile; without it the c3 dip is still 90% to 75%. LF programs: 82-89% every year. CNCF alone fell to 73% in 2024, so this is not entirely “LF is immune”. Excluded: non-US foundations, CNCF and the LF umbrella list (they mix members and landscape entries), OpenStack and Eclipse (page scope changes), OWASP (list reset in Nov 2023). Question for the audience: why would the same companies cut standalone charities first? Also: help needed to beef up this data and look for more trends.

template: fullheader name: taxoverview

Tracking REAL Money

Analyzing Foundation tax returns (US 990s)

??? Let’s change direction

template: logorb

Two Ways to Count Money

.left-column-equal[ Sponsorship Models

  • Listed sponsors × level price
  • What a program should bring in
  • Snapshot of a sponsor page
  • Includes in-kind at list price
  • Per program (for LF projects) ] .right-column-equal[ IRS 990 tax returns
  • Audited-style accounting
  • What a foundation did bring in
  • One fiscal year
  • In-kind support mostly not shown
  • Per US legal entity ]

.bottomnote[Sponsorship is “contributions” for c3, but “program service revenue” for c6]

??? While we’re still counting dollars, these are slightly different dollars. Sponsorships are estimated incomes from published sponsor lists and prospectuses - but we know some programs have discounts or the like. 990 data is submitted to the IRS, so it should be exact; however it uses the IRS definitions of “contributions” or “program service revenue”. Detailed 990 data shows how c3 public benefit organizations typically rely on “contributions” (i.e. donations), but c6 business leagues often rely on program services: running events, certification/training programs, or formal corporate Memberships.

template: logorb

Foundation Finances 990s

US IRS 990 tax forms capture:

  • Revenue: contributions, dues, program services
  • Expenses: program vs management vs fundraising
  • Net assets, cash, deferred revenue
  • Employees, volunteers, board
  • Officer pay; top contractors
  • Grants given; related organizations

??? With apologies for the US focus here - getting EU finance/tax data at scale takes a lot more effort. In the US, with many thanks to the Nonprofit Explorer by ProPublica. NEW this year: ProPublica lags 1-2 years and only extracts ~70 summary fields. FOSS990Parser reads the IRS’s own e-file XML directly (GivingTuesday data lake + IRS monthly zips): 398 filings for 65 foundations, 94 of them not yet in ProPublica, including fiscal year 2025. CAUTION: FY2025 data is partial. Calendar-year foundations file their 2025 990s from May to November 2026 (with extensions), so most FY2025 numbers in this talk come from early filers and non-calendar fiscal years like the ASF (April) and Conservancy (February). Trends use FY2024 as the last complete year.

template: logorb

Nonprofit Types (US focus)

.left-column-equal[ 501(c)(3)

  • Charitable Organization
  • Public Benefit Organization
  • Tax-deductible for donors
  • Restrictions on activities
  • Files IRS form 990 yearly ] .right-column-equal[ 501(c)(6)
  • Business League
  • Not tax-deductible
  • For “common business interest”
  • Files IRS form 990 yearly ]

.bottomnote[How do we map EU charities to this?]

??? So let’s change focus to real money, and track tax data of foundations.

template: logorb

Finance Data - 501(c)(3)

.left-column-equal-vsm[

  • Electronic Frontier Foundation
  • Signal Technology Foundation
  • Creative Commons Corporation
  • Numfocus Inc
  • The Freebsd Foundation
  • Software Freedom Conservancy Inc
  • Opencollective Foundation
  • Python Software Foundation
  • Beneficent Technology Inc
  • Apache Software Foundation
  • Plone Foundation
  • Software In The Public Interest Inc
  • Kuali Foundation Inc
  • Open Source Robotics Foundation Inc
  • Open Education Global Incorporated
  • Free Software Foundation Inc
  • Drupalcon Inc
  • United States Postgresql Association
  • Wikimedia Foundation
  • Mozilla Foundation
  • Open Technology Fund ] .right-column-equal-vsm[
  • Internet Systems Consortium Inc
  • Gnome Foundation Inc
  • Open Information Security Foundation Inc
  • Llvm Foundation
  • Wordpress Foundation
  • Participatory Culture Foundation
  • Open Molecular Software Foundation
  • Open Source Election Technology Institute
  • Ruby Central Inc
  • Linux Expo Of Southern California
  • Yet Another Society
  • Open Source Initiative
  • Haskell Org Inc
  • Biobricks Foundation Incorporated
  • Netbsd Foundation
  • Django Events Foundation North America
  • Tex Users Group
  • Software Freedom Law Center Inc
  • Sahana Software Foundation
  • Raspberry Pi Foundation North America Inc ]

template: logorb

Finance Data - 501(c)(6)

.left-column-equal[

  • Linux Foundation
  • AlmaLinux OS Foundation
  • Bytecode Alliance
  • Clojurists Together
  • Commonhaus Foundation
  • .Net Foundation
  • OASIS Open ] .right-column-equal[
  • Open Compute Project
  • Open Connectivity Foundation
  • Open Source Collective
  • OpenID Foundation
  • OpenInfra (OpenStack) Foundation
  • Rails Foundation
  • Rust Foundation ]

??? Recall that the Linux Foundation is the parent to two or three hundred foundations, like the CNCF, in case you’re thinking something’s missing here. They are directed funds inside the LF’s single 990, so they have no separate tax filings. Note: OSGeo and OpenJS file as 501(c)(4) organizations, which are slightly different than 501(c)(6)s.

template: fullheader

Foundation Finances

Let’s explore some 501(c)(3) finances! Public Benefit Organizations

template: logorb

C3 - Revenues

??? Here are the total revenues for 4 comparable US c3 foundations that host multiple projects. Years are fiscal year end (ASF’s year ends April 30, SFC’s in February). 2025 points are shown where the 990 has been filed; most calendar-year foundations file 2025 returns in late 2026.

template: logorb

C3 - Net Assets

??? Now we’re looking at the net assets at end of year for those four foundations.

template: logorb

C3 - Revenues - Outliers

??? Here are the same four foundations, plus the related-but-not-quite C3 foundations Wikimedia and Mozilla. Note that Signal Foundation and the EFF are both also highly endowed.

template: logorb

C3 - Net Assets - Outliers

??? Here are the same four foundations, plus the related-but-not-quite C3 foundations Wikimedia and Mozilla. Note that Signal Foundation and the EFF are both also highly endowed.

template: logorb

C3 - Annual Average Finances

.left-column-equal[ Average Revenues

  • All 44 tracked c3 orgs:
    • 9.2M USD
  • Not including Mozilla, Wikimedia, EFF, Signal:
    • 3.1M USD ] .right-column-equal[ Average Net Assets
  • All 44 tracked c3 orgs:
    • 11.5M USD
  • Not including Mozilla, Wikimedia, EFF, Signal:
    • 1.65M USD ]

.bottomnote[Fiscal year 2024 990s; fossfoundation.info + FOSS990Parser]

??? Fiscal year 2024 (the latest year nearly all foundations have filed). Last year’s deck showed FY2023: 2.6M average revenue and 1.9M average net assets for the smaller c3s.

template: logorb

Spending More Since 2023

.bottomnote[FY2024: 22 of 39 c3 foundations spent more than they took in]

??? This is the same set of 34 US 501(c)(3) FOSS foundations that filed every year 2019-2024, excluding the big four outliers and the Open Collective Foundation (which wound down in 2024, spending down ~$19M of net assets). Since FY2023, expenses have run ahead of revenue. Across 39 c3s (excluding OCF), net assets fell from 76M to 65M USD, about -14%, in one year. Examples: NumFOCUS 8.4M to 5.7M, Conservancy 6.8M to 4.8M, PSF 4.7M to 3.4M, FreeBSD 4.9M to 4.0M. Question for the audience: is this planned spend-down of pandemic-era reserves, or a funding gap?

template: fullheader

Foundation Finances

Let’s explore some 501(c)(6) finances!

template: logorb

C6 - Revenues

??? Here are the majority of US C6 foundation revenues.

template: logorb

C6 - Net Assets

??? Here are the majority of US C6 foundation net assets.

template: logorb

C6 - Revenues - Oh, Wait

??? Oh, I apologize, I forgot to include the Linux Foundation!

template: logorb

C6 - Net Assets - Oh, Wait

template: fullheader

Foundation Finances

Let’s compare c3s versus c6’s!

template: logorb

Foundation Net Asset Sums

Total net assets of 40 c3 foundations (excluding Wikimedia, Mozilla, EFF, Signal)

66M USD

Net assets of the Linux Foundation in 2024:

145M USD

.bottomnote[https://github.com/Punderthings/fossfoundation/blob/main/_data/p990/foundations_990_common.csv]

??? Figures are rounded, from US 990 forms for fiscal year 2024. (Last year’s slide compared total assets: the LF’s total assets are now 224M USD.) The c3 total includes the Open Collective Foundation’s wind-down; even excluding it, the smaller c3s lost about 14% of net assets in one year, while the LF grew about 16%.

template: fullheader name: spending

Where Does the Money Go?

??? New this year: the full IRS e-file data lets us look past top-line revenue - into how foundations earn, spend, and save.

template: logorb

How Foundations Spend

Foundation Expenses Program Staff Volunteers
Apache (FY Apr 2025) 2.3M 87% 4 8,200
NumFOCUS (2024) 11.1M 94% 25 383
Conservancy (FY Feb 2025) 3.8M 78% 8 5,000
Python SF (2024) 5.8M 86% 14 1,200
FreeBSD (2024) 2.6M 81% 6 5
OSI (2024) 1.0M 65% 1 20

.bottomnote[IRS 990 Part IX functional expenses; Part I staff and volunteer counts as reported]

??? IMPORTANT: These are high level accounting definitions from IRS 990 forms - so be sure to understand what’s being reported.

Program percentage = program services share of total functional expenses (vs management and fundraising). IRS definition of the program services column (Form 990 Part IX, column B): https://www.irs.gov/instructions/i990#en_US_2025_publink11283jd0e12386 (2025 instructions; the anchor changes with each year’s edition). 501(c)(6)s like the LF and Rust aren’t required to split expenses this way, so they’re not shown. Volunteer counts are self-reported estimates - but 8,200 for the ASF tells its own story.

template: logorb

Apache: Volunteers + Contractors

.left-column-equal[ FY ending April 2025

  • Revenue: 2.22M
  • Expenses: 2.33M
  • Net assets: 4.21M (~1.8 years)
  • Deferred revenue: 1.23M ] .right-column-equal[ Who gets paid?
  • Officers & directors: $0
  • Salaries: $0 (since FY2021)
  • Contractors over 100K: 3
    • System administration
    • Services
    • Publicity ]

.bottomnote[ASF 990 for FY2025 (TY2024), IRS e-file]

??? The ASF reports 4 employees but no salary expense: staff work is paid through contractors (990 Part VII-B), and every officer and board member is unpaid. Deferred revenue of 1.2M is mostly sponsorship paid in advance. A small operating deficit this year, against healthy reserves of about 1.8 years of expenses. IMPORTANT: Note that the ASF’s has a large in-kind infrastructure donation budget, which is not reflected here (i.e. cloud credits, hosting, CDNs that are provided directly)

template: logorb

Years of Runway

.left-column-equal[ Net assets ÷ annual expenses

  • GNOME: 3.2 $3.2M
  • SPI: 2.7 $2.6M
  • Apache: 1.8 $5.5M
  • FreeBSD: 1.6 $4.1M
  • Conservancy: 1.1 $5.4M ] .right-column-equal[  
  • OSI: 0.8 $1.0M
  • Linux Foundation: 0.7 $224M
  • Python SF: 0.6 $4.3M
  • NumFOCUS: 0.5 $6.4M
  • OpenInfra: 0.2 $1.4M ]

.bottomnote[Latest filed 990 per foundation (FY2024 or FY2025); small italics: total assets]

??? A rough resilience measure; years = NET assets / expenses, while the small italic figure is TOTAL assets (before liabilities). CAUTION: mixes FY2024 and FY2025 filings, depending on what each foundation has filed so far; FY2025 is partial. Note that net assets include restricted funds - for fiscal sponsors like NumFOCUS and Conservancy, much of that money belongs to specific projects. IMPORTANT: Note again, this does not include ASF’s in-kind infrastructure services donations, which are a separate critical support need for us.

template: logorb name: renew

Ask Your Employer to Renew

  • Does your employer sponsor the foundations you rely on?
  • At what level, and since when?
  • Renewing, or holding a level, is worth more than a new logo
  • Look up any foundation’s sponsor history:

.code[https://fossfoundation.info/sponsorships]

.bottomnote[Apache: https://www.apache.org/foundation/sponsorship.html]

??? Call to action built on the squeeze and the short tenure numbers: most sponsors who leave do so within two years, and few step down a level - they just stop. If you are inside a sponsoring company, the most useful thing may be making sure the renewal happens.

template: thanks name: closingslide class: inverse-header

Foundations Data

FOSS Foundations metadata and models

.code[https://fossfoundation.info/]

Do your own US funding research

.code[https://projects.propublica.org/nonprofits/]

.code[https://github.com/Punderthings/foss990parser]

Open Source Sustainability research

.code[https://fosssustainability.com/]

name: last-page template: thanks

Thank You & Questions

@shanecurcuru@fosstodon.org

.code[https://shaneslides.com]

Apache v2.0

template: logorb

Other topics to investigate

??? There’s a lot more to talk about out there - I look forward to your ideas!

template: logorb

Open Source is a ______

  • Sustainable contribution model.
  • Model for broad innovation.
  • NOT a Business model.

??? Yes, this is a trick question. https://medium.com/@stephenrwalli/there-is-still-no-open-source-business-model-8748738faa43

template: logorb name: publicmoney

Some Public Money is Out There

Funder Amount Reach
EU Next Generation Internet €140M (2019-2024) 1,200+ projects
Sovereign Tech Fund (DE) €23.5M since 2022 ~€19M/yr agency budget
Prototype Fund (DE) €12.3M 293 projects
STF: Log4j €596K (2023) one project
STF: PHP Foundation €205K (2023-24) one foundation

.bottomnote[Osborne et al. 2024, arXiv 2411.06027]

??? Osborne, Sharratt, Foster & Boehm, “A Toolkit for Measuring the Impacts of Public Funding on Open Source Software Development”, arXiv 2411.06027, 2024. Public funding is now real money, but mostly goes directly to projects and maintainers, not to foundations; the PHP Foundation grant is the notable exception. Prototype Fund: over half of funded projects struggled to secure follow-up funding afterwards - the same “what happens after the grant” problem foundations face. NGI uses cascade funding: 20% to intermediaries, 80% to third parties; an additional €32M is allocated for 2024-2027. Compare scale: the ~$66M total net assets of 40 US c3 FOSS foundations (slide on net asset sums). TODO: validate and clarify messages on this slide.

template: logorb name: appendix-lf-members

Linux Foundation Member History

.bottomnote[Appendix · LF umbrella only: Platinum 16 to 12, Gold 19 to 11, Silver 619 to 1,463]

??? LF umbrella membership only, not sub-foundations (CNCF, OpenSSF and others have their own lists). Counts are members listed on July 1 each year (2026 = June 1), from 80 versions of the LF landscape file in git, Dec 2019 to Oct 2026. Platinum: 16 (2020) to 12 (2026). Gold peaked at 19 in 2022 and is now 11. Silver more than doubled (619 to 1,463) and Associate grew eightfold (77 to 626). Many companies join the LF at Silver because a sub-foundation membership requires it, so Silver growth tracks the growth of LF projects as much as interest in the LF itself. Latest (Oct 7 2026): 13 Platinum, 11 Gold, 1,686 Silver, 717 Associate. TODO: validate and clarify messages on this slide.

template: logorb

Sponsorships - NumFOCUS

.left-column-equal[ Funding Sponsors

  • First: 3 Principal
  • Second: 7 Sustaining
  • Third: 3 Contributing ] .right-column-equal[ Other Sponsors
  • Community: 8
  • Grants: 3 ]

.bottomnote[book values (USD): Cash ~0.7M (Sept 2026); grants ?]

??? NumFOCUS has a smaller set of projects and a different focus, but otherwise is a c3 like the ASF that provides various project services. Here we see a notably smaller number of sponsoring organizations. Book value at list prices: 3x100K + 7x50K + 3x10K + 8x1K community = ~0.69M, from the Sept 10 2026 capture of numfocus.org/sponsors. A year earlier (Sept 2025) it was 5/8/3 and ~0.94M; the drop is two fewer Principal sponsors. The value is likely low overall, because their grant programs aren’t assigned specific values. Anyone know if they publicly disclose those levels of finances?

template: logorb

Sponsorships - CNCF

.left-column-equal[ Funding Sponsors

  • First: 15 Platinum
  • Second: 25 Gold
  • Third: 607 Silver ] .right-column-equal[ Other Sponsors
  • Academic: 4
  • Community: 24
  • End User: 86 ]

.bottomnote[book value (USD) 39M]

??? CNCF, which is a division of the Linux Foundation shows a couple of interesting differences. First, the sheer number of organizations listed as sponsoring: over 700 total. Plus, the presumed book value of those sponsorships: 39M (down from 42M just about a year ago). Note that various funding sponsors may well have different agreements, and also that other sponsors here likely have very little financial expectations vs. corporate sponsors. Level changes across all tracked foundations (same sponsor, same foundation, moving between paid levels): downgrades outnumbered upgrades in 2023 (20 down vs 12 up), 2025 (30 vs 22) and 2026 so far (24 vs 20); upgrades led every year 2019-2022. At CNCF itself it is roughly even: 37 downgrades vs 35 upgrades over the whole history.

template: logorb

Sponsorships - Linux Foundation

.left-column-equal[ Funding Sponsors

  • First: 13 Platinum
  • Second: 11 Gold
  • Third: 1,686 Silver
  • Fourth: 717 Associate ]

.bottomnote[book value (USD) ~41M · Oct 2026]

??? The thing I find most surprising at looking at the LF’s sponsorship list is that number on the third silver level - well over one thousand companies sponsor the LF at the silver level. This is partly because many of the LF’s sub foundations each require an LF silver membership. Overall LF sponsor numbers and funding numbers have gone up in the past year: Sept 2025 was 12/13/1,392/553 (book value ~35M); Oct 7 2026 is 13/11/1,686/717. Book value at list prices: 13x500K + 11x100K + 1,686x20K = ~41.3M. Silver is a sliding scale by company size (20K is the top), so this overstates Silver; Associate members pay nothing. See the appendix slide for the history since 2020: the top tiers shrank while Silver and Associate grew. Also note that the total book value is an approximation: it’s clear that there are sponsorship discounts when companies sponsor multiple LF subfoundations. It’s also clear that many software companies sponsor multiple LF subfoundations.

template: logorb

Contributions, per the IRS

.left-column-equal[ Linux Foundation, 2024

  • Total revenue: 221M
  • Contributions: 6.6M
  • Project & event fees: 149M
  • Training: 29M
  • Membership dues: 20M ] .right-column-equal[ Other c6 examples
  • OpenInfra: dues 4.2M booked as contributions
  • Open Compute: conferences 14M of 19M
  • Rust Foundation: dues 2.1M of 5.3M ]

.bottomnote[IRS 990 Part VIII line 2: program service revenue by type]

??? “Contributions” on a 990 means what IRS tax law says it means (Part VIII line 1: gifts, grants, membership dues treated as gifts), not the everyday sense of “all the money people give us”. Program service revenue (line 2) is money earned by doing the mission: https://www.irs.gov/instructions/i990#en_US_2025_publink11283jd0e10859 c6 trade associations book most corporate money as “program service revenue”, not contributions - and they don’t all book it the same way. So comparing “contributions” across c3s and c6s is apples to oranges. The detail lines in the e-file data fix that. TODO: This is a deep dive slide for future use.