But Is It Open?The online edition
Type & theme

Make yourself comfortable

Page color
Text size
20 px
Reading width
Ask about this chapter ↗

Your settings and place stay in this browser.

Chapter 6 · Timothy O’Brien

Who Pays for Open Source?

7 minute read16 source notesFree online edition

I drive a Jeep Wrangler.

I was poking through the menus on the in-vehicle screen when I found a link to view licenses, and I tapped it. What came up was overwhelming: Apache Software Foundation entries, the Linux kernel, and more, page after page. Most drivers think of that screen as part of the car. Underneath, it’s a pile of open-source software with a dashboard on top.

I start there because open source is in almost everything you use: your car, air-traffic-control systems, the app you order food from. Most people never open the license screen, so they never find out.

Free to use, not free to create

A driver can use all that software without ever paying its developers directly. That can hide the cost of making it.

Somebody wrote the code in my Jeep. Somebody reviewed the patches, fixed the bugs, and answered the security reports. That work cost something. I never paid for it directly, so someone else did.

The economic insight, stated bluntly

Shared work can produce better outcomes and profit. The biggest companies in the world contribute to and support open source because a world that doesn’t depend on brittle, proprietary standards and core components lets them do bigger and better things together. That’s self-interested cooperation, and I don’t mean it as an insult.

Companies fund much of the work on heavily used open-source infrastructure, often by paying their own engineers to contribute. That does not tell us how support is distributed across the much larger population of projects.

A 2024 funding report from the Linux Foundation, GitHub, and Harvard’s Laboratory for Innovation Science extrapolated from the organizations it surveyed and estimated that organizations put about $7.7 billion a year into open source. Eighty-six percent of that value was employee labor: people on payroll whose jobs include contributing. The authors point out that they surveyed the organizations most likely to support open source, so scaling up their answers directly would overstate the total. The $7.7 billion is a statistical estimate that tries to correct for that.1 The Linux kernel has the longest record of who pays. The Linux Foundation’s 2017 kernel development report found that unpaid developers contributed 8.2 percent of changes, down from 11.8 percent in 2014. In that period, most identified kernel contributions were paid work, with Intel, Red Hat, and Google among the sponsors. The kernel is an important example, not a representative sample of every open-source project.2

Releasing your company’s own stack is a different contribution from paying people to work on a shared dependency. Walmart has released projects from its engineering work; that fact alone does not establish how much it contributes to infrastructure such as the kernel.3

Who controls what gets funded

Corporate support can also concentrate control. Companies employ contributors, fund foundations, and sit on boards. A project can have a permissive license while its priorities remain effectively those of one company. The question is how far the community’s decision rights extend.

EU law approaches the roles separately. The Cyber Resilience Act excludes free and open-source software developed or supplied outside commercial activity, while creating a separate, lighter set of duties for qualifying open-source software stewards. Manufacturers integrating components have their own obligations.4 The revised Product Liability Directive excludes free and open-source software developed or supplied outside commercial activity, but manufacturers can be liable for products that incorporate it.5 The AI Act uses different tests: its exemption for open-source AI systems has exceptions for high-risk systems and certain prohibited or transparency-regulated uses. General-purpose models have a separate, partial exemption; copyright-policy and training-summary duties remain, and systemic-risk models do not get that exemption.6 You can read that as protecting volunteers. You can also read it, as some legal scholars have, as handing whoever takes on the compliance role a new position of control over open source.7 Open-source organizations pressed for changes while the Cyber Resilience Act was being negotiated.8

Corporations aren’t the only ones doing the work. Bug reports, fixes, documentation, and answers on a mailing list contribute value too. But use does not automatically produce contribution. A 2024 Tidelift survey found that about three in five maintainers are unpaid.9 Corporate money tends to follow the projects companies depend on and worry about, like the kernel, OpenSSL, Python, and Rust. The long tail of smaller projects gets very little. Small contributions matter. They do not by themselves balance the books.

Who actually pays: a spectrum

From my own experience, support looks like a spectrum.

You may get into open source before any company supports you, as a way to learn and become active in a community. That work may never become paid work.

Then you get a job and work on open source as part of it. What happens next depends on the project. A company-owned project is one thing. A project at the Apache Software Foundation or the Linux Foundation is another: your relationship with it may outlast your relationship with your employer. Plenty of people were paid to work on a project, saw the job end, and still feel emotionally and professionally invested. That happens more often than people realize.

Some people do open source purely for the community and the technical interest. They’re usually the ones who feel overwhelmed, because they have little or no support around them. Foundations are the exception: you can sometimes get a position with a foundation or a similar group. That’s rare, though, and foundations have their own bills to pay. The Linux Foundation forecast about $311 million in revenue for 2025, mostly from memberships, project services, events, and training, and much of its spending goes to project support and shared infrastructure.10 The Apache Software Foundation says roughly 70 percent of its budget, a little over $2.5 million a year, goes to infrastructure.11 The Open Source Initiative reported revenue of about $667,000 in 2025.12 Those budgets do not guarantee a salary for the person maintaining the small library your service depends on.

Corporate money follows direct commercial interest. Graphics-card and GPU companies fund device drivers for the Linux kernel, which is about as direct as it gets. An information-security vendor might pay for cryptographic tooling. You can usually predict where the money goes by asking who sells something that needs the code to work.

General-purpose projects are the hard case: their value is spread across many users, with no single buyer responsible for keeping them healthy. Curl, a tool and library for moving data across networks, shows one way to connect maintenance with a business. Since 2019 its founder and lead developer, Daniel Stenberg, has worked on curl full time at wolfSSL, paid through the commercial curl support that company sells, and the project takes donations and sponsorships to cover its own expenses.13 That arrangement pays for sustained work, but it depends on customers choosing support. Widespread use alone does not produce revenue for a project.

When the volunteers stop

The risk here isn’t abstract. In 2024 a lot of people found out how much depended on xz Utils, a compression library maintained largely by one volunteer. Over roughly two and a half years, someone using the name Jia Tan built trust in the project, took on maintenance work, and slipped a backdoor into it (CVE-2024-3094). Andres Freund found it while investigating unusual SSH performance.14 Afterward CISA, the U.S. cybersecurity agency, said that every technology manufacturer that profits from open source must contribute back, and that the burden shouldn’t fall on an individual maintainer.15

Companies have put some money toward this. Alpha-Omega, a Linux Foundation security initiative backed by AWS, Google, Microsoft, and others, had made more than 70 grants totaling over $20 million by March 2026.16 That is one initiative’s cumulative grant total, not the whole security budget. Companies, distributions, auditors, and other initiatives fund security work too. To judge whether the support is enough, we need to know which critical projects receive it and what maintenance work remains unpaid.

I don’t have a clean fix. What I’ve noticed is that the projects with the worst sustainability problems tend to look the most like infrastructure. Everyone uses them, and almost no one thinks about who keeps them running until something goes wrong.

Beyond software

This chapter has stayed with software, and the picture is messy enough there. But a lot of what makes open source work isn’t code. Competing companies also meet to agree on specifications, data formats, and shared ways of doing things, and that work has its own funding and its own governance.

Notes

  1. Sam Boysel, Frank Nagle, Hilary Carter, et al., “2024 Open Source Software Funding Report,” GitHub, the Linux Foundation, and Harvard Laboratory for Innovation Science, 19 November 2024, https://opensourcefundingsurvey2024.com. Survey respondents reported $1.7 billion a year; the $7.7 billion figure is a calibrated bootstrap extrapolation to all contributing organizations. See also Linux Foundation Research, “2024 Open Source Software Funding Report,” https://www.linuxfoundation.org/research/open-source-funding-2024. ↩

  2. The Linux Foundation, “The Linux Foundation Releases Annual Kernel Development Report,” press release on the 2017 Linux Kernel Development Report by Jonathan Corbet and Greg Kroah-Hartman, 25 October 2017, https://www.linuxfoundation.org/press/press-release/linux-foundation-releases-annual-kernel-development-report. Unpaid developers contributed 8.2 percent of changes, “significantly down from the 11.8% reported in 2014.” The top ten sponsoring organizations in that period were Intel, Red Hat, Linaro, IBM, Samsung, SUSE, Google, AMD, Renesas, and Mellanox. ↩

  3. Walmart Global Tech, GitHub organization “walmartlabs,” https://github.com/walmartlabs; Electrode, “Web applications with node.js and React,” copyright Walmart, Apache License 2.0, https://github.com/electrode-io/electrode. Walmart does not appear among the top kernel sponsors listed in the source for note 2. ↩

  4. Regulation (EU) 2024/2847 of the European Parliament and of the Council of 23 October 2024 (Cyber Resilience Act), Official Journal L, 20 November 2024, Art. 2(12) (noncommercial free and open-source software), Art. 3(14) (open-source software steward), Art. 13(5) (manufacturer due diligence when integrating free and open-source components), and Art. 24, https://eur-lex.europa.eu/eli/reg/2024/2847/oj/eng; European Commission, “Cyber Resilience Act - Open source,” Shaping Europe’s digital future, https://digital-strategy.ec.europa.eu/en/policies/cra-open-source. ↩

  5. Directive (EU) 2024/2853 of the European Parliament and of the Council of 23 October 2024 on liability for defective products, Art. 2(2) (excludes free and open-source software “developed or supplied outside the course of a commercial activity”) and recital 15 (a manufacturer that integrates such software into a product can be held liable), https://eur-lex.europa.eu/eli/dir/2024/2853/oj/eng. ↩

  6. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 (Artificial Intelligence Act), Arts. 2(12) and 53(2), https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng; article text via “Article 2: Scope,” EU Artificial Intelligence Act, https://artificialintelligenceact.eu/article/2/. The exemption covers AI systems released under free and open-source licences unless they are placed on the market as high-risk systems or fall under the prohibited-practice or transparency provisions. ↩

  7. Jennifer Tridgell, “Open or closing doors? The influence of ‘digital sovereignty’ in the EU’s Cybersecurity Strategy on cybersecurity of open-source software,” Computer Law & Security Review 56 (April 2025), 106078, https://doi.org/10.1016/j.clsr.2024.106078. ↩

  8. Max Roveri, “The Cyber Resilience Act and Open Source,” Linux Professional Institute blog, 9 September 2025, https://www.lpi.org/blog/2025/09/09/the-cyber-resilience-act-and-open-source. Secondary source. ↩

  9. Tidelift, “Top 5 findings from the 2024 maintainer survey,” 2024, https://assets-eu-01.kc-usercontent.com/ef593040-b591-0198-9506-ed88b30bc023/23071823-ff61-4de9-87ca-e8a67c533c56/survey-one-pager-24-.pdf (60 percent of maintainers not paid); David Rubinstein, “The state of open source maintainers,” SD Times, 2024, https://sdtimes.com/open-source/the-state-of-open-source-maintainers. ↩

  10. The Linux Foundation, “2025 Linux Foundation Annual Report,” December 2025, https://www.linuxfoundation.org/hubfs/Publications/2025%20Linux%20Foundation%20Annual%20Report_121825a_lr.pdf. Forecast gross revenue of $311.3 million: memberships and donations $133.3 million, project services $83.6 million, events $58.6 million, training and certification $29.6 million; forecast spending includes $181.9 million for project support. ↩

  11. The Apache Software Foundation, “ASF 2025 Year in Review: Building for Resilience and Growth,” The ASF Blog, December 2025, https://news.apache.org/foundation/entry/asf-2025-year-in-review-building-for-resilience-and-growth. ↩

  12. Open Source Initiative, “2025 Annual Report,” https://opensource.org/2025-annual-report (total revenue $667,000 in 2025, down from $1,091,000 in 2024). ↩

  13. Daniel Stenberg, “Five year full time curl anniversary,” daniel.haxx.se, 2 February 2024, https://daniel.haxx.se/blog/2024/02/02/five-year-full-time-curl-anniversary; curl, “Commercial Support,” https://curl.se/support.html; curl, “Donate to the curl project!,” https://curl.se/donation.html (donated funds are used only for project expenses). ↩

  14. Russ Cox, “Timeline of the xz open source attack,” research!rsc, 1 April 2024, updated 3 April 2024, https://research.swtch.com/xz-timeline (Jia Tan’s first patch on 29 October 2021; Andres Freund’s discovery on 28 March 2024; public disclosure on 29 March 2024); National Vulnerability Database, “CVE-2024-3094 Detail,” NIST, https://nvd.nist.gov/vuln/detail/CVE-2024-3094; Kevin Beaumont, “Inside the failed attempt to backdoor SSH globally, that got caught by chance,” DoublePulsar, 31 March 2024, https://doublepulsar.com/inside-the-failed-attempt-to-backdoor-ssh-globally-that-got-caught-by-chance-bbfe628fafdd. ↩

  15. Jack Cable and Aeva Black, “Lessons from XZ Utils: Achieving a More Sustainable Open Source Ecosystem,” CISA blog, 2024, https://www.cisa.gov/news-events/news/lessons-xz-utils-achieving-more-sustainable-open-source-ecosystem. ↩

  16. OpenSSF, “Linux Foundation Announces $12.5 Million in Grant Funding from Leading Organizations to Advance Open Source Security,” press release, 17 March 2026, https://openssf.org/press-release/2026/03/17/linux-foundation-announces-12-5-million-in-grant-funding-from-leading-organizations-to-advance-open-source-security. The “About Alpha-Omega” section gives the cumulative total (“over 70 grants totalling over $20M”) and lists Alpha-Omega’s backers as Anthropic, AWS, Citi, GitHub, Google, Google DeepMind, Microsoft, and OpenAI. ↩

From But Is It Open? by Timothy O’Brien (Discursive, 2026). © 2026 Timothy O’Brien. All rights reserved.

Chapter 67 min leftNext →

The book, chapter by chapter.

Source note

Go to the notes & sources ↓

The vocabulary.

A short guide to the book’s recurring terms. Follow each chapter for the argument and its sources.

Open weights
A release makes a trained model’s learned parameters available. The accompanying terms decide what you can do with them. Access to weights alone does not provide the training data, code or recipe. Read the chapter →
Copyleft
A way of using license conditions to preserve recipients’ freedoms when covered software is passed on. The obligations depend on the license and on what you distribute or make available. Read the chapter →
Permissive licenses
Licenses such as MIT and Apache 2.0 grant broad reuse rights with comparatively few conditions. Notice obligations still apply, and a software license does not automatically cover a release’s data or other components. Read the chapter →
Reproducibility
The ability to reconstruct and examine how an artifact was produced using the resources that are published. Running an existing checkpoint is a different achievement from rebuilding its training. Read the chapter →
Governance
Who can participate, who makes decisions, and how authority over a project or release is shared. A right to use the code does not by itself create a voice in upstream decisions. Read the chapter →
Training data
The examples used to train a model. Publishing the data, documenting its provenance and describing how it was filtered are distinct kinds of disclosure; privacy and other rights may limit what can be shared. Read the chapter →

A passage worth sharing.

Select a passage in the chapter, or share the one you’re reading.