Mostly, nobody pays. Of the twenty-odd projects in a typical homelab stack, maybe five have someone working on them full-time for money; the rest run on evenings and goodwill. The 2024 xz-utils backdoor — a burned-out solo maintainer socially engineered into handing commit access to his attacker — is what that model looks like when it fails, and it nearly shipped a backdoor into every SSH server on earth. Here's how the software you rely on actually gets funded, and the short list of things a user can do that measurably help.
The five funding models behind your stack
| Model | Examples | Failure mode |
|---|---|---|
| Volunteers + donations | Syncthing, FreshRSS, Vaultwarden | Burnout; bus factor of one |
| Open-core company | Gitea, n8n, Portainer | Features drift to the paid tier; relicense risk |
| Patron company | Immich (FUTO, since 2024) | Patron's priorities change |
| Foundation | Home Assistant (Open Home Foundation, 2024) | Slow; but the sturdiest |
| Paid cloud funds the code | Plausible, Bitwarden, Nextcloud GmbH | Incentive to gate convenience features |
The failure-mode column is the forecast. Volunteer projects die quietly when life happens to one person. Open-core companies under revenue pressure move features behind the paywall, then sometimes relicense — the progression covered in licences explained for self-hosters. Foundations are slow and unglamorous and outlive everything else.
The maintainer math is worse than you think
Public data points are remarkably consistent: donation income for even very popular infrastructure projects is typically hundreds of dollars a month, not thousands — visible in the open budgets on Open Collective. Tidelift's maintainer surveys have repeatedly found a majority of open source maintainers earn nothing at all from that work. Meanwhile the demand side assumes enterprise SLAs: the same unpaid person is expected to triage a CVE within days, review drive-by PRs, and answer "any update on this?" comments from companies with nine-figure revenues.
Two counterexamples show what funding actually buys. Immich was a nights-and-weekends project until FUTO started paying a full-time team in 2024 — the release cadence, mobile app quality, and issue-close rate changed visibly within months. Home Assistant moved its code and trademarks into the Open Home Foundation in 2024, funded substantially by Nabu Casa's $6.50/month cloud subscription — a product many subscribers buy explicitly as a funding channel with a convenience wrapper. Funded projects ship; that's the whole story.
What actually helps, ranked by value per hour
Money is not the only currency, and for most projects it isn't even the scarcest one. Ranked by what an hour of your effort is worth to a maintainer:
- A reproducible bug report — versions, logs, minimal steps — saves the maintainer the hour they'd spend extracting those from you. A vague one costs them that hour instead.
- Answering other users' questions in the forum or Discord. Support load, not code, is the burnout driver on big projects.
- Documentation PRs. Perpetually understaffed, gratefully merged, and they don't demand deep codebase knowledge.
- Testing release candidates and betas on your real setup. Every bug caught in RC is a bug that didn't generate fifty duplicate issues.
- Recurring donations — even $2/month. Recurring income is plannable; one-off spikes are not. GitHub Sponsors, Open Collective, and Liberapay all handle this; many projects list their preference in a
FUNDING.yml.
And the highest-value move available to anyone employed in tech: if your company uses a project, get it to buy the enterprise tier or a support contract, even unnecessarily. One business licence routinely equals hundreds of individual donors.
Audit your own stack tonight
Ten minutes to find out who you're free-riding on:
# Funding links for the projects you actually run
for r in immich-app/immich jellyfin/jellyfin louislam/uptime-kuma \
paperless-ngx/paperless-ngx syncthing/syncthing; do
echo "== $r"
curl -s "https://raw.githubusercontent.com/$r/HEAD/.github/FUNDING.yml"
doneWhile you're there, check the contributor graph. If the last 90 days of commits come from one person, you're looking at the project's bus factor, and it's 1.
The free-rider math, honestly
A self-hosted stack replacing $40/month of subscriptions runs on software that cost you nothing. Redirecting 10–15% of the savings — $5/month to each of three projects — costs $180/year, less than a single seat of most SaaS you left, and lands in budgets where a few hundred dollars a month is the difference between "hobby" and "one paid day a week". Nobody owes this. It's still the best price-to-impact ratio in your entire budget.
What I'd do
Pick the three projects whose disappearance would hurt most — for most homelabs that's the photo app, the automation platform, and the backup tool — and set up $5/month to each. File one properly reproducible bug report the next time something breaks instead of a "me too" comment. And if you're employed: one internal message to whoever owns tooling budget, suggesting a support contract for the open source your team already depends on, likely outweighs everything else on this page.