The 8 Ways to Make Money from Open Source (Ranked by How Long They'll Survive AI)

· playbookai-era
The 8 Ways to Make Money from Open Source (Ranked by How Long They'll Survive AI)

There are exactly eight ways people make money from open source. Some of them are about to get much harder. Some were never about the code to begin with — and those are the ones that will outlive the current shift.

Here is the full landscape, ranked by how well each model survives a world where any competent developer with an agent can rebuild your product in a weekend.

1. Selling support and operations

The Red Hat model. Nobody ever paid Red Hat for Linux — Linux was free. They paid for accountability: someone to call, someone to escalate to, someone the auditors can put on a slide.

Agents can generate code. They cannot be on the hook at 3am when your deployment eats itself. Survives AI untouched.

2. Hosting and managed services

Uptime is a track record, not a feature. An agent can generate a Supabase clone in a weekend. It cannot generate five years of 99.99% uptime history, a SOC2 report, or a security team that has handled a real breach.

Survives AI — arguably gets stronger.

3. Sponsorships

Sponsors fund humans and track records, not repos. Nobody sponsors a codebase. The donation button on a maintainer profile converts because of years of visible, verifiable work.

Survives AI untouched.

4. Selling the whole product to enterprises (governance, not features)

The quiet winner. SSO, audit logs, RBAC — companies pay for these not because the code is hard, but because their compliance regime requires someone accountable to provide them. “My agent built our own SSO” does not survive a SOC2 audit.

Survives AI — the compliance surface is the product.

5. Grants and foundations

NLnet, Sovereign Tech Fund, Open Collective fiscal hosting. Grant bodies fund public goods and stewardship — a category agents cannot fabricate.

Survives AI.

6. Paid work and bounties

Algora-style issue bounties. Works, but watch the new failure mode: agent-generated low-effort bounty claims flooding your queue. Viable with screening.

Survives AI with friction.

7. Open core with feature gates

The paid tier exists because the features took months to build. But a clone doesn’t care about your months. Feature-gating weakens badly when generation is free — gate on operations and governance instead, or lose the tier.

Weakened by AI.

8. Dual licensing and license fees

The “pay or we get copied” threat only worked when copying was expensive. It isn’t anymore. Selling legal certainty still works for compliance-driven corporates — but as a threat, it’s hollow.

Most weakened by AI.

The pattern

Every model that survives is selling something that is not the code: accountability, uptime, a track record, a compliance surface, stewardship. Every model that dies is selling the code itself.

That is not a coincidence. It is the whole thesis of this blog: code is free now. Someone who will still be fixing it in three years is not — and that asymmetry is the business model.

Before you pick a model, check how projects already running it score on maintenance health over at the open source scorecards. The winners of the AI era will be the projects that are still alive in three years.

Before you copy any playbook, check how the projects in this cohort actually score on maintenance health: open source scorecards.