Why I'm leaving GitHub for Forgejo

Most developers treat their Git host like a utility—something that just works until it doesn't. I used to be one of them. But after watching GitHub transform from a developer-first platform into a data-harvesting engine for Microsoft’s CoreAI division, I realized that convenience is a trap. I’ve moved my code to a self-hosted Forgejo instance, and it’s the best decision I’ve made for my digital autonomy.

If you’re still hosting your primary work on GitHub, you aren't just using a tool; you’re feeding a machine. The recent shift where interaction data from Copilot users is used for model training by default—with no repository-level opt-out—should be a massive red flag. As a maintainer, you no longer have control over whether your code becomes training material for the very AI that might eventually replace your workflow.

The Illusion of Independence

The argument for staying on GitHub used to be that Microsoft kept it at arm's length. That changed in August 2025 when GitHub ceased having its own CEO and was fully absorbed into Microsoft’s CoreAI division. When you push code to GitHub today, you aren't pushing to an independent platform; you’re pushing to a unit of an AI organization.

This isn't just about outages, though the 257 incidents logged between May 2025 and April 2026 are impossible to ignore. Those outages are a symptom of a platform struggling to scale under the weight of "agentic AI" workflows. When the infrastructure breaks, it’s because the platform is prioritizing AI features over core reliability. If you want to see what happens when a platform stops serving developers and starts serving models, look at the current state of GitHub.

Why Self-Hosting is the Only Real Solution

Moving to a self-hosted Forgejo instance on a single NUC isn't just a hobbyist project; it’s a strategic move for digital sovereignty. By hosting your own code, you regain control over your data, your privacy, and your jurisdiction.

Here is why this shift is gaining momentum:

1. True Ownership: You control the hardware and the software stack. No one can flip a switch and start training models on your private repositories.
2. Jurisdictional Security: By hosting outside of US-jurisdictional reach, you mitigate risks associated with FISA Section 702 and the CLOUD Act.
3. Stability: You aren't at the mercy of a third-party's "agentic AI" load. Your CI/CD pipelines run on your terms, not on an overloaded shared cluster.
4. Open Source Integrity: Forgejo is fully open source, ensuring that your hosting platform won't be sold or absorbed into a corporate AI strategy.

How to Reclaim Your Workflow

If you’re tired of the platform-as-a-service model, start small. You don't need to migrate everything overnight. Begin by setting up a Forgejo instance on a local server or a private VPS. Use it for your personal projects first. Once you’re comfortable with the workflow, start archiving your public GitHub repositories and pointing them to your new home.

Here’s where most people get tripped up: they think they need a massive team to manage their own infrastructure. You don't. A single, well-maintained NUC with a hardened setup is more than enough for most individual developers. The real question isn't whether you can afford to host your own code—it's whether you can afford the cost of losing control over it.

If you’re ready to take back your digital sovereignty, start by auditing your current repository settings and looking into self-hosting alternatives. The move to a self-hosted Forgejo instance is the only way to ensure your code remains yours.