What's next?

TL;DR

GitHub's alternatives are mostly worse or unready, but Forgejo/Codeberg is the best current option while Pierre builds next-generation Git infrastructure.

Key Points

• 1.GitHub is losing trust due to reliability failures. Random merge reversions and downtime measured in days rather than minutes are pushing developers to explore alternatives.
• 2.GitLab is the most-cited alternative but has deeply poor UX. Its 1.6M-line Ruby codebase, broken navigation, useless release pages, and Vue 2 frontend make it 'a worse GitHub with slightly better uptime.'
• 3.Bitbucket's only real value is Jira integration. Atlassian markets it primarily on cost savings ($15/user vs GitHub Enterprise), not on being a better product — 'Jira' appears five times on their homepage.
• 4.Forgejo and Codeberg are the strongest Gen 2 open-source alternatives. Forgejo is MIT-licensed Go code (~400K lines vs GitLab's 3.8M Ruby lines), has transparent Mastodon status updates, and actions files are nearly 1-to-1 compatible with GitHub Actions.
• 5.The host endorsed Forgejo by donating $1,200 upfront and pledging $400/month. He was surprised by how usable it was, praised its release tab and issue system, and plans to migrate some projects after the stream.
• 6.Pierre is building Gen 3 Git infrastructure via code.storage. It stored 9 million repos in 30 days, peaked at 15,000 repos/minute for 3 hours with no downtime — designed for high-throughput agentic workflows GitHub can't handle.
• 7.Pierre also ships open primitives: diffs.com and trees.software. Diffs.com powers T3 Chat's diff viewer; trees.software renders file trees — building foundational libraries for whoever creates the next GitHub.
• 8.Graphite was acquired by Cursor and may become a fundamentally different paradigm. Originally built to bring Facebook-style stack diffs to GitHub, it began mirroring repos internally to escape GitHub API slowness before the Cursor acquisition.
• 9.Entire, founded by ex-GitHub CEO Thomas, raised a $60M seed round. Its first product is a CLI tracking agent context — preserving *why* code changed, not just what changed — addressing context loss in agentic coding sessions.