[2/3] Octoverse 2026: Languages, Industries, and Open Source
In part one of this look at Octoverse 2026, we covered GitHub's scale and geography. Now let's look at what users do, which languages they choose, and which projects they gather around.
💻 Languages: TypeScript stays on top
The top three languages by number of contributors are unchanged: TypeScript, Python, and JavaScript. From September 2025 to September 2026, their monthly communities grew by 3.62 million, 3.23 million, and 1.82 million respectively. The ranking counts different forms of participation, from pushes to issues and reviews, and associates each account with the repository's primary language. It measures activity around a language, and one person can belong to several communities. The audiences can't be added together. TypeScript was the primary language in 35% of new repositories with detected signs of agent use and 24% of those without. GitHub relates the difference to the ease of checking agent-written code, though those shares don't separate the effect of agents from the choice of tasks or teams' habits.
🏭 Industries: the repository as a work tool
This year GitHub added an industry breakdown. A model classified random samples of public repositories created in 2022–2026 according to NAICS using their names, descriptions, topics, and READMEs. The authors estimate that:
– Professional, scientific, and technical services moved from fourth to second in repository count, behind the information sector.
– Repositories for mining, oil, and gas grew around 65× from a small 2022 baseline.
– Manufacturing repositories grew almost threefold.
What interests me here is the perspective: a repository as a tool for solving an industry problem. A sample of public projects doesn't tell us how much industrial software exists or what share of companies have adopted AI.
🌐 Open source
There were 2.5 billion public contributions in the year, up from 1.12 billion. GitHub split the rankings into independent or community-led projects and projects under the strategic control of one company. Fast-growing independent projects include llama.cpp, SGLang, and vLLM. Claude Code, Codex, and OpenCode lead the corporate top ten by absolute contributor growth.
A separate engagement ranking combines positions for stars, forks, and PR and issue authors. PyTorch, Rust, Home Assistant, and CPython are still among the independent projects there. Interest in agents coexists with work on languages, frameworks, and home automation.
I find this distinction useful when picking dependencies. Access to source code and influence over a project's direction are different things. It matters to a team who decides on compatibility, licensing, and the tool's future.
Now that we've looked at project choices, we'll see how people use these tools: delegating tasks, choosing models, and checking changes.
#Engineering #Software #OpenSource #AI #Metrics