Key takeaways
- India now has 32.0 million developers on GitHub, second only to the US (38.3 million), according to GitHub's Octoverse 2026 report, published 28 October. China is a distant third at 12.4 million.
- India and the US each added more than 10 million developers in twelve months. Because India started from a smaller base, its growth rate is higher. Digit's analysis of the report puts it at about 46% for India and 37% for the US.
- Agents have moved from novelty to normal. By September 2026, developers started more than 40% of pull requests with an AI agent, up from 17% in March.
- Activity roughly tripled in a year. There were 34.3 billion commits, 17.9 billion pushes and 1.3 billion merged PRs, so the total number of PRs has grown along with the share that agents start.
- What to do: stop measuring yourself by how much code you type. Get good at specifying, reviewing, testing and shipping work that agents help produce, and have a public GitHub profile that shows it.
1. What GitHub published
Octoverse is GitHub's yearly report on how developers use the platform. The 2026 edition covers the twelve months ending 30 September 2026 and, unlike earlier years, counts activity in both public and private repositories. That change in method is one reason some year-on-year jumps look so large, so compare trends rather than exact multiples.
The headline is that the platform now has more than 225 million developers. The second big story is AI agents: tools like Copilot's coding agent, Claude Code, Codex and similar tools now open, edit and merge code at scale.
2. The activity numbers, side by side
All figures are from GitHub's report and cover the 12 months to 30 September of each year.
| Metric | 2024 | 2025 | 2026 | 2026 vs 2025 |
|---|---|---|---|---|
| Commits | 10.0B | 12.9B | 34.3B | ~2.7× |
| Pushes | 5.2B | 6.7B | 17.9B | ~2.7× |
| Merged pull requests | 402M | 538M | 1.3B | ~2.4× |
| GitHub Actions jobs | 8.0B | 14.0B | 45.4B | ~3.2× |
Some other figures from the report:
- Public contributions: about 2.5 billion, up from 1.12 billion the year before.
- Dependabot: more than 30 million merged pull requests, about double the 2025 pace. Automated dependency updates are now a large share of everyday repo traffic.
- Agent configuration files (files that tell an agent how to work in a repo): new repositories that had one logged about eight times as many commits as new repositories without one.
- Model choice: 61% of Copilot users in September used more than one AI model.
The pattern is clear. More code is being produced, more of it is machine-assisted, and CI pipelines (Actions jobs grew fastest) are doing the checking. If agents write more of the code, your test suite and review process become the real quality gate.
3. India's position: second, and closing the gap
| Country | Developers (Sep 2026) | Notes |
|---|---|---|
| United States | 38.3M | Added 10M+ this year |
| India | 32.0M | Added 10M+ this year; projected 57.5M by 2030 |
| China | 12.4M | — |
| Brazil | 10.2M | Added 3.28M, the third-largest increase |
| Japan | 7.15M | — |
| United Kingdom | 6.47M | — |
| Indonesia | 6.34M | — |
Source: GitHub Octoverse 2026.
Three points matter for Indian readers:
- The 2030 projection still holds. GitHub forecasts 57.5 million Indian developers by 2030, ahead of a projected 54.7 million in the US. One year into that window, India is 28.3% of the way to its target and the US is 38.6% of the way to its own. India is still on course, but the crossover is not guaranteed, and Digit notes that GitHub does not give a crossover date.
- India ranks 2nd for pushes to public repositories (adjusted by profile location), behind the US and ahead of China, Brazil and Germany. Indian developers are visible in open source, not only in private enterprise repos.
- Raw headcount is not the same as earning power. A large share of these 32 million are students and early-career developers. The supply of "can write code" is huge and growing, so that alone will not get you hired. That ties back to what we saw in October's hiring data: openings are rising, but entry-level roles are a small slice of them.
4. Agents at 40% of PRs: what changes in day-to-day work
When four in ten pull requests start with an agent, the job shifts. Here is the practical comparison:
| Skill | Value before agents | Value with agents (2026) |
|---|---|---|
| Typing boilerplate, CRUD, glue code | High for juniors | Low, since agents do it in seconds |
| Writing a clear issue/spec | Medium | High, since the spec is the agent's input |
| Reading and reviewing a diff | Senior-only | Expected from everyone |
| Writing tests and CI checks | Often skipped | Core, since tests are what catch agent mistakes |
| Debugging production / reading logs | High | Still high, and agents are weaker here |
| Security and dependency hygiene | Specialist | Everyday work (Dependabot volume shows this) |
| Repo conventions (README, agent config, contribution rules) | Nice to have | Measurably linked to more activity |
GitHub's sector data adds a reality check. In industries like healthcare, agriculture and administrative services, more than 4 in 10 repos mention AI, but agent-started PRs there are under 3%. Most Indian services and GCC work for those clients will still need people who can modernise legacy code by hand, with agents as helpers, not autopilot.
5. What to do: a plan for Indian students and engineers
If you're a student or fresher (0–1 year):
- Make your GitHub profile your CV. Pin 3–4 projects that each have a README, tests and a working CI badge (GitHub Actions is free for public repos). Recruiters cannot tell a hand-written repo from an agent-written one, but they can tell a maintained one from an abandoned one.
- Use an agent on purpose, and show it. Add an agent instructions file (for example
AGENTS.mdor.github/copilot-instructions.md) to one project and write in the README how you used AI and what you checked yourself. That shows judgement, which is the scarce skill now. - Contribute one real PR to an open-source project used by Indian companies (a docs fix, a test, a small bug). India ranks 2nd for public pushes, but one merged upstream PR still stands out in a pile of tutorial clones.
- Pick a stack the data supports. The report ranks TypeScript, Python and JavaScript as the top three languages by monthly contributors. TypeScript plus Python covers most web, backend and AI-glue roles in Indian job listings.
If you're working (2–8 years):
- Become the reviewer. Volunteer to own review rules, branch protection and CI on your team's repos. As agent output grows, whoever controls the quality gate becomes more valuable.
- Learn to measure agent work. Track how many agent PRs are merged, reverted, or cause incidents. Teams that can show these numbers will make better tooling decisions.
- Keep costs in mind. AI coding tools such as Copilot are priced per seat in US dollars (see GitHub's plans page), so Indian teams also pay 18% GST and card FX markup on top. For a 20-person team, even a mid-tier seat adds up to several lakh rupees a year, so measure the gain on one team before you roll it out everywhere.
If you run a team or a startup:
- Add
AGENTS.md/agent config, CODEOWNERS and required checks to your main repos before you widen agent access. - Turn on Dependabot and secret scanning. With this much automated code being merged, catching leaked keys and vulnerable packages automatically is no longer optional.
6. How to read these numbers carefully
- Method change: 2026 includes private repositories and uses a different twelve-month window than older reports, so some of the "tripling" comes from what is now being counted.
- "Developer" means a GitHub account located by profile, not an employed engineer. Ignore older articles still quoting 2025 figures (about 22 million Indian developers). The current number is 32 million.
- Agent-started is not agent-written. A PR "started with an agent" may still be heavily edited by a human. The 40% figure shows how widely agents are used, not that they replaced 40% of developers.
Bottom line
India is now the world's second-largest developer community on GitHub and still growing faster than the US in percentage terms. In the same year, agents became a normal part of how code reaches production. For Indian engineers that is both good and bad news: the market is bigger, but "I can write code" is no longer enough to stand out. The people who stand out will be the ones who can specify, review, test, secure and ship work, whether it was written by them or by an agent.
Planning your next step? Follow our software developer roadmap or DevOps & SRE roadmap, see the CI/CD and DevOps toolchain, and update your CV with the free resume builder.