What Is GitHub?

Most people who use software every day have never heard of GitHub. That is not surprising, it operates quietly, behind the scenes, in the same way that electricity operates behind the appliances that depend on it. GitHub is a website where software developers store their code, track every change they make to it, and collaborate with other developers anywhere in the world. It now hosts over 180 million developers globally and surpassed one billion stored projects in June 2025. To put that number in perspective: if every project were a book, GitHub’s library would dwarf the collections of the world’s largest physical archives many times over.

Understanding why GitHub exists requires understanding a problem that software developers have faced since the beginning of the industry. Writing code is not like writing a document that one person finishes and hands to another. It is a continuous, overlapping, often chaotic process in which many people make changes simultaneously, mistakes get introduced and need to be undone, and the same project may branch in several directions at once before anyone knows which direction is the right one. Managing all of that without a structured system is, at scale, impossible.

The solution is called version control. A version control system records every change made to a piece of code, who made it, when, and what exactly was altered, and allows developers to move backward or forward through that history at any point. Think of it as the “track changes” feature in a word processor, but far more powerful, applied to entire projects containing millions of lines of code, and shared across dozens or hundreds of contributors simultaneously.

How Git and GitHub are different things

GitHub is built on top of a tool called Git, and the two are frequently confused. They are not the same.

Git is the version control system itself — the underlying technology. Linus Torvalds, the Finnish-American engineer who also created the Linux operating system, built it in just over a week in 2005. He needed a replacement for BitKeeper, a version control tool he had been using for the Linux project, after its free licence was revoked. Torvalds built Git to solve a specific frustration: older version control systems stored all the code in one central place, which meant every developer had to connect to that central server to do anything. If the server was slow, everyone was slow. If it went down, work stopped entirely.

Git took a different approach. Instead of keeping one master copy that everyone draws from, Git gives every developer their own complete copy of the entire project and its history, stored on their own machine. Work can happen offline. Changes are merged together when developers are ready. Nothing is lost because no single server holds the only copy. This was a fundamental shift in how collaborative software development could work, and Git became the dominant version control system in the industry within a few years of its release.

GitHub, by contrast, is a company and a website launched in February 2008 by Chris Wanstrath, P. J. Hyett, Tom Preston-Werner, and Scott Chacon that built a friendly, visual platform on top of Git. Where Git is a set of command-line tools that require technical knowledge to operate, GitHub wraps those tools in a website that makes the key functions visible, navigable, and social. Developers can browse code in a browser, comment on specific lines, propose changes, and discuss problems in a format that resembles, in some ways, a professional social network organised entirely around software projects.

Microsoft acquired GitHub in 2018 for $7.5 billion. The acquisition drew concern from developers who worried that a large corporation owning the primary hub of open-source software would compromise the culture of openness that made it valuable. Those concerns have not, in the years since, produced a catastrophic outcome. The platform has grown faster under Microsoft’s ownership than it did before, adding features, expanding access to free accounts, and integrating AI tooling at every layer of the development workflow.

What developers actually do on GitHub

The core unit of GitHub is the repository, usually called a repo. A repository is a project folder that contains all the files belonging to a piece of software the code itself, documentation explaining how to use it, configuration files, and the complete history of every change ever made. Repositories can be public, meaning anyone in the world can view and contribute to them, or private, meaning access is restricted to specific people.

When a developer wants to work on a project, they make a local copy of the repository on their own computer. They then make their changes, test them, and when ready, send those changes back to the shared version on GitHub. Before those changes are incorporated into the main project, they typically go through a process called a pull request, a formal proposal that says, in effect, “I have made these specific changes; here is why; please review them before merging them in.” Other developers can read the proposed changes, leave comments, ask questions, and either approve or request further revisions. This review process, now standard across the industry, is one of the main reasons that large software projects can maintain quality across many contributors who may never meet in person.

Branches extend this further. Rather than making changes directly to the main version of a project, which could break things for everyone immediately, developers create a branch, a parallel copy of the project in which they can experiment freely. If the experiment works, the branch gets merged back in. If it does not, it gets discarded, and the main project is unaffected. The ability to branch, test, and merge without risk is one of the features that makes large-scale software collaboration manageable.

GitHub also supports a culture of open-source development, in which the source code of a project is made publicly available for anyone to read, use, modify, and redistribute. The majority of the software tools that power the modern internet, web frameworks, databases, operating systems, AI libraries, were built this way, with hundreds or thousands of contributors working voluntarily on shared repositories hosted on GitHub.

What the numbers reveal about the present moment

GitHub publishes an annual report called the Octoverse, which tracks activity across the platform. The 2025 edition contained several findings that point beyond platform growth toward broader shifts in the global software industry.

More than 36 million new developers joined GitHub during the 2024–25 period, a rate of more than one new developer every second, sustained across the entire year. Developers pushed nearly one billion commits to their repositories, a 25% increase on the prior year. India overtook the United States in total open-source contributor count for the first time in the platform’s history, adding over 5.2 million developers in 2025 alone. Brazil, Indonesia, Japan, and Germany also posted substantial growth. The centre of gravity of global software development is shifting, and GitHub’s data captures that shift in precise terms.

The language data are equally instructive. TypeScript, a version of JavaScript that adds a strict type system, catching errors before code is run rather than after, overtook Python and JavaScript as the most-used language by contributor count in August 2025, growing at a rate of 66.63% year-over-year. This matters because it reflects a maturing industry response to scale: as codebases grow larger and AI-generated code becomes harder to verify by hand, developers are moving toward languages that surface mistakes earlier in the process.

The fastest-growing projects on the platform in 2025 were not consumer applications or entertainment products. They were AI infrastructure tools, frameworks for running large language models, systems for building AI agents, libraries for orchestrating AI workflows. GitHub has become the primary distribution channel for the open-source AI stack. When a research lab releases a new model, or a startup publishes a new agent framework, they publish it on GitHub. The platform does not just host the tools that developers use. It now hosts the tools that AI systems use.

Why this matters beyond software

For most of computing history, GitHub was a specialist concern. It mattered to software engineers and, increasingly, to data scientists. The question of whether someone understood repositories and pull requests was irrelevant to nearly every other profession.

That is no longer the case. The rise of AI coding tools, GitHub Copilot being the most prominent, launched in 2021 and now embedded across the platform, has begun collapsing the distance between people who write code professionally and people who direct AI systems to write code on their behalf. The skill of navigating a repository, reading documentation, understanding what a project does and how to modify it, is becoming relevant to a much wider range of workers than it was five years ago. GitHub is not simply the place where software is stored. It has become the terrain on which the tools that are reshaping every other industry are assembled, tested, and distributed. Those who can read that terrain have a material advantage. Those who cannot are working at a remove from the infrastructure that is reorganising the economy around them.

Phil