History of Linux: from Torvalds' Kernel to Distributions, Summary

The origins: a student, a newsgroup and a name born by accident

In 1991 Linus Torvalds was a computer science student at the University of Helsinki, in Finland. For his coursework he used MINIX, a small operating system designed for teaching, and it was from that base that he started writing a kernel of his own, the program that acts as the go-between for a computer and its hardware. He built it with help from the GNU C Compiler, but independently of the operating systems that already existed.

On 25 August 1991, at 21, Torvalds announced the project on the comp.os.minix newsgroup, in a message that has become famous in computing history:

“I’m doing a (free) operating system (just a hobby, won’t be big and professional like gnu) for 386(486) AT clones.”

In that same post he noted that the kernel was “free of any minix code” and had a multi-threaded file system, though he warned that, in that first version, it wasn’t yet portable to other architectures. The first public release, version 0.01, followed shortly after, on 17 September 1991.

The final name came about almost by accident. Torvalds wanted to call the project “Freax”, a blend of “free”, “freak” and “x”, and kept the files under that name for about six months; he had also considered “Linux”, but dropped it as too self-centered. It was Ari Lemmke, at the Helsinki University of Technology, who renamed the project’s folder on the FTP server to “Linux” in September 1991, without asking Torvalds first, who accepted the name after the fact. Five years later, in 1996, the mascot arrived too: Tux the penguin, drawn by Larry Ewing, inspired by a bite Torvalds got from a penguin during a visit to the National Zoo & Aquarium in Canberra, Australia. The name was suggested by James Hughes, as a play on “Torvalds’ UniX” and a nod to the color of a tuxedo.

Just the kernel: why people also say GNU/Linux

This is where one of the most common mix-ups about this operating system comes from: “Linux”, strictly speaking, isn’t a complete system, just the kernel. A kernel alone doesn’t let you switch on a computer, open a program or browse the web, it also needs the whole set of tools around it, from compilers to basic utilities to the interfaces users actually interact with.

Much of that toolkit comes from the GNU project, announced by Richard Stallman on 27 September 1983 on Usenet, with development starting in January 1984. Stallman’s goal was to build a Unix-compatible operating system that was entirely free, after an episode that had shaped his thinking in the late 1970s at MIT’s AI Lab: when the institute installed a printer with proprietary firmware, he couldn’t modify its code to fix a practical problem, and saw in that restriction a wider issue, not just a technical one. The GNU Manifesto listed four essential freedoms for anyone using a program: running it for any purpose, studying and changing how it works, redistributing copies, and improving it and sharing those improvements with others. To protect these freedoms, Stallman wrote the GNU General Public License, the GPL.

When Torvalds’ kernel met the GNU tools, the result was the first entirely free, complete operating system. Torvalds announced that the GPL would take effect on 1 February 1992; version 0.95, from 7 March of that year, was the first release actually distributed under that license. He later called it “the best thing I ever did”. From there came the habit, born in projects like Debian, of calling the complete system “GNU/Linux” rather than simply “Linux”: a technical distinction more than a political one, useful for remembering where the pieces of the system really came from. It’s a different story from other operating systems born from the start as closed corporate products, as our piece on the history of Microsoft Windows explains.

Distributions: one kernel, many different packages

A kernel plus the GNU tools still isn’t quite what a user installs on their computer. It also takes a package manager, a boot system, often a graphical environment: that whole bundle of choices, packaged and ready to use, is called a distribution. Several were born starting in the early 1990s, each with its own priorities.

Distribution	Year founded	Feature
Slackware	1993	The oldest surviving distribution
Debian	1993	The largest community-based distribution
Red Hat	1994	First 1.0 release
SUSE	1994	First 1.0 release, developed in Germany
Ubuntu	2004	Derived from Debian, built for ease of use
openSUSE	2005	Community variant of SUSE, backed by Novell

Each distribution makes different choices on top of the same underlying material. Debian focuses on stability and community control; Ubuntu, which derives from it, focuses on easy installation and commercial support; Red Hat and SUSE targeted businesses from the start. The sense that “there are too many versions of Linux” is more apparent than real: they all share the same kernel, and picking one popular distribution doesn’t require knowing all the others. Choosing between two distributions, for that matter, echoes the choice people face between Windows, macOS and Linux as their computer’s operating system, a comparison we cover in our article on the differences between Windows, macOS and Linux.

Practical example: picture having to pick a computer for a relative who isn’t tech-savvy, between two options with the exact same Linux kernel underneath: a “pure” Debian install, built for someone who wants to manage every package by hand, and a distribution like Ubuntu, which adds a graphical app center, preconfigured drivers and a guided setup that takes a few minutes on top of that same kernel. The engine, the kernel, is identical in both cases; what changes is everything else, that choice of packages, tools and interfaces that makes each distribution suited to a different kind of use, not unlike two cars sharing the same engine but with a completely different body, gearbox and dashboard.

Where the Linux kernel runs: servers, supercomputers, phones

As of September 2026, according to W3Techs tracking, Linux ran on 62.2% of websites with a known underlying operating system, and remained the dominant platform within the broader “Unix” category, which covered 92.0% of tracked sites. Sites running Linux included names such as Microsoft.com, X.com and The New York Times. Wikipedia’s entry on Linux distributions cites a similar estimate, around 64.5% of the web server market, without a precise date: rather than a contradiction between sources, it reflects different measurement methods. It’s a significant piece of the infrastructure behind the sites we visit every day.

On the supercomputer side, the growth has been even sharper. Linux first appeared on the TOP500 list in June 1998, with a single system, the Avalon Cluster, ranked 314th, just 0.2% of the total. By November 2002 it had reached 14.2%, with 71 installations; by November 2009, 89.6%, on 448 of the 500 systems, a 531% growth in Linux installations in just seven years. In November 2017 the last two non-Linux systems dropped off the list, pushing the share to 100%; by November 2024, for a seventh consecutive year, all 500 supercomputers on the list still ran Linux, including the three exascale systems from the US Department of Energy: El Capitan, Frontier and Aurora.

Then there’s the particular case of Android. Google’s mobile operating system is built on the Linux kernel to talk to a phone’s hardware, but from there the path diverges from traditional GNU/Linux distributions: apps run on the Android Runtime (ART) instead of as ordinary compiled binaries, the C library is Bionic instead of the GNU C Library, and the Android framework relies on its own system services, like SurfaceFlinger, instead of the traditional Unix daemons. One reason Android and Linux sometimes get confused is precisely that shared origin in the kernel; market statistics, though, treat them as separate categories: according to Statcounter, as of August 2026 Android covered 67.61% of the worldwide mobile market, while “Linux” tracked as its own category stood at just 0.01%. Anyone who wants to dig deeper into their Android phone, beyond the standard interface, usually starts right there, by turning on the hidden developer options panel, one chapter among many in the history of computers.