WhereIsAtlas · Open source

What Happened to cURL?

Nothing — which is the story. The tool Daniel Stenberg wrote in 1996 to fetch a web page became libcurl, and libcurl is now inside almost everything that talks to a server.

cURL is alive, actively maintained and installed billions of times over. Created in 1996 by Daniel Stenberg and given an embeddable library, libcurl, in 2000, it never needed a rescue: it became the quiet layer that most of the internet's data transfers run through.

A Swedish developer and a currency bot

cURL started as a problem about exchange rates. In 1996 Daniel Stenberg, a Swedish developer, wanted users of an internet relay chat bot to be able to ask for currency conversions, and he needed a small program that could fetch a URL from the command line and print the result. He called it httpget, released the source, and then — as users asked for more protocols, more options and more control — renamed it curl and kept building.

That origin explains a lot about the tool. It was written by someone who wanted a simple, scriptable way to move bytes across a network, and it has stayed true to that: one command, an enormous number of options, and behaviour predictable enough that people have been writing scripts around it for nearly three decades.

libcurl and the invisible layer

The turning point came in 2000, when the project separated its transfer engine into a reusable library called libcurl. That single decision turned curl from a utility into infrastructure. Programmers no longer had to shell out to a command-line tool; they could link a library and get HTTP, HTTPS and a long list of other protocols inside their own application.

This is why the honest answer to "who uses curl" is: almost everyone, usually without knowing. libcurl is compiled into applications, operating systems, devices and appliances, and it is the reason a smart television, a car's infotainment system and a server-side script can all make the same kind of request with the same battle-tested code. The command-line tool, meanwhile, became the default way to test an API, check a header, download a file or debug a certificate problem.

Why nothing replaced it

Plenty of software has tried to do parts of curl's job, and none has displaced it. The reasons are mundane and powerful: it supports an enormous range of protocols and platforms, it is free under a permissive licence that allows commercial embedding without negotiation, it is backward compatible to a fault, and it is boringly reliable.

It is also everywhere already. Once a library is that deeply embedded — inside languages, frameworks, package managers, installers and devices — replacing it means rewriting a dependency that nobody thinks about. curl benefits from the oldest advantage in infrastructure software: being the thing that already works.

Where curl actually ends up

The installed base is the real measure of curl's place in the world, and it is hard to overstate. Microsoft ships curl.exe as part of Windows, so a copy sits on hundreds of millions of PCs. Programming languages and frameworks bundle libcurl to give their users HTTP, package managers rely on it, and phone apps, televisions, set-top boxes, cars and industrial equipment carry it because embedding a proven library is cheaper than writing your own.

By the project's 25th anniversary in March 2023, Stenberg estimated that curl had been installed more than ten billion times over its life. Almost none of those users know the name. That anonymity is the point: curl succeeded so completely that making a network request is no longer a specialised task, it is a default capability.

The maintenance problem

The flip side of ubiquity is burden. curl is used by banks, airlines, phone makers and governments, but it is maintained by a small team led by Stenberg, funded mainly by sponsors and support arrangements rather than by the companies that rely on it. He has spoken and written publicly about the pressure that comes with maintaining critical open-source code — including the security reports, the sheer volume of email and the difficulty of finding paid time for unpaid work.

The project's response has been transparency rather than secrecy: security issues are tracked, published and fixed in the open, and releases arrive on a steady cycle of roughly every eight weeks. At the 25th anniversary in March 2023, Stenberg noted that curl had by then been installed more than ten billion times — a number that says as much about the fragility of the arrangement as about its success.

Where cURL is now

cURL is not a nostalgia entry. It is in active development, still led by its creator, and still adding support for whatever the internet invents next — HTTP/3, new authentication schemes, new protocols. Microsoft ships it with Windows; every major platform builds it; millions of developers type its name every day.

The unusual thing about curl is that its story has no ending to write. Encarta needed a shutdown, Sega needed a pivot, but curl just kept being needed. It is the internet's plumbing: unglamorous, largely uncredited, and working — which is exactly why the question "what happened to curl?" has such a short answer.

Entries in this story

• cURL

Sources

1. cURL - Wikipedia — Wikipedia (2026-09-15)
2. FOSSified: curl 25 years — Daniel Stenberg (2026-09-15)
3. curl project history — curl on GitHub (2026-09-15)
4. curl mailing list archive — curl.se (2026-09-15)