Here's a historical look at the Linux market share on Steam, gathered using Valve's publicly provided data. Be sure to look over all the information, don't just take the first chart. If you have feedback, let us know in this feedback Forum.

Note: Data from before September 2018 has been removed, due to a flaw Valve reported in the survey. Windows 7 machines were being over-counted since around August 2017, so there's no point keeping false data. See here for info. Then, Valve also reported another flaw in October, where smaller distributions were not being counted and the data for September was updated.

When buying games from Steam, how is the desktop platform counted? From what Valve told us (February 2020), they first take the desktop Steam client platform used to buy it, then whatever has the most playtime in two weeks. However, Android and Web Browser purchases default to Windows and then what has the most playtime in two weeks. If you want to ensure developers see a Linux sale: play it only on Linux in the first two weeks, don't buy it and leave it.

You can interact with these graphs. Click a label to disable/enable it to fine-tune your experience when looking at it

This chart shows what happens if we only take English only on Steam as a language for Linux, compared with Linux across all languages. The point here is to show the market share of just Linux/English as a comparison to the overall share.

For an estimation of the total number of Linux users on Steam, Valve reported they had 120.4 million "monthly active users" in January 2021 (source).

Using the latest months recorded share (Apr-2021 - 0.85%): 1,023,400 estimated "monthly active users" for Linux+Steam.

To be clear, that is not the total, that is monthly active.