Valve published the August 2026 Steam hardware and software survey on September 1, and Linux came in at 3.90 percent, back below the four percent line it crossed in July. The month over month move is 0.11 percent, which is noise. The distribution split underneath is not. SteamOS accounts for roughly 21 percent of Linux users on Steam and CachyOS for 15.4 percent, up more than a point in a month, which means an Arch derivative that ships no hardware of its own is now the second most common way people game on Linux. That is the number worth looking at.
The short answer
Valve's August 2026 Steam hardware and software survey, published September 1, puts Linux at 3.90 percent, below the four percent line it had crossed in July, a move of about 0.11 percent. Windows holds 93.95 percent and macOS 2.14 percent. Among Linux users, SteamOS Holo leads at 0.82 percent of all Steam users, roughly 21 percent of the Linux total, with CachyOS second at 0.60 percent or about 15.4 percent of Linux users, up 1.16 points month over month. Arch Linux, Linux Mint 22.3 and Bazzite follow. AMD processors account for the upper 60 percent range among Linux gamers against about 45 percent on Windows.
Every month the Steam survey produces the same two reactions to the same Linux number, and neither of them is especially useful. A tenth of a percent is not a trend. What is worth reading is the layer below, where August 2026 shows something genuinely new: a distribution that ships on no hardware at all has become the second most common way people run Steam on Linux.
The headline, in full, so we can move past it
Linux is at 3.90 percent of Steam users for August 2026. Windows is at 93.95 percent, macOS at 2.14 percent. Inside the Windows figure, Windows 11 64 bit holds 70.97 percent and Windows 10 64 bit 22.90 percent, which is its own quiet story about a migration that has taken years.
Linux crossed back above four percent in July and has now dropped 0.11 percent below it. In March the figure reached 5.33 percent, the high for the year, and it has bounced in a three to four percent band since.
A share is a ratio, and Steam's denominator grows. Phoronix put the caveat plainly in its coverage: in absolute numbers the Linux gaming population may be flat or higher even as the percentage falls. Reading a 0.11 percent monthly move as a signal about Linux adoption is reading noise.
The distribution split is where the information is
Here is the August breakdown as a share of all Steam users, alongside what each represents within the Linux total.
SteamOS Holo leads at 0.82 percent of all Steam users, about 21 percent of the Linux total. That share is not really a preference measurement. SteamOS ships preinstalled on the Steam Deck and the Steam Machine, so its number tracks hardware sales, and a Deck owner who never thinks about their operating system counts exactly the same as an enthusiast who chose one.
CachyOS is at 0.60 percent, roughly 15.4 percent of Linux users, up 1.16 points month over month. Nobody ships CachyOS on hardware. Every one of those installs is a person who downloaded an ISO, made a decision, and did the work. Arch Linux itself follows at 0.34 percent, Linux Mint 22.3 at 0.33 percent and Bazzite at 0.30 percent.
Put those together and the picture is that Arch and its derivatives dominate Linux gaming. SteamOS is Arch based. CachyOS is an Arch derivative. Arch proper is fourth. The distributions built around stability and long support windows, which is what most people would recommend to a newcomer, sit lower than the rolling release family. On a gaming machine the tradeoff runs the other way from a server: new Mesa and a new kernel matter more than a package set that will not move for two years, which is the same reasoning behind the tooling choices we keep seeing in this space.
One methodological note that consistently distorts these lists. The survey reports individual point releases as separate rows, so a distribution with many concurrent versions in the wild appears more fragmented than a rolling release that only ever has one. Linux Mint 22.3 at 0.33 percent is one row among several Mint rows. Read the list with that in mind before concluding anything about a family's total.
The CPU split says something about who these users are
Among Linux gamers on Steam, AMD processors account for the upper 60 percent range. On Windows, AMD is around 45 percent.
Part of that gap is the Steam Deck and Steam Machine, both of which are AMD based, and that part is just hardware sales again. But the gap is large enough that it does not reduce to Valve hardware alone. It also reflects the state of the drivers: AMD graphics support is in the mainline kernel and Mesa, maintained upstream, and it works on a fresh install without a proprietary module. Anyone assembling a Linux gaming machine deliberately runs into that fact early, which is why the coverage of things like the AMDGPU display work and the amd-pstate scheduler behaviour has a real audience rather than a niche one.
What this survey is good for, and what it is not
It samples Steam clients that opt in. That means it describes people who game on a computer, which is a specific slice of computing and not a proxy for anything else. It says nothing about servers, nothing about workstations, nothing about the machines in an office.
Where it earns its keep is tracking the same population month after month. The absolute percentages carry sampling error and a moving denominator, but the mix within Linux is measured consistently, and a distribution gaining more than a point of Linux share in a month with no hardware channel behind it is a real observation. That is the line to follow into September, not the third decimal place of the headline.
Sources and further reading
- Steam Hardware and Software Survey, Valve, August 2026
- Steam On Linux Dips Below 4% For August, Phoronix, September 1, 2026
- Linux back over 4% in the July 2026 Steam Survey, GamingOnLinux, August 2026
Frequently asked questions
What are the headline numbers for August 2026?
Linux sits at 3.90 percent of Steam users, Windows at 93.95 percent and macOS at 2.14 percent. Within Windows, Windows 11 64 bit holds 70.97 percent and Windows 10 64 bit 22.90 percent. The Linux figure is down about 0.11 percent from July, when it had crossed back above four percent. The high water mark for Linux this year was 5.33 percent in March, and the figure has moved in a roughly three to four percent band since.
Which Linux distributions do Steam users run?
SteamOS Holo leads at 0.82 percent of all Steam users, which is around 21 percent of the Linux total. CachyOS follows at 0.60 percent, roughly 15.4 percent of Linux users and up 1.16 points month over month. Then Arch Linux at 0.34 percent, Linux Mint 22.3 at 0.33 percent and Bazzite at 0.30 percent, with assorted Ubuntu and Fedora versions making up much of the remainder. Note that the survey reports individual point releases separately, so distributions with many concurrent versions look more fragmented than they are.
Why did the share drop if the story is positive?
Because a percentage is a ratio and Steam's denominator moves. The survey samples an enormous and growing user base, so a share can fall while the absolute number of Linux users stays flat or rises. Phoronix made this point directly in its coverage: in absolute numbers the Linux gaming customer base may well be similar or higher than the previous month despite the percentage decline. A 0.11 percent monthly move is inside the range you should expect from sampling and from seasonal shifts in who is playing.
What makes the CachyOS number interesting?
SteamOS has a hardware channel. Every Steam Deck and Steam Machine arrives with it installed, so its share reflects units sold as much as it reflects preference. CachyOS has no hardware channel at all. It is an Arch derivative that people install deliberately, one machine at a time, and it is now the second most common Linux distribution on Steam at 15.4 percent of Linux users. That growth is the clearest available signal of what desktop Linux gamers actually choose when nothing is chosen for them.
Is the Steam survey a reliable measure of Linux adoption?
It is reliable for what it measures and frequently over read beyond that. It samples Steam clients that opt into the survey, so it describes people who game on a computer, which is a specific and unrepresentative slice of all computer users. It says nothing about servers, workstations, embedded systems or the machines in an office. What it is genuinely good for is tracking the same population over time and seeing the distribution mix change, which is why the CachyOS trend line carries more information than the headline percentage.