I wouldn't read too much into the exact numbers when looking at changes from one month to the next. For example, the survey shows MacOS with an increase in market share of 0.25 percentage points from the previous month. Coming from 2.20% up to 2.45% that would be huge, implying an 11% increase in Mac users on Steam. However, the nearly 13 year-old Windows 7 (32/64 bit) seemingly saw an even larger increase, going from 2.53% in May up to 3.17% in June, implying a 25% increase in Windows 7 users, or a 0.64 percentage point increase overall. Now, obviously, that scenario seems unlikely, so it's clear that Steam's sampling of systems is likely affecting these results.
And actually, if we check the previous month's hardware survey (on archive.org), we see that Windows 10 actually saw an increase of 0.32 percentage points that month, while MacOS share plummeted by 0.35 percentage points, implying a 14% reduction in Mac users. And Windows 7 users dropped by 0.57 percentage points that month, implying an 18% reduction in users. And while Windows overall appeared to lose 0.31 percentage points to Mac and Linux this month, the previous month it gained 0.37 percentage points, resulting in it being up overall when both months are considered. So in reality, Mac and Linux seem to not be making any notable gains against Windows on Steam, or at least the short-term survey data isn't implying that.
Really though, as these numbers jumping around should indicate, you shouldn't read much into monthly changes, as there is just too much noise for the results to be meaningful. I would rather see Valve report something like a 12-month average change to better indicate trends in a way that is less likely to be skewed as much by sample changes from one month to the next. But even there, things like increased expansion into certain markets can affect the demographic being sampled. So if Windows 7 is more popular in a country like China, for example, and the Chinese user-base grows (or things like Internet cafes mess with results), you could see an increase in that OS's market share, even if install numbers are actually falling overall. Being able to filter results to certain regions, or to systems with certain classes of hardware, could make the survey results a lot more meaningful, but Valve isn't bothering with that, or at least they are not making those tools publicly available.
So articles like these comparing monthly changes are a bit silly. Or at the very least, the author should take the time to analyze trends beyond one month if they actually want to report some meaningful data on trends. There's little doubt that Windows 11 installs will continue increasing as people move to new hardware or install the free update, but the Steam Hardware Survey isn't a particularly accurate way to measure that.