Obviously there's gonna be special software in certain fields that's going to be around for quite a while, like the pharma example from this other reply, but even there the point was rather that upgrading from the XP version is too expensive....
And ultimately the question is at which point maintaining or even improving Windows becomes economically unfeasible. With Windows 10 being "the final Windows" there was already a big shift in paradigm. Microsoft is very busy expanding in the cloud/SaaS business, so it's not like they don't see how Windows might lose relevance over time. This in turn sends signals to software vendors who still make software that requires desktop machines.
Ok, in this scenario, why would I choose Linux? Linux is literally irrelevant if "all you need is Chrome" (among the more dystopian futures I can imagine for computing, btw). Chrome is the platform then and everything is just a Chromebook. In this scenario no one reaches for Linux, they just buy some horrible proprietary locked down no-user-serviceable-parts thing like they do with phones and personal computing is declared dead.
I don't buy that that is true anyway. People have been predicting the death of desktop PCs for a long time too and it still hasn't happened. Cloud is just the latest in the thin-client/fat-client cycle of hyped up computer industry bullshit as far as I'm concerned.
> Seems to me that DirectX is the last remaining barrier to finally getting the majority of the Windows user base to migrate to Linux.
To which my point was simply "you are very wrong". As far as I can tell, you agree that there are a lot more barriers to moving to Linux from Windows than just DirectX (again, if that were true why haven't office workers moved? Oh yeah, lots of other reasons!).
So what exactly is the point you're trying to make?