There is no reason to switch to Linux, considering that they would need to reinvent the whole UI stack of Linux like MacOS did. Keeping a 50 people hired for Kernel development sounds cheaper.
The only reason Linux is so successful is actually the fact that it's open source. For individuals (like me) because that gives you much more flexibility in customizing the system, tinkering with it, and more importantly, debugging problems and understanding inner workings.
But even for big corps (which are mostly running the show when it comes to Kernel dev these days) it's much easier to whack the Linux kernel into shape for whatever you want to achieve, than trying to work with Microsoft to get what you want from NT.
I agree a significant part (necessary part?) of Linux success among hackers and developers is it being FOSS... I don't know much about NT kernel but it would be quite interesting to see the alternate reality in which it was also FOSS and independent of the rest of windows, I wonder if it would have a similar success in the arenas Linux does.
There are some other really interesting FOSS kernels out there today like Minix3 (quite different from the Minix that inspired Linux), they are sure to have deficiencies when directly compared to something like Linux due to the difference in the shear size of community driving them, but they have noteworthy designs, superior in various aspects, and they make me wonder, is there only enough space in the world for so many popular FOSS kernels? i.e how much of Linux is now inertia much like windows desktop? (and I say this as a full time Linux user)
You would need extensive partnership agreements like what Google has with Microsoft regards Chrome or Samsung has with Microsoft regards .NET Core. Challenging for smaller vendors.
Linux has a more pure (not owned) FOSS style making it successful.
MacOS X is not based on Linux or GNU and it's UI evolved out of a combination of Next technology and features combined with re-implemented classic Mac OS features. So it wasn't exactly a clean slate.
More relevantly though, they put some effort into maintaining compatibility with programs written for their previous OS by both porting their APIs and providing CPU emulation! And that is on top of bringing the rest of nextstep up to standard for what consumers were expecting out of a desktop OS at the time.
> Keeping a 50 people hired for Kernel development sounds cheaper.
I doubt it, Microsoft can and has rewritten their UI multiple times... if they wanted to, they could write it for Linux atop their own ported windows API with little effort. I think they just lack the incentive to do so, Linux users wont care because they wont use it, Windows users wont care because they don't care about Linux... who will benefit? not Microsoft that's for sure.
From a business perspective MS behavior makes sense, they are embracing Linux because that's the way one half of the market is headed whether they like it or not... meanwhile their desktop business still has a hold on the market mostly out of inertia and legacy, so why would they do something as radical as change the core OS to mess that up?
However as the parent said "one day", maybe, in the future, when windows makes zero sense to run on metal, then perhaps microsoft will create an official and FOSS windows API... if only to relieve themselves from security support hell.
The rest of their business is where the money is. Office runs one way or the other out of the web, on Android, iOS or macOS.
I'm no sure the powers that be at Microsoft have any great love for Win32.
Back when Windows 8 was released they introduced UPM (Universal Windows Platform) and that looked a lot like an attempt to actually kill off Win32.
They only backed away from this strategy after the backlash they got from Win32 Windows developers.
Move on to today and I think they are now using platforms like .NET Core 3.0 to gently move developers away from Win32.
Nothing is stopping them from making Linux server distros either since they do produce SQL Server for Linux as it is. They ported a lot of VS code to Mac (not it isnt just MonoDevelop) and were surprised by how much just worked due to the OO design.
Microsoft is embedding the Linux kernel already. Whats next? A linux friendly file system? Things are going to get interesting overtime.
I have no knowledge of what they are up to truly just a bit of speculation. But given what has happened in the last decade that is "not Microsofts MO" none of this would surprise me.