Tried in 8, met with utter failure, reverted by 10.
> embracing open source and move nt kernel to linux kernel, reducing the costs of development
...and throwing out pretty much everything that makes Windows better for a Desktop? Including their driver model, their backwards compatibility, etc? Unlikely. Replicating all of that to an appreciable standard would be significantly costly.
> slowly move its on premise infrastructure to the cloud (office, exchange)
This does seem to be happening, much to the chagrin of many, many smaller business that don't like to pay rent on their tools.
> move windows to the cloud and leave only smaller part local, just to pick up network (and potentially some other devices like graphic card)
Possible. Microsoft already has some kind of Windows Desktop as a Service, and RDP is damned good.
> stop developing/selling locally installed windows and move to subscription based in cloud infrastructure
They can try, but at best that just means that a whole lot of people will just stick with the last version of non-subscription-only Windows (as many did with 7 instead of moving to 8-10). I imagine it will be a large enough group that third parties pick up the slack and provide solutions to things MS is no longer fixing. At worst, there will be lynchings.
They're already doing that on mobile, they stopped fighting android and are now trying to be there in the form of apps and services for both android and ios.
Personally I think it makes sense: An OS is becoming more and more a layer to access the web. Most people use just a web browser and a couple other apps. It makes sense for their focus to be not in that layer, but in being the app/web that the user searches for.
That's an odd way of putting "have been winning so hard for so long that they probably can't even lose now if they tried".
...unless of course you mean the server OS wars, in which case that statement is a lot less wrong.
- windows moving to metro, to get its users used to much simpler GUI elements
- embracing open source and move nt kernel to linux kernel, reducing the costs of development - slowly move its on premise infrastructure to the cloud (office, exchange)
- move windows to the cloud and leave only smaller part local, just to pick up network (and potentially some other devices like graphic card)
- stop developing/selling locally installed windows and move to subscription based in cloud infrastructure
We will see if Satya is stellar or just devil. I have moved after 20 years of system developmeny on Windows to Linux. We will see what will happen.