The problem is not trust, but direction
This applies to pretty much all companies including Microsoft, but they're very popular, so it's worth bearing in mind. Apple are probably worse, but have less market share and business usage. Phone related operating systems are substantially more execrable.
If you're using relatively standard Win32 and other core technologies, including core parts of .NET, I would not be worried about developing and using a Microsoft solution. They have a solid OS, with underpinnings that generally improve in each release, and an excellent commitment to backwards compatibility.
As soon as anything whatsoever outside the above (i.e. something that is not too big to be changed) is used, the company's vision becomes important. If your way of working or product design doesn't ally with that, then there is a problem.
If the platform supporting your product isn't open source or you have insufficient internal expertise to maintain an open source platform, and your product or way of working is indelibly marked on that platform, you have a splendid 'opportunity' to frantically change your environment.
Need to use Remote Storage Manager? That lasted all the way from Windows 2000 to Vista, and then got dropped. Bits of Exchange have changed radically between 5.5 and 2000, and again between 2003 and 2007. The Microsoft vision for a client OS is for a frequently updated client, with a constant moderate speed Internet connection.
It's even worse if you're using minority technologies, such as with Windows Phone 7, or new technologies that have not proven themselves in the market place. Expect to have the rug pulled from under you.
None of this should be a surprise. The mobile direction of Windows has been happening for years. Telemetry has increased with each Windows release and is generally a good idea. Windows 8 has had a considerable number of patches that changed it, and the Windows Store apps regularly. Automatic updates has defaulted to 'download and apply' for years, so it's clear that Microsoft sees the trade off of patched systems vs a (relative) minority of broken systems as acceptable.
However, it's not going to change unless people pay for it, and by pay I mean 'deliberately go through the cost and manpower to re-implement on a platform that allies with your aims for the foreseeable future'. It's all about the apps, it always was, and always will be. The Internet connected world has considerably increased the amount of activities that are possible solely in a browser, but native apps are still necessary.
It may also - specifically talking about moving off Windows - involve more pain, and paying more for fewer, higher quality features. It also needs a compelling feature for people to move, and I should point out that anything mass market has a similar Internet connected, data/telemetry reporting, automatically updated design to Windows (I do not include any non end user mass market Linux distributions, or stuff like BSD, even if I personally like it)
(I'm tempted to put a VMWare rant in there as well, but the post is already long enough)