With Windows, not only do things break, the solutions are nonsense, or in the best cases, idiotic.
For me, that is the main complaint. With a good architecture and design, things may still break, but they break in places where you can reasonably expect them to and hence you can prepare. With Windows it is all over the place because they lack that good architecture and design and never understood thinks like KISS, resilience, the Principle of Least Surprise and basically all the things that are required to make engineering good.
Essentially, Windows is a souped-up toy at this time and used for things way outside of what it can do reliably. And the requirements are growing and the gap between what Windows would need to be and what it is is getting larger and larger.
I do agree that many Windows users and basically everybody that decided for Windows (or Microsoft) for a professional application scenario are not professionals by common standards. Of course they do not want to hear that, because nobody wants to hear they are not qualified for their job, especially when that is the only skill they bring to the table. My guess is that the increasing regulation and reliability (at this time spear-headed by the EU) will essentially kill off Windows and other Microsoft products, because they really cannot compete once real quality requirements come into the picture. This may still take 10-20 years to become blatantly obvious, although the LLM mess may accelerate things.