If your hardware does the job it's generally meant to do and doesn't catch fire, and then you run software that makes it pull 20% more power which DOES make it catch fire... well, apparently the hardware wasn't built for the job you were trying to make it do. Shitty hardware? maybe, but if you hadn't ran the software, your computer wouldn't be on fire. Read a thread somewhere yesterday about someone's RMA on their dead video card being denied because it was primarily used for inference which caused the memory to heat up more than the gpu, so the gpu's fans didn't stay active enough to keep the memory cool.
Is that in any way comparable to what a windows update does? I would fucking hope not, but clearly software can break hardware. You can't say no to updates, you can't see what they're doing, and they sure as fuck eat my computer when they're running. I know, I know, "show me on the doll where the updates touched you"... I've had more bricked machines from windows updates than from all other causes combined, so right here, they touched me right here in this stack of brokenass computers. /BelieveAllAngryWindowsUsers
As I said: Microsoft is going to Microsoft, and your Windows install isn't necessarily safe from Windows Update but it isn't doing anything to your RAM or any other hardware that any other software wouldn't do. Also, Microsoft itself recognizes this; that's what System Restore points are for, and generally, before it does any Windows Updates, Windows will typically create a System Restore Point so you can roll back before the Update. Microsoft has gotten better about this over the years but A) there's a long time before something becomes mandatory B) Let other people go first and be the guinea pigs. Microsoft typically fixes things or pulls things back when they see them breaking things, particularly in the OS space. Not because they love their customers; it saves them money in Customer Support calls.
A great place to look for issues is this website:
Microsoft 365 – Service Status
It's the Status page for the Microsoft Tenant for Earlham University, and if something is going wrong with any Microsoft Service to which they subscribe (and it looks like they subscribe to them all) , they will post the latest status update here [Edit: by here, I mean to one's MS Service Tenant admin center; Earlham just has created a website that surfaces this for everyone, probably just intended for folks at the university but I found it and am now sharing with you.]
I found this when one of our suppliers (and much of the rest of the world) was having an issue with Exchange Online, while we weren't and if didn't have access to a MS Tenant Admin Center with an issue, you couldn't get timely updates. [Edit: They also post to Twitter/X for big issues, and for really global issues they have another dashboard] I'm trying to get my company to build a page like this, although I guarantee it won't be exposed to the world like this. Thank you very much, Earlham.edu. It's a great website.
One can always go Mac (my M6 Mac mini is awesome but has soldered-in RAM now, GLHF) or some flavor of Linux (I still need to spend time with Omarchy; the little bit I have dabbled with it seems cool). The key point is that using hardware will cause it fail. It always has and always will. Hardware fails, and it sucks. It sucks supremely now because we are in the middlle of another of these periodic hardware crises (was just GPUs with Crypto, now its everything with memory because of AI) and lingering M2 money supply inflation. I still remember 2017 when I bought my 1080Ti for $700 on launch day and no one batted an eye. What Sweet Summer children we were.
One might make the case for some hardware, like video cards/GPUs/some NVMEs, one might argue that some "software", really more the activity, pushes the hardware power/heat envelope and you see more failures in that circumstance: Gaming in the extreme, mining, LLM/AI, Compiling/processing video, etc. I personally wouldn't think of it as "software" but I would leave it the individual as to how they define that. Did the software kill that GPU or the fact it was mining for x years? I will leave this here.