I’m running linux happily on an i7 860 which is a 10 year old pc I got for free, and it has plenty of power to run any application I have.
These days you can get a much more modern second hand PC for less then EUR200 or even a complete laptop including a windoze license if you want it.
A quick check like:
gives plenty of results.
My previous PC was a DualCore E6600 which was 12 year old when the power supply died. It was also adequate for KiCad (Olinuxino A64 was getting a bit slow, but I do not make PCB’s of that complexity) and I was looking into buying more modern hardware when someone gave the i7 to me. The speed difference does make it more comfortable to work with the i7 and it made me realize I kept hanging on to the old DualCore far too long. I just kept using it out of habit, and had no good reason to spend money on an upgrade. Maybe even the i7 should be recycled as it’s to old and more modern hardware would pay for itself through a lower electricity bill.
I have some difficulty in imagining that EUR200 would be big obstacle to prevent an upgrade and keep hanging on to obsolete hardware. The market with 5year old PC hardware is full of cheap options and such second hand hardware still has plenty of life left.
So my opinion stays the same. I have no objection against keeping support for 32 bit KiCad as long as it’s simple to do, but when it starts needing a “significant” amount of effort to keep it supported, or when progress of KiCad gets hindered by 32 bit limit it’s time to let it go.
From what I know of KiCad’s direction it’s goal is to be a professional EDA suite and to be on par with “professional” commercial programs. 32-bit compatibility seems to be a courtesy to … (I do not know how to say it nicely).