What killed the Amiga apart from the Commodore bankruptcy was hardware costs and lack of viable 3D support on the low end far more than Windows 95.
I still have mine, suped up with a 68010, that'll get you 5% in a tight loop.
What seemed to sink the Amiga in the long run was that to go from OCS to ECS or AGA you basically had to buy a new computer.
But a PC could go from EGA to VGA by swapping out a card.
Never mind that the first 3D accelerators were pass through concoctions, where the VGA card had a short cable run over to the 3D card that in turn was connected to the screen.
That, and of course that Commodore management consistently messed up and delayed the better options.
What is a real shame is that Hombre didn't get finished before he bankruptcy [1]: A 64 bit 3D graphics chipset with an integrated PA-RISC and DSP, coupled with an AGA-Amiga-on-a-chip for backwards compatibility, and with Windows NT compatibility (back in the days when NT ran on multiple hardware architectures; though at that time I probably would have thought Commodore had sold it's soul to the devil if I'd seen the Win NT compatibility).
Now completely unusable, of course, like its Hercules cousin, because it was an ISA bus device.
People made code that depended on those hacks and magic. If the coprocessor doesn't work the exact same way in the future then everything breaks. That's fine if it's doing something simple like displaying a big array where each byte was a pixel or RGB value because you can easily make that backward compatible. Instead it was doing things like running code between scan-lines and using 5 bits per pixel.
A new display chip would basically have to include the old chip in its entirety just to be backward compatible. All the Amiga hardware worked like this.
Linux was slightly convenient on PC in 1995 if you really cared about: (1) a basically Unix-y userland, with all the good and bad that implies. Shells and stuff worked better than cygwin (2) tcp stack (3) gcc
That's really it. X was a fucking nightmare to get to work "correctly" and then, as now, "correctly" on X means "it handles a remote networking case no one cares about better than the alternatives but still poorly, and everything else about it is totally broken."
Hardware support for every feature that was becoming interesting in PCs in 1995 (basically anything where economies of scale were making interesting stuff cheap) was absolutely terrible on Linux. Drivers for PC hardware, even the kind of PC hardware nerds liked (SCSI CD-ROM drives) were super flaky or non-existent. Application software was horrendous. There was no support at all for anything that still had a licensing regime around codecs, which, at the time included all the media software.
Linux was "quite good" in that you could set it up in a corner of your house and run httpd. That was dope in 1995. But compared to Amiga as a user-facing hacking desktop, it sucked shit. Hell, it sucked shit compared to a PC running Windows or a Mac.
Seriously, find a 486SX/25 w/ 4MB ram, a 160MB disk, a decent modem, and an S3 video card and put like, Yggdrasil on it. It will be way less fun to fuck around w/ than an old Amiga/Mac/NeXT/BeBox/SGI....at least in 1995 an Intel machine running DOS could run DOOM. It's probably the least-fun hardware/os combo from 1995 you could dredge up to play with.
And of course it would be worse than BeBox, SGI, NeXT. SGI and NeXT were high end work stations (or above). We had SGI Indy's at uni that they'd gotten at near cost from the distributor at about 3x the cost of the my souped up Amiga 3000... BeBox was a PPC SMP box - it was where many of us had hoped the Amiga was headed.
I held on to the Amiga 3000 for 2-3 more years, but mostly out of loyalty and nostalgia at that point. I still loved (and love) the OS, but already in '95 it was on borrowed time and it cost a fortune to tack on enough third party add ons to make it viable. And the lack of software was becoming a problem - a colleague and I even spoke o Escom at the time about writing a browser for them, because the available Amiga web browsers were already then lagging behind he times.
And in '95 people did care about the "remote networking case no one cares about". At university, we also had old Sun servers that served up apps for dozens of dumb X11 terminals each. At work, by '96 5 of us ran all our applications off a single 120MHz Pentium w/128MB RAM, using our 486's as terminals.
The old 1995 Linux box is less interesting to play with than the stuff in your list because all the stuff in your list became extinct and aren't just retarded versions of their modern incarnations.
My IDE CD-ROM wasn't reckognized, I had to copy all floppies into the MS-DOS partition and install from there.
X Windows could not do more than 800x600, and that was after lots of fighting with the X Config file.
So quite good?! No.
What it was, was a cheap way to have UNIX at home, because we were having DG/UX and Aix at the university.
One of the reasons I kind of lost my way into the game dev culture, was because I had this silly idea of trying to invest into the Linux demoscene that went nowhere.
With my interest for demoscene, gaming and graphics, which I already had before jumping into GNU/Linux, I could never get fully away from what those communities use.
The demoscene and gaming cultures are more focused into results, how to get there doesn't matter and there is a certain pride in being the only one capable of doing X.
That year is when I realized it was time to sell my Amiga and buy a PC because the future had just jumped to a different hardware architecture.
To me, booting Windows 95 was as jaw dropping as seeing the ray traced juggler demo and Marble Madness on an Amiga 1000 Kickstart 1.0 in 1986.