AROS does run on ARM. As for AmigaOS 4.x I don't think Hyperion have the resources, and they're very dependent on their hardware partners bearing part of the cost.
The only problem with PPC from an Amiga point of view is that most of the SOCs are targetting networking and automotive markets and similar embedded use rather than desktops (total PPC sales in terms of number of chips is still in a similar region to x86 sales; revenue per unit is way lower, though). It's not like people will buy these machines looking for top end performance, and the OS is still incredibly light weight, but if you pay for it you can still get PPC chips that perform very well.
But AmigaOne's are an very expensive hobby for most people who buy them, not bought to be cost effective.
More comparable to the FireBee you have a multitude of M68k accelerators and FPGA Amiga reimplementations that are much cheaper.
There's been various discussions about trying to run AmigaOS or AROS on ColdFire like the FireBee too. Problem is people would expect to run their old binary-only software, and AFAIK the ColdFire can't trap all the unimplemented instruction combinations fully, so it'd require "fun" hacks (e.g. dynamic translation; I think it probably could be done reasonably easily for AmigaOS, but finding someone with both the time and the skills is another matter) so the Amiga community focus has been on FPGAs.
Yeah ColdFire is also a dead end because Freescale isn't going to be releasing any more models and certainly not at any higher clock speeds. And it's really being made for specialized markets (automotive, ethernet switches, etc.)
The FireBee is a neat hack I don't mind supporting, but if people really want hardware and emulators don't cut it, then FPGA is probably the way to go.
Or switch architectures and port the OS (EmuTOS and/or AROS) + some sort of CPU translation/emulation layer for 68k stuff. It would probably have to be something big endian, so that rules out RISC-V but OpenRISC, J2 (SuperH) and OpenSPARC are some open source candidates.
The only problem with PPC from an Amiga point of view is that most of the SOCs are targetting networking and automotive markets and similar embedded use rather than desktops (total PPC sales in terms of number of chips is still in a similar region to x86 sales; revenue per unit is way lower, though). It's not like people will buy these machines looking for top end performance, and the OS is still incredibly light weight, but if you pay for it you can still get PPC chips that perform very well.
But AmigaOne's are an very expensive hobby for most people who buy them, not bought to be cost effective.
More comparable to the FireBee you have a multitude of M68k accelerators and FPGA Amiga reimplementations that are much cheaper.
There's been various discussions about trying to run AmigaOS or AROS on ColdFire like the FireBee too. Problem is people would expect to run their old binary-only software, and AFAIK the ColdFire can't trap all the unimplemented instruction combinations fully, so it'd require "fun" hacks (e.g. dynamic translation; I think it probably could be done reasonably easily for AmigaOS, but finding someone with both the time and the skills is another matter) so the Amiga community focus has been on FPGAs.