Hello
Allow me to dream in technicolor.
I could see a future where, regardless of the underlying hardware (Arm64, RiscV or x86, 16-bit support or not) we boot a small linux kernel with a stripped down userland, just enough to run qemu/kvm which would run OS/2 or ArcaOS. In fact, this would also allow you easily to do block-device backups and ship them remotely (s3 or else) without much additional work; so I would even consider this as a better option.
I also support the idea. If it is emulated it's emulated, but I want to dream further.
I dream of a future where OS/2 is natively interpreted under a different 64bit kernel. For example grabbing the Linux kernel (just because it is famous, not because I like it

) and have the OS/2 binaries running PM, SOM, WPS and OS/2 applications thinking they are still under the OS/2 kernel.
Ryan C. Gordon tried to create an OS/2 emulator/interpreter
called 2ine. But it is too much to try to clone all the APIs. What it will be a great idea is to have the OS/2 binary interpreter (2ine) and see the components that are missing to make all the rest of the OS/2 APIs binaries to work as they are on a different kernel.
For example, Fuchsia OS (Zircon Kernel) has "
Starnix" (or they are trying to make it work), that will run unmodified Linux binaries on Fuchsia OS. That will run Linux apps under a different non-Linux kernel. 100% compatibility may never be possible, but to get to 70%, 80% is a win too.
Maybe we can run OS/2 binaries, the PM, SOM, WPS interpreted on a different kernel and keep growing from there, and later trying to find open source replacement for more closed components with more time. Open source is not the goal, open source is the resource to have a long term future. We are using OS/2 today, thanks to open source and the support of developers.
Regards