This has some interesting performance implications. Specifically there is no need for a traditional "kernel bypass" when you want to work with hardware from userspace.
For instance in Linux we have stacks like the Infiniband networking stack and some userspace TCP stacks (usually based on PF_RING) that are quite a hack to build in Linux that would be "just the way it is" in TempleOS. (though it's worth mentioning that TempleOS has not networking capability, apparently Gods decision)
These sorts of operating systems were common before multi-user computing but died out because of the need to protect the operating system from it's users and the applications they may want to run.
However the world is coming full circle again. Unikernels are back, sometimes they are written in C, sometimes in OCaml (MirageOS), sometimes in Erlang (ZergVM or something I think).
TempleOS may actually provide insights for how to do these sorts of things as it's been built this way for a long time which can't be said of many other modern operating systems.
Alas, people would probably need to see past the somewhat spiky exterior (Terry himself) to appreciate his work.
I wish the source code (http://www.templeos.org/Wb/Accts/TS/Wb2/LineRep.html) were easier to read though. It's very dense with a lot of abbreviated variable names and no comments. It's the main reason why I haven't installed TempleOS yet, because of course I'd want to hack on it, and the code seems tough to learn.