The problem is less about cost, and more about opportunity cost. If we spend time to develop a fully-fledge VR driving mode, how many sales will that get us? If we spend the same amount of time on some other feature X, how many sales? Being a small company, we can't simply say "why not both?"
There's also a notable difference between building a simple product from the ground up for VR, and retrofitting an established game. If we were to step into the VR market with Trainz, it's likely that there would be some substantial prior expectations about what we should deliver, which could substantially increase our costs, or equally, decrease our ability to innovate and make a good "VR game".
Sidestepping this by making a spin-off game might be one approach, but heads more squarely into the "opportunity cost" territory. Another option would simply be to dabble - ie. release a partially formed experience to hard-core VR users to experiment with, rather than actually trying to develop a marketable product from the get-go. This would assumedly only work if there were such users within our existing community.
chris