We’re definitely close to the point where it would make sense out of experimental, but if you look at the next steps in #103248: Geometry Nodes Volume Features Implementation, you’ll see they add some tricky things like modifying voxel values with fields, referencing grids with anonymous attributes, etc. I think that sort of thing is going to be essential in the future, but it’s hard to predict exactly how it will work for people.
So the reason for the caution and for moving slowly is that we don’t want to end up going down a dead end. We did that with the pre-fields geometry nodes and the prospect of doing that again is a bit freaky!
Also, I think it would be great to make volume nodes more public at a point where people can be pleasantly surprised by their performance. That’s still a bit of an open question technically.