I think we will. I consider Unreal interesting for people who want a 1 min+ animation but don't have 10-15 000$ to spend. Since doing an animation does not take much more time than doing a still (except the compo/storyboard/post prod) it could be more affordable. You completely remove the rendering costs. It's win-win for everybody, except rebusfarm hehe!
I wouldn't do a Unreal4 project unless it was at least with 15 000 budget to start with though. Even then I would hesitant. Not worth the effort it takes to reach commercial quality comparable with raytracing yet.
You might have noticed, but there isn't a single quality commercial archviz project yet, the absolutely stunning works are all personal. That is because if client suddenly asks you to use 20 custom furniture pieces you won't be able to just buy quickly from DesignConnected. They need to work with Unreal first, you need to be sure they can be nicely unwrapped. And then do the actual unwrapping.
It won't work for any existing vegetation on market either. You need to create all of it from scratch, with heavy limitations. Bespoke vegetations is super hard and expensive to make now, but for Unreal, it's one step further.
So for one or two more years, it will be minor. But than there is this possibility it will be a big nieche :- )
The latest Rafael's work above is STUNNING !! Love the quality, the smoothness...