CPU bottlenecks are a problem in games that aren't targeted to specific performance levels at a specific resolution, or in games that are designed to take advantage of more than 4 cores (console ports). VR games are 90 FPS at ~1080p right now, and they'll be designed to do that with a 290 and a 4690K, and since there are very traditional graphics settings that affect CPU resources, the focus will be on the GPU only. Highest settings has little to do with CPU bottlenecks.
The choice of CPU is really not that important; single-thread performance is what's important, and nearly every Intel CPU performs fine in that regard. Using 6+ threads is a task that I don't think most VR developers are going to take on when they don't need to.
It's all about the software, and VR games are designed with plenty of restrictions and targets in mind.
What about for ports of traditional games to VR? Most people are struggling to run pcars in VR, and while no data has been conclusively collected on the issue, my guess would have been cpu bottlenecks. Fallout and doom are both slated for VR ports in the near future. I would expect cpu will be a big factor for ruining these properly.
don't you think the benchmarks provided by RS here are relevant?
http://forums.anandtech.com/showthread.php?t=2442910
Note especially the severe dropoff of min fps in those benches, which is what results in annoying judder in VR.
Assuming rx480 is in fact 3515 fsu points, I think the fact that most cpus are bottlenecks for gtx 980ti (3800 fsu points) in modern games at 1080p is relevant.
I'm no expert on the degree to which games are adapted to multithreaded performance, but the logic that single threaded performance is all that matters doesn't seem to be reflected in benchmarks. If that were true wouldn't a heavily overclocked i3 be the best thing since sliced bread?