When you get to scale, you can start doing things like manufacturing displays with higher PPI near where the lenses are centered. That way you have the same number of total physical pixels but a higher resolution contributing to the image.

You can further reduce the computational costs by using foveated rendering(http://research.microsoft.com/en-us/um/people/johnsny/papers...). You can use low-latency eye-tracking to spend more time rendering the portions of the frame that the user is actively looking at.

I think the combo of falling gpu prices, and clever hardware design means we will probably see headsets with effectively higher and higher resolutions sooner rather than later.

That's going to be difficult given how often and how fast eyes move and how sensitive they are.

You'd need a new lens design for that really. Right now the lenses that are used aren't really good if you start looking to the side too much. When I get going in VR I generally just look with my head instead of my eyes to avoid those lens issues like chromatic aboration.