I think people are overly optimistic about the ability to endlessly increase the performance of wireless networks. The existing networks make fairly efficient use of the spectrum they're allocated. That means you really only have two options to make it go faster.

1) More spectrum. But it's massively in demand (read: expensive), nobody wants to give any up, and even if you had "all" the spectrum there are still practical physical limits about how much data you can transmit without using a wavelength that won't penetrate walls.

2) More towers that each use lower power. This is the one that can get you almost arbitrarily large amounts of wireless bandwidth, but it's also the one whose cost converges on the cost of building a new fiber optic network as the number of towers you need approaches the number of users you have.

Neither one of those is going to make for an inexpensive roll out of a wireless network capable of handling Netflix's video traffic to millions of customers simultaneously in the same city.