The whole reason it's called a "cellular" network is because they are "cells" - one for each tower (give or take multiple carriers sharing a tower). These cells overlap, and operate in different parts of the frequency spectrum. The important point about understanding this "cell" nature is that there is no reason a cell in rural Kansas has to be the same size as a cell in downtown NYC. In fact, they generally aren't the same. That's related to why they previously prohibited (and how they now can technologically allow) cell phones on planes - there is a "femtocell" placed on the plane so that cell phones use their minimum power. The use of higher power levels can cause an overlap of a flying cell phone with multiple distant cells (using the same frequency) on the ground... and cause a whole chain of confusion (for a system effectively designed to be 2-dimensional).
You make it out like additional bandwidth (in the form of more parallel downloads) is not an economic problem. Additional bandwidth has a cost: More tower density (or at least more radio antenna if you can mount on existing buildings) using smaller cells, and that's absolutely an economic problem. (It's also a permitting problem, and in suburban areas likely a ditch-digging problem, both of which are likely worse than buying the equipment)
A few other notes:
You need to calculate 3X/Y, not X/Y, as 4G Cell towers will likely use Tri-sectored directional antenna. It's widely deployed in 3G environments, and is basically a requirement in any dense area (and also facilitates cellular 911 location when more accurate location isn't available)
There are also additional technologies that could be deployed that change the rules of spectrum usage. MIMO comes to mind. (The major problem of MIMO is much like other things in the cellular world, in that signal reflections (and interference in general) are somewhat of a royal pain and the resulting demand of processing power makes the basestations and phones infeasible to deploy at this time.)
Net is that smaller cell sizes and using additional technologies could absolutely reduce congestion, but the expense in doing so is enormous. It doesn't really change why the cap needs to exist at some level (they are up against wall, even if a financial loss wall rather than a physics wall), but we're nowhere near the fundamental laws of physics.