SpaceX is based on a fairly different principle, of taking proven designs from the past, updating them, and building them for less. For example SpaceX's Grasshopper VTVL is following in the footsteps of the 1990s McDonnell Douglas DC-X, not some kind of completely untested idea. He isn't taking a previously speculative technology idea, like a space elevator or something, and trying to build it.

An analog here would be if Musk had a proposal to take an existing, working mode of transit (at least working in prototype) and build it for less. But the current proposal is more like hard sci-fi: goes into enough depth to be back-of-the-envelope plausible, but no detailed engineering or working demonstration, and very hand-wavy cost estimates.

Currently SpaceX doesn't offer significantly lower LEO delivery costs -- they are on par with Ukrainian/Russian space programs [1].

[1] http://en.wikipedia.org/wiki/Comparison_of_orbital_launch_sy...

The Falcon 9 is the cheapest Medium sized launch vehicle it's more expencive to LEO per LB than a heavy launch system or retrofitting old ICBM's. However, we are considering US construction costs and SpaceX spent far less on R&D and spends less on construction than any US equivalents.