Unmanned USA military aircraft travels from NYC to LA in 12 minutes.

That page concludes with:

"Had the latest test flight gone to plan, the Falcon HTV-2 would have separated from its rocket high above the atmosphere and entered a steep dive before levelling out and performing a series of subtle manoeuvres to test its aerodynamic performance. At the end of the flight the plane would have rolled upside down and steered a graceful arc into the ocean."

The official press release at http://go.usa.gov/KLe says that they confirmed separation and that they achieved high-speed aerodynamic flight, not that they "would have separated". They lost contact 9 minutes into the aerodynamic tests, and the plane (if you can call it that, the CNN article calls it a "triangular wedge of zoom" which seems more appropriate) went down sometime after that.

>(The speed is about the same.) I'd imagine it's cost.

Both system consist of booster (which need to provide 20 Mach delta-v to its payload) and payload - upper stage which is warhead carrying vehicle. ICBM is cheaper because it follows dumb ballistic trajectory which puts no specific structural requirements on the warheads and their enclosure. The warhead vehicle traveling at 20 Mach in the upper atmosphere would cost enormously more ( and possibly would be much heavier, thus requiring bigger booster). I.e. ICBM is cheaper.

The dumb ballistic trajectory was a weak spot of the ICBM system, and thus modern ICBM warheads perform evasive maneuvers during final atmosphere re-entry.

It's not just expense; due to nuclear early-warning systems mostly being keyed off ICBM launches, it's risky to launch one. You're gambling that other nuclear-armed powers will believe you that it's conventionally-armed and not heading at them, which is a decision they only have several minutes to make. So in practice ICBMs have usually not been considered a normal part of the conventional-force arsenal.