It's not that high. A lot of the losses in take-off aren't the consequence of drag. I think I saw on wikipedia that the aerodynamics losses are only 1-200 m/s of the delta-v budget.
Going by raw math, if the Earth were a perfect magic sphere with no atmosphere, you could attain "orbit" at ground-level by accelerating to 7.90 km/s. Then you could Hohmann to LEO with one 0.06 km/s burn to transfer up and another 0.06 km/s burn to achieve a circular orbit, so that's only 8.02 km/s in an ideal world. [Oh, I forgot that you start with about 0.40 km/s from rotation if you are near the equator.]
Oh, finally found a nice reference, on Wiki: http://en.wikipedia.org/wiki/Low_Earth_orbit#Human_use The delta-v needed to achieve low earth orbit starts around 9.4km/s. Atmospheric and gravity drag associated with launch typically adds 1.5–2.0 km/s to the delta-v launch vehicle required to reach normal LEO orbital velocity of around 7.8 km/s (28,080 km/h). It doesn't break those out, unfortunately.