I think the suborbitor in 20 million range is the most realistic if you want to build rockets.
Personally I like idea of building launchpads.
Cheap launchpads. Launchpads that add speed to the space launcher.
A Huge extremely Dumb launchpad that floats in the water.
Material used: mostly big 1/2" thick "plate" steel pipe, about 20 yards of concrete, and miles of welding. Major cost: about 1000 tons steel pipe and labor cost of welding pipe into a big long cylinder. Other costs: rent of shipyard, then towing to a location, deploying it, and testing it.
Total cost: about 5 million.
Or it can be scaled down.
Maybe it could be made for a sub-orbital vehicle- say less than 100 tons of steel pipe, rent a barge, etc. Maybe cost somewhere around 1/2 million. Wouldn't add any significant speed, but you could launch from shallower water- say in Gulf of Mexico in the military launch area. If you were going to a number of test flights it could be worth the cost of making it- then again maybe not. One advantage of the sub-orbital is possiblitity launching from airport-like conditions.

Yup. He talking about that pipelauncher again. Gawd!
:)