Why Not Start With the Most Off-the-Stelf Reactor?

2040 for the first commercial scale one seems pretty far out, and it seems like a pretty high price to pay. Why not build something that could be built in just three or four years and just get something up and running? I've got a fondness for working backwards. Our ultimate aim here, if we can do this, is that we can get reactors out as propulsion units on large ships.