I seem to be missing the implication as well. I am under the impression that Haskell relies on boxed values primarily due to its parametric polymorphism. There is a Simon Peyton Jones talk somewhere discussing the acrobatic GHC requires for the type theory to deal with unboxed types and their implicit requirement for an alternative Kind at the type level.

The laziness as a default just makes boxing closure values a sensible default, when the type system requires the boxing of everything in general.