Inconsistency in the Energy of polarized linear dielectric

There is apparent inconsistency in Griffiths’ interpretation of the energy of polarized dielectric.

Griifiths says there is some extra energy contained in the "springs" that hold the dipoles.
However, if perfect dipoles are considered, The "plus" of one dipole stands on the "minus" of another dipole (that’s the reason we can compute potentials with considering only bound and free charges)
And there are really no dipoles within the polarized dielectric (because every plus stands upon another minus) , so how can we claim that there is energy contained in those "dipoles".
What is wrong with this reasoning?