I don't believe this is true. NiCl42- has a tetrahedral arrangement (Ni(CO)4 does as well, I believe).
On a side note, why is it that d8 transition metal ions tend to form 4-coordinate metal complexes that give total electron counts of 16, which seem to disobey the "18-electron rule"?
Crystal field theory explains this pattern. In a square planar environment around a d
8 metal's 5 d-orbitals, the d
x^2-y^2 is not split by the coordinating ligands, therefore raising its energy. On the other hand, the ligands will split the d
xy, d
xz, d
yz and d
z^2 orbitals, which will place them at a lower energy. Filling these four orbitals with the 8 electrons increases their gap from d
x^2-y^2, resulting in a stable complex. Google should bring up some more in depth intros to crystal field theory involving various geometries.