The load capacitance in CMOS logic circuits mainly consists of the capacitance between the gate and the substrate (often called Bulk) \$C_{GB}\$.
In newer technologies the gate area (length x width) generally gets smaller making the gate capacitance \$C_{GB}\$ smaller as well. That's because this capacitance is simply a capacitance per area. So more area means more capacitance.
Is total input capacitance for the FET not increased with the smaller PN junction gap?
The only PN junctions that are involved are the drain to substrate and source to substrate capacitances \$C_{DB}\$, \$C_{SB}\$. Remember that in CMOS logic circuits, MOSFETs are used, not JFETs. These substrate capacitances are directly related to the width of the transistors. In newer technologies, the width generally gets smaller and that makes these capacitances smaller as well (assuming the doping levels in the PN junctions don't change drastically, which usually they don't).
I don't know what you mean by "PN junction gap", my guess is that you mean the depletion region? The size of the depletion region does mainly depends on doping levels and applied voltages. If we kept those the same between technologies, the size of the depletion region would not change and the \$C_{DB}\$, \$C_{SB}\$ capacitances would stay the same if we kept them the same size. However, in newer techologies, they can be made smaller in size as the transistors can be made smaller.