This has to do with economics. Typically, each new process generation shrinks by a factor 0.7 from the previous. Eg 0.5 um, 0.35 um, 0.25 um, 0.18, 0.13, 0.09, 0.065, 0.045 ... The reason this is done, is to justify the cost of all the new equipment, and the new tools, there needs to be enough improvement in terms of area density.
You see then, that on top, there are 'minor' shrinks that 'optimize' a given node, keeping the existing equipment, but just pushing the limit a bit. These gives 10-15 % shrink. (eg 90 nm to 80 nm, 65 nm to 55 nm, 45 nm to 40 nm.) Note that things are sometimes misleading. 55 nm is eg a 90 % shrink of 65 nm (keeping the equipement), but here the math does not add up ... .
For some organizations, the cost of following all the nodes is sometimes still too expensive. I have seen organizations going straight from 180 nm to 90 nm, and then likely to 40 nm.