I don't think this is generally true, there are lots of examples where you measure something with higher resolution than the tool's manufacturing precision (for instance, an STM) and similarly lots of examples where you use a lower precision tool to build a higher precision tool (for instance, a mill or lathe but optical focusing would also allow for a shrink if it weren't for the diffraction limit).
But moving a generation forward in precision is good, you only ever get close to perfect with diminishing returns.
But moving a generation forward in precision is good, you only ever get close to perfect with diminishing returns.