it doesn't work that way at all.
Old designs would be for different process technologies. As the tech changes the DRCs (Design rule checks) change as well.
You can't run a design on a process it wasn't made for, the resulting product simply won't work correctly (if at all).
If the CPU was designed for a gate width of 35nM then it was designed with biasing around that gate width's leakage. If you then try to spin that part on a 14nM fab the biasing of the gates is all wrong and it will (likely catch fire) not work at all because of such high leakage.
Additionally, price doesn't scale the way you imply. A wafer start costs about $1K. Doesn't matter what process you run on it (it does, but not really all that much). The cost per part is based on the number of functional parts per wafer at the end. Thus going from an 8 to a 12 inch wafer lowers cost even though the process change requires a $2.2bn fab to be built, you have gone from 201 sq inches to 452 square inches, over *double* the yield.
Same thing from process shrinks, you cut the area used by your transistor gates and you make the die smaller, then you can fit more on a wafer.
Thing is, Intel may not even have fabs capable of making the older parts any more, even if they wanted to. Process tech has evolved, IDK if they even have an 8" fabs left...
To just "redesign" the part for the new process is not realistic either.
TLDR: To make an old part will cost the same or more than it did when it was the latest and greatest.