Perhaps an analogy to Moore's law?

Bitter lesson #1: don't waste time optimizing code when a faster processor is around the corner.

What countered it: Moore's law stopped working.

Bitter lesson #2 similarly relies on scaling laws that might have diminishing returns wrt model runtime vs intelligence. Runtime matters for turnaround on the problem you're solving.