I wonder how often a C program does something wrong because of a compiler bug versus a program bug. I’d think the former is extraordinarily rare for e.g. gcc. Although, if you spend an hour studying every line of code (as is customary when writing safety-critical code), it’s possible you may arrive at a point where the primary source of bugs is the compiler.

However, for non safety-critical software I’d say buying a proven correct C compiler would be a waste of money.