Unless a company is running an engineering team composed of monkeys, it will be hard to achieve 1% optimization per month, even without compounding. There are a lot of optimizations which look good on paper but in practice bring other unforeseen cons
Anyway, a profitable software company will be looking at getting a big multiple of 10x the salary of their engineers
I work in perform engineering, although in a client rather than backend role. We usually do better than 1% a month. In fact we do this even as >1% of regressions make it into the build from the software development process.
It depends what you're optimizing. Mature c++ codebase that is routinely profiled? Large legacy codebase with few tests? Mixed workloads?
All hard. But there's also a lot of easy cases out there - even good engineers might code like monkeys if they're under time pressure and trying to grow the business.
I've done a lot of performance work, and while of course no one individual could achieve a 1% savings per month on the entirety of FAANG compute loads, there is lots of code out there that nobody thought would run at scale or be on the critical path of something important, but now it is and it has gotten expensive.
Anyway, a profitable software company will be looking at getting a big multiple of 10x the salary of their engineers