It's not about salaries because those aren't the major issue. The COL difference is minor in comparison to the execution risks. In fact, the traditional reason why startups "had to be" in NYC, Boston, and San Francisco is job-market latency. Yes, you can find great engineers elsewhere, but for some ridiculous purple squirrel query (production experience in the 3 NoSQLs you use) that needs to be filled tomorrow, you need to be in the Valley or New York (due to Wall Street, there's a lot of technical talent, albeit expensive). If you're in Austin, you can find great people but probably not the guy who matches your purple squirrel query exactly.
That said, a place like NYC has the urban dating problem (too many alternatives, so people treat each other like shit). We'd write great offers to people only to have them picked off based on minutiae. We were ridiculously picky and rejected people for dumb reasons; only fair for them to be.
However, I feel like the grow-fast-or-die red-ocean nonsense is played-out and the next wave is going to be in the mid-growth (5-30%/year headcount growth, focus on technical excellence) space which doesn't need to be in the Valley or New York.