how confident are we that spending $1B on real estate in a market where you already have a strong footprint is optimal?

I guess at what point does it make more sense to locate to a less pricy metro and pay engineering talent to come (basically pick Amazon's reject list other than the VA and NY winners)?