Linear scaling is more the exception than the rule, no? And it not just for "tech" either: it's unlikely you could learn the names of 15k employees in 2 months, for instance.