Non-exact computation is useful for some stuff: machine vision, deep-learning, and probably other types of machines learning. Those have a wide variety of uses

Non-exact computation is useful for some stuff: machine vision, deep-learning, and probably other types of machines learning. Those have a wide variety of uses and they demand lots of compute. Also in places where performance is critical and money is spent on optimization, isn't lots of this bureaucracy removed ?