The usual calculus here is that they don't capture ALL the upside of that training, but they capture plenty of it.