Out of 40 servers since 2013, we've had zero SSDs fail (2+ per server) and two motherboards fail. That, I would not have predicted.