Or they're going to pay to get a service like Google GCE or Amazon AWS to scale to their needs simply by using a friendly interface or via REST APIs. At work we use Google and if one of the cloud servers gets funky (even the database,) we can destroy it, re-create it, and point san storage towards everything that matters.
In defense, there are some things that are on premises that matter a little more like a phone (PBX) server, gateway, firewall, VPN, etc. but, that really depends on the business, how small or big it is and it's demands on networking infrastructure. So for the PBX, if you don't have a RAID that has redundancy then if a drive fails, you lose everyone's voicemail. Now wouldn't that suck? Either way, there is a time and a place for RAID.