The fact of the matter is that almost any hardware you buy is overkill for any given application in a development function. Even chroot is a primitive form of virtualization. You talk of incompetent sysadmins reverting their mistakes, but what of the mistakes of others? While the distinction and processes between "production" and "development" are made on a company wide basis, the development teams don't have the time or money to spend on production environments, but if their development environments go down then they cry just like any production level environment would. You have thousands of developers waiting for an environment to come back up? What is the cost to the company? The company will see that as a lot of money going down the drain, even if it's labeled "development'. Virtualization gives me the ability to maintain development levels of flexibility and production levels of stability. So yes that strange fantasy that "this shit is going to fly" is not only my job, it's what the Fortune 15 company that employs me expects of our environment.