Sure, FOSS-based, off-the-shelf products exist, but companies usually have this nagging problem of generating revenues somehow, and usually if they can't make it by selling support, they don't make the software at all. And if this software is designed to generate revenues based on how much trouble you have using the software and need to call support, one should think really hard about whether or not the software is right for one's organization.
I think we've been over this enough times. This alleged benefit is not a benefit at all, especially if we're talking about government agencies that don't have MIT graduates on staff to examine the code let alone modify it.
So there is this big private-public partnerships around FOSS. OpenStack is a great example here.
I've mentioned this before. It's not necessarily a comparison between what is cheaper. The case for public development of FOSS is no different then the case for public scientific research in general. It's like saying the government could save money by shutting down NIH, NSF, NASA, etc. Well that would miss the point. It's not all about saving money, it's about the advancement of humanity.
Like public science, FOSS provides a baseline commons anyone can derive from. We should not be making a comparison between the costs of FOSS and proprietary software. The government should be developing and cultivating free software for its own sake.