My take on this is that server-based audio is still in its infancy. Server software should be regarded like an OS that requires a large user base in order to assure continued development and support. Hardware should be modular because it is changing rapidly. You never want to have to re-rip or convert your library so it should be stored on external dedicated devices, e.g., NAS. Outputs should be digital and card-based to allow for changing communications technology. DACs should also be external because they, too, are changing rapidly. Perhaps, some time in the future, enough dust will have settled for a single box solution.
I was timid about the complexity of doing this until I borrowed and played with a PC-based server and it gave me the courage to go in that direction. It also inspired at least 2 of my friends to do the same. We now have libraries of recordings in many formats with resolutions up to 4xDSD in mono/stereo/multichannel with storage, renderer and DAC as independent modules affording the ability to track the progressing technology. Investing in an expensive, completely integrated and closed system is only for those with an irrational fear of anything hand's-on.