We have researchers that literally cannot get work done efficiently because they don't know that core labs exist on campus to serve them, and labs that spend tens of thousands to buy instruments that they seldom use for the same reasons. At a department level (let alone an institute level), we have no idea what instruments or services people need, and no usage statistics for instruments that we already have. This means that it is likely that shared facilities are sub-optimally serving the community as a whole, and labs are buying multiple copies of the same pieces of equipment when one unit could do if it were shared. The lack of central indexing also means that most labs have no idea what other labs are working on, which hinders collaboration.

For training, right now EACH core lab forces researchers to do similar training courses for the same instruments; there is no way to prove you know how to use an instrument without taking each course. Similarly, every core lab employs different billing software which financial admins / lab admins have to sign up for and deal with.

How is this at all productive or efficient for anyone? If you were to suggest a similar setup for interacting business units, you would literally be laughed out of the room at a company.

The reason people like this current system is precisely because it's inefficient and confusing. This makes it difficult to regulate at a high level and makes it ripe for abuse by powerful people.