This is a huge over simplification, but it is worth noting that Universities, unlike companies, are closed systems that generally do not make a profit. Giving additional benefits from someone means taking them from someone else. For a company it is usually the share holder that loses out. For academic departments added benefits to graduate students hurts professors.
There is no question in my mind that while graduate student unions may provide a short term benefit to graduate students, those students who wish to continue on in academia are effectively shooting themselves in the foot. Those increased benefits come directly from departmental budgets. This means heavier teaching loads and less discretionary money for academic staff. To offset this you need either high grant overheads which will make getting a grant even more competitive or higher tuition. Basically if you support graduate student unions, you lose your right as a faculty member to complain about high teaching loads, not having a grant, and high tuition.
As a graduate student in response to "threats" about unionization my university increased the PhD student stipend. The stipend was above the NIH recommendation. This means that any student funded by an NIH grant needed to find an additional source of non governmental funding. This essentially boils down to PIs using their overhead accounts to supplement the stipend, but PIs did not get a bigger slice of the overhead pie to cover this new cost. At the same time the University got a site license for Matlab. To cover the cost the PI slice of the overhead pie was reduced. Universities take, but rarely give.