It will disproportionately help the well-off, period; not just "some of them". The dominant distributional effect will be to shift dollars that could have been spent on services, tax relief, or debt relief for less-advantaged households to far-better-advantaged households. It's pretty simple to see why; it's the same reason why the mortgage interest deduction is regressive.

I don't see how advocates for this policy can escape that logic without declaring math irrelevant.

If we want to improve access to and financing of higher education in the US, we should radically expand Pell Grants, not write a 12-figure check to wealthy families.