Why pay money to make a better product when you can pay zero money for a worse product and no change in subscriptions? What are your customers gonna do, go get the paper somewhere else?
They can of course choose short term profits over long term viability, which wouldn't be all that surprising, but that changes the explanation from "more profits" to "short-sightedness/incompetence"
When we looked at well-run noncommercial journals, like The Physical Review, the cost was justified by exception handling in the peer review process. The average peer review case goes smoothly but if somebody has a complaint and there a lot of appeals and reviews the cost can skyrocket.
Our cost structure was $3-$5 a paper but we struggled to get even that. arXiv was unfunded at Los Alamos and I think Cornell never appreciated the value that it created for the world, had we found a way to capture a few percent of the value we created we would have been "sustainable" but I think that is incompatible with running it on a shoestring the way we did.
I am not enthusiastic about arXiv being independent, I don't have any problem with the high salary they want to pay the director, you are going to pay that for a good non-profit manager in NYC but you could just as easily pay that much for a bad non-profit manager.
I have nightmares though that had arXiv been independent in the 2000s somebody I know might have wound up at Epstein's island not because I think he's evil or perverse but rather because he's naive [1]. arXiv is a gem that would be attractive to somebody like Epstein and would be very possible for somebody like that to have funded it 100% back then. As it is it will be sucked into a somewhat corrupt NGO-industrial complex and end up spending $30-$50 a paper just on fundraising. It's sad.
[1] so many people who got in his circle strike me like children who were playing in the street and got hit by a car, and you'd hope people in those leadership positions should have better judgement