There was a time in my career when I would have struggled to pay standard open access fees (in $2-3 k range). However that was at the time when open access was uncommon. Universities still spend a lot of money for subscriptions, one way or another. Departments at many universities get internal incentive funding for their quantity of publications. And, producing a scientific publication costs in excess of $20k in total (and usually way more than that), in salaries and overhead, even if you are a theoretician working at a desk. There is money in the system.
Being unable to pay for open access only penalises self-funded researchers. There are relatively few of them, and nobody can work without salary for long. If I were in that situation today and could not convince the head of my affiliated department to shell out $2k for an extra journal paper in our annual statistics, I would either let my articles stay in arXiv forever (a few important papers in my field are) or go to a journal with low fees. There is at least one such journal in my field, with rigorous peer review but no other expenses. It chardes $100 per article. It has no copyediting and keeps its accepted articles on arXiv. My point is, while the transition to the world with no subscriptions may be a bit turbulent, more economic publishing options will appear and be recognised as legit when there is demand for them.
Making sci-hub legal is going to kill the subscription model, fast. Already I know one research institute that no longer buys library subscriptions, and you know why, because its researchers nowadays can read all the journal articles for free.