Nature has re-used existing folds all over the place (partially because of genetic mutation but also because it's improbable to come up with stable folds from scratch by evolution I would guess), this was encapsulated by earlier award-winning systems like Rosetta. There is probably a finite number of folds in nature, with most of the difference being in the outward-facing amino acids "citation needed" :)

So an extremely large DL network would have a good chance to find and integrate ("compress") all the existing folds and sub-folds that human researchers or Rosetta missed or just hadn't the time to investigate and characterize yet (I'm not an expert on Rosetta by far btw so please expand if you are :).

I would venture to say it's a good problem fit for DL methods (as was impressively demonstrated).

Regarding your question, "does it understand something about the structure of the protein folding problem" - expanding on the above, I would say it understands enough, but it probably doesn't understand the generics of chemistry as protein and their folding is a biased subset. The output is (as far as I remember) an atom distance matrix and not atom trajectories etc. so folding dynamics is not part of the model (this is btw an important part of protein science as well).