Underestimating Trump for the third election cycle in a row would be an existential crisis for the industry, particularly as pollsters (as Nate Cohn has documented in various pieces, for instance) have gone to great lengths to try to fix whatever errors they think may or may not have been present. And particularly as response rates and getting representative samples are getting worse and worse.
This isn't even getting into the the fact that there is less high quality state polling this year than in previous cycles (once again, well documented)... which stands as evidence on it's own as well.
I don't know what will happen, we will have to see how it works out in a couple of weeks, maybe they are bang on.... but I'm not going to ignore the human psychology at play here.