I made a huge technical mistake in a paper I published on arXiv... How bad is it in academia?
Mistakes are "bad" in academia if you fail to disclose them once you know about them (which constitutes unethical behavior), or if they affect something that counts towards a formal credential (like a degree; which is unlikely for an arXiv submission).
So, the way forward is:
Then, if you think you can correct the error and rescue the rest of the paper, get to work on that.
Academics make mistakes all the time. Acknowledge your error, disclose it, and move on.