Ive done this a few times. Depending on the startup (size, domain, technical complexity, resources) it could just a coder with a fancy title. Now nothing wrong with that if you are getting to build freely, solving hard/interesting problems or working on problems that somebody is willing to pay tangibly for or you are trying to just get your startup creds up for your next play.
But I have found often startup CEOs look for CTOs for a few tactical and myopic reasons. They cannot code (or worse cannot design beyond copy-pasta from stackoverflow) or just to bolster their "look we have a faang-er as CTO" creds. Just as often the CEO has very little business or product sense (not to mention any leadership) but just needs a coder to build out their ideas (usually half baked ones in a very reactive fashion). Here you will also find that there is a new direction each week based on what a lead told them in passing.
Worst of all you will have very little resources (are you the sole coder/CTO?) but will be blamed for all the wrong choices (dont worry about writing tests - we dont have time ===> why are things breaking?).
The pendulum in this case is not a bad choice. Typically after you are burnt by one startup it doesnt meant all startups are bad. Just that you could use that big-co IC role as a charging station. Or if you actually like management (and there are lots of reason to) you could always switch to management in a big co and then go back to a startup more worthy of you. Now obviously you are the mercy of the hiring committee/team/market so a better story for why the switch definitely helps.