Those 12 terminals still need to talk to satellites. Only a certain number will be in reach, and if the ship is significantly out to sea, those satellites will need to pass data along to others to reach the ground.
We know a densely populated land area can saturate the satellites overhead; it's part of the reason we don't use Starlink in, say, NYC. The same math applies to a thousand cruise ship passengers trying to use it at the same time.
(It will absolutely be much better than the previous state-of-the-art, though.)
Absolutely the density of 12 terminals over what amounts to skyscraper on edge, is high. But cruise ships are not packed into city blocks. The density of terminals over ship-in-ocean sized areas is quite low. The density is very much more sparse than a countryside or farm area. So it feels unlikely to me that you are going to saturate the visible satellites. And those satellites will be very underutilized at their locations not quite over land yet.
Only recently (2025) has Starlink been documented to pass data between satellites and only for new ones. Cruise ships have had starlink for several years so the most common case will likely still be a single hop. Most cruises hug closely enough to the shore (within 100 miles) that a second hop isn't required.
It's not really a thousand passengers, it's just 12 terminals. Sure, those terminals are well oversubscribed vs a normal user terminal, but given the documented terminal-to-ground-station ratio (quite high), I'm pretty doubtful a cruise ship or 2 significantly impacts (or is impacted by) ground bandwidth. The density of users is just at the wrong scale for it to be significant.