Two point: the consensus is that the 'most difficult' (not longest) chain which follows the agreed consensus rules is the 'real chain'. This is not necessarily the one with the most computing power associated with it, especially in the case of a fork, where two different sets of consensus rules diverge. The main issue with being the smaller side of the fork is that if you are still using the same proof of work mechanism then you are vulnerable to 51% attacks from the larger fork, which tends to lower prices and so discourage mining which makes the situation worse (and this has happened to multiple bitcoin forks, including 51% attacks). This may in principle be less of a concern with PoS, because the 'mining resource' also splits with the fork and stakers can continue to stake both sides, reducing the strength of the positive feedback loop (though not completely, since the less popular and lower priced coin is more vulnerable to someone trying to buy enough of it to execute a 51% stake attack).