Trying to spot the peak of a bubble is an exercise in futility. Things can always stay irrational for longer than you can expect
And in the case of the AI Bubble, it’s much too early to gauge anything until we see the response to GPT-5, whenever that comes out