Currently their moat is history. Why I keep coming back to ChatGPT is it ‘remembers’ our previous chats, so I don’t have to explain things over and over again.
And this history builds up over time.
Not sure how this argues their moat. The context window in pretty small (at best 192k on 5 with the right subscription). Once you run past it, history is lost or becomes gimmicky. Gemini 2.5 Pro by contrast offers 1M. Llama 4 offers 10M (though seems to perform substantially worse).