My take on this is that, despite an ever-increasingly connected world, you still need an assistant like this to remain available at all times your device is. If I can’t rely on it when my signal is weak, or the network/service is down/saturated, its way of working itself into people’s core routines is minimal. So either the model runs locally, in which case I’d argue OpenAI have no moat, or they uncover some secret sauce they’re able to keep contained to their research labs and data centres that’s simply that much better than the rest, in perpetuity, and is so good people are willing to undergo the massive switching costs and tolerate the situations in which the service they’ve come to be so dependent on isn’t available to them. Let’s also not discount the fact that Apple are one of the largest manufacturers globally of smartphones, and that getting up to speed in the myriad industries required to compete with them, even when contracting out much of that work, is hard.