Apple has sent out the invites for its September event, the one with the glowing ring artwork, and the timeline has decided this is the only thing worth talking about for the next week. I've watched enough of these to know roughly how it goes, and yet here I am, reading the rumour roundups like everyone else.
The interesting part this year isn't the phone. It'll be a bit faster, the camera will be a bit better, there'll be a new button or a moved one, and a slab of the keynote will be devoted to a colour. The interesting part is whether the on-device AI features they trailed back in June actually ship in a state you'd want to use, or whether they ship as a "coming later this year" asterisk that quietly becomes next year.
That asterisk matters more than usual because the whole pitch leans on it. If the headline reason to buy the new hardware is a set of AI features, and those features arrive in stages over the following months, then the launch is really a pre-order for a promise. We've all bought one of those before. Sometimes the promise lands. Often it lands smaller and later than the keynote implied, with the rough edges left in.
I'm not cynical about it, or I'm trying not to be. Putting real inference on the device, where your data doesn't have to leave the phone to be useful, is the right direction and genuinely hard to do well. If they've pulled it off at a quality that survives contact with actual users, that's worth the noise. The bit I'll be watching for, once the confetti settles, isn't the demo on stage. It's what people say a fortnight later when they've lived with it.
For now I'll do what I always do. Skip the livestream, read the bullet points the next morning, and wait for the teardown and the first wave of real-world reports before forming an opinion worth holding. The keynote is marketing. The interesting writeups come a week after, when someone's actually used the thing on a train with bad signal and a half-charged battery, which is where all of our software eventually has to work.