I think the private cloud compute stuff isn’t quite “real” yet. My strong suspicion is that they did not have the capital expenditures in place to make that the fallback— because it obviously should be. I also suspect feedback just isn’t good enough yet to rush that spend.

@jsonbecker That sounds right. In theory it could work, but they will need a huge number of these servers, and because it's custom hardware they can't just lean on other cloud providers.

maybe the PCC isn’t ready?

@numericcitizen Maybe. Because it is supposed to just transparently work, it's hard to know if it's being used at all right now.

@jsonbecker or maybe they are waiting for Apple to release the M5 chip? Oh, wait... they are Apple!

@numericcitizen Apple is made of people who don’t always coordinate with each other. Other than that, designing separate chiplets takes time, as well as designing them into a single package with interconnects. Actual fabrication of each single M5 chip takes months from silicon ingot to final product. If nothing goes wrong.

@numericcitizen @jsonbecker Apple is building their own chips. As @renevanbelzen said, it is not that easy. Apple's current ML models were trained on third-party platforms, not PCC yet.

I think Apple is spread too thin - [1] manufacturing chips for PCC [2] designing and manufacturing new chip models and [3] reducing the cost of chips to enable Apple Intelligence on all devices

I was able to trade in some old tech and upgrade to the latest iPad Mini from the previous generation, I did this specifically for the Apple Intelligence and the extra amount of base storage. I have yet to fill the storage and have yet to use Apple Intelligence on it in any meaningful way. I hope Apple figures out something soon as Apple Intelligence is not turning out to be a huge differentiator thus far.

@SteveSawczyn The latest iPad Mini is a great device so hopefully a good upgrade regardless. But yeah, Apple Intelligence isn't yet a differentiator. I like the summaries but can also live without them.