In John Gruber’s article about Apple Intelligence, he writes about the server impact:

As things stand, with only devices using M-series chips or the A17 or later eligible, Apple is going to be on the hook for an enormous amount of server-side computation with Private Cloud Compute. They’d be on the hook for multiples of that scale if they enabled Apple Intelligence for older iPhones, with those older iPhones doing none of the processing on-device.

Also, people are going to use this more than today’s Siri. How often will it need to go to the cloud? Seems very difficult to predict the scale.

@manton Processing on device is an excellent solution and plays to Apple’s strengths vs. competitors who rely on centralized data centers and must bring all of the compute power there. Distributing energy use to end compute devices reduces the strain on the power grid while making those who use the features pay for the power directly.

@fgtech @manton However, for on-device processing, people may need to upgrade to newer phones, which may spark the usual controversy of "Why is Apple making me upgrade my perfectly good phone?" Or they have to live without Apple Intelligence features that require external servers. Can or should Apple invest in selectively offering AI services depending on how old phones are or should they just offer no AI services to older phones?

@fgtech No argument from me there. The main thing we don't know is how often on-device models can be used, because they are quite limited, but after this ships I'm sure someone will figure it out.

@pratik I'm sure people will complain about upgrades. It's going to be hard for Apple to explain this in a satisfying, non-technical way. And users won't really be wrong asking why they can't have the better Siri.

@manton I think it's very similar to slowing down phones for degraded batteries incident. Apple tries to set a bar for optimal experience and makes changes to variables that affect that experience without making the user tweak around in the settings.