Apple CEO Tim Cook delivers a keynote address (Photo by Justin Sullivan/Getty Images)
Tim Cook and his team introduced its AI suite to the world at Apple’s recent Worldwide Developer Conference, yet there was a sting in the tail. Apple Intelligence would not be available on the vast majority of existing iPhones.
The reasoning behind the decision would be confirmed in the traditional “Talk Show Live From WWDC,” hosted by John Gruber. He raised the question of Apple restricting its Apple Intelligence AI software to the iPhone 15 Pro and 15 Pro Max with his guests from Apple’s management team. John Giannandrea answered:
“So these models, when you run them at run times, it’s called inference, and the inference of large language models is incredibly computationally expensive. And so it’s a combination of bandwidth in the device, it’s the size of the Apple Neural Engine, it’s the oomph in the device to actually do these models fast enough to be useful. You could, in theory, run these models on a very old device, but it would be so slow that it would not be useful.”
A look back through the history of the iPhone shows that Apple has always kept the RAM levels as low as possible. This was one of the advantages that iOS has over Android. Apple’s tight integration over software and hardware meant iOS could be heavily optimized for a single footprint. Android needed to run across many different chipsets and I/O devices, which meant it required more memory.
That has served Android devices nicely at the dawn of the AI smartphone. That extra memory can now be used to help process generative AI locally. Apple cannot back-port its AI suite to iPhones older than last year’s flagship model. Samsung has introduced Galaxy AI not only to the S24 family but also to the older Z Fold 5, Z Flip 5 and Galaxy S23 family.
Yet Samsung cannot go back any further because memory is not the only limiting factor. Generative AI is an intensive process that requires significant processing power to achieve results in an acceptable timeframe. Going back one generation of Qualcomm’s Snapdragon 8 series is an impressive feat, but Qualcomm (like all of the main Android chipset manufacturers) has added the necessary hardware to the current chipset to assist the rollout of AI.
Apple will be able to roll out its AI to 2020’s M1-powered MacBook Air and 2021’s iPad Pro but not 2020’s iPhone 12, even though the M1 and the iPhone’s A14 Bionic are built around the same architecture. Both chipsets have Apple’s NPU and offer 11 TOPS over 16 neural cores. Clearly, Apple Silicon had the capability to support 2024’s AI with 2020 technology, and Apple’s decision to offer lower specs to its smartphone rather than its tablet or laptop has resulted in a loss of AI expansion on mobile.
Apple is now playing catch-up with generative AI. Google started the smartphone AI race in October 2023, Microsoft brought AI to Windows 11 this week through its Copilot+PC program and there are various online services (such as Grammarly) that have harnessed AI to improve their offerings.
Would Tim Cook and his team like to go back to 2020 to change the specs so every iPhone from that point on could support Apple Intelligence? Given just how much of a lead in AI Apple could claim the day after iOS 18 was released to the public, I think it would like to revisit that decision.
Now read the latest iPhone headlines in Forbes’ weekly Apple Loop news digest…