Apple’s new AI strategy might also play a key role in its upgrade-your-iPhone strategy.

The company used its annual developer conference today as a platform to announce Apple Intelligence, a decidedly non-generative nomenclature for a suite of new AI features that, like other generative AI tools, are trained on massive datasets.

Apple’s approach is an additive one, relying on the influence and footprint of its existing apps rather than spinning up a new chatbot or web browser that spits out humanlike responses. Once Apple Intelligence rolls out to iPhones, Macs, and iPads later this year, it supposedly will turn sketches into images, sort through photos and videos, rewrite emails, change the tone of messages, and allow its voice assistant Siri to tap into different apps to string together smarter responses.

There’s just one catch: It won’t work on your old iPhone.

Apple is limiting these new AI features to the iPhone 15 Pro and Pro Max, which run on Apple’s A17 Pro chip; the iPads Pro and Air, which run on Apple’s M1 chip or later versions; and various Mac computers that run on the M1 chip or later.

The company hasn’t said exactly why it is limiting its Apple Intelligence features to the newest, and most expensive, hardware, though industry experts surmise that less-powerful chips would possibly create a less-performant AI and that Apple would draw the line at a lackluster tool. (At the time of writing Apple hadn’t responded to inquiries from WIRED.) But whether this limitation is a technical requirement or a product-differentiation strategy, it might be Apple’s best shot at convincing customers to upgrade to newer iPhones this fall.

An iPhone powered by large language models and focused on new AI-personalization features “should change the growth trajectory of Cupertino–spur an AI-driven iPhone upgrade cycle starting with iPhone 16,” tweeted Dan Ives, managing director of Wedbush Securities and a longtime Apple analyst.

The new AI features are being rolled out at the same time Apple’s iPhone sales have slumped. In May, when the company reported its earnings for the January through March period, Apple saw a 10 percent year-over-year decline in iPhone sales and the biggest drop in sales since the summer of 2020, when some factories had to close. Its revenue last quarter decreased 4 percent from a year ago on account of the iPhone sales dip. Apple’s gross margin last quarter was still healthy, but that was due largely to its growing services business.

Apple’s phone hardware has enabled plenty of new features and functions over the past several years, but some of these tools have become commoditized or are layered in abstraction. Its custom-designed iPhone and iPad chips are bleeding edge and help sell top-of-the-line phones to the nerdiest or wealthiest consumers, but “CoreML on Apple silicon” isn’t as big of a selling point for consumers seeking midrange phones. And while new iPhones have iteratively better camera systems each year, high-end phones from rivals like Samsung and Google also have advanced photo and video systems. The same is true of convenience features, like wireless charging.

This trend bears out in secondary market data: Shipments of used smartphones increased nearly 10 percent, to 309.4 million shipments, in 2023, up from 282.6 million units the year prior, according to research firm IDC. For a lot of people, a good phone really is just good enough.

Apple is also selling privacy as part of its generative AI package, saying that Apple Intelligence “is integrated into the core of your iPhone, iPad, and Mac through on-device processing.” Apple’s AI tools use Apple-developed large language models, instead of relying on another entity’s models or a patchwork of LLMs, as confirmed by Axios. In instances where an iPhone isn’t capable of processing a user’s actions or queries on their device, Apple Intelligence will send the user’s data to a server running on Apple silicon, which will keep that user’s personal data secure, the company says.

Which raises the question: If Apple is already planning to offload some of the processing to its cloud, then couldn’t a slightly older iPhone—like the iPhone 14 Pro, which is powered by a slightly older chip—also get the AI glow-up?

Michael Gartenberg, a consumer technology analyst at Flash Advisory & Research who previously worked at Apple, says he can’t technically say at this point whether Apple “is being disingenuous about what devices can run this. But I do know iPhones can already run ChatGPT and an awful lot of Google’s AI features, so I suspect this is the opportunity Apple has been waiting for to tell you that the iPhone 13 really isn’t good enough anymore,” he says.

Another question the introduction of Apple Intelligence raises, pertaining to iPhone sales, is whether it gives consumers a reason not to buy an iPhone before this upcoming fall, Gartenberg says, which stalls the current iPhone buying cycle. (And that’s assuming buyers want the generative AI features at all; Pew survey results suggest Americans are slightly more concerned than excited about generative AI.)

And, since Apple Intelligence will be available only in US English to start, it’s unlikely to immediately boost iPhone sales elsewhere—like in China, one of Apple’s most important markets—unless Apple does some critical “futureproofing,” says Carolina Milanesi, founder of the research firm Heart of Tech.

“It depends on how they roll out experiences to other countries,” she says, like if AI-generated Genmoji are offered as a feature before text editing or other language-based features. “The bigger update cycle will happen next year, when more languages are added,” Milanesi predicts. And in China specifically, Apple has to not only develop language support but determine how it will handle data storage, she says.

Either way Apple now has a new way to compel iPhone buyers to upgrade come September. This time it’s not just selling them on a new camera jammed into the same container; it will undoubtedly try its hardest to convince customers that any newer iPhone is a much smarter smartphone, one that offers a flavor of generative AI much more palatable than the AI chat platforms still in search of application.