Apple has 2.5 billion active devices, TIME reports, and an AI model that sits far from the technology's frontier. In an interview the magazine published on 15 October 2026, Greg Joswiak, Apple's senior vice president of worldwide marketing, said the company is not building for the people chasing superintelligence.
'We love them, and they can use it too, but that's not our target,' he said, adding that the aim is to make AI better for everyday people. The remark is a product decision and a wager on how much of AI's value comes from the underlying model and how much comes from the device that makes it useful.
Apple Builds for Personal Context, Not Raw Model Power
The new Siri, which launched in the US in September, can draw on messages, emails, calendars and documents stored on a user's devices, with permission. It is designed to help prepare for a meeting, find a flight confirmation buried in an inbox or interpret health data from an Apple Watch.
It cannot code like OpenAI's Codex or write as elegantly as Anthropic's Claude. Apple's answer is that the value comes from what surrounds the model: the chip, the operating system and what the device knows about its owner.
Privacy sits inside that argument. Siri processes requests on-device where possible and sends heavier ones to Private Cloud Compute, which Apple says it designed so that queries are not stored or made available to the company. Independent researchers can inspect parts of the system, and some who praise it note limits to its transparency.
AI Enthusiasts Are Already Buying Apple Hardware for Local Models
Earlier this year, a wave of AI enthusiasts bought Mac Minis, at $600 each, to run personal AI agents through the open-source project OpenClaw. Apple had nothing to do with the project.
Users found that the energy-efficient Apple silicon, compact form, and unified memory let them run AI workloads locally and continuously for minimal electricity costs. Demand left some Mac Mini models unavailable for months.
TIME reads the episode as support for a particular future: if smaller, capable models soon run cheaply on devices, the scarce asset becomes the device, the operating system and the personal context through which a chatbot turns useful. That is the magazine's inference, not Apple's claim.
The surge shows appetite for local-AI hardware without showing that Apple's consumer strategy works. Chief executive John Ternus holds a version of the view: 'People want physical things,' he says, and he sees no scenario in which devices disappear.
Analyst Neil Cybart adds that rivals' success leaves the ecosystem story intact, since people still reach Spotify and Netflix through Apple devices.
Apple Owns the Context While Leaning on Google's Technology
The same Siri runs on an Apple AI model that leverages Google's Gemini technology, which leaves Apple relying on a rival for a key building block. Steve Jobs went to great lengths to own the whole product, from hardware through software to services.
Ternus rejects the idea that this is a retreat. Apple already relies on suppliers for screens, memory and modems, he says, and it makes 'the things where we can really bring differentiation'.
Apple did not unveil Apple Intelligence until June 2024, after ChatGPT's November 2022 release, and then spent more than two years struggling to deliver the personalised Siri it had promised.
Gadget Review frames the exposure in terms of what users notice: if frontier models grow far more capable and the gap becomes visible, a consumer-first position could turn from a strength into a liability. That is the outlet's inference, not an Apple admission.
The bet therefore faces a test: whether users who see what frontier models can do still find enough value in Siri's integration, privacy and personal context.
