Apple's New Chief Defends Restrained AI Strategy Amid Industry Rush

A Philosophy at Odds with the Moment

John Ternus assumed the role of Apple chief executive in September 2026, inheriting a company that had spent the better part of two years fielding criticism over its artificial intelligence roadmap. In a recent conversation, the longtime hardware executive made clear he has no intention of changing course. The company will integrate machine learning and large models where they solve real problems, he said, but it will not chase capabilities for their own sake.

That stance places Apple in an unusual position. Across the consumer technology industry, companies have raced to embed generative AI into every surface: search bars, productivity suites, customer-support chat windows, even hardware devices with no clear use case. Apple, by contrast, spent much of 2025 and early 2026 refining a voice assistant that many observers had written off as obsolete. The more capable iteration of Siri finally reached users in the northern summer of 2026, years after competitors had begun layering conversational models into their own products.

Ternus acknowledged the external commentary but showed little concern. "Everyone has an opinion," he remarked, adding that he remains satisfied with the company's trajectory. The framing is consistent with Apple's historical pattern: arrive later than rivals, ship fewer features, and rely on ecosystem lock-in and design polish to retain customers.

Utility Over Velocity

The central argument Ternus put forward is that artificial intelligence should be invisible to the user. If a feature works well, he suggested, the underlying technology becomes irrelevant. Whether a recommendation engine relies on a transformer model or a hand-tuned heuristic matters less than whether the recommendation feels right. "We don't want to ship AI for AI's sake," he said. "When we're doing it right, people won't have to think about whether it was AI that made this happen or more traditional software."

That philosophy has merit in principle. The industry has produced a glut of features marketed as AI-powered without clear benefit, from washing machines with "smart" cycles to text editors that autocomplete sentences users never intended to write. But the approach also carries risk. Rivals are not standing still. Google has woven its Gemini models into Android and search; Microsoft has embedded Copilot across Windows and Office; even Samsung has begun shipping devices with on-device inference for translation and photo editing. If Apple waits too long to deliver capabilities users have come to expect elsewhere, brand loyalty may prove insufficient insulation.

At Opentechwire, we have tracked the diffusion of on-device AI across the Android ecosystem over the past eighteen months. The gap in capability between flagship devices running Qualcomm's latest Snapdragon silicon and Apple's A-series chips has narrowed, particularly in real-time translation and computational photography. Apple retains an advantage in silicon design and vertical integration, but that lead is no longer insurmountable.

The Agent Question

The more revealing exchange came when Ternus was pressed on the possibility that AI agents might render traditional applications obsolete. The premise is not fanciful. A capable agent, given access to APIs and a sufficiently large context window, could in theory handle tasks that today require opening multiple apps: booking travel, comparing prices, drafting messages, scheduling meetings. If that vision materialises, the App Store, a pillar of Apple's services revenue, could face structural decline.

Ternus deflected. He acknowledged that such a system might make life easier, but pivoted quickly to a defence of product creation as a "deeply human and personal thing." The phrasing is telling. It suggests Apple views the agent paradigm not as an inevitability to prepare for, but as a scenario that may or may not unfold, and one the company is not eager to accelerate.

That reluctance is understandable from a business perspective. Apple earned approximately USD 24 billion in App Store revenue in fiscal 2025, much of it from the 30 per cent commission it extracts on digital goods. A world in which users interact primarily with an agent, rather than with individual apps, threatens that stream. But strategic caution and user interest do not always align. If a rival platform offers a genuinely useful agent experience, iOS users may find themselves envying Android, a reversal of the dynamic that has prevailed for much of the past decade.

The Cost of Incrementalism

Apple's approach to AI mirrors its approach to other technologies that required cannibalising existing revenue. The company was slow to embrace larger iPhone screens, slow to add stylus support to the iPad, and slow to allow third-party watch faces on the Apple Watch. In each case, the delay was justified internally as a commitment to quality, but externally it looked like an unwillingness to disrupt profitable product lines.

The current AI moment may be different in kind. The technology is advancing rapidly enough that a twelve-month delay can mean the difference between leading and following. Inference costs are falling, model capabilities are improving, and the user experience of conversational interfaces is becoming more fluid. Competitors are iterating in public, learning from user behaviour, and refining their models accordingly. Apple, by contrast, is iterating behind closed doors, shipping only when it believes a feature meets its internal bar.

There is a defensible logic to that method. Shipping half-baked AI features can erode trust. Google's early Bard demos produced factually incorrect answers; Microsoft's initial Copilot integration felt intrusive to many Office users. Apple has avoided those missteps by shipping less. But the company has also avoided the learning that comes from deployment at scale. Without millions of users stress-testing its models in the wild, Apple may find itself optimising for the wrong use cases.

What Comes Next

Ternus did not outline a detailed roadmap, but the contours of Apple's strategy are becoming clearer. The company will continue to integrate machine learning into its hardware and software stack, but it will do so selectively, prioritising features that align with its existing product philosophy. Expect improvements to Siri, better on-device photo and video editing, more capable predictive text, and perhaps some form of summarisation in Mail and Messages. Do not expect Apple to launch a ChatGPT competitor, open-source its models, or allow third-party agents deep access to iOS.

Whether that strategy proves sufficient depends on how quickly user expectations evolve. If the broader market remains content with incremental AI features, Apple's caution will look wise. If, however, a rival platform delivers a step change in utility, Apple may find itself in the unfamiliar position of playing catch-up in a category that defines the next decade of computing.

For now, Ternus appears untroubled. He has inherited a company with enormous financial resources, a loyal customer base, and a track record of succeeding on its own timeline. The question is whether the AI era will allow Apple the luxury of moving at its own pace, or whether the velocity of change will force the company to choose between its principles and its market position.