Last year, Apple found itself in a situation where its artificial intelligence efforts looked like a huge mess. The company that once set trends in consumer technology was struggling to catch up in AI. It delayed promised Siri features and had to rely on Google's Gemini to fill gaps in its Apple Intelligence platform. Wall Street investors took it badly—the company's shares fell in the first half of the year because they feared Apple had missed the decade's most important technological shift.
Then came leadership changes. During one week in December, AI chief John Giannandrea retired because the delays had made his position untenable. The head of interface design moved to Meta, and the company announced the retirements of its general counsel and environmental chief. Dozens of engineers fled to competitors. OpenAI bought Jony Ive's design studio for more than CZK 138 billion and hired dozens more engineers from Apple. Meta lured AI researchers with enormous salaries, including people from the team that worked on Apple's models.
CEO Tim Cook, who took over from Steve Jobs in 2011 and is now 65, faces succession debates among analysts and investors. Nevertheless, Apple shares have risen 40% since June, giving the company a market value of CZK 92 trillion, the second highest after Nvidia. Meta and Microsoft shares, by contrast, declined even though those companies poured billions into AI infrastructure.
Apple is playing a different game
What looked like falling behind now seems smart. Apple avoided the AI race and the massive spending that goes hand in hand with it. At WWDC in 2025, it announced the Foundation Models framework, which gives developers access to AI directly on the device with just three lines of code.
This approach addresses three major problems. API costs make AI features too expensive for smaller developers. The need for an internet connection limits where and when AI works. Companies fear liability and data security risks when sending information to the cloud. With on-device AI, Apple removes these obstacles and could thus trigger the mass adoption of AI that has yet to materialize despite massive investment.
Apple's model, with 3 billion parameters, cannot match leading systems such as GPT-4, which reportedly has more than a trillion parameters. But that is beside the point. With more than a billion iPhone users and millions of developers in the App Store, Apple is betting that broad access to decent AI will beat the most powerful model that few can afford. Even if Apple's current models are weaker than those from OpenAI or Google, allowing millions of developers to experiment with free AI on their devices could produce breakthrough applications that Apple itself could not develop.
Apple is betting on winning in AI the same way it did in mobile phones. It does not need to have the most powerful models if it has the best applications. Millions of developers testing on-device AI could make the iPhone the best place for AI. And it is far cheaper than building models that could be obsolete in three months.
As one portfolio manager told Bloomberg, Apple "kept a cool head and is controlling spending while all its peers went in the opposite direction." Investors increasingly doubt whether AI investments will pay off.
The chaos may be overstated
Some departures were overdue. John Giannandrea's departure acknowledges the problems with Apple Intelligence. Others could be improvements, such as appointing a new head of interface design with industry experience to replace someone from fashion and advertising. Still others look more like unfortunate timing in the media than an exodus from a sinking ship—such as the retirements of the general counsel and environmental chief after many years.
The talent losses are real, especially in AI research, where Meta and OpenAI are aggressively recruiting people. But this week Apple confirmed that Johny Srouji, whose teams developed the M-series processors and in-house mobile modems, is staying despite rumors of his departure.
Apple still faces challenges. It has not launched a successful new product category in a decade. Its AI products have disappointed. The leadership changes bring uncertainty at a time when it needs to prove that it can compete in the AI era.
But Tim Cook remains at the helm, the team is being renewed with internal talent, and the company is not promising features it cannot deliver. Whether this cautious approach succeeds depends on execution. For a company supposedly in crisis, the market is betting on stability.
Siri's delay
Apple postponed advanced AI features for Siri in 2025, contributing to its broader AI failures. According to CNBC reporter Steve Kovach on the December 31, 2025, episode of Squawk Box, Siri was expected to "be able to do all sorts of great things," but they had to postpone it. This dampened investor enthusiasm, as shares had risen 35% on the hype, but the new AI-enabled phones proved to be more of a nice bonus than the main driver of sales.
Kovach emphasized that 2026 must deliver amazing AI results to revive Apple, including integrations with AirPods and a focus on iPhone upgrades. The year 2025 was pivotal, with Apple failing to meet expectations despite its CZK 92 trillion market value.
