Apple AI Strategy: Why the Crisis Is Smaller Than Thought Apfelpatient

Apple has been under close scrutiny regarding artificial intelligence for some time now. Delays at Apple Intelligence, the departure of several executives, and a gradual restructuring of internal teams have given the impression that Apple is struggling to keep up in the AI race. However, a closer look reveals a much more nuanced picture. Apple's AI team is larger than publicly perceived, its strategic direction remains stable, and the recent changes suggest a reprioritization rather than a failure.

The public discourse surrounding Apple's AI strategy focuses heavily on individuals and departures. John Giannandrea's resignation, in particular, has often been interpreted as a symbol of structural failure. However, this perspective overlooks Apple's fundamental understanding of AI. Apple does not view AI as a standalone product, but rather as an integrated function within software and devices. This understanding is precisely what shapes the current restructuring.

Apple's AI strategy according to Giannandrea

The departure of John Giannandrea came as no surprise to many observers. Internally, he was considered by some to be a key figure at the center of criticism regarding the slow development of Apple Intelligence. Nevertheless, his resignation has not resulted in an abrupt change of course. The existing AI teams continue to work on the same foundations, and there is no indication that Apple has abandoned its long-term strategy.

If Apple had fundamentally changed its direction, this would have been evident in the numerous leaks surrounding the launch of Apple Intelligence in the spring. Instead, evidence is mounting that Apple is continuing to implement exactly what it previously announced. The focus is clearly on AI systems that primarily run on the device. The so-called Apple Foundation Models form the technical foundation. These are supplemented by third-party models accessed via private cloud computing.

Restructuring of AI leadership

Apple recruited Giannandrea from Google in 2018 to consolidate internal AI and machine learning efforts as Senior Vice President. From the outset, his approach was clearly focused on on-device AI, meaning models that operate as closely as possible to user data. Over time, his responsibilities expanded considerably. In addition to Siri and AI, he also oversaw the development of the Apple Car and later led the robotics team.

In 2025, however, Apple began a gradual restructuring. Robotics was placed under the responsibility of the hardware chief, and Siri was organizationally transferred to the Vision Pro team. This allowed the remaining AI team to fully concentrate on developing the Apple Foundation Models. This structure remained in place even after Giannandrea's departure, strongly suggesting that the reorganization was planned independently of him.

Internally, the delay at Apple Intelligence likely played a key role. The restructuring can be seen both as a reaction to missed deadlines and as a necessary step to realistically meet new deadlines.

New responsibilities and familiar approaches

Amar Subramanya, also a former Google employee, now plays a key role. As Vice President of AI, he reports directly to Craig Federighi. The new title underscores that AI is firmly embedded in Apple's software development and not viewed as a separate product segment.

Subramanya's team is focusing on the further development of Apple's Foundation Models. Parts of the previous AI team have been reassigned to other departments. The search and knowledge area will now fall under the responsibility of Eddy Cue.

The Foundation Models team is led by Zhifeng Chen. His team includes several former Google AI researchers, among them Li Xiao, Christopher Fifty, and Jun Xu. According to a report by The Information, more than half of Chen's team comes from Google, many of whom joined Apple within the last two to four years. There is speculation that some of these employees might leave the company following Giannandrea's resignation, but so far there is no reliable evidence to support this.

Departures, fluctuation and their significance

Last year, there were about a dozen reports of managers and employees from Apple's AI division leaving the company. These reports provide little context. While some of those affected were reportedly offered severance packages in the millions, the specific reasons for their departures remain unknown.

These isolated incidents were often portrayed as signs of a crisis. What's overlooked is that Apple's AI efforts don't hinge on one person or a small group. Many of the senior executives who left had been with Apple for six years or more. At that level, that's a normal turnover rate. At the employee level, turnover in the AI field is high across the industry, sometimes occurring as frequently as every nine months.

The actual size of Apple's AI teams

Apple is known for its extreme secrecy. Employees are sometimes even encouraged to conceal their team affiliation. This makes it difficult to realistically estimate the team size. However, the numerous scientific publications in which Apple is involved offer some clues.

An attempt to identify the individuals involved in this research quickly revealed the scale of the operation. Reviewing just four of the 96 pages of publications yielded a list of nearly 200 named Apple employees. A complete inventory would have been virtually impossible. This clearly demonstrates that Apple's AI teams are significantly larger than often assumed and, despite some departures, continue to conduct intensive research and publish extensively.

The fact that this work receives little attention in the wider AI community is probably due to Apple's tendency to translate research findings into products only late or indirectly.

Outlook on Apple Intelligence from 2026

No drastic strategy changes are expected in the short term. Giannandrea's work on powerful on-device models will become visible in the near future. A comprehensive overhaul of Apple Intelligence is planned for early 2026 with iOS 26.4.

A key element will be a new, LLM-powered version of Siri. This version will be able to directly access Apple Foundation Models on the device and perform complex tasks via app intents. Apple also plans partnerships that integrate third-party models via private cloud computing. For example, a specially adapted Gemini model could be used to perform web searches, leveraging data prepared by an Apple model.

Apple plans to integrate Gemini into Siri

Apple is deliberately taking a different approach to AI

Apple continues to pursue its own, relatively low-key approach to artificial intelligence. The recent restructuring doesn't indicate failure, but rather a deliberate focus. While other companies make headlines with exaggerated promises and mass-produced AI-generated products, Apple is working on an integrated, private, and efficient AI ecosystem deeply embedded in the iPhone. Short-term turbulence doesn't change the fact that Apple is consistently and long-term implementing its AI strategy. The best products for you: Our Amazon storefront offers a wide selection of accessories, including those for HomeKit. (Image: Shutterstock / Ekaterina_Minaeva)

Have you already visited our Amazon Storefront? There you'll find a hand-picked selection of various products for your iPhone and other devices – enjoy browsing !

This post contains affiliate links .

Add Apfelpatient to your Google News Feed.

Was this article helpful?

YesNo

Apple's AI strategy: Why the crisis is smaller than expected

iPhone Fold: Why the outer display is much smaller

Apple's AI strategy: Why the crisis is smaller than expected

iPhone Fold: Why the outer display is much smaller

Apple allows alternative app stores on iOS in Brazil

About APFELPATIENT

Legal

service