Apple Ousts Giannandrea, Installs New AI Chief
Amar Subramanya’s arrival signals a strategic reset as Apple races to repair stalled AI progress.
News
- OpenAI Hits ‘Code Red’ as ChatGPT Upgrades Take Priority Over Ads
- Apple Ousts Giannandrea, Installs New AI Chief
- Gemini Named Most Culturally Aligned AI in UAE’s First-of-its-Kind Index
- UAE Approves National Encryption Policy for a Post-Quantum Shift
- OpenAI API Users Hit With Third-Party Security Breach
- Uber and WeRide Launch First Driverless Robotaxi Service Outside the U.S. and China
[Image source: Chetan Jha/MITSMR Middle East]
Apple has issued a carefully phrased statement announcing that John Giannandrea, its head of artificial intelligence since 2018, will step down and transition into an advisory role. His departure marks a consequential shift for a company that has struggled to keep pace in the AI competition.
The company has appointed Amar Subramanya, a senior Microsoft leader with a long tenure at Google, where he oversaw engineering for the Gemini Assistant, to assume leadership of its AI efforts. The choice is strategic and significant: Subramanya arrives with knowledge of rival architectures, organizational structures, and roadmaps.
The leadership change follows a turbulent year for Apple Intelligence, the company’s attempt to articulate a response to the generative-AI moment. The system’s rollout, beginning in October 2024, was marred by reliability failures and public-facing errors.
A notification-summary feature produced inaccurate, sometimes sensational headlines, prompting formal complaints, including two from the BBC. In parallel, the much-promised revamp of Siri faltered. Weeks before launch, Apple’s own software chief, Craig Federighi, reportedly found that many core features underperformed or failed entirely. The delay led to class-action complaints from customers who had purchased iPhone 16 devices expecting AI capabilities.
According to Bloomberg, Giannandrea had been progressively stripped of key responsibilities months before Monday’s announcement. Oversight of Siri was transferred to Vision Pro architect Mike Rockwell; Apple’s robotics initiative was reassigned as well. The report described an AI organization struggling with inadequate communication between engineering and marketing groups, mismatched budgets, internal skepticism about leadership, and the departure of researchers to companies such as OpenAI, Google, and Meta. Some employees reportedly referred to the unit as “AI/MLess”.
Perhaps most striking is Apple’s open willingness to rely on Google’s Gemini models for future versions of Siri — a move that would have been nearly unthinkable given the companies’ long-standing competition.
Apple has long prioritized on-device processing using Apple Silicon, emphasizing privacy and minimal data collection. While this approach offers strong security guarantees, it restricts model size and limits the breadth of training data.
Rivals, meanwhile, have invested heavily in cloud-scale systems powered by massive datasets. Whether Apple’s privacy-centric architecture can be reconciled with expectations for AI performance remains one of the company’s most pressing questions.
