Apple has announced a significant leadership change in its artificial intelligence division, with John Giannandrea, the company's AI chief since 2018, stepping down. His replacement is Amar Subramanya, a seasoned tech executive with extensive experience at both Google and Microsoft, most recently leading engineering for Google's Gemini Assistant. This strategic appointment comes as Apple faces mounting challenges with its AI initiatives, including Apple Intelligence and Siri.
Characterized as an inevitable shake-up, Giannandrea's departure follows a tumultuous period for Apple's AI efforts. Apple Intelligence, the company's answer to the generative AI boom, has struggled since its October 2024 launch, drawing reviews from "underwhelming" to "alarmed." Early missteps included a notification summary feature that generated embarrassing and false headlines in late 2024 and early 2025. Notably, the BBC twice reported on erroneous information from Apple Intelligence, including a false report of a suspect's suicide and a premature championship win for a darts player before the final even began.
The promised overhaul of Siri also proved to be a significant setback. A May Bloomberg investigation exposed the extent of Apple's AI struggles, revealing that weeks before its anticipated April launch, many new Siri features failed to work during testing by Craig Federighi, Apple's software chief. This led to an indefinite delay in the launch and prompted class-action lawsuits from iPhone 16 buyers who had expected an AI-powered assistant.
According to Bloomberg, Giannandrea had already been sidelined prior to this announcement. Reports indicated that Apple CEO Tim Cook had removed Siri from Giannandrea's oversight in March, reassigning it to Vision Pro creator Mike Rockwell. Apple's secretive robotics division was also reportedly moved out of Giannandrea's control.
The Bloomberg report detailed a culture of organizational dysfunction within Apple's AI division, citing poor communication between AI and marketing teams, budget misalignments, and a leadership crisis. Some employees reportedly derided Giannandrea's group as "AI/MLess." The investigation also highlighted an exodus of AI researchers to rival companies such as OpenAI, Google, and Meta.
In a potentially humbling development, Apple is now reportedly considering using Google's Gemini model to power the next iteration of Siri. This move underscores the intensity of the rivalry between the two tech giants, which spans over 15 years across various domains, including mobile operating systems, app stores, and cloud services, and now extends to artificial intelligence.
Giannandrea, who joined Apple from Google where he led Machine Intelligence and Search, was responsible for Apple's overall AI strategy, machine learning infrastructure, and Siri development. Amar Subramanya now inherits these critical responsibilities, reporting directly to Craig Federighi, with a clear mandate to accelerate Apple's progress in the competitive AI landscape.
This leadership transition occurs at a pivotal time for Apple's AI strategy. Unlike competitors who invest heavily in large AI data centers, Apple has championed a privacy-first approach, focusing on processing AI tasks directly on user devices via its custom Apple Silicon chips. While this method minimizes user data collection, routing complex requests through Private Cloud Compute for temporary processing, it presents significant trade-offs. On-device AI models are inherently smaller and less powerful than the vast cloud-based systems of rivals. Furthermore, Apple's reluctance to collect extensive user data means its researchers train models on licensed and synthetic data, potentially limiting their capabilities compared to systems fueled by massive real-world data troves. The long-term efficacy of this unique AI philosophy remains a critical unanswered question for the company.
