Apple just named a new AI chief with Google and Microsoft expertise, as John Giannandrea steps down

In a carefully worded announcement on Monday, Apple said John Giannandrea, who has been the company’s AI chief since 2018, is “stepping down” to, well, not work at Apple anymore. He’ll stick around through spring as an advisor.

His replacement is Amar Subramanya, a highly regarded Microsoft executive who spent 16 years at Google, most recently leading engineering for the Gemini Assistant. It’s a savvy hire, given that Subramanya knows the competition intimately.

The move is being characterized as a shake-up. It was seemingly inevitable in retrospect. Apple Intelligence, the company’s answer to the ChatGPT moment, has been stumbling since its October 2024 launch. Reviews have ranged from “underwhelming” to outright alarmed.

Its first months were some of the roughest. A notification summary feature meant to condense multiple alerts into digestible snippets generated a series of embarrassing, untrue headlines in late 2024 and early 2025. Among other missteps, the BBC complained twice after Apple Intelligence falsely reported that Luigi Mangione, the man accused of killing UnitedHealthcare CEO Brian Thompson, had shot himself (he hadn’t) and that a darts player, Luke Littler, won a championship before the final even began.

Then there was Siri’s promised overhaul, which became a black eye for Apple.

A Bloomberg investigation published in May revealed the depths of Apple’s AI struggles. For instance, when Craig Federighi, Apple’s software chief, tested the new Siri on his own phone just weeks before its planned launch in April, he was dismayed to find that many of the features the company had been touting didn’t work. The launch was delayed indefinitely, triggering class-action lawsuits from iPhone 16 buyers who’d been promised an AI-powered assistant.

By that point, Giannandrea had already been sidelined, according to Bloomberg. The news organization reported that Tim Cook had stripped Siri from Giannandrea’s oversight entirely back in March, handing it to Vision Pro creator Mike Rockwell. Apple removed its secretive robotics division from Giannandrea’s control, too.

Bloomberg’s investigation painted a picture of organizational dysfunction, with weak communication between AI and marketing teams, budget misalignments, and a leadership crisis severe enough that some employees had taken to mockingly calling Giannandrea’s group “AI/MLess.” The report also documented an exodus of AI researchers to competitors, including OpenAI, Google, and Meta.

Apple is reportedly now leaning on Google’s Gemini to power the next version of Siri, an astonishing and also, presumably, humbling twist considering the intense rivalry between the two companies that dates back more than 15 years, across mobile operating systems, app stores, browsers, maps, cloud services, smart home devices, and now AI.

Giannandrea came to Apple from Google, where he ran Machine Intelligence and Search. At Apple, he oversaw the AI strategy, machine learning infrastructure, and Siri development.

Now Subramanya inherits those responsibilities, reporting to Federighi with a clear mandate to help Apple catch up in AI.

It’s an interesting moment for the company. While competitors have been pouring billions of dollars into massive AI data centers, Apple has focused on processing AI tasks directly on users’ devices using its custom Apple Silicon chips, a privacy-first approach that avoids collecting user data. (When more complex requests require cloud processing, Apple routes them through Private Cloud Compute, servers that promise to process data temporarily and delete it immediately.)

Whether that philosophy pays off or whether it has permanently left Apple behind is an outstanding question. Apple’s approach comes with clear trade-offs. Among them, on-device models are smaller and less capable than the massive models running in competitors’ data centers, and Apple’s reluctance to collect user data has left its researchers training models on licensed and synthetic data rather than the giant troves of real-world information that fuel its rivals’ systems.