According to 9to5Mac, Apple has announced that John Giannandrea, the company’s senior vice president for Machine Learning and AI Strategy, is stepping down. He will serve as an advisor before officially retiring in the spring of 2026. The company is bringing in Amar Subramanya, a former Microsoft AI and Google researcher, as the new vice president of AI reporting to Craig Federighi. Subramanya will lead Apple Foundation Models, ML research, and AI Safety. The rest of Giannandrea’s team is being split between executives Sabih Khan and Eddy Cue. The report frames this as a move following Apple’s “AI missteps” over recent years.
A Shift in Command
So, what does this shuffle actually tell us? First, it signals a pretty significant consolidation of AI power under Craig Federighi, Apple‘s senior VP of Software Engineering. Having the new AI VP report directly to him, instead of being a separate SVP-level role, suggests Apple wants AI development welded directly to its core OS teams—iOS, macOS, etc. That makes sense. But splitting the existing organization between hardware ops (Sabih Khan) and services (Eddy Cue) is more curious. It could mean they’re trying to bake AI deeper into silicon and into apps like Siri and Apple Music simultaneously. The risk? A lack of a single, unified vision. Without a central figure like Giannandrea, can they avoid internal silos?
The Google-to-Microsoft Pipeline
Here’s the thing about the new hire, Amar Subramanya: his resume reads like a map of the AI wars. He’s a former Googler, just like Giannandrea was. But he’s coming directly from Microsoft AI, which is the company that has absolutely sprinted ahead in the generative AI race thanks to its partnership with OpenAI. Apple is basically hiring from the team that has been eating its lunch in public perception. Is this Apple trying to inject some of that aggressive, product-focused AI DNA? Probably. But corporate culture is a hell of a thing. Microsoft’s “partner and integrate” fast-follow strategy is very different from Apple’s “perfect it in secret” ethos. Can Subramanya navigate that?
The Foundation Model Gap
Let’s talk about that “Apple Foundation Models” team Subramanya is now leading. This is the core of the issue. While OpenAI, Google, and Meta have been launching and iterating on large language models for years, Apple has been conspicuously quiet. We’ve seen research papers, but no flagship product. The rumored “Apple GPT” internal tool hasn’t materialized for consumers. They are years behind in scale and public testing. Building a foundation model that is both powerful *and* privacy-centric—which Apple insists on—is an immense technical challenge. It’s a trade-off between capability and their core selling point. They can’t just clone ChatGPT. They need their own moat.
What “Retirement” Really Means
Now, a “retirement” in the spring of 2026, with an “advisor” role until then, is a very long, soft exit. That’s nearly two years. This isn’t a sudden firing; it’s a managed transition. It gives Subramanya time to get his bearings while Giannandrea’s institutional knowledge is still on tap. But let’s be real. It also feels like an acknowledgment that the previous strategy, under Giannandrea, didn’t produce the results Apple needed at the speed it needed them. The pressure is publicly on now. With WWDC 2024 just months away, where AI is expected to be the star, this leadership change is Apple’s way of telegraphing a new chapter. The question is, is it too late to change the narrative?
