According to 9to5Mac, Apple’s head of AI, John Giannandrea, is retiring next year and stepped down immediately, replaced by industry veteran Amar Subramanya. The company is also restructuring its AI initiatives, with Subramanya leading critical areas like Apple Foundation Models and AI Safety. Meanwhile, OpenAI CEO Sam Altman has declared an internal “code red,” telling employees all efforts must focus on improving ChatGPT’s quality, personalization, speed, and reliability. This push is a direct response to fears that Google’s Gemini has overtaken ChatGPT. The report states Apple’s AI team is at an “all-time low morale” and Subramanya faces an “uphill battle.” All this raises the pressure for Apple’s new Siri, expected to launch sometime next year.
The Scale of Apple’s AI Problem
Look, losing your AI chief is never a good sign, especially when you’re already perceived as years behind. But here’s the thing: the internal details are even more telling. The report mentions a “heavily shorthanded team” with rock-bottom morale. That’s not just a staffing issue; it’s a cultural one. When you’re trying to reboot a massive, foundational technology like your AI stack, you need a motivated, focused team. Apple’s move to split Giannandrea’s old org between hardware lead Sabih Khan and services boss Eddy Cue makes sense on paper—it aligns AI work with the products that will use it. But in practice, it risks creating more internal friction and competing priorities right when they need singular focus. Can Subramanya, a newcomer to Apple’s famously siloed culture, really unify this effort in time?
What OpenAI’s “Code Red” Really Means
So OpenAI is hitting the panic button. Sam Altman delaying other products to throw everything at ChatGPT’s quality is a huge admission. For months, the narrative has been about GPT-5 and their next big model. But this “code red” shifts the focus to the boring, brutal work of refinement: making it faster, more reliable, and actually personal. That’s the hard part. Anyone can demo a flashy new capability. Building a product that millions use every day without hallucinating, lagging, or feeling generic? That’s a whole different engineering marathon. This move basically confirms that Google, with the integrated might of its search empire and Gemini, has landed some serious blows. The AI race isn’t just about who has the smartest model anymore; it’s about who can build the most usable, dependable product.
The 2025 Siri Showdown Looms
Now, put these two stories together and the stakes for next year become crystal clear. While OpenAI and Google are in a sprint, iterating weekly, Apple is trying to rebuild its engine mid-race. The expectation for the “new Siri” in 2025 isn’t just to be good by 2024 standards; it needs to be competitive with whatever Gemini and ChatGPT have evolved into by *then*. That’s a terrifying gap to close. Apple’s traditional advantage has been deep hardware-software integration and privacy. But will that be enough if the core intelligence feels a generation behind? I think the pressure isn’t just on Subramanya’s team to ship something. It’s on them to ship something that doesn‘t feel like a catch-up product on day one. That’s a nearly impossible task, but it’s the one they’ve been handed.
