According to Gizmodo, Apple will reportedly pay Google about $1 billion per year to power Siri with a 1.2 trillion parameter Gemini-based LLM in what’s being treated as a temporary fix. The upgrade, internally codenamed “Linwood,” will launch this coming spring as part of iOS 26.4 and focus on personalization by tapping into user data and on-screen activities. Apple apparently regards Google as just a “behind-the-scenes technology supplier” rather than an AI partner. The company will still use its own proprietary models for simpler queries that can run on-device, with Google’s Gemini handling complex tasks and summarization while running on Apple’s servers to preserve privacy.
Sponsored content — provided for informational and promotional purposes.
Siri’s long road
Here’s the thing: Siri has been the embarrassing cousin at Apple‘s family reunion for years now. Released back in 2010, it was supposed to be this revolutionary voice assistant that would transform how we interact with our phones. Instead, it became the thing you accidentally activate and then immediately cancel because it’s just going to disappoint you. We’ve all been there – asking Siri something simple and getting “I’m sorry, I can’t help with that” or being directed to a web search. It’s basically become a symbol of Apple playing catch-up in the AI race while everyone else sprinted ahead.
Behind-the-scenes struggles
Now, this billion-dollar Band-Aid tells a pretty revealing story about what’s been happening inside Apple’s AI division. They were reportedly talking to OpenAI and Anthropic back in July about potentially using GPT or Claude models. When those AI announcements didn’t materialize, it was pretty clear something wasn’t going according to plan. And then you’ve got key staff like Ke Yang, who was leading Apple’s in-house chatbot effort, jumping ship to Meta last month. That’s not exactly a vote of confidence in Apple’s internal AI capabilities. So they’re essentially renting Google‘s brain while trying to build their own – and paying through the nose for it.
Privacy and power balance
The privacy angle here is actually pretty clever. Apple’s making sure Google’s models run on Apple’s own servers without external data sharing. On-device personal data will get processed using Apple’s Foundation Models. It’s a way to maintain their privacy-first branding while essentially admitting they can’t compete on the raw AI power front. But let’s be real – when you’re paying another company $1 billion annually for their technology, how much of a “behind-the-scenes supplier” are they really? That’s serious money, even for Apple. It suggests this might be less temporary than they’re letting on.
What’s next for Apple AI
So where does this leave Apple in the long run? They’re clearly still committed to building their own AI system eventually. But with staff defections and what appears to be significant technical hurdles, that “eventually” might be further away than they’d like to admit. Meanwhile, they’re essentially funding their biggest competitor’s AI development. It’s an awkward position for a company that usually insists on controlling its entire tech stack. The spring iOS 26.4 release will be the first real test of whether this expensive patch job can make Siri actually useful rather than just a weather-telling novelty. Honestly? I’m skeptical, but I’d love to be proven wrong.
