Siri Reborn: ‘Linwood’ Project to Use $1B Google AI for ‘Planner’ Functions

Apple’s new Siri, code-named “Linwood,” is on track for a spring release, and it will be powered by a $1 billion-a-year deal with Google. This “ultrapowerful” 1.2 trillion parameter Gemini AI model will be integrated into Siri’s core to handle its most complex cognitive tasks: the “summariser” and “planner” functions. This will allow the new assistant to synthesize information and strategically execute multi-step user requests, capabilities that have long eluded it.
This move is part of the “Glenwood” project, an internal Apple initiative led by Craig Federighi and Mike Rockwell to fix the lagging assistant. After an extensive evaluation of models from OpenAI and Anthropic, Apple selected Google’s Gemini as the best “interim solution.” This alliance allows Apple to leverage a competitor’s superior AI to bridge its own technology gap while its teams work on a long-term in-house replacement.
The technological leap is significant. The 1.2 trillion parameter model dwarfs Apple’s current 150-billion parameter cloud AI, offering a vast expansion in power and contextual understanding. This raw capability is what Apple needs to make Siri competitive with other modern AI assistants. However, other parts of Siri and Apple Intelligence will continue to run on Apple’s own proprietary models.
Privacy is a non-negotiable aspect of this deal. The Gemini model will not run on Google’s cloud but will instead be hosted on Apple’s own Private Cloud Compute servers. Apple has already allocated the necessary server hardware to run the model in this “walled-off” environment, ensuring that Google has no access to any Apple user data.
This partnership will not be publicly promoted. Google will serve as a “behind-the-scenes” technology supplier, a very different arrangement from its visible role as the Safari default search engine. While Apple is working on its own 1 trillion parameter model to eventually replace Gemini, it is in a race against Google’s rapidly improving, top-tier AI, making this $1 billion “temporary fix” a potentially long-lasting one.