Apple’s ‘Project Glenwood’ Taps Google in $1B Deal to Fix Siri’s Brain

“Project Glenwood” is the internal code name for Apple’s high-stakes mission to fix Siri, and its primary tactic involves a $1 billion-a-year deal with Google. This initiative, led by top executives Mike Rockwell and Craig Federighi, is finalizing a pact to license Google’s 1.2 trillion parameter Gemini AI model. This “ultrapowerful” model will serve as the new “brain” for Siri’s most complex functions, acting as an “interim solution” while Apple’s own AI matures.
The “Glenwood” project’s new assistant, code-named “Linwood” and planned for iOS 26.4 next spring, will be a hybrid. Google’s massive model will be deployed for the “summariser” and “planner” components, allowing Siri to synthesize information and execute multi-step tasks. This is a monumental upgrade from Apple’s current 150-billion parameter cloud AI. Other Siri features will continue to run on Apple’s in-house models.
The decision to use Google’s technology came after Apple’s extensive testing of offerings from OpenAI (ChatGPT) and Anthropic (Claude). Google’s Gemini was deemed the most capable of delivering the generational leap Apple needs immediately. This deal represents a significant, though reluctant, reliance on a chief competitor to remain relevant in the generative AI race.
To protect user privacy, the Gemini model will be hosted on Apple’s own Private Cloud Compute servers. This “walled-off” architecture is a critical part of the deal, guaranteeing that no Apple user data is ever processed by or shared with Google’s infrastructure. Apple has already dedicated significant server resources to run this computationally intensive model.
This partnership will be intentionally low-profile. Apple will not promote Google’s involvement, treating it as a “behind-the-scenes” supplier. This is a temporary measure, as Apple’s management is still pushing its AI teams to develop a 1 trillion parameter in-house model to eventually replace Gemini. However, catching Google’s top-tier, rapidly evolving AI remains a formidable challenge.