Google has reportedly won a $1 billion-a-year contract from Apple, as its 1.2 trillion parameter Gemini model was selected over offerings from OpenAI and Anthropic to power the next generation of Siri. Apple conducted an extensive evaluation period, testing Gemini, ChatGPT, and Claude to find an “interim solution” for its generative AI shortcomings. The company “zeroed in on Google earlier this year,” concluding that its technology was the most capable of rebuilding Siri’s underlying framework.
This decision is a significant win for Google and a core component of Apple’s “Glenwood” project, the internal effort to fix Siri led by executives Craig Federighi and Mike Rockwell. The new Siri, code-named “Linwood” and set for a spring release, will use the “ultrapowerful” Gemini model to run its “summariser” and “planner” functions. This will enable Siri to process complex, multi-step requests, a massive upgrade from Apple’s current 150-billion parameter cloud AI.
The partnership, while substantial, will be kept “behind-the-scenes.” Apple will not publicly promote Google’s involvement, treating it as a technology supplier. This is a different strategy from the public-facing Safari search deal and is also separate from earlier, failed discussions about integrating Gemini as a standalone chatbot. Apple’s CEO has, however, suggested that other chatbots beyond the current ChatGPT option could be offered in the future.
A critical requirement for the deal is Apple’s “walled-off” privacy system. The 1.2 trillion parameter Google model will be hosted and run on Apple’s own Private Cloud Compute servers. This ensures that no user data is ever processed by or visible to Google, allowing Apple to leverage its rival’s superior AI without compromising its core privacy values. Apple has already allocated the necessary server hardware for this task.
While this $1 billion deal provides a much-needed short-term boost, Apple’s long-term ambition is to replace Gemini with its own 1 trillion parameter in-house model. However, Apple’s AI teams are in a difficult race against Google, whose Gemini 2.5 Pro model continues to top AI performance leaderboards, making this “temporary” reliance a potentially long-term and costly engagement.