Apple’s AI Failure: $1B Paid to Rival Google to ‘Fix’ Siri

In a stunning admission of its lag in the artificial intelligence race, Apple has agreed to pay its chief rival, Google, $1 billion annually. This massive fee is to license Google’s Gemini AI, which will serve as the “behind-the-scenes” brain for the next version of Siri.
This move is a reluctant but necessary “interim solution” for Apple, whose “Glenwood” project is tasked with overhauling the company’s struggling assistant. Apple’s own 150-billion parameter models are proving insufficient for the complex tasks users now expect.
The new Siri, codenamed “Linwood,” will be a hybrid. It will rely on Google’s “ultrapowerful” 1.2 trillion parameter model for all heavy lifting, including complex “summariser” and “planner” functions. Apple’s tech will be relegated to simpler tasks.
Google secured the deal after its AI dominated an extensive “bake-off,” beating both OpenAI’s ChatGPT and Anthropic’s Claude. The project is being overseen by top Apple executives Craig Federighi and Mike Rockwell, who are under pressure to deliver a competitive assistant.
Despite using a rival’s technology, Apple is enforcing strict privacy rules. The Gemini model will run on Apple’s “walled-off” Private Cloud Compute servers, guaranteeing that Google gets the $1 billion fee but no access to Apple’s user data.