Apple has struck a $1 billion-a-year deal to use Google’s 1.2 trillion parameter Gemini AI for Siri, but it’s all happening on Apple’s “Private Cloud Compute” servers. This “walled-off” architecture is the key to the “interim solution.”
This setup allows Apple to use Google’s “ultrapowerful” AI—which beat OpenAI and Anthropic in a “bake-off”—without compromising its non-negotiable privacy stance. Google gets its $1B, but no user data.
This “interim solution” is part of Apple’s “Glenwood” project to fix Siri. The new “Linwood” Siri, due in the spring, will be a hybrid. Apple’s 150-billion parameter models will handle simple tasks, while Google’s AI will tackle complex “summariser” and “planner” functions.
The partnership is a reluctant admission of Apple’s AI lag, overseen by executives Craig Federighi and Mike Rockwell. It’s also a huge win for Google’s “AI supplier” business.
While Apple is pushing its teams to build a 1T+ model to replace Gemini, this “temporary” fix could last for years.