Apple May Use Google Servers for AI Siri
Apple is reportedly considering using Google's servers to store data for its upgraded, AI-powered Siri assistant. The potential move would mark an about-face for Apple, which has historically prioritized in-house infrastructure for privacy, signaling the immense challenge of scaling cutting-edge AI.
This potential partnership is part of a broader initiative, internally dubbed "Project Greymatter," to embed a suite of AI tools across Apple's core apps like Safari, Photos, and Notes. The overhaul for Siri aims to make the assistant more contextually aware, factoring in people, locations, and dates from user data to generate smarter responses. The collaboration involves integrating Google's powerful Gemini AI models to serve as the foundation for Apple's next-generation Siri. Under the multi-year deal, Apple is expected to pay Google approximately $1 billion annually. This arrangement is structured as a cloud-computing contract rather than a simple licensing agreement, giving Apple access to a specially modified 1.2 trillion-parameter Gemini model. Apple's move to partner with a competitor follows internal delays and challenges in developing its own large-scale AI models. The company evaluated models from other major players, including OpenAI and Anthropic, before landing on Google's Gemini as the most capable foundation for its ambitions. This allows Apple to accelerate its AI development and reduce execution risk by leveraging mature technology. To address privacy concerns, the companies plan to use a "Stateless AI" system, meaning user queries are processed without being stored on Google's servers. This aligns with Apple's overarching AI strategy, which prioritizes on-device processing for most tasks and uses a secure "Private Cloud Compute" system for more complex requests that require off-device computation.
