Apple Reportedly Nearing $1 Billion Annual Deal with Google to Power Revamped Siri

Apple is nearing a deal to pay Google approximately $1 billion annually for a custom Gemini AI model to power a revamped Siri voice assistant, according to a report from Bloomberg’s Mark Gurman. The custom model features 1.2 trillion parameters and would run on Apple’s Private Cloud Compute servers, with a spring 2026 launch targeted.

The partnership represents Apple’s largest AI licensing agreement to date. After evaluating models from OpenAI, Anthropic, and Google, Apple selected Google’s Gemini based primarily on pricing rather than raw performance, Bloomberg reported. The custom Gemini model’s 1.2 trillion parameters would mark an eight-fold increase over the 150 billion parameters powering Apple’s current cloud-based Intelligence features.

Apple will maintain control over the infrastructure by running Google’s model on its own Private Cloud Compute servers, preserving the company’s privacy-first architecture. The servers use custom Apple silicon and implement end-to-end encryption, ensuring user data remains protected during processing. Apple will also keep certain Siri functions powered by in-house models, creating a hybrid system that prevents total dependence on Google.

Neither Apple nor Google has officially acknowledged the partnership. Bloomberg reported both companies plan to keep the arrangement private, with Apple marketing the upgraded Siri as proprietary technology without promoting Google’s involvement. The approach mirrors Samsung’s integration of Google Gemini features into Galaxy AI while branding them under its own name.

Strategic Shift Following Talent Losses

The reported deal comes as Apple faces mounting pressure after losing key AI talent, including Ke Yang, who led the Siri AI search initiative before departing for Meta in October. The talent exodus has raised concerns about Apple’s ability to deliver its ambitious March 2026 Siri overhaul using only internal resources.

Apple’s Private Cloud Compute infrastructure launched as part of the company’s strategy to handle computationally intensive AI workloads while maintaining privacy guarantees. The system uses custom-built server nodes powered by Apple silicon, running a specialized operating system derived from iOS and macOS. Each node’s Secure Enclave stores decryption keys that cannot be duplicated or extracted, and the system implements stateless processing to ensure no user data persists after a request completes.

The custom Gemini model would handle tasks like summarization and complex planning that exceed the capabilities of Apple’s on-device models. Apple Intelligence currently processes personal data locally on iPhones and Macs using smaller foundation models, while routing more demanding requests to cloud infrastructure.

Launch Timeline and Market Impact

Bloomberg reported the overhauled Siri is slated for spring 2026 as part of iOS 26.4, though plans could change before release. The partnership positions Apple to compete more directly with advanced AI assistants from OpenAI and Anthropic, which have demonstrated superior performance in complex reasoning tasks.

The $1 billion annual price tag reflects the strategic importance of AI capabilities to Apple’s product ecosystem. General Motors recently announced plans to integrate Google’s Gemini across all vehicles starting in 2026, signaling broader adoption of Google’s AI technology across consumer platforms.

Apple’s decision to rely on a third-party model for core Siri functionality marks a departure from the company’s traditional approach of developing critical technologies in-house. The hybrid architecture aims to balance cutting-edge AI capabilities with Apple’s commitment to user privacy and data security.