Apple is reportedly closing in on a deal worth around $1 billion annually to licence Google’s Gemini AI model to power its long-delayed next-generation Siri voice assistant, following months of speculation.

According to Bloomberg, the agreement would give Apple access to Google’s 1.2 trillion parameter Gemini system to run key Siri capabilities such as summarisation and task planning. The custom Gemini model would mark a significant upgrade from Apple’s current 150-billion-parameter system.

Apple plans to host the Gemini model on its Private Cloud Compute servers, keeping user data isolated from Google’s systems. The company has already set aside dedicated AI server hardware to support the new model.

The iPhone-maker tested several AI systems including OpenAI’s ChatGPT and Anthropic’s Claude before opting for Google’s Gemini as an interim solution while it develops its own 1 trillion parameter cloud-based model. The in-house system could be ready for consumers next year.

Apple and Google reportedly declined to comment.

Unlike earlier talks to plug Gemini into Siri as a chatbot, the current deal does not extend to integrating Google’s AI search or chatbot directly into Siri.

The move is part of a broader effort to overhaul Siri’s foundations and integrate it with the company’s wider AI-powered Apple Intelligence platform.

In an August staff call, CEO Tim Cook told employees that Apple had to lead the AI revolution, adding that a more personalised version of Siri is slated to launch next year. The enhanced voice assistant is tipped to debut with iOS 26.4 next spring, Bloomberg reported.

Apple’s Siri overhaul has faced persistent setbacks since it first teased the upgrade last year. In March, Siri chief Robby Walker reportedly lamented the delays to new AI features, stating the company was still unsure when the revamped voice assistant would launch.