Apple explores using Google Gemini AI to support its revamped Siri, Alphabet rises 3.8%

On Friday local time, according to media reports, Apple is in talks with Google for a collaboration, engaging in preliminary discussions to explore the use of Google's Gemini AI to support its new version of the Siri voice assistant. This marks a key step for Apple in outsourcing artificial intelligence technology. Following the news, Google's stock price rose by 3.8%.

Recently, Apple has been in contact with Google, a subsidiary of Alphabet, to discuss the possibility of creating a custom AI model for the new version of Siri expected to launch next year. Google has begun training a model that can run on Apple's servers.

This effort is part of Apple's attempts to catch up in the field of generative AI. Apple started late in this area and has struggled to make breakthroughs. Earlier this year, Apple also explored collaborations with Anthropic and OpenAI to assess whether Claude or ChatGPT could become the "new brain" for Siri.

Apple still has several weeks before making a final decision, and it has not yet determined whether to continue using its self-developed model or to choose an external partner. Spokespeople for both Apple and Google declined to comment.

This potential shift stems from delays in Siri's upgrade plans. The upgrade was originally scheduled for release this spring, aiming to execute commands by calling on users' personal data and allowing users to control devices entirely through voice. However, due to engineering setbacks, the release has been postponed by a year.

After the setbacks, Apple excluded AI chief John Giannandrea from the Siri development process. The project is now led by software chief Craig Federighi and Mike Rockwell, the creator of the Vision Pro headset, who are considering breakthroughs through external technology.

Apple's software chief Federighi has expressed interest in external models. The initial design for the Siri upgrade was based on technology from Apple's foundational model team, which is also developing the large model that drives "Apple Intelligence" for device-side tasks such as text summarization and custom emoji generation.

To fix Siri's shortcomings and push delayed features into production, Federighi, Rockwell, and the corporate development team led by Adrian Perica have begun discussions with Anthropic and OpenAI to explore potential collaborations.

Currently, Apple is conducting an internal "competition," simultaneously developing two versions of the new Siri: one codenamed Linwood, based on Apple's self-developed model; the other codenamed Glenwood, which runs on external technology