Apple is shifting its AI strategy. The company plans to use Nvidia chips to power a new version of Siri based on Google's Gemini model. This marks a departure from Apple's usual approach of controlling core technology in-house.
Apple will deploy Nvidia Blackwell B200 GPU clusters via Google Cloud.
According to a report from The Information, Apple will deploy Nvidia Blackwell B200 GPU clusters via Google Cloud. Some user queries for the updated Siri will be processed on Google Cloud using a licensed version of Gemini. The arrangement is part of an agreement between Apple and Google.
The Blackwell B200 offers significant improvements in inference performance, memory bandwidth, and multi-GPU scalability over the previous Hopper architecture, according to Nvidia. Apple will also enable Nvidia's confidential computing for data encryption, ensuring user privacy during processing.
How Apple's Private Cloud Compute will integrate remains unclear.
How Apple's own Private Cloud Compute system will integrate with future Siri products remains unclear, the report notes. The company had previously announced its own server infrastructure for AI tasks, but the partnership with Nvidia and Google suggests a hybrid approach.
This move represents a strategic pivot for Apple. Historically, the company has favored developing its own chips and services to maintain tight control over user experience and data. Partnering with a rival like Nvidia for core AI hardware is unusual for Apple.
The Information's report was also covered by 9to5Mac. No pricing or availability details for the new Siri features have been disclosed. Apple has not officially commented on the report.
What is confirmed: Apple will use Nvidia Blackwell B200 chips via Google Cloud for Gemini-powered Siri queries. The deal includes confidential computing for encryption. The integration with Apple's Private Cloud Compute is still uncertain.
Discussion
0 comments
Log in to join the thread with a thoughtful take, question, or correction.