Apple Reportedly Plans to Use Nvidia Chips for Gemini-Powered Siri
▼ Summary
– Apple will use Nvidia’s Blackwell B200 data center chips in Google Cloud for some Siri queries.
– The setup will employ Nvidia’s confidential compute feature, which encrypts data during processing.
– Apple’s use of Nvidia chips through Google Cloud diverges from its usual strategy of controlling key product components.
– It remains unclear how Apple’s existing Private Cloud Compute system will integrate with the upcoming Siri product launch.
– The Blackwell B200 is a flagship Nvidia GPU designed for large-scale AI training and inference, succeeding the Hopper architecture.
A fresh report from The Information offers new insight into the technical workings of Apple’s upcoming Gemini-powered Siri, revealing how the company plans to use Nvidia chips for certain AI queries. The details shed light on Apple’s evolving approach to artificial intelligence and cloud infrastructure.
Just days ago, The Information reported that Apple has an agreement with Google, under which some Siri queries will run on a licensed version of Google’s Gemini model inside Google Cloud. Now, the publication adds that Apple has recently approved the use of a privacy technology from Nvidia in that setup, indicating the company will rely on Nvidia AI hardware for at least part of its cloud computing needs.
Specifically, Apple will tap into Google’s fleet of Nvidia Blackwell B200 data center chips, according to people familiar with the matter. The company will also enable Nvidia’s confidential compute feature, which encrypts data while it is being processed on the chips.
The Nvidia Blackwell B200 is a flagship GPU designed for large-scale AI training and inference. Built on Nvidia’s Blackwell architecture, the successor to Hopper, it offers significant improvements in inference speed, memory bandwidth, and multi-GPU scaling. Nvidia markets Blackwell as a platform capable of running and training trillion-parameter AI models.
The confidential compute feature is a hardware-based security system that protects data during active processing on Nvidia GPUs. According to Nvidia, it “preserves the confidentiality and integrity of AI models deployed on Rubin, Blackwell, and Hopper GPUs,” enabling “sensitive AI workloads to run securely at scale with near-native performance, even in shared or cloud environments.”
The Information notes that Apple’s decision to use Nvidia chips through Google Cloud marks a shift from its usual strategy of controlling all critical components of its products. The report also raises questions about how Apple’s own Private Cloud Compute server system will fit into the upcoming Siri launch.
You can find the full report from The Information at the link provided.
(Source: 9to5Mac)
