A report from The Information reveals that Apple is joining forces with Google and Nvidia to power the next generation of Siri, which now incorporates AI capabilities. Breaking away from its traditional approach of in-house development or a diversified supply chain strategy, Apple is set to rely more heavily on Nvidia's Blackwell series chips and Google's Gemini large language model for support.

Specifically, Apple will leverage Nvidia's Blackwell B200 data center chips, available through Google Cloud, to handle certain Siri queries. Built on the Blackwell architecture, these chips are tailored for AI models with trillion-parameter scales, delivering substantial enhancements in inference performance, memory bandwidth, and multi-GPU scalability. To bolster data security, Apple has activated Nvidia's confidential computing feature, ensuring data encryption during processing.

Previously, Apple's in-house Private Cloud Compute solution failed to meet expectations in terms of model processing speed, prompting the company to collaborate with Google to boost performance. Although Apple intends to maximize on-device processing for the new Siri, intricate requests will still necessitate cloud-based processing, with Nvidia chips on Google's servers providing the necessary computational resources.