Reported 3 months ago
Apple has chosen to employ Google's tensor processing units (TPUs) instead of Nvidia's processors to train two key AI models, as revealed in a recent research paper. This decision signifies a strategic shift toward leveraging Google's cloud infrastructure, particularly with examples of using 2,048 TPUv5p chips for mobile AI applications and 8,192 TPUv4 processors for server models. While Nvidia dominates the AI chip market, Apple indicated potential for more advanced models using Google's technology, marking an important development in its AI ecosystem.
Source: YAHOO