“`html
TLDR:
- Apple’s AI models were pretrained on Google’s custom chips, not Nvidia’s GPUs
- Apple released a new AI system called Apple Intelligence with improved features
Apple recently revealed that its AI models powering Apple Intelligence were pretrained on Google’s TPU chips instead of Nvidia’s GPUs, signaling a potential shift in the industry. The decision comes as Nvidia’s GPUs have been in high demand for AI training but difficult to procure.
The new AI system, Apple Intelligence, was introduced with features such as a redesigned Siri, enhanced natural language processing, and AI-generated text summaries. Over the next year, Apple plans to add functionalities based on generative AI, including image and emoji generation.
In a technical paper, Apple disclosed that its models were trained on Cloud TPU clusters, highlighting the efficiency and scalability of the process. The company also mentioned that inferencing would happen on its own chips in data centers.
Apple’s move to utilize Google’s TPUs for AI training could influence other tech companies to explore alternatives to Nvidia’s GPUs. The tech giant is set to report quarterly results following this development in the AI space.
“`