Apple made headlines recently when it announced that its artificial intelligence models in Apple Intelligence were pretrained on processors designed by Google. This move raised eyebrows in the tech industry as it signaled a shift away from Nvidia, a dominant player in the training of cutting-edge AI models.
Apple’s decision to use Google’s Tensor Processing Units (TPUs) for training its AI models was detailed in a technical paper released by the company. This move indicates that Big Tech companies are exploring alternatives to Nvidia’s pricey GPUs, which have been in high demand in recent years.
It is interesting to note that while Apple opted for Google’s TPUs, other tech giants like OpenAI, Microsoft, and Anthropic are still relying on Nvidia’s GPUs for their AI models. Additionally, companies like Google, Meta, Oracle, and Tesla are also actively purchasing Nvidia’s GPUs to enhance their AI systems.
Meta CEO Mark Zuckerberg and Alphabet CEO Sundar Pichai have expressed concerns about the industry’s overinvestment in AI infrastructure. They acknowledge the importance of staying abreast of technological advancements to remain competitive in the market.
Although Apple did not explicitly mention Google or Nvidia in its technical paper, it did reveal that its Apple Foundation Model (AFM) and AFM server were trained on “Cloud TPU clusters.” This indicates that Apple rented servers from a cloud provider to carry out the necessary calculations efficiently and at scale.
Apple’s unveiling of Apple Intelligence comes after other companies embraced generative AI technology. The new system boasts features like a revamped Siri interface, improved natural language processing, and AI-generated summaries in text fields. Apple plans to introduce additional functionalities based on generative AI in the coming year.
Google’s TPUs, which are among the most mature custom chips for AI, offer a cost-effective solution for training AI models. Apple’s AFM on-device was trained on 2048 TPU v5p chips, while AFM-server utilized 8192 TPU v4 chips. Apple’s decision to leverage Google’s TPUs for training showcases its commitment to innovation in the AI space.
Apple’s utilization of Google’s TPUs for training its AI models marks a significant departure from the industry norm of using Nvidia GPUs. This strategic move highlights Apple’s willingness to explore alternative solutions to stay ahead in the competitive landscape of artificial intelligence. Only time will tell how this decision will impact Apple’s AI capabilities in the long run.
Leave a Reply