Apple recently revealed that it had chosen Google’s Tensor Processing Unit (TPU) for training its artificial intelligence models, a surprising deviation from the dominance of Nvidia in the high-end AI training chip market. This decision sheds light on the increasing competition among Big Tech companies for cutting-edge AI capabilities.
Apple’s AI system, Apple Intelligence, was detailed in a technical paper that highlighted the use of Google’s TPUs for training. The company also released a preview version of Apple Intelligence for selected devices. While Apple did not explicitly mention Google or Nvidia in its paper, it did reveal that its Apple Foundation Model (AFM) and AFM server were trained on “Cloud TPU clusters.” This approach allowed Apple to efficiently train its models, including on-device and larger models.
Nvidia’s GPUs have been the go-to choice for high-end AI training chips, but their high demand has led to procurement challenges in recent years. Apple’s decision to opt for Google’s TPUs could signal a shift in the industry, with other tech companies like Meta, Oracle, and Tesla also investing in Nvidia’s GPUs for their AI systems. This move highlights the growing competition and demand for AI infrastructure among tech giants.
Google’s TPUs offer a cost-effective solution for AI training, costing under $2 per hour when booked for three years in advance. They are among the most mature custom chips designed for artificial intelligence and have been used by Google internally since 2015. The TPUs provide a scalable and efficient way to train AI models, making them an attractive option for companies like Apple looking to enhance their AI capabilities.
Apple’s introduction of Apple Intelligence includes new features such as a refreshed Siri interface, improved natural language processing, and AI-generated summaries in text fields. The company plans to roll out functions based on generative AI, including image and emoji generation, as well as a more advanced Siri that can access personal information and perform actions within apps. This focus on generative AI aligns with industry trends and showcases Apple’s commitment to advancing its AI technology.
Apple’s decision to use Google’s TPUs for training its AI models underscores the evolving landscape of AI infrastructure among Big Tech companies. The competition for cutting-edge AI capabilities is driving companies to explore alternative solutions beyond Nvidia’s GPUs. Google’s TPUs offer a cost-effective and efficient option for training AI models, making them an attractive choice for companies like Apple. As the AI landscape continues to evolve, it will be interesting to see how companies like Apple leverage these technologies to enhance their AI systems and offerings.
Leave a Reply