In a surprising development, Apple has unveiled that it utilized Google’s custom-designed chips to train its sophisticated AI models, which are pivotal to its Apple Intelligence system. This revelation, disclosed in a research paper released on Monday, signifies a major shift in the landscape of Big Tech’s reliance on AI technology. Traditionally, Nvidia’s graphics processing units (GPUs) have dominated this sector, controlling around 80% of the AI chip market.
According to the research, Apple used two distinct versions of Google’s proprietary tensor processing units (TPUs) to meet its AI training requirements. Specifically, Apple deployed 2,048 TPUv5p chips for training AI models intended for iPhones and other consumer devices. In contrast, for server-side AI models, the company utilized 8,192 TPUv4 processors. This dual use of Google’s TPUs highlights Apple’s strategic move to leverage cutting-edge technology beyond the widely used Nvidia GPUs.
The significance of this move cannot be overstated, given Nvidia’s longstanding dominance in the AI chip market. Nvidia’s GPUs have been in high demand due to their crucial role in AI development and the rapid expansion of AI technologies. However, Apple’s decision to employ Google’s TPUs indicates a shift in the competitive dynamics of the AI chip industry.
In addition to this announcement, Apple offered a glimpse into its eagerly awaited Apple Intelligence system. This system is poised to introduce a range of groundbreaking features and tools, as showcased during Apple’s recent WWDC conference. Among the exciting developments is the integration of OpenAI’s ChatGPT into Apple’s software ecosystem. This feature is expected to be rolled out to select beta users within the coming week, further enhancing the functionality of Apple’s AI offerings.
Recently, several leading tech companies, including Google, Meta, Microsoft, and Amazon, have begun to develop their own AI chips. This shift is driven by the high costs associated with training extensive AI models. Google’s TPUs, which are accessible through the Google Cloud Platform, require users to build their software on Google’s cloud infrastructure to fully utilize these advanced chips.
The research paper also highlights that Google’s chips have the potential to support even more complex and expansive AI models beyond those currently in use. This potential underscores the growing competition in the AI chip market and signals a new era of innovation and collaboration among major tech giants.
Overall, Apple’s strategic use of Google’s TPUs not only illustrates a shift in the AI chip market but also emphasizes the evolving landscape of technological collaboration and competition. As companies continue to push the boundaries of AI technology, developments like these will play a crucial role in shaping the future of the industry.
