Reports on Tuesday cited a new research paper from the iPhone maker saying Apple (NASDAQ:) is using chips designed by Google (NASDAQ:) instead of industry leader Nvidia (NASDAQ: NVIDIA) in two key components of its artificial intelligence software infrastructure. NASDAQ:) designed chips.
Even though Nvidia is dominant, accounting for about 80% of the market with chips from Google, Amazon (NASDAQ: NASDAQ: ) and other cloud computing companies, Apple has chosen Google’s cloud infrastructure.
According to reports from Reuters and CNBC, the research paper does not explicitly state that Nvidia chips are not used, but Nvidia hardware is conspicuously absent from the description of Apple’s AI tools and infrastructure. Instead, Apple relies on two types of Google Tensor Processing Units (TPUs) to train its AI models, which are organized into large clusters of chips.
To develop AI models for iPhones and other devices, Apple uses 2,048 TPUv5p chips. For server-based AI models, Apple deployed 8,192 TPUv4 processors.
According to the Google website, Google’s latest TPU is cost-effective, priced at less than $2 per hour of use if reserved three years in advance. Originally launched by Google in 2015 for internal purposes and publicly released in 2017, these TPUs are now among the most advanced custom chips designed specifically for AI.
In contrast, Nvidia focuses on designing graphics processing units (GPUs) that are widely used in artificial intelligence applications. Unlike Nvidia, which sells its chips and systems as separate products, Google provides access to the TPU through its Google Cloud Platform, requiring customers to build software in Google’s cloud ecosystem to use the chips.
In the research paper, Apple engineers noted that it would be possible to develop larger and more complex models than those discussed using Google’s chips.
Apple will roll out part of its Apple Intelligence suite to beta users this week.
The company announced a number of new AI features at its developer conference in June, including integrating OpenAI’s ChatGPT technology into its software.
Next year, the tech giant plans to roll out AI-based generation features such as image and emoji generation, as well as an enhanced version of Siri that can access user information and perform actions within apps.
