Apple Opts for Google’s AI Chips Over Nvidia’s

In a recent development that has caught the attention of the tech industry, Apple Inc. (NASDAQ:AAPL) has been reported to utilise Google-designed chips for training its artificial intelligence (AI) models.

This strategic move is detailed in a research paper which points out that instead of using industry leader Nvidia Corporation’s (NASDAQ:NVDA) hardware, Apple turned to a rival powerhouse, Alphabet Inc.'s Google (NASDAQ: GOOGL), for its AI infrastructure needs.

Apple deployed a substantial amount of Google’s tensor processing units (TPUs) — specifically, 2,048 TPUv5p chips for its device AI model, and a staggering 8,192 TPUv4 processors for its server AI model.

Google’s TPUs are custom-built for machine learning workloads and are the tech giant’s response to the growing AI chip market. This is in contrast to Nvidia, which specializes in graphics processing units (GPUs) that also serve as a backbone for AI computational needs.

Apple's decision to employ Google’s TPUs signifies a substantial shift, where traditionally Nvidia’s GPUs have been perceived as the de facto standard for AI training and inference tasks. It highlights Apple’s confidence in Google’s hardware capabilities when it comes to AI model training.

This strategic direction was previously undisclosed until the publication of the research paper, which for the first time revealed the extent of Apple’s reliance on Google's hardware for AI. The paper further elaborates that Apple engineers believe even more sophisticated AI models could be built using Google’s chips.

In June, at its developer conference, Apple unveiled new AI features which included integrations of OpenAI's ChatGPT technology, suggesting an aggressive push into AI.

The implications of Apple's strategic hardware choice could ripple across the market, potentially influencing future partnerships and technology developments in the field of AI. This decision may signal a broader trend of tech giants opting for specialised, in-house, or alternative solutions to power their burgeoning AI endeavors.