Apple employs Google’s chips for AI model training

Apple employs Google’s chips for AI model training is profiled by BTW Media because published evidence links it to internet infrastructure, governance, operational dependencies, or market visibility.

• Apple has decided to use Google’s chips to develop two key constituent parts of its artificial intelligence software infrastructure.
• This collaboration highlights the application of Google’s tensor processing units (TPUs) for developing key AI components, as Apple rolls out its new AI suite.

OUR TAKE
Apple’s decision to use Google’s TPUs instead of Nvidia’s GPUs for AI model training marks a significant shift in its AI infrastructure strategy, leveraging Google’s tailored AI chips. This move highlights Apple’s aim to diversify its hardware reliance and explore alternative technologies amidst Nvidia’s market dominance.
-Vivienne Xie, BTW reporter

What happened

According to a research paper published on Monday, Apple has opted to use chips designed by Google, rather than those from industry leader Nvidia, to develop two essential components of its artificial intelligence (AI) software infrastructure. The decision to rely on Google’s cloud infrastructure is notable given Nvidia’s dominance in the AI processor market.

Nvidia, known for its highly sought-after AI processors, commands roughly 80% of the market alongside chips from other cloud computing companies like Google and Amazon. Despite Nvidia’s prominence, Apple’s research paper did not mention the use of Nvidia hardware. Instead, it underlined its reliance on Google’s tensor processing units (TPUs).

To train its AI models, Apple utilised two types of Google’s TPUs, organised in large clusters. Specifically, Apple used 2,048 TPUv5p chips for the AI model intended for iPhones and other devices, and 8,192 TPUv4 processors for its server AI model. Unlike Nvidia, which focuses on graphics processing units (GPUs) sold as standalone products, Google offers access to TPUs through its Google Cloud Platform, requiring customers to build software within this platform to use the chips.

Also read: Apple commits to AI safety in White House Initiative
Also read: Amazon develops AI chips to challenge Nvidia’s market leadership

Why it’s important

Apple’s move is significant as it represents a shift away from Nvidia’s GPUs, which are widely used for AI applications. Google tailors TPUs for specific AI tasks, making them a viable alternative for companies like Apple.

This week, Apple is rolling out portions of its new AI suite, Apple Intelligence, to beta users. Monday’s research paper fully disclosed the extent of Apple’s reliance on Google’s hardware for the first time, despite Reuters reporting the use of TPU chips in June. Both Google and Nvidia declined to comment on the matter.

Apple engineers noted in the paper that the entity could develop even larger and more sophisticated models by using Google’s chips. This announcement follows Apple’s unveiling of new AI features at its June developer conference, including the integration of OpenAI’s ChatGPT technology into its software.

Following this revelation, Apple’s stock saw a slight decrease of 0.1%, closing at $218.24 on Monday.