Apple Admits Relying On Google's Custom Chips To Train AI Models

(RTTNews) - Apple (AAPL) used Google's Tensor Processing Units or TPUs to pre-train its artificial intelligence models, which powers Apple Intelligence, the company stated in a technical paper.

The tech giant revealed that it used TPUv5p chips to build AI models and TPUv4 processors for server AI models.

"Pre-training data set consists of … data we have licensed from publishers, curated publicly available or open-sourced datasets and publicly available information crawled by our web crawler, Applebot," Apple wrote. "Given our focus on protecting user privacy, we note that no private Apple user data is included in the data mixture."

Notably, Apple has not named Google specifically in the paper, but stated that its Apple Foundation Model and AFM server are trained on Cloud TPU clusters.

"This system allows us to train the AFM models efficiently and scalably, including AFM-on-device, AFM-server, and larger models," Apple stated.

The paper also explained how the TPUs helped in training large AI systems efficiently, and highlighted the company's ethical considerations while developing AI.

Apple had emphasized that it has not used private user data to train its AI models, and has always adhered to responsible data practices.