What are the two new Google Cloud AI chips announced?

Google Cloud announced two new chips: the TPU 8t for model training and the TPU 8i for inference.

What performance improvements do Google's new TPUs offer?

The new TPUs offer up to three times faster AI model training and 80% better performance per dollar.

How do Google's new AI chips relate to Nvidia's hardware?

Google's new TPUs are designed to supplement, not replace, existing Nvidia-based infrastructure, and Google will continue to offer Nvidia's latest chips.

Home / Technology / Google Splits AI Chips for Training and Inference

Google Splits AI Chips for Training and Inference

23 Apr

•

Summary

Google Cloud launched new TPUs for training and inference.
New chips offer up to 3x faster AI model training.
Google's chips supplement, not replace, Nvidia's infrastructure.

Google Splits AI Chips for Training and Inference

Google Cloud announced its eighth generation of custom-built AI chips, known as tensor processing units (TPUs), will be offered in two specialized versions. The TPU 8t is designed for AI model training, while the TPU 8i is optimized for inference, the process of using trained models with new data.

These new TPUs reportedly deliver substantial improvements over previous generations. Google claims they provide up to three times faster AI model training and an 80% enhancement in performance per dollar. Additionally, the infrastructure can support over one million TPUs working together in a single cluster, promising increased compute efficiency with reduced energy consumption and costs for customers.