Google Cloud TPUs v5e for large-scale AI inference

Designed to be efficient, scalable, and versatile, the new Cloud TPU v5e delivers high-throughput and low-latency inference performance.

Read more here: External Link