Why it matters
TPU is Google's AI hardware. Understanding shapes GCP AI strategy.
Advertisement
The architecture
MXU (matrix mult unit): systolic array.
HBM.
Fast inter-chip via ICI (torus).
Advertisement
How it works end to end
Pods: many TPU chips + fast interconnect.
JAX: preferred framework. PyTorch via XLA.
Availability: GCP only.
Cost: competitive with GPU for large training.