Google TPU v5e
Chip
AI Accelerator
Shipping
The efficiency TPU: cost-per-token optimized inference and mid-scale training.
You're looking at a chip record: what this processor is, its key specs, who designs, manufactures and supplies its parts, and where it gets used.
The v5e trades peak performance for economics — Google's public positioning was performance per dollar, with pods of 256 chips aimed at serving and fine-tuning rather than frontier training.
Specifications & key facts
Each figure is labeled by basis: vendor spec, vendor claim, reported, estimate, or official. "Not publicly disclosed" means exactly that — we never invent values.
Overview
Designer
Google
Announced
August 2023
Memory
Memory
16 GB HBM2, 819 GB/sVendor spec
Compute
BF16
197 TFLOPSVendor spec
INT8
394 TOPSVendor spec
Physical
Pod scale
256 chipsVendor spec
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Designed by
Available from
Succeeded by
Sources
Google Cloud TPU documentation ↗
Record reviewed as of 2026-01-15 — see Methodology.