# What is Cloud TPU? Definition and meaning

Google's custom accelerators for machine-learning workloads.

## Definition

**Cloud TPU**: Google's custom accelerators for machine-learning workloads.

In short: Google's tensor processing units.

## When it is used

Large-scale training, especially with JAX.

## Where it fits

Cloud TPU belongs to **Hardware** in the [MLOps and AI infrastructure](/topics/mlops-ai-infrastructure) topic. These are the other terms in the same group:

- **[CUDA](/glossary/cuda)**: A programming model and toolkit for running general-purpose code on NVIDIA GPUs.
- **[NCCL](/glossary/nccl)**: A library that implements fast all-reduce and other communication between GPUs.
- **[NVIDIA H100](/glossary/nvidia-h100)**: A GPU with Transformer Engine and fast interconnect, widely used for training and inference.
- **[NVIDIA H200](/glossary/nvidia-h200)**: A Hopper GPU with 141 GB of HBM3e memory.
- **[NVIDIA B200](/glossary/nvidia-b200)**: A GPU with a new architecture, larger memory and support for lower-precision formats.

## Learn more

The full guide to **[Cloud TPU](/tools/cloud-tpu)** explains how it works, when to use it and how it compares with the alternatives, with examples. The [glossary](/glossary) lists every term on the site from A to Z.

Previous term: [CLIP](/glossary/clip) · Next term: [Clustering](/glossary/clustering)
