Cloud TPU

Train and run machine learning models faster than ever before

View Documentation Get Started

Accelerated Machine Learning

Machine learning (ML) has the power to greatly simplify our lives. Improvements in computer vision and natural language processing help all of us interact more naturally with technology. Businesses rely on ML to strengthen network security and reduce fraud. Advances in medical imaging enabled by ML can increase the accuracy of medical diagnoses and expand access to care, ultimately saving lives.

Speed Up Machine Learning Workloads

These revolutionary Cloud TPUs were designed to accelerate machine learning workloads with TensorFlow. Each Cloud TPU provides up to 180 teraflops of performance, providing the computational power to train and run cutting-edge machine learning models. Cloud TPUs can help you transform your business or create the next research breakthrough.

On-Demand ML Supercomputing

Access powerful machine learning accelerators on demand—no up-front capital investment required. Whether your task requires Cloud TPUs for hours or weeks, you can get exactly the machine learning acceleration you need without creating your own datacenter.

Easy On-Ramp to Cloud

Because TensorFlow is open-source, it’s easy to take ML workloads you’re already running in TensorFlow and try them on Cloud TPUs. Use TensorFlow’s high-level APIs to move models across CPUs, GPUs, and TPUs with minimal code changes. If you have image classification or machine translation needs, we provide a number of models that run well on TPUs - just bring your data, download a reference TensorFlow model and train.

Access Google’s AI Innovation

Access the same accelerators used by Google to develop world-class machine learning products. Cloud TPUs are purpose-built specifically to accelerate state-of-the-art machine learning workloads, including both training and prediction.

Cloud TPU Features

High Performance
Each Cloud TPU offers up to 180 teraflops of computing performance as well as 64 gigabytes of ultra-high-bandwidth memory.
The Benefits of Google Cloud
Run machine learning workloads on Cloud TPUs and benefit from Google Cloud Platform’s industry-leading storage, networking, and data analytics technologies.
Reference Models
Solve image classification and object detection problems on Cloud TPUs with models such as ResNet-50 and RetinaNet. Try out machine translation and language modeling with the Transformer model from Tensor2Tensor.
At their core, Cloud TPUs and Google Cloud’s Data & Analytics services are fully integrated with other GCP offerings, providing unified access across the entire service line.
Connect Cloud TPUs to Custom Machine Types
You can connect to Cloud TPUs from custom VM types, which can help you optimally balance processor speeds, memory, and high-performance storage resources for your individual workloads.
Preemptible Cloud TPU BETA
Save money by using preemptible Cloud TPUs for fault-tolerant machine learning workloads, such as long training runs with checkpointing or batch prediction on large datasets. Preemptible Cloud TPUs are 70% cheaper than on-demand instances, making everything from your first experiments to large-scale hyperparameter searches more affordable than ever before.
Beta: Cloud TPUs are a Beta release. This feature is not covered by any SLA or deprecation policy and may be subject to backward-incompatible changes.