Ironwood Tensor Processing Unit (TPU)

Ironwood: Google Cloud's 7th Generation TPU Engineered for Inference

Train, fine-tune, and serve larger models and datasets efficiently with the most powerful TPU yet.

Request more information about Ironwood.

First name

Last name

Business email

Calling Code

Business phone

Job title

Company name

Sign me up to receive news, product updates, event information, and special offers about Google Cloud from Google.

I understand my personal data will be processed in accordance with Google's Privacy Policy.

Is your infrastructure ready for the age of inference?

Ironwood is Google’s most powerful, capable and energy efficient Tensor Processing Unit (TPU) yet, designed to power thinking, inferential AI models at scale. Building on extensive experience developing TPUs for Google's internal services and Google Cloud customers, Ironwood is engineered to handle the computational and memory demands of models such as Large Language Models (LLMs), Mixture-of-Experts (MoEs), and advanced reasoning tasks. It supports both training and serving workloads within the Google Cloud AI Hypercomputer architecture.

Optimized for Large Language Models (LLMs): Ironwood is specifically designed to accelerate the growing demands of LLMs and generative AI applications.
Enhanced Interconnect Technology: Benefit from improvements to TPU interconnect technology, enabling faster communication and reduced latency.
High-Performance Computing: Experience significant performance gains for a wide range of inference tasks.
Sustainable AI: Ironwood continues Google Cloud's commitment to sustainability, delivering exceptional performance with optimized energy efficiency

Ironwood integrates increased compute density, memory capacity, and interconnect bandwidth with significant gains in power efficiency. These features are designed to enable higher throughput and lower latency for demanding AI training and serving workloads, particularly those involving large, complex models. Ironwood TPUs operate within the Google Cloud AI Hypercomputer architecture.

Cloud AI products comply with our SLA policies. They may offer different latency or availability guarantees from other Google Cloud services.