Ironwood, Google's seventh-generation Tensor Processing Unit (TPU), is the first AI accelerator designed for inference and the most scalable and performant
Ironwood is the most powerful, competent, and energy-efficient TPU. It is also built for large-scale inferential AI models
Ironwood AI Chip uses Google's Pathways software stack to let developers easily and consistently use tens of thousands of Ironwood TPUs
Ironwood AI Chip supports "thinking models," including complicated reasoning tasks, Mixture of Experts (MoEs), and Large Language Models, in communication and computing
Ironwood AI Chip delivers massive parallel computing capacity for the hardest AI tasks, such as ultra-dense LLM or MoE models with thinking skills for training and inference
Ironwood includes an upgraded SparseCore accelerator for ultra-large embeddings, which are common in advanced ranking and recommendation workloads
Pathways, Google DeepMind's machine learning runtime, enables distributed computing across TPU devices
Google Cloud, which integrates AI computing into Gmail, Search, and other services for billions of users everyday, is the only hyperscaler with over ten years of experience enabling cutting-edge research
AI applications are more cost-effective due to performance advancements and power economy. Ironwood has twice the performance per watt of Trillium, a sixth-generation TPU introduced last year
Advanced liquid cooling methods and improved chip architecture can retain up to twice the performance of conventional air cooling under constant, heavy AI workloads. Compared to the 2018 Cloud TPU, Ironwood AI Chip
Ironwood AI Chip enhanced processing power, memory capacity, ICI networking innovations, and dependability make it a singular breakthrough in the age of inference