Google joins competitors Amazon and Microsoft in the edge computing market, with the announcement of the launch of Edge Tensor Processing Unit (TPU) and Cloud IoT Edge.
TPU is an application specific integrated circuit, specifically designed to deal with accelerating machine learning workloads, whilst Cloud IoT Edge is Google’s edge computing platform extending Google Cloud’s data processing and machine learning to edge devices.
Cloud TPUs offer the power needed to train complex machine learning models based on deep neural networks, delivering an incredible 180 teraflops of performance with 64 GB High Bandwidth Memory (HBM). Edge TPU is a purpose-built ASIC designed to run high-accuracy AI at the edge.
Google’s new chips will be used in a range of AI applications and use-cases including predictive maintenance, anomaly detection, machine vision, robotics, voice recognition, and loads more.
Cloud IoT Edge will be used in cutting-edge new robotics hardware and applications and in industrial applications in wind turbines and oil rigs. It will run either cloud-based IoT platform Android Things or Linux OS.
Google’s Edge TPU development kit for developers will be available from October, including a system on module (SOM) that combines Google’s Edge TPU, an NXP CPU, Wi-Fi, and Microchip’s secure element in a compact form factor.