Google goes after IHVs with Cloud IoT Edge

Google’s Cloud IoT Edge hardware-software package for edge devices, announced on July 25, aims to be a comprehensive bundle for the edge ― for devices and for gateways. In this offering, Google leverages its two big assets in machine learning, TensorFlow software and the tensor processing unit (TPU) processor, to stake a position in edge hardware and software.

TBR believes the edge is the leading edge of Internet of Things (IoT) growth. There is competition for both edge hardware and edge software, but few vendors can offer both. There will be consolidation in hardware and software, and the companies left standing will have large and growing businesses and opportunities to expand. In the case of Google, as well as Microsoft and Amazon, capturing the edge helps drive the core cloud offering. By staking a big claim on the edge, Google is better positioned to compete with the other big clouds.

TensorFlow and the TPU processor are the keys to Google’s offering. TensorFlow is one of the most popular machine learning software libraries while the TPU processor is optimized for machine learning. Google claims advantages of the TPU over GPUs for machine learning tasks include lower power consumption and better performance on inference as well as learning tasks. These two benefits, power consumption and inference performance, are critical on the edge. Power consumption is important in edge devices, especially mobile and remote devices. Machine learning training is best suited to the cloud; edge devices need fast inference.

Google is targeting this offering to companies making IoT hardware, devices and gateways, ranging from narrowly specialized to broadly applicable, from custom-built to off the shelf. Companies producing off-the-shelf products are independent hardware vendors, and their offerings range from components for IoT solutions to end-to-end hardware and software solutions. Google’s Cloud IoT Edge is attractive to this market; it is a hardware-software solution with differentiating hardware and familiar software.

In the enterprise market for custom-built devices, Microsoft will often leverage its incumbency. However, there remain many market opportunities, especially in off-the-shelf smart devices with built-in machine learning. Video is a likely market for this technology, and Google will continue to make it easier and less expensive to build smart cameras.

Google’s Cloud IoT Edge is a well-conceived response to the challenge of the edge, and there is potential additional upside. The new Edge TPU is very small, and Google claims very low power consumption. Google will introduce tools and applications that leverage the processor to provide tangible benefits on smartphone, tablet and PC platforms. If successful, Google could own the IP to be a necessary component of edge computing.