Google, one of the best companies in the hot area of artificial intelligence, will start to let customers use their own processor for technology starting in October.
Google's TPUs or tensor processing devices accelerate AI tasks like understanding voice commands or recognizing objects in images. Today, Google lets you pay to do this kind of work on cloud computing infrastructure. However, through a program called Edge TPU announced Wednesday, Google will let programmers install the TPUs in their own machines.
"There are also many benefits to gain from real-time intelligent decision making at the point where these devices connect to the network," without waiting for a trip across the network to Google's machines, Injong Rhee, vice president of Google Clouds internet of things, in a blog post. "Your sensors become more than data collectors ̵
AI, also commonly called machine learning or deep learning and using a brain-spun technology called neural networks, is a major change for data processing. It allows people to train computers using real world data and find out patterns for themselves, like what a pedestrian in front of a self-propelled car looks or how to choose the right exposure for a sunset image.
Although Google's chips are escaping their data centers, they do not go as far as anyone. There is a run of AI chips that will work on phones and other mobile devices.
Google offers AI services on its cloud computing infrastructure so customers pay when they go to use the technology. Allowing customers to use Google's AI chips in their own computers can significantly increase the number of customers and the number of jobs that are interested in Google's AI technology. There is a good chance that they already use Google's TensorFlow software that the TPUs are designed to run.
Google's move also makes it a tighter competitor with Microsoft's Azure computing service, which plans to expand its own AI processing technology to customers through an effort called Project Brainwave.
Google TPUs will be available in a hardware module that can be connected to a computer using a PCI Express expansion slot – usually on servers – or a USB port. Interested Google Cloud AI customers can search online to access Google's AI chips.
Follow the money: How to change digital cash how we store, shop and work.
CNET en Español: Get all your technical news and reviews in Spanish.