Google working on its own chips to power AI bots: Report

During Google I/O, CEO Sundar Pichai stated that the company has been testing a new custom chip, which is specially designed for artificial intelligence and machine learning that will work with TensorFlow, a software engine that drives the company’s deep neural networks.

According to a report by Wired, Google’s aim with the new chip is, unprecedented efficiency. Also, Google will not sell its chips to other companies. This move clearly threatens commercial chip makers such as Intel and nVidia.

A custom ASIC (application-specific integrated circuit), called a TPU (Tensor Processing Unit) are the chips that Google has been using in its own data centers for almost a year. The report points out that in building its own chip, Google has taken yet another step along a path that has already remade the tech industry in enormous ways. 

Norm Jouppi, Distinguished Hardware Engineer at Google says, “We’ve been running TPUs inside our data centres for more than a year, and have found them to deliver an order of magnitude better-optimised performance per watt for machine learning. This is roughly equivalent to fast-forwarding technology about seven years into the future (three generations of Moore’s Law).”