Google introduces TPU v3.0, the next generation of machine-learning processors

During the I/O 2018 Keynote, Google announced the release of its new TPU 3.0 chip.

Following a string of demonstrations of the power of machine learning, including Smart Compose for Gmail and intelligent editing suggestions in Google Photos, Sundar Pichai took a moment to put the spotlight on the new chip that brings these possibilities to life. ‘Tensor processing units’ or TPUs are the hardware components that enable most of Google’s AI and machine learning capabilities, including AlphaGo.

These TPUs can be leased to developers though Google Cloud. The third-generation Tensor processing unit pod has 8 times the performance of the previous generation, with speeds reaching 100 petaflops. The new chips are so powerful that Google had to add liquid cooling to their data centers to compensate for the additional heat.