Ray Lucchesi of RayOnStorage Blog comments:
At Google IO conference this week, they revealed (see Google supercharges machine learning tasks …) that they had been designing and operating their own processor chips in order to optimize machine learning. They called the new chip, a Tensor Processing Unit (TPU). According to Google, the TPU provides an order of magnitude more power efficient machine learning over what’s achievable via off the shelf GPU/CPUs. TensorFlow is Google’s open sourced machine learning software.
When it comes to machine learning, hardware is still a necessity to do some of the really complicated things fast.
Read more at: TPU and hardware vs. software innovation (round 3)