Infrastructure
Tensor Processing Unit
TPUA chip designed by Google specifically for machine learning, as an alternative to general-purpose GPUs.
A TPU is custom hardware built from the ground up for the kinds of math machine learning needs, rather than adapted from graphics like a GPU. Google developed TPUs to train and run its own models efficiently and offers them through its cloud. They are one example of the broader move toward purpose-built AI chips as demand for compute keeps climbing.
Reference: Wikipedia: Tensor Processing Unit
Related terms
Graphics Processing Unit
The specialized chip that does the heavy parallel math behind training and running AI models.
Compute
The raw processing power used to train or run a model, a key driver of cost and capability.
Inference
Actually running a trained model to get an answer, as opposed to training it in the first place.