Machine learning, deep learning are THE hot topics these days. Most machine learning tasks are handled on fast&expensive dedicated hardware. Complex structures based on larger models with lots of training data are in dare need of the any GPU cycles and local memory available. Yet, is already practical to run (infer) already trained NN models on your mobile today. The Neural Network API (NN API) - introduced for Android 8.1 - is available on some handsets already and a major step towards a new breed of applications and use cases. This talk provides an overview into the current state of machine learning/deep learning and crosses the bridge to some typical uses cases, which can be run on the mobile handset. Dedicated NN hardware will shift the scenario even further. Highly specialized and dedicated NN hardware will allow for more and more complex local models running on the handset - without the need to call back home to let the model run in the cloud.