Data center energy optimization: with bigger firm like Google, Facebook etc. the data center size and complexity of information it receives from various data source, which might be from an individual user to various sensor data from vivid devices increase dramatically.
Deep learning algorithm are used in energy optimization of data center, to identify the various set of scenarios that a human might miss with such abundant data in first place and when we add various combinations, it make the problem interesting.
Let me explain the scenarios I am talking about, consider the example of monitoring the heat inside the server rooms or efficient energy usage, we need to cool it off efficiently. With complex and sophisticated machinery being interacting with each other, environment tend to change non linearly. External factor like weather, response time expected from machinery may vary based on time of peak user usage hours etc. this adds up to the conclusion that it would be difficult to deal with such situation using traditional engineering model which might involve hard-coded factors to it.
These process are an apt problem to be taken care and solved by deep learning algorithm. Identifying a preset of sensor data that can be used to identify vivid patterns to take appropriate actions is the way solve the problem now.  Google is already taking care of energy usage in its  data center using deep learning algorithm more precisely by feeding huge amount of sensor data its had to the DeepMind AI . I am not sure of the numbers but it’s a figure of 30–40% reduction in bill for google data center. I consider this as an important application of Deep learning when it comes to data center, simple fact is data is important but now so do as our energy resources.
Other emerging application using deep learning for gathering and cleansing data,  data modelling and analysis : interesting example is comparison of data files by deep learning algorithm, in case of huge amount of data deep learning algorithm will be able to find correlation between files more efficiently. More importantly wont miss out and would be able to cover a wider range or scope.

I would like to add for “Speech recognition? Image processing?” which is able to thrive by the usage of deep learning? Answer is:  As of now Image processing is leading the race. However we can see that recently deep learning algorithm have changed the way we interact with our mobile phones specifically Android, there has been a huge improvement to speech processing in android device.

There might be others example too, If you can add-on please comment below I would like to hear.