If an iPhone user sends a command to Siri, where is this information processed or where does speech recognition take place? Answer: Until recently this voice stream was processed in the computer centers of Apple (here: Mesa / Arizona). Speech processing required at the time quite some computational power, so this process didn't happen on the mobile device itself.
However, as chips on mobile phones get ever more powerful, data storage gets cheaper and AI-algorithms get ever faster the voice stream processing is shifted to the smartphone itself. It's called edge computing. If data processing is decentralized (or also: in the network periphery), then this is called edge computing or fog computing.
Actually, we'll see many more (business) applications shift to the cloud in the next years; Public Cloud Service Providers enjoy two-digit growth rates at the moment and this growth rate will keep up for a while. However, for reasons of data protection, data processing on the edge will remain important (or even: get more important).