soon, important tasks and functions will operate directly on device…with some fine-tuning a number of these models will enable many creative and on-device applications.
 
doesn’t hurt to 💭🤏

Apple will fit many LoRa adapters into devices, with the base models. The vector indexer will have multi-modal capabilities, allowing it to fine-tune texts, images, videos, and usage behavior daily to provide a more creative experience.

tasks and functions = draw in wattage

which really begs the question, which tasks & functions can be displaced to fog compute, and what capabilities/radio range exists in sparser areas?

tough to predict anything until robotics makes another leap.