If this hypothetical future is one where mixtures of experts is predominant, where each expert fits on a node, then the nodes only need the bandwidth to accept inputs and give responses — they won't need the much higher bandwidth required to spread a single model over the planet.
Future versions will get more capable and smaller, portable.
Can be used to train new types models (not just LLMs).
I assume the GPU can do 3D graphics.
Several of these in a cluster could run multiple powerful models in real time (vision, llm, OCR, 3D navigation, etc).
If successful, millions of such units will be distributed around the world within 1-2 years.
A p2p network of millions of such devices would be a very powerful thing indeed.