Has anyone developed a fog computing-based AI solution yet? I'm surprised that nobody has come up with the idea of having computers share resources to run an LLM.