I think it will be less of a local versus cloud situation, but rather one where both complement each other. The next step will undoubtedly be for local LLMs to be fast and intelligent enough to allow for vocal conversation. A low-latency model will then run locally, enabling smoother conversations, while batch jobs in the cloud handle the more complex tasks.

Google, at least, is likely interested in such a scenario, given their broad smartphone market. And if their local Gemma/Gemini-nano LLMs perform better with Gemini in the cloud, that would naturally be a significant advantage.