I think if people will people know how accessible it is to run local LLMs on their device they will consider buying devices with more memory that will be able to run better models. Local LLMs in the long run are game changers