Intel and AMD both need to make LLM inference tools plug n play with their GPUs.

Otherwise Nvidia is the default choice when selecting a GPU for local LLM inference.

Aurora is working with OpenAI and Azure which is a very positive step in this direction.