Probably because every large LLM provider is trying to develop their own silicon to run their models.
Probably because every large LLM provider is trying to develop their own silicon to run their models.