8G VRAM for LLM, are you sure? I thought you need way more, 20GB++ Nvidia doesn't want peasants running own LLMs locally, 90% of their business is supporting AI

8G VRAM for LLM, are you sure? I thought you need way more, 20GB++ Nvidia doesn't want peasants running own LLMs locally, 90% of their business is supporting AI bubble with a lot of GPU datacenters