With 96gb of VRAM, will you really need to? Personally I think apple servers might be adopted for AI and LLM workloads soon.