Comment by ilaksh

One big issue actually is that HF has the llama.cpp team, which is (was?) arguably the biggest threat to Nvidia's dominance through it's closed CUDA software.