Chamath argues NVIDIA's GPU architecture is uniquely dominant for the compute-bound prefill phase of LLM inference and that Jensen Huang is operating at a level of industry insight unmatched by anyone other than Elon Musk, reinforcing NVIDIA's long-term leadership in AI infrastructure.
He is operating at a level of insight about what's happening in this industry that I've really never seen with other folks in other industries other than Elon in his industries. It's a level of mastery and a level of skill that is really” ⚑