“Jensen Huang asserts that NVIDIA's platforms provide the lowest cost per token for AI inference in the world by orders of magnitude.”