Introduces interview with Nvidia CEO Jensen Huang after CES keynote, noting analysts pressed him on China and overall demand.
John Ford
Jensen Huang
Nvidia works with every AI company globally through an open strategy, avoiding proprietary lock-in to collaborate across all domains of science.
Asks what makes this moment different for robotics and physical AI, specifically about humanoid robots.
John Ford
Jensen Huang
Timing is everything - we've been waiting for the enabling technology moment, similar to other breakthroughs like real-time ray tracing that took 30 years.
Jensen Huang
Computers don't care what tokens they generate - language, video, or physical actions. The breakthrough came when seeing text-to-video models generate realistic physical actions like picking up a cup.
Asks about Rubin chip performance vs cost and value, referencing charts showing Hopper-Blackwell-Rubin progression.
John Ford
Jensen Huang
AI factories produce tokens/numbers and serve three purposes: 1) Train next frontier models faster (4x leap with Rubin), 2) Generate tokens cost-effectively (10x cost reduction), 3) Increase factory throughput (10x increase).
Jensen Huang
In AI factories, going faster equals going cheaper. Faster completion costs less.
Jensen Huang
The value is unquestionably incredible. Nvidia achieved this with only 1.7x more transistors through co-design innovation across CPU, GPU, networking, switches, and data processing.