Jensen Huang: 2027 AI bottleneck is 35GW of power, revenue justifies the capex
BG2 Pod · rss · 2026-10-15
Jensen Huang joins Brad Gerstner and Clark Tang for a third BG2 conversation, recorded days after the White House State Dinner and his National Medal of Science.
- 2027 bottlenecks: 50 GW of chips vs only 35 GW of power — power constraints will shape AI's next phase
- Capex question: argues AI revenue justifies the spending and that demand extends far beyond just two AI labs — "too much pressure on two labs" is unsustainable
- Open vs closed models and NVIDIA's stance on openness
- Singularity & cost of intelligence: uses Navier-Stokes as an example of falling intelligence costs, and what to do with abundance
- Physical AI: robotics as the next wave; even floats SpaceX putting data centers in orbit
- Safety: an "airbags" analogy for building fast but responsibly; explains America's "three-legged stool" superintelligence framework
- Business: Vera Rubin Ultra orders, Europe/Japan expansion, and a review of Wall Street consensus
- Closes 99.99% optimistic about the decade ahead
More from Companies & People
- A 6-person team shipped healthcare prior-auth AI agents in 3 months — here's the playbook — alex_verem · 2026-10-15
- Scale AI exec: data companies profit because frontier labs don't own the product surface — sarahookr · 2026-10-15
- Energy company laid off programmers after adopting Claude agents, dad reports — Ok_Boat_3855 · 2026-10-15
- OpenAI kicks off first DevDay Exchange of the year in Bengaluru — paw_lean · 2026-10-15
- Dan Shipper: run product and labs in parallel, even with a labs team of one — danshipper · 2026-10-15
- AI Safety PhD Quits in Final Year After 15 Papers at Technion — IgorCarron · 2026-10-15