NVIDIA’s Jensen Huang on Reasoning Models, Robotics, and Refuting the “AI Bubble” Narrative
NVIDIA’s Jensen Huang on Reasoning Models, Robotics, and Refuting the “AI Bubble” Narrative
NVIDIA’s Jensen Huang on Reasoning Models, Robotics, and Refuting the “AI Bubble” Narrative
In a forward-looking conversation at the start of 2026, NVIDIA’s Jensen Huang joins Sarah Guo and Elad Gil to reflect on the transformative advances in AI over the past year and what lies ahead. Moving beyond the hype, the discussion explores how AI is reshaping industries, economies, and global infrastructure.
Jensen Huang highlights 2025’s breakthroughs in AI reasoning, reliability, and inference token economics, emphasizing that falling compute costs—driven by hardware innovation far outpacing Moore’s Law—are enabling widespread adoption. Rather than displacing workers, AI is transforming jobs, increasing demand in fields like radiology and healthcare, while robotics helps address labor shortages in manufacturing and logistics. The future lies in specialized, vertical-specific AI models rather than monolithic 'God AI' systems, with open source playing a crucial role in fostering innovation across sectors. Advances in self-driving technology and embodied AI are accelerating, supported by a layered AI stack spanning hardware to applications. Despite alarmist narratives, AI progress has improved safety through distributed monitoring and synthetic data. Looking ahead, digital biology and multi-modal models are poised for a 'ChatGPT moment,' while U.S.-China tech interdependence remains vital for global progress. With real revenue growth and expanding infrastructure, the AI revolution is grounded in tangible momentum, not speculation.
00:24
00:24
AI is emerging as a trusted tool in various fields
04:16
04:16
AI is infrastructure because it affects all applications, companies, industries, and countries.
09:08
09:08
AI automation increases productivity and creates more jobs
15:14
15:14
China's open-source AI models are now scoring highly on benchmarks.
18:39
18:39
The advantage in AI lies in the diversity of the entire stack, not a single monolithic model
21:56
21:56
The concept of God AI is unrealistic and not likely to appear soon
27:01
27:01
Lower AI costs enable millions of AIs to monitor each other for safety
29:30
29:30
Cost of building ChatGPT-level models dropped from billions to weekend projects on non-PC systems
35:20
35:20
Training is where compute scaling translates to intelligence
37:49
37:49
Coding is likely the first AI-native app to reach a billion in ARR
43:21
43:21
Foundation models for proteins and cells will revolutionize biological understanding and design
48:51
48:51
Everything that moves will be robotic, powered by AI as a general-purpose technology.
54:11
54:11
AI is the biggest driver for sustainable energy, surpassing policy-driven initiatives.
58:49
58:49
Decoupling from China is naive; the two countries are too deeply coupled.
1:13:04
1:13:04
The next five years will be extraordinary, especially next year.
