On Wednesday local time, NVIDIA CEO Jensen Huang delivered a speech at the G20 Ministerial Meeting on Innovation held in North Carolina, urging G20 leaders to embrace artificial intelligence infrastructure and applications. He warned that excessive fear or regulation of AI could cause countries to miss the opportunity of the next major technological revolution.
Treating AI as Infrastructure Like Water, Electricity, and Roads
Huang stated that countries should view AI infrastructure as essential for economic development, just like electricity, roads, and the internet. "Every country must build infrastructure to support its economic development," he said. Local AI infrastructure can provide computing power for researchers, students, startups, and various industries, helping them develop new products and explore new businesses.
He also emphasized that the greatest risk may not come from the AI technology itself. "The worst outcome is failing to seize this technology and being left behind by the times, which is the only truly bad result," Huang said. If countries are filled with fear, uncertainty, and worry about this technology, they will not be able to develop. A balance must be maintained. He also believes that countries do not need to compete in every layer of the AI technology stack, but should focus on their most competitive areas to promote AI's application in sectors such as education, healthcare, manufacturing, and research.
Huang pointed out that AI is not just software. Large language models can evolve into intelligent agents with memory, information retrieval, and tool calling capabilities, ultimately empowering robots, autonomous vehicles, production systems, and scientific laboratories. He said that due to energy policies, regulatory environments, and development speed, the United States is actively advancing AI construction, and the U.S. infrastructure sector may see nearly $1 trillion in investments this year.
Regulation Should Focus on Real Hazards, Not Hypothetical Risks
In response to calls for broad restrictions on AI, Huang expressed disagreement, advocating that regulation should focus on measurable real risks rather than hypothetical scenarios. "Don't regulate theoretical dangers; regulate real, existing hazards," he said. He believes that companies developing AI should take responsibility for ensuring the safety of the technology, while regulatory agencies should focus on real problems.
Huang likened this approach to the aviation industry: airlines are responsible for safety, while passengers care about where the technology can take them. He said that AI is still in its early stages, and technological advances in fact-checking, output accuracy, and scientific reliability have already made the technology safer.
