Nvidia CEO says he would cut off buyers who build unsafe AI

Nvidia CEO Jensen Huang stated in an interview with Bloomberg’s Emily Chang that he would stop supplying chips to a customer if they were developing unsafe AI software. However, he does not believe that current concerns about the technology are justified. In the interview conducted in London, Huang explained that this would be a hypothetical scenario and that the company would first have a conversation with the responsible party before taking such a decision.

He emphasized that Nvidia does not control the actions of other companies using their technology, and their chips are general-purpose. Nvidia is the leading supplier of artificial intelligence accelerators, which are crucial for training and running AI models. The company's dominance has placed it at the center of a debate over the potential risks of AI software growing too powerful and becoming uncontrollable.

AI executives, such as Anthropic's Dario Amodei, have suggested a temporary slowdown in AI development to address these concerns. Huang, who is on a European tour in preparation for a Berlin event, believes that AI safety is an issue of growing pains for a rapidly expanding industry. He reaffirmed Nvidia's commitment to supporting AI research and pledged a $250 million donation over five years to UK institutions, including the University of Oxford and University College London, to study AI safeguards and control mechanisms.

Huang acknowledged that some of the concerns are self-imposed and that the industry has been poor at explaining the benefits of AI. Despite some negative market reactions, Huang remains confident that the demand for AI infrastructure will outpace the supply of components for years to come, eventually leading to a balanced supply-demand equilibrium.