Kai-Fu Lee: Enterprise AI Adoption Should Not Wait Amid Safety Debate

Kai-Fu Lee, CEO of 01.AI, said in a Bloomberg interview on The China Show that frontier AI requires prudent exploration, but enterprises should not pause their AI transformation because of safety concerns. He suggested that if the latest models are already at the "tenth generation," while many companies are still at the "second generation," the priority is to put existing AI capabilities to use.

Lee speculated that U.S. tech leaders calling for slower frontier AI development may have produced very powerful and concerning technologies in their labs. He argued that regulators need to move faster, and that companies should discuss responses and strengthen self-discipline before formal rules are in place. Chinese AI companies, he noted, are more focused on open-source models, applications, and commercialization.

Lee also predicted that GPUs will mainly be used for inference rather than training, and that inference demand will keep growing rapidly. He said enterprise AI transformation should be a "No. 1 engineering project" led by the CEO, and that 01.AI is building an AI decision-making hub for enterprise leaders. While U.S. companies currently lead in AI adoption, he said the picture could change if AI is proven to improve corporate profits.

On agents, Lee said they cannot assume responsibility because they lack self-awareness and intrinsic motivation, so clear boundaries are needed. They should not replace CEOs in major business decisions. He also stressed that AI must move beyond peripheral tasks and into core business operations.