The amount of compute required to develop artificial intelligence models could increase by a factor of 10 because of rigorous safety evaluations, Jensen Huang stated. During a discussion on The Ezra Klein Show, the Nvidia chief argued that laboratories such as OpenAI or Anthropic should be shut down if their researchers believe their models are not safe or if the organizations have no way to contain their experiments.

Huang rejected the premise that external regulatory pressure is needed to slow the pace of development, asserting that executives have the power and responsibility to halt unsafe launches. He noted that firms are already incentivized to prioritize safety to avoid the loss of customers and the threat of civil or criminal lawsuits stemming from negligence.