Jensen Huang says uncontrollable AI labs should shut down

By Splitfeed EditorialLast updated Sep 26, 2026, 19:17 UTC

On a podcast published around September 23, 2026, Nvidia CEO Jensen Huang stated that AI companies should not receive antitrust exemptions or liability carveouts, and went further to say that if a lab cannot contain its models, it should be shut down. The remarks were made publicly and in a direct enough form that they were carried by Reuters and Gizmodo on the same day, with The Economic Times and others picking them up the following morning. In a separate context, Huang also described the rental cost of a 1-gigawatt AI factory as roughly $50 billion per year, a figure he said equals what it costs to build one.

The significance of these remarks comes partly from who is saying them. Nvidia supplies the chips that power the AI industry, making Huang one of the most commercially central figures in the field. When the CEO of a company that profits from AI expansion publicly opposes regulatory relief for AI developers and sets a concrete threshold, lab containment failure, as grounds for shutdown, it cuts against the lobbying posture that many AI firms have adopted in Washington and Brussels. The antitrust and liability questions are live policy debates, not abstract ones, and Huang's position adds a notable voice to the side favoring accountability over exemption.

The Pro-AI camp has not yet reacted publicly to this story, though it would typically argue that heavy regulation stifles innovation and that Huang's shutdown threshold, however well-intentioned, could be applied arbitrarily in ways that harm competitive development. The Anti-AI camp has also not weighed in, but would generally be expected to welcome Huang's containment language while questioning whether voluntary industry standards or self-policing would ever be sufficient to enforce it. The Middle Ground camp, which has reacted, frames Huang as occupying a careful space: accounts including BizToc noted he rejected the view that AI poses an existential threat to humanity even as he insisted on accountability, while Know Mouth pointed to the underlying structural question of whether safety responsibility belongs to companies, regulators, or both. Impact Newswire highlighted Huang's acknowledgment that AI could erode some basic cognitive skills, adding texture to his nuanced position.

The debate is unlikely to resolve without concrete legislative movement on the liability and antitrust questions Huang addressed. Any formal regulatory proposal, committee vote, or agency rulemaking on AI liability carveouts in the United States or the European Union would serve as a practical test of how much weight remarks like his carry in actual policy outcomes.