On October 14, 2026, Jensen Huang said NVIDIA could hypothetically stop supplying a customer developing unsafe AI—but he would speak with the customer first. The NVIDIA CEO also said the company’s technology is general-purpose and that it has limits on judging how customers use it. Bloomberg reported the exchange.
What Huang said about a possible cutoff
Huang’s answer described a possible response, not a decision to stop shipments. He said NVIDIA does not run its customers’ businesses and that its general-purpose technology can be used in different ways.
How the answer fits Huang’s AI safety approach
The hypothetical fits a broader position Huang had already described publicly. On September 15, 2026, he told CNBC that safety and testing are engineering problems. He argued that new AI-specific laws or antitrust waivers were unnecessary if companies properly engineered and tested products before release, according to CNBC’s report.
In an earlier CNN interview, Huang recommended containing AI agents in secure environments and monitoring their activity. He also said products judged unsafe should not be released. Those are his proposed controls; the remarks do not turn the hypothetical supply cutoff into an NVIDIA enforcement policy. CNN’s interview with Huang records his recommendations.
For a separate industry discussion, NeoTeo previously covered Sam Altman’s comments linking a possible OpenAI IPO to model safety.
