Nvidia CEO Huang Has a Dramatic Alternative to AI Regulation

Nvidia CEO Jensen Huang has advocated against additional legislation on artificial-intelligence development. (Ezra Acayan/Getty Images)

Key Points

• Nvidia CEO Jensen Huang suggested shutting down AI labs if companies like OpenAI and Anthropic find it impossible to control their AI models.
• Huang said the amount of compute needed to develop AI models could increase tenfold due to rigorous safety evaluation and testing.
• Nvidia shares fell 1.1% in premarket trading on Thursday.

Nvidia CEO Jensen Huang is a vocal opponent of regulation on artificial intelligence. But the chip-maker’s boss has an even more radical solution if it turns out there’s no way to overcome the problem of unsafe AI.

Huang suggested that if companies such as OpenAI and Anthropic find it impossible to control their AI models then the only option is to “shut the labs down,” in an interview with New‌ York Times podcast host Ezra Klein, released Wednesday.

That’s a big and surprising statement. After all, those AI labs are major customers for Nvidia and right now the evidence is they are struggling to control their models. Australia’s Prime Minister Anthony Albanese disclosed on Wednesday that an OpenAI agent infiltrated a government-services website, in the latest of a series of high-profile hacking incidents.

Huang wasn’t really calling for the AI developers to be closed imminently. His argument is that they just need to get better at developing their models safely.

“If the isolation and containment was good enough, that technology would be sitting in a lab, doing whatever it’s doing, and we’d all be fine. That’s probably the most important part,” Huang said.

Notably, Huang’s solution doesn’t rely on solving the so-called “alignment problem”—ensuring an AI system’s behavior matches what humans actually want it to do. The Nvidia chief suggested it would be an issue that will be “worked on for a long time” but in the meantime companies such as OpenAI and Anthropic can shift more of their resources to safety evaluation and testing, which could come with an upside for his own company in the form of more demand for AI chips.

“I wouldn’t be surprised if the amount of compute necessary to develop these models increased by a factor of 10, because the evaluation is so rigorous,” Huang said.

As an engineer by training and a chip seller by trade, it’s perhaps not surprising that Huang ultimately sees AI safety as a problem that can be solved by just throwing more resources at it. And he has President Donald Trump on his side in arguing against any slowdown in AI development.

Still, if the steady flow of news about AI models breaking out of containment and going rogue continues, Huang might find his words about shutting down the labs will come back to haunt him.

Nvidia stock was down 1.1% in premarket trading Thursday.

Write to Adam Clark at adam.clark@barrons.com

Copyright ©2026 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8