Nvidia CEO Jensen Huang urges US, China collaboration on AI safety

Jensen Huang has a message for Washington and Beijing: keep talking. The Nvidia CEO told reporters in London that the US and China need to cooperate if artificial intelligence is going to advance safely.

Coming from the man whose chips power much of the AI boom, that is a notable stance. It is also a pointed one, given that Nvidia currently sells almost nothing to China because of US export rules.

What Huang said in London

The remarks came during a media briefing in London held October 14-15, 2026. Huang described AI as a technology with two faces: enormously powerful and, in his words, “potentially dangerous.”

His proposed fix was not a new rulebook. It was conversation.

“Dialogue, dialogue.”

That repetition was Huang’s way of stressing that more engagement with Chinese researchers is essential. He argued that talking across borders beats building walls around the science.

Huang also pushed back on the idea of new regulations or pauses in AI development. He suggested the industry managing itself, and collaborating across borders, would work better than government-imposed brakes.

He framed the US-China AI relationship as both competitive and collaborative. The two countries are racing each other, but Huang believes their researchers still need to compare notes on safety.

The China revenue problem

Huang’s call for cooperation lands awkwardly against Nvidia’s actual business in China. The company has faced US export restrictions on advanced AI chips to China since mid-September 2026.

Huang’s guidance to investors was blunt. He advised them to model Nvidia’s revenue from China as zero.

Still, he drew a line between commerce and science. Nvidia’s sales prospects in China may be limited, but Huang said personal and research ties with Chinese counterparts continue to flourish.

For scale, Huang referenced Nvidia’s market capitalization at approximately $5.5 trillion during discussions in September.

Why this is happening now

Huang’s comments arrive amid rising global concern about AI safety. Recent incidents involving rogue AI agents have sharpened the debate over how much autonomy these systems should have, and who keeps watch over them.

The US and China have also been holding bilateral discussions on AI risks. Talks between the two countries on AI safety were planned for mid-September 2026, the same window in which the chip export restrictions took hold.

Nvidia has also been positioning itself on safety. In late September 2026, the company announced its Open Agent Safety Platform, a tool designed to help monitor AI agents.

A different tune from other AI labs

Huang’s position stands apart from figures at other major US AI labs. Some of them have called for stricter regulation and even pauses in AI development.

Huang is effectively arguing the opposite. His view: slowing down or adding rules is less useful than keeping research channels open, including with China.

What this means

For investors, the clearest signal is the zero-revenue guidance on China. By setting expectations at nothing, Huang removes a source of potential disappointment. Any future easing of export rules would then show up as upside rather than a recovery from a forecast miss.

The Open Agent Safety Platform is worth watching as a test case. If tools like it gain adoption, they strengthen the argument that the industry can handle safety without heavy-handed rules.

Then there is the diplomatic track. The bilateral AI risk discussions between Washington and Beijing are where Huang’s call for dialogue either finds a home or gets ignored. Nvidia cannot sell advanced chips into China right now, but its CEO is clearly betting that the conversation about AI safety is too important to be cut off along with the shipments.