‘Run as fast as you can’: Nvidia boss rejects the idea AI safety means slowing down

Just days after some of the biggest names in artificial intelligence called for the industry to slow down, Nvidia CEO Jensen Huang arrived at Dreamforce with a rather different message. Run faster.

Speaking alongside Salesforce CEO Marc Benioff on the opening day of annual tech summit in San Francisco, Huang said safety must remain paramount as AI becomes more powerful, but rejected the idea that innovation and safety sit on opposite sides of the equation.

“Safety is paramount in a lot of ways. It’s job one,” Huang said. “However, safety is an engineering problem.”

His comments land in the middle of an unusually public debate over whether the breakneck development of frontier AI is beginning to outrun the industry's ability to control it.

Anthropic CEO Dario Amodei this month called for the industry to “pace the frontier”, arguing AI capabilities are advancing so quickly that safety systems need more time to catch up. OpenAI CEO Sam Altman subsequently backed the push, including Amodei's call for independent evaluators to have deeper access to frontier AI companies.

The debate has become significant enough to rattle investors, with AI-linked stocks, including Nvidia, coming under pressure this week amid concern that greater scrutiny could ultimately slow the extraordinary levels of investment flowing into the sector.

Huang is not arguing companies should simply throw caution aside. If a system cannot be demonstrated to be safe, he said, don't release it.

Companies should build proper test environments, assess products carefully and “pace yourself until you are confident” in what is being put into the market.

But he drew a clear distinction between that and slowing the broader development of AI.

“We don’t need any new laws. We don’t need new regulations,” Huang said. “I think innovation, speed and safe products is a false choice. You could definitely have both at the same time. So run as fast as you can.”

If at any point a company believes its technology is unsafe or getting out of control, Huang added, “take a pause and make sure you get ready”.

Two views of the same problem

The contrast at Dreamforce was difficult to miss.

Amodei had appeared with Benioff earlier in the day arguing that AI is moving at a pace even he did not expect and that businesses today are extracting perhaps only 5–10% of the value already available from existing technology.

Huang agrees wholeheartedly on the first point. He simply arrives at a different conclusion on the second.

His message to business leaders was not to wait for the technology, regulation or debate to settle down before acting.

“You don’t want to be left behind,” he said. “This is too important of a technology revolution. It is too extraordinary of capabilities. Engage the technology, learn about it.”

And if an AI system does not quite work today, he argued, try again soon.

“If you just tested it today and you felt that it did 80% of what you expected, test it again in a couple of weeks. It’s going to get much, much better. The most important thing is don’t get left behind.”

Benioff, speaking to media after the keynote, positioned himself somewhere between the two camps.

“I think that we heard both sides of the pendulum,” he said. “In my view, technology is never good or bad. It’s what you’re doing with the technology that matters, and I think that that remains the key thing.”

The responsibility, he said, ultimately falls back on the companies building and deploying it.

“These companies obviously have to be responsible for what they’re building,” he said. “That is going to be the key to them being ethical companies.”

We're all AI companies 

Huang's wider prediction is that AI will ultimately become an infrastructure layer in much the same way electricity and the internet did before it.

“With electricity, we could power everything. With the internet, we could find anything,” he said.

“And now with artificial intelligence as an infrastructure layer across the planet, we can know everything and do anything.”

From there, Huang said the distinction between an AI company and a traditional company begins to disappear.

“Every company, every enterprise, every country should become an AI company,” he said.

Nvidia itself has already undergone that transformation.

Once overwhelmingly understood as a GPU company, Huang now describes Nvidia as a “full-stack AI infrastructure company”, spanning chips, enormous computing systems, frontier models, agentic software and the secure environments required to deploy agents inside organisations.

That evolution matters because Huang believes AI has crossed an important threshold over the past six months. It is no longer predominantly something being demonstrated in laboratories. It is becoming useful.

“Companies are going to become more ambitious than ever,” Huang said. “The sky’s the limit for every industry, for every single country. So engage AI. Don’t get left behind.”