"Run as fast as you can": Nvidia CEO responds to those calling for a slowdown in AI

Jensen Huang, CEO of Nvidia, the market leader in AI chips, rejected the need for additional regulation of AI safety on September 15, according to Bloomberg. “We don’t need new laws—we don’t need new rules,” he said in a conversation with Salesforce CEO Marc Benioff at the Dreamforce conference (as quoted by Bloomberg).

For Huang, AI is not a new form of “alien intelligence,” as one OpenAI employee described it, but rather hardware and software created by humans, according to TechCrunch. This means, the Nvidia CEO argues, that humans and existing laws are capable of controlling AI. “Safety is an engineering problem, not a legal one,” he explained. “After all, we’re developing software. <…> It’s a complex computing system, but ultimately it’s [nothing more than] a computing system” (quoted in TechCrunch).

According to Huang’s reasoning, the free market will prevent AI developers from releasing unsafe products. “If you’re creating a product or service and aren’t confident in its functionality, capabilities, or safety, then don’t release it. That’s a completely obvious step,” TechCrunch quotes him as saying. A company should proceed without haste until it is certain it is releasing exactly what the market needs, the Nvidia CEO added. “Market forces are already at work,” the businessman emphasized.

Huang also reiterated his previous point: the choice between development speed and AI safety is a false dilemma, according to Bloomberg. “It’s entirely possible to achieve both at the same time. So run as fast as you can. But if at any point you feel that the company is getting out of control or the product won’t be safe, take a break and make sure you’re doing everything right,” TechCrunch quotes him as saying.

The day before, on September 14, Huang gave an interview to the All-In podcast on stage at the conference of the same name in Los Angeles. According to a video circulating on social media, U.S. President Donald Trump called him in the middle of the conversation, Axios reports. The Nvidia CEO put the call on speakerphone so the audience could hear the conversation.

“Robots won’t take over,” Trump told Huang. “AI won’t take over the rest of the world. This whole thing is a hoax.” The president was referring to a wave of warnings about the catastrophic risks of AI, Axios explains. The head of Nvidia replied, “We’ll make sure everyone wins in the American AI race.” Huang called predictions that artificial intelligence could bring about the end of the world “scientifically unfounded” (quotes via Axios).

Nvidia has a direct stake in keeping the AI boom going, Axios notes. According to Silicon Analysts, the company controls about 80% of the market for AI accelerators—chips for data centers. TechCrunch suggests taking a skeptical view of Huang’s position: additional regulation could prevent Nvidia from selling more and more AI systems and software.

Huang himself sees commercial value in some of the cyberthreat warnings issued by AI: in his view, they can help the cybersecurity industry, according to Axios. “Is there a better way to create demand than to create a problem?” — he argued at a Goldman Sachs conference in San Francisco. “Who wouldn’t want their product to cause a frenzy in the market, with a line stretching around the corner?” Forbes quotes him as saying.

Anthropic and other major AI labs have discussed a coordinated slowdown in development, according to Bloomberg. Anthropic CEO Dario Amodei voiced support for slowing the pace of AI development in an essay published following the high-profile departure of a company employee. The employee claimed that Anthropic and OpenAI are “putting our lives on the line”: in his view, their technologies could pose a threat to humanity. Meta CEO Mark Zuckerberg proposed another approach: to verify the safety of their models, labs should bring in independent experts and advisors, the agency reports.

This article was AI-translated and verified by a human editor