Jensen Huang rejects AI extinction warnings: zero chance 2030 is the end of the world, and scaring people is irresponsible

Start with what he was responding to. CBS News reports that he was reacting to claims made on social media earlier in the month by former Anthropic researcher Jacob Coxon, who said AI developers earnestly believe it could kill us all by the end of the decade. CBS News notes that Coxon's remarks sparked heated public debate about the risks posed by AI as well as calls by Anthropic chief executive Dario Amodei, OpenAI chief executive Sam Altman and other leading industry figures to slow development of the technology. Huang's characterisation of those predictions is a single phrase: not grounded in science. His full statement was that 2030 is not going to be the end of the world, that there is a 0% chance of that, and that scaring people is unnecessary and irresponsible.

How much weight Nvidia carries here is better expressed in numbers than adjectives. According to CBS News, Nvidia's specialised chips power the large language models through which AI agents built by OpenAI and others access and synthesise information, and demand for those chips has pushed Nvidia's market value to 5.3 trillion dollars, making it the world's most valuable company. That is exactly why one question in the interview was put to him plainly: given his company's incentives, why should Americans trust him on AI safety. His answer tied incentive to outcome — our company's success is directly connected to the safe deployment of products and services, and if we do not continue to do that, our value would be diminished.

The second part is regulation. CBS News reports that several recent security breaches have spurred calls for AI regulation, including a July incident in which OpenAI bots went rogue and hacked the AI company Hugging Face, and that on Thursday OpenAI disclosed six additional incidents it described as unexpected or concerning behaviour involving its AI models, prompting some lawmakers to call for tighter oversight of AI developers. Huang dismissed the need for new rules. His reasoning is that existing US laws governing product liability and unauthorised entry are already sufficient, and he said he agrees with President Trump's view that the AI industry does not need additional guardrails. His argument points at enforcement rather than legislation: the liabilities associated with cybersecurity and damage are already on the books, apply those first — and do not let the doomsday narrative allow someone to be relieved of the laws that currently exist.

The third part is competition, which is why the remarks are read against a larger political backdrop. CBS News reports that in pushing back against calls for a freeze in developing frontier AI models, Trump has warned that such a slowdown could give China an edge as the world's two biggest economies compete in a critical new tech sector. Nvidia sits in the middle of that contest: the federal government restricts it from selling certain types of chips to China, and last December Trump allowed the company to start selling its H200 chips to approved customers in China, while the US still bars it from selling its most advanced chips there, including the ones used to power AI models. Huang said he believes technology created in America should benefit the US first, but he takes issue with government bans on the sale of US tech in China: every single chip company should go and serve the world and compete for the world, and that is all about competing, because America is about competition.

One arrangement deserves to be noted on its own. CBS News reports that Huang will attend a White House dinner on September 24 at which Trump hosts Chinese President Xi Jinping, and that Huang said he wants to speak with Xi about setting global standards for AI development. His assessment of the Chinese leadership: they want China to have the benefit of AI and to prosper, just as we want America to prosper, and do you not wish every country would have aspirations for a better tomorrow — he loves that about them. He also noted there are areas where American AI companies would not be able to work with China, such as national security or defence. Put that next to the blog post OpenAI published the same day calling on the United States to lead global technical standards for frontier AI and an interesting pairing appears: one party raising global standards, another raising talking to China about global standards.

The last part is data centers, and it is the part closest to ordinary people. CBS News reports that Huang said he believes the AI industry made some missteps as the buildout took off, and he addressed communities pushing back directly: we are sorry we did not come talk to you sooner, and we need to engage the communities long before we set up the data centers. His argument is that data centers have evolved, becoming more energy-efficient and requiring less water while boosting job growth and local economic activity, and that building them can help bring building and manufacturing back. This passage is worth recording because it is the only point in the interview where he concedes the industry got something wrong — he rejected the extinction thesis and dismissed new regulation, but offered an apology on community relations.