Will AI Destroy Humanity?
Jensen Huang, the man whose chips power almost every serious artificial intelligence system on the planet, has decided the debate about AI wiping out humanity is over. There is, he says, a 0% chance of it happening by 2030. The trouble is that his own industry, and even his own recent remarks, tell a rather less certain story.
Speaking to CBS News senior business and technology correspondent Jo Ling Kent, in an interview recorded in the days before CBS published it on Sunday, the Nvidia chief executive was unambiguous. “2030 is not going to be the end of the world,” he said. “There is 0% chance that’s going to be the end of the world.” He went further, dismissing the researchers behind the warnings rather than just their conclusions. “I believe the claims of the end of the world, stirring fear across America, and doing it by the people who are doing it makes no sense to me, so they must be doing it for ulterior reasons,” he said. He described the predictions as “not grounded in science” and added: “Scaring people is unnecessary. It is irresponsible.”
The man he did not name
Mr Huang did not need to name his target. Days earlier, Jacob Coxon, a researcher who had worked at both OpenAI and Anthropic, resigned his position at Anthropic and gave up equity due to vest within two months, publishing a multi-part message on X that has reportedly drawn more than 170 million views. In it, he warned that AI companies were “racing straight to self-improving superintelligence” and “gambling with our lives,” concluding that development was speeding up while remaining, in his words, “not under control.” Evan Hubinger, who leads Anthropic’s Alignment Science work, did not soften the alarm; he echoed it, saying the company “earnestly” believes AI could kill everyone, and separately putting the odds of AI causing human extinction within the next decade at “more than 10 per cent.” Anthropic’s own chief executive, Dario Amodei, has previously put the chance of a civilisation-scale AI catastrophe at somewhere between 10 and 25 per cent.
These are not fringe figures, still less cranks shouting from the sidelines. Mr Coxon and Mr Hubinger worked, and in Mr Hubinger’s case still work, inside one of the AI industry’s most prominent safety-focused laboratories. Mr Huang chose to treat their concern as a symptom rather than a signal, arguing that existing law should be applied to AI companies before lawmakers rush to write bespoke rules, or, as he reportedly put it, “apply that first.”
A commercial interest in the answer
None of this makes Mr Huang wrong. But it is worth being clear about where he sits in this argument. Nvidia’s revenue depends overwhelmingly on the world continuing to buy ever larger quantities of the chips that train and run these systems. Mr Huang told CBS that “our company’s success is directly connected to the safe deployment of products and services,” framing Nvidia’s commercial interest as aligned with caution. It reads just as easily the other way round. A slowdown driven by genuine safety concerns is the one outcome that would directly damage Nvidia’s business, which makes his blanket dismissal of the researchers raising those concerns rather more convenient than it first appears.
The timing adds a further layer. Mr Huang gave the interview shortly before he was due to attend a White House dinner with President Trump and Chinese President Xi Jinping, at which he has said he wants to discuss establishing shared global standards for AI development, and where he has separately lobbied for the loosening of the US ban on exporting the most advanced AI chips to China. A chief executive pushing simultaneously for faster global adoption of his product, and for the removal of the export controls that limit its sale, has every reason to want the safety debate settled, and settled in the direction of “go faster.”
His own words tell a different story
Even setting Nvidia’s commercial position to one side, Mr Huang’s own recent statements do not sit easily together. Weeks before the CBS interview, he had reportedly struck a noticeably more cautious note, suggesting that AI labs should pause or slow their work if they felt they were losing control of it. That is not the position of a man certain the risk is zero. A confident “0 per cent chance” and an earlier warning that companies should be ready to hit the brakes are hard to hold at the same time, and only one of them can really be the operating principle behind how fast Nvidia and its customers actually build.
What it means for business
For anyone running a business built on AI tools, or investing in the companies that make them, the practical lesson is not to take either extreme at face value. Mr Huang has a balance sheet that benefits enormously from confident, uninterrupted growth in AI spending. Anthropic’s most vocal critics from within its own walls have no such financial stake, and some, like Mr Coxon, have walked away from significant sums of money specifically because they judged the risk to be real. Neither side is a neutral witness, and a White House dinner invitation and a chip export licence are not evidence about the science of machine intelligence.
The verdict on the numbers themselves
A precise “0 per cent” is, on its own terms, an odd thing for an engineer to say about a technology whose long-term behaviour nobody can fully predict, including the people building it. It is exactly the kind of certainty that would be reassuring coming from a disinterested party, and considerably less reassuring coming from the chief executive of the company that sells the picks and shovels for the entire AI gold rush. Whatever the true odds of catastrophe by 2030 turn out to be, they are unlikely to be found in the confident soundbite of a man whose entire product line depends on nobody slowing down.
