Nvidia CEO suggests chipmakers could stop supplying customers developing unsafe or irresponsible AI
finance.yahoo.com ● Covered by 5 sources
Jensen Huang says Nvidia would stop selling chips to a customer if its AI was clearly unsafe. That’s a big line from the company supplying the picks and shovels for the whole boom.
Based on reporting by finance.yahoo.com — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Nvidia chief Jensen Huang told Bloomberg the company would cut off a customer if it knew the chips were being used to build harmful AI. He framed it as a simple line: if an application of Nvidia’s technology is bringing harm to people, the answer is yes, stop selling. But he also said the company would first have a conversation, because Nvidia does not run other people’s businesses.
The interview, in London with Emily Chang, was as much a defense of the AI industry as a warning label. Huang said the industry’s message has gotten “out of control,” but argued that this is just what growth looks like when young companies scale quickly. He compared the current moment with the internet’s early years and said the sector is only now growing into the accountability that comes with building powerful tools.
He pushed back hard on the idea that safety and capability are separate debates. In his view, they are the same conversation. That means more work on containment technology, monitoring systems, and AIs that watch other AIs, not less development. Safety, he said, is paramount. So is speed.
Huang also said the market story around AI has changed fast. He described Anthropic and OpenAI as the two fastest-growing technology companies in history, with real revenue and real profit from that revenue. He said the economics are improving too, claiming the cost of a token is falling by 30 to 40 times a year, which makes the technology cheaper for customers even as demand for compute keeps rising.
The broader point was clear enough: Nvidia wants to be seen as both supplier and referee, but only up to a point. Huang spent a lot of time talking about sovereign AI, control of data, and the need for countries and companies to avoid a denial of service from critical infrastructure. That sounds neat. It also sounds exactly like the kind of argument a chipmaker makes when everyone else is worried the boom is getting too big, too fast.
My take — AI-written commentary, not fact-checked reporting
This is the classic Silicon Valley move: build the engine, then announce a conscience once the smoke gets noticed. Nvidia is right that safety can’t be an afterthought, but “we’d stop selling” is easy to say when the money is still flowing. The real test is whether chip suppliers ever use that power before the scandal is already on the front page.
Read more about this at: finance.yahoo.com
