Nvidia CEO Jensen Huang has stated that the company would cut off customers developing unsafe AI, though he believes current concerns about AI safety are overblown. Speaking to Bloomberg in London, Huang described the scenario as hypothetical but emphasized that Nvidia would not hesitate to halt chip supplies to customers engaged in risky AI development. He noted that while Nvidia does not manage other companies, it has already implemented strict vetting processes, such as an approved-buyer list in Singapore, Malaysia, and Japan, which led to more than half of previous customers being excluded.
Huang argues that the fears surrounding AI are largely self-imposed and suggests accelerating the development of containment and monitoring technologies as the solution. Nvidia is building a safety team to vet AI agents before deployment, focusing on trust as the key to broader adoption. He reiterated his stance from a previous event, Dreamforce, where he opposed calls for new AI regulations, advocating instead for companies to proceed rapidly unless a product becomes unsafe.
This perspective contrasts with that of Dario Amodei, who has pushed for an industry-wide slowdown with shared standards. Meanwhile, Nvidia is investing $250 million in UK research to study AI safeguards, with institutions like Oxford and University College London involved. The company also plans to announce $2 billion in European deals within a week, following discussions in Paris, London, and Berlin.