Palantir and Nvidia build US-controlled AI system, raising sovereignty fears abroad

Palantir and Nvidia have developed an air-gapped AI system that allows US government agencies to run advanced AI models entirely on American hardware. The setup keeps all data and model weights under Washington's direct control, with no external access. Shortly after the system's unveiling, Palantir published an 'AI sovereignty' manifesto warning that relying on a foreign power's AI infrastructure risks ceding national autonomy. The message carries a pointed implication for other countries considering adopting US-built AI platforms. Nations now face a strategic dilemma over whether building on American AI infrastructure could expose them to a potential technological kill switch.