Jensen Huang argues AI progress needs guardrails, not a halt

The brief

Jensen Huang joins the panel to challenge AI doomerism without dismissing safety, arguing that frontier labs need better controls, engineering, measurement, and evaluation.

The discussion moves from recursive self-improvement to open and closed models, with Huang presenting broad participation and practical deployment as strategic advantages.

Huang says winning the AI race means exploiting models effectively, while automation should redirect engineers toward more ambitious work rather than simply eliminate jobs.

A surprise live call from Donald Trump interrupts the Nvidia business discussion, turning data centers, jobs, AI leadership, and opposition to stopping the industry into an onstage political exchange.

Afterward, Huang describes Nvidia as an infrastructure builder coordinating capital, power, land, supply chains, data centers, and regional cloud providers.

The closing argument is expansive: specialized forms of AGI or superintelligence may already exist, and America should move quickly while developing AI prudently.