Huang Tells Trump on All-In Stage That AI Slowdown Will Not Happen
Nvidia CEO Jensen Huang received a phone call from President Donald Trump while Huang was on stage at the All-In Summit. Huang put the call on speaker so summit attendees could hear the conversation. The appearance was in Los Angeles, as reported Sept. 14, 2026. TechCrunch The New York Times
The call interrupted a discussion about the pace of AI capability work. Huang had been speaking with All-In hosts Chamath Palihapitiya, Jason Calacanis, David Friedberg and David Sacks about a call by Anthropic CEO Dario Amodei to slow the pace of improving AI capabilities. Trump called as that topic was under discussion.
Trump said AI work must be done prudently but added "that doesn't mean we're going to stop an industry" and "We're going to lead." He also said slowing AI could play into the hands of political people or China and described it as a hoax that "we're not going to let" happen.
Huang answered, "You're right. We're not going to let that happen, sir," as the crowd applauded. The reply was brief. The crowd response was audible.
Two other chief executives have taken a different public position. SpaceX CEO Elon Musk and OpenAI CEO Sam Altman have publicly said they agree with Amodei's call to slow AI capability improvements. Huang, separately, has said he believes AI will increase productivity and create jobs rather than eliminate them. Reuters
The broader context here is a split among AI leaders about the speed of capability work, not about whether the work continues. One side treats slower capability gains as prudent management. The other treats continued capability gains as necessary for productivity and competitive position. For practitioners, that distinction matters because it shapes release cadence, safety review, deployment controls and procurement expectations.
In my view, the onstage format compressed a complicated engineering and policy question into a binary choice. Capability improvement covers pretraining, post-training, evals, tooling and production operations. Each can be paced separately. A broad commitment to avoid a slowdown still leaves open how teams allocate effort across those layers, how they test frontier behavior and how customers absorb new function.
Looking at what this means for technical teams, the near-term signal is continuity. Roadmaps that assume steady model iteration, larger deployment footprints and tighter integration into workflows are unlikely to pause. That puts weight on internal discipline. Versioning, rollback, access controls, monitoring and incident response become the practical place where prudence is expressed, even while capability work proceeds.
In this author's view, public alignment between a sitting president and a vendor chief executive does not settle the underlying tradeoff. Productivity gains and job creation are outcomes to be measured in hiring, wages, output per hour and task automation, not asserted on a stage. The optimistic case remains plausible over the long arc, and technology has often expanded the scope of work even as it changed specific roles. The work ahead is to instrument that claim carefully.
