Nvidia CEO Jensen Huang publicly rejected the AI slowdown movement championed by Anthropic's Dario Amodei and backed by OpenAI's Sam Altman — calling Trump onstage at the All-In Summit to make the point as loudly as possible.
Key takeaways
Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier," calling for deliberately slower AI development and third-party safety audits; Sam Altman and Google DeepMind's Demis Hassabis loosely aligned with his position.
Nvidia CEO Jensen Huang told President Trump at the All-In Summit that "we're not going to let [an AI slowdown] happen," putting him in direct opposition to the Amodei coalition.
The split matters practically: Nvidia supplies the GPUs that power nearly every frontier model, giving Huang real leverage over the pace of development regardless of what lab CEOs agree to.
Critics — including analysts cited by Ars Technica — argue the slowdown pact could function as a competitive moat, locking in incumbents while blocking new entrants.
Trump and House Speaker Mike Johnson have both dismissed the slowdown push, framing it as industry overreach.
The fracture at the top of the AI stack
The weekend's events began with Amodei's essay, which argued that the risks of rapidly scaling large language models — from economic disruption to existential danger — justify a deliberate pause in frontier development. Altman and Hassabis signaled agreement, and the three represent the labs closest to the current capability frontier. That alignment looked, briefly, like a rare industry consensus.
Then Huang walked onstage at the All-In Summit and took a live speakerphone call from Trump. According to TechCrunch, Huang told the president directly: "We're not going to let [an AI slowdown] happen." Trump, who has consistently framed AI deceleration as ceding ground to China, was a receptive audience.
Nvidia CEO Jensen Huang takes a live call from President Trump at the All-In Summit, publicly opposing the AI slowdown push.
The optics matter as much as the substance. Huang isn't just a chip vendor — Nvidia's H100 and Blackwell GPUs are the physical substrate on which Anthropic's Claude, OpenAI's GPT series, and Google's Gemini all run. When the person who controls compute supply says there will be no slowdown, that's not a PR statement; it's a structural reality.
What the cartel argument actually says
The Verge notes that skeptics immediately flagged an ulterior motive: the three lab CEOs who signed onto "pace the frontier" happen to lead the three organizations best positioned to benefit if frontier development slows. A coordinated deceleration would freeze the competitive landscape in their favor, making it harder for smaller labs or open-source projects to close the gap. MIT Technology Review called it the industry's "doomer turn" while questioning whether safety is the real driver.
For creators who rely on frontier image and video models — whether through APIs or platforms built on top of them — the Huang-versus-Amodei fracture is the more consequential story than any single safety essay. If Nvidia continues to ship next-generation silicon on its current cadence and labs keep training on it, the practical effect of the slowdown pact is limited. New models will arrive; capabilities will expand.
The more immediate risk is regulatory: if Trump's alignment with Huang produces federal policy that preempts state-level AI safety rules, the compliance environment for every tool you use could shift within months. Conversely, if the Amodei coalition succeeds in establishing voluntary — or eventually mandatory — capability thresholds, the gap between what frontier models can do and what gets released to consumers could widen.
Either way, the next few months of model releases will be a cleaner signal than any executive statement. Creators tracking which labs ship and which ones pause will have a better read on who actually won this argument than any summit appearance can provide. The Charmloop model catalog is one place to watch that in real time.
