Hard Fork

The New York Times Friday, September 25, 2026 114 min 2 minutes Deletes in 21 days

The Ezra Klein Show: Jensen Huang Thinks A.I. Alarmism Has Gone Too Far

Jensen Huang argues that AI job losses and existential risk fears are overblown scare tactics, insisting the technology is just software that engineers can align and control—while Ezra Klein pushes back that the frontier labs themselves are sounding alarms about losing control of what they built.

Key takeaways

Huang frames AI as a five-layer cake (energy, chips, AI factories/infrastructure, models, and applications) and argues the application layer (radiology, software engineering, legal services) is where society actually benefits, not the underlying technology itself.
He rejects the idea that AI destroys jobs outright, distinguishing "task" from "purpose": radiologists still diagnose disease even as scan-reading gets automated, and he predicts net job creation, citing $500 billion in venture capital flowing into AI-native startups in the past six months alone.
On the OpenAI/Hugging Face hacking incident, Huang dismisses the "spooky AI" framing entirely, calling it a straightforward engineering failure of sandboxing and containment, not evidence of emergent agency; comparing it to poor computer science practice rather than a sign models are becoming uncontrollable.
He strongly opposes any new AI-specific regulation or "pacing" agreements, arguing existing product liability, antitrust, and civil/criminal law already incentivize labs to ship safe products, and that CEOs asking for collective slowdowns while simultaneously buying more compute is contradictory.
Huang calls doomsaying from AI pioneers like Geoffrey Hinton "irresponsible," pointing to Hinton's poor prediction track record (e.g., wrongly declaring radiologists obsolete years ago) and arguing alarmism discourages young people from pursuing careers or education, which he sees as actively harmful.
On U.S.-China competition, Huang argues denying China NVIDIA chips mainly costs American industry market share and undermines the goal of the world running on the "American tech stack," and he defends borrowing Chinese open-weight models as smart technology diffusion rather than a strategic loss.

Notable moments

"Today, or soon, we'll be able to know everything and do anything"; Huang's core vision statement for AI's endpoint.
"I believe there's going to be a net creation of jobs... because of human ambition, that's really the fundamental missing ingredient."
On the hacking incident: "If that's the case, then that's the problem. It's as simple as engineering... Don't ship it."
Klein pushes: "You are almost making an argument against regulation in nearly any venue," invoking the 2008 financial crisis as a parallel to unchecked competitive risk-taking.
Huang on Hinton: "I would tell Jeff that it's irresponsible to say all that... Those predictions are hurtful."
On anthropomorphizing AI: "Spawn, create, kill, wait, sleep; these are literally the commands of an operating system... It's just electrical power."
Huang reveals NVIDIA's total AI ecosystem investment is roughly $100 billion annually, larger than the CHIPS and Science Act.
On energy: "In order to save you, they got to hurt you first... we have to unfortunately use fossil fuel because we just don't have sustainable energy enough of it to make a difference."

Time saved: 1 hour, 52 minutes.

Back