Jensen Huang Says Nvidia Has 'Achieved AGI' — On His Own Terms

Summary

In an episode of Lex Fridman's podcast released on Monday, March 24, 2026, Nvidia CEO Jensen Huang made a statement that immediately stirred controversy across the tech industry: "I think we've achieved AGI."

The declaration came in the context of a question from Fridman about how long it would take AI to innovate, find customers, and manage a team to build a $1 billion company. When asked whether that milestone was five to twenty years away, Huang asserted that for a company of that scale, the era of AGI is already here.

The statement is notable — and contested — because it rests on an explicit redefinition of the concept. Huang is not claiming that AI has reached general human-level reasoning; he is saying that AI can build and run a ten-figure enterprise, and that this is sufficient to declare AGI achieved.

In practice

To illustrate his position, Huang used a concrete example: "It is not out of the question that a Claude model was able to create a web service, some interesting little app that all of a sudden, a few billion people used for 50 cents, and then it went out of business again shortly after." He compared this type of company to what was seen in the internet era — simple websites that achieved fleeting viral success — and argued that today's AI models are already capable of doing the same.

Huang's logic is strictly capitalistic: AGI does not mean, for him, an AI that reasons like a human in a broad sense, but rather an AI capable of executing and monetising a business idea autonomously, even if that monetisation is temporary and driven by virality.

However, Huang himself acknowledged the limits of this view when confronted with the case of Nvidia. Asked whether AI agents could eventually run a company as complex as his own, he was direct: "The odds of 100,000 of those agents building Nvidia is zero percent." He added that the number of software engineers at Nvidia will grow, not decline — because what he wants from his engineers is to solve problems, not to write a certain number of lines of code.

Context

The tech industry has struggled for years with a precise definition of AGI. Google DeepMind defines it as an AI capable of performing most cognitive tasks at least at the level of a skilled adult. OpenAI and Microsoft have a contractual agreement that considers AGI achieved when an AI generates a certain level of profit. Neither definition is consensus.

Huang's claim has a clear strategic dimension: by declaring that AGI has already been achieved, the CEO of the company that manufactures the chips powering virtually all major AI systems reinforces the narrative that demand for high-end hardware will continue to grow indefinitely. If AGI is "now," then companies like Google and Microsoft need to keep scaling their data centres — and they need Nvidia's products to do so.

Analysts and commentators have been sceptical about the substance of the claim. Huang's definition reduces AGI to the ability to monetise temporary virality, which represents a specific and narrow type of success, far removed from the general and sustained reasoning most researchers associate with the concept. As coverage of this event notes, the redefinition conveniently aligns the AGI milestone with the current state of model capabilities — and with Nvidia's commercial interests.

Why it matters

• Jensen Huang's declaration is the highest-profile claim yet from an industry leader publicly asserting that AGI has been achieved, even if under a highly particular and contested definition.
• The choice of a capitalistic metric — autonomous management of a billion-dollar company — rather than a cognitive one represents a meaningful shift in the public debate on AGI, with implications for regulation, investment and social expectations.
• The tension within Huang's own statement — simultaneously claiming AGI has arrived while insisting Nvidia will still need more human engineers — illustrates the real limits of current AI model capabilities.
• For Nvidia, declaring AGI "already happened" is also a positioning move: it sustains the narrative that investment in AI infrastructure is not speculative, but a response to a reality that is already here.