“I think we’ve achieved AGI.”

That was Jensen Huang on the Lex Fridman podcast this Monday. The CEO of Nvidia—a company now worth roughly $4 trillion on the back of the AI hardware boom—had just declared the arrival of artificial general intelligence.

Then he explained what he meant. And the explanation revealed more about the state of the AI industry than the claim itself.

The Billion-Dollar Technicality

Fridman had offered a specific threshold for AGI: an AI system capable of starting, growing, and running a technology company worth more than $1 billion. He asked whether that milestone was five, ten, or twenty years away.

“I think it’s now,” Huang replied.

His reasoning was precise in a way that undercut the headline. “You said a billion,” Huang told Fridman, “and you didn’t say forever.”

The scenario Huang described wasn’t an AI building the next Nvidia. It was an AI creating a viral app—“some social application that, you know, feeds your little Tamagotchi or something”—that generates a billion dollars in value before fading into irrelevance. He cited the dot-com era as precedent, noting that most of those websites were no more sophisticated than what AI agents can produce today.

Then came the caveat. “The odds of 100,000 of those agents building Nvidia,” Huang said plainly, “is zero percent.”

That’s not a small qualifier. It’s the entire distinction.

The Moving Goalpost

This isn’t Huang’s first attempt at defining AGI. In 2023, at the New York Times DealBook Summit, he defined it as software capable of passing tests that approximate normal human intelligence at a competitive level. He predicted AI would clear that bar within five years.

Now he’s saying it’s already here—provided you define “here” as “a flash-in-the-pan app that monetizes briefly and dies.”

The through-line isn’t a consistent theory of machine intelligence. It’s a consistent pattern of defining the threshold in whatever way makes “yes, we’re there” the easiest possible answer.

Huang isn’t alone in this game. The industry has spent the past year backing away from “AGI” as the term accumulated too much baggage—too many unmet promises, too much skepticism from investors who’ve watched capital burn rates climb into the stratosphere. Tech leaders have started inventing new terminology, phrases they insist are less hyped and more useful, even as they describe essentially the same thing.

Why Definitions Matter

The semantic game isn’t academic. It has real money attached to it.

Microsoft’s partnership with OpenAI contains what insiders call “The Clause”—a provision that voids the companies’ agreement if OpenAI achieves AGI. The contract defines AGI as “a highly autonomous system that outperforms humans at most economically valuable work,” with a separate threshold of “sufficient AGI” tied to generating roughly $100 billion in investor returns.

According to Wired, Sam Altman has suggested AGI could arrive soon, while Microsoft has reportedly been skeptical that AGI is close to being realized. The vagueness of the definition gives OpenAI’s board significant leverage to decide when the partnership terms no longer apply.

In other words: billions of dollars hinge on what these words mean.

The Timing of the Claim

Huang’s declaration comes at a particular moment for Nvidia. The company faces intensifying competition in the AI chip market, regulatory scrutiny on multiple continents, and the perpetual pressure of justifying a valuation that makes it one of the most valuable companies in history.

A CEO claiming that the technology his company powers has reached its ultimate milestone serves obvious purposes. It reinforces Nvidia’s centrality to the AI boom. It shifts the conversation from “when will this pay off” to “we’re already there.”

But the claim only works if you don’t examine it too closely. Huang’s definition of AGI—an AI that can create a billion-dollar app once, without the ability to build something lasting or manage complex organizations—is so narrow that it’s essentially a tautology. If a billion-dollar viral hit counts as AGI, then the threshold for “general intelligence” has been lowered to “successful marketing campaign.”

The Opening Argument

Treat Huang’s claim as an opening statement, not a verdict. The CEO of a $4 trillion company doesn’t misspeak by accident on a podcast. He chose his words carefully, and he chose his definition more carefully still.

The real story isn’t whether AGI has arrived. It’s that the most powerful figure in AI hardware felt compelled to argue that it has—using a definition so specific that it renders the term nearly meaningless.

When a CEO has to shrink the milestone to fit inside it, the milestone probably hasn’t been reached.

Sources