NVIDIA CEO Jensen Huang’s definition of AGI is telling

Artificial general intelligence, or AGI, has become the AI ​​industry’s favorite buzzword over the past year or so. With the industry’s leading companies burning through capital at historic rates, racking up energy costs and investor expectations becoming increasingly difficult to meet by the quarter, the promise of imminent human-level machine intelligence has become a useful thing to have in your pocket.

Whether we’re actually close to this milestone depends almost entirely on how you define it. It turns out that this definitional flexibility represents a lot of work.

Take, for example, Jensen Huang, CEO of NVIDIA – a company currently valued at around $4 trillion, built largely on the GPU hardware that’s fueling the AI ​​boom – who recently sat down with podcaster Lex Fridman for a wide-ranging conversation covering data centers, geopolitics, and whether AGI is here yet. Huang thinks so. The reasoning behind this claim is quite dubious, however.

As Fridman points out, Huang has previously stated that the timing of the IGY depends on what defines it. At the New York Times DealBook Summit in 2023, Huang defined AGI as software capable of passing tests that approximate normal human intelligence at a reasonably competitive level. He expects IA to pass this mark within five years.

For his part, Fridman offered Huang a generous definition to work with: True AGI, in Fridman’s framework, would look like AI capable of starting, growing, and running a tech company worth more than a billion dollars. He asked whether this would be feasible in the next five to 20 years, given the recent proliferation of agentic AI tools like OpenClaw.

Huang didn’t need five to twenty years. “I think it’s now. I think we’ve reached the AGI,” he replied to Fridman.

This, however, relies on a narrow interpretation of what Fridman asked. According to Huang, AI doesn’t need to build anything sustainable. There is no need to manage people, run a board, or maintain a business. You only need to reach a billion dollars once.

“You said a billion,” Huang told Fridman, “and you didn’t say forever.”

In either case, the guideline does not constitute a coherent theory of artificial intelligence. It’s a consistent pattern of setting the threshold in whatever way makes “yes, we’re there” the simplest possible answer. His illustration of what this might look like is revealing.

After his initial response, Huang lays out his thoughts, describing a scenario in which an AI creates a simple web service – an app that goes viral, is used by a few billion people at 50 cents a pop, and then silently folds. He then cites the Internet era as precedent, arguing that most of these websites were no more sophisticated than what an AI agent could generate today.

Huang was also candid about the ceiling of this vision. “The chances of 100,000 of these agents building NVIDIA,” he said plainly, “is zero percent.” This is no small warning. That’s the whole ball game.

What Huang actually describes – a viral app that briefly monetizes and dies – is a far cry from the transformative, economy-reshaping AGI that dominates public debate. So, by its own admission, the kind of complex institutional intelligence needed to build something like NVIDIA is not yet available.