Nvidia CEO declares AGI achievement then calls it senseless

In an investor call for Nvidia on Wednesday, chief executive Jensen Huang announced that the company had “achieved AGI”, one of the industry’s most proclaimed goals, and in the same breath described the milestone as “senseless”. He offered no precise definition or benchmark, but said the models have moved from responding to simple prompts to autonomous agents that learn new skills and improve themselves recursively. According to Huang, the crucial point is that the AI “does productive and useful work” and “generates profitable tokens”, with more compute producing more tokens and inevitably more profit.

It is not the first time Huang has made such a claim. In March, on Lex Fridman’s podcast, he said simply “I think we have achieved AGI”. Fridman offered his own definition, a system capable of “doing your work” in the sense of founding, growing and managing a tech company worth over a billion dollars. Huang immediately retreated: “The chance that 100,000 such agents would build Nvidia is zero percent”. The underlying issue remains: the industry lacks consensus on what AGI actually is, let alone how to know it has been reached, rendering any declaration inherently arbitrary.

Competing definitions coexist with substantial financial stakes. OpenAI’s charter defines AGI as “highly autonomous systems that perform far better than humans on most economically valuable work”. Sam Altman himself said last year that it is “not a super useful term”. A separate financial definition was struck between OpenAI and Microsoft, apparently for systems able to generate at least $100 billion in profit. In a recent Time story, research director Mark Chen estimated OpenAI is “80 % of the way” to AGI, while Altman said the company expects to have something by the end of the year that he would be willing to call AGI.

The ambiguity is not hidden: CEOs acknowledge it while basing forecasts on the concept. Dario Amodei, CEO of Anthropic, called AGI “inaccurate”, even “a marketing term”, preferring to talk about “powerful AI”. Others have adopted their own labels for similar ideas: Meta speaks of “personal superintelligence”, Microsoft of “humanistic superintelligence”, Amazon of “useful general intelligence”. Demis Hassabis of Google DeepMind has also shifted language; the excerpt cuts off, but the trend is clear—everyone is inventing their own definitions for the same vague notion.

Bottom line: profitable tokens. Huang’s statement reveals what truly drives Nvidia—not a scientific breakthrough but a business model. More compute yields more tokens, which translate into more revenue. As long as customers purchase GPUs to run these autonomous agents—whether they call it AGI, powerful AI, or simply “the next thing”—Nvidia’s machines keep working. The AGI announcement is essentially background noise.