Nobody Knows What AGI Means, and That's the Point
Nvidia's Jensen Huang declared his company had 'achieved AGI' on an earnings call, then called the milestone 'senseless' in the same breath. He's not wrong. The term has no agreed definition, and tech leaders like it that way.
Key points
- Nvidia CEO Jensen Huang said on a May 2025 earnings call that the company had "achieved AGI" for "many tasks," then immediately dismissed the milestone as "senseless."
- AGI, or artificial general intelligence, has no agreed scientific definition across the industry.
- OpenAI defines AGI as systems that "outperform humans at most economically valuable work" but its CEO has called that standard "not a super useful term."
- OpenAI reportedly has a separate, financial definition of AGI agreed with Microsoft: systems that generate at least $100 billion in profits.
- Anthropic CEO Dario Amodei has called AGI a "marketing term."
Jensen Huang, the CEO of Nvidia, the company that makes the chips most AI systems run on, dropped a remarkable claim on a recent earnings call. "For many tasks," he said, "we could say that we've already achieved AGI."
AGI stands for artificial general intelligence, the idea of an AI system that can match or beat human thinking across a wide range of subjects, not just one narrow task like chess or translating text. It is widely treated as the ultimate goal of the AI industry.
Huang then called the milestone "senseless." That second part, at least, is defensible.
What does AGI actually mean?
No one agrees. That is the honest, short answer.
OpenAI, the company whose entire founding purpose was to build AGI, defines it in its charter as "highly autonomous systems that outperform humans at most economically valuable work." OpenAI CEO Sam Altman admitted last year that this is "not a super useful term." The Verge AI first reported that OpenAI also has a separate, private definition worked out with Microsoft: systems capable of generating at least $100 billion in profits. OpenAI's chief research officer Mark Chen said in a recent interview that the company is "80% of the way" to AGI, while Altman has predicted he would be calling something AGI by the end of 2025.
Anthropics's Dario Amodei prefers the phrase "powerful AI" and has called AGI "imprecise" and a "marketing term." Meta talks about "personal superintelligence." Microsoft uses "humanist superintelligence." Amazon says "useful general intelligence." Google DeepMind's Demis Hassabis has described the current moment as the "foothills of the singularity," a hypothetical point where AI growth becomes self-sustaining and uncontrollable.
New vocabulary. Same fog.
Has Huang said this before?
Yes. In March 2025, on the Lex Fridman podcast, Huang said plainly: "I think we've achieved AGI." When Fridman suggested a definition, specifically an AI that could build and run a billion-dollar tech company, Huang backed off. "The odds of 100,000 of those agents building Nvidia is zero percent," he said.
On the earnings call, Huang shifted the goalposts entirely. What actually matters, he argued, is AI "doing productive and useful work" and "generating profitable tokens," meaning billable AI outputs that earn money. More computing power produces more outputs, which produces more revenue. That framing is useful for a company that sells the chips powering all of it.
What does this mean for ordinary people?
Practically, nothing has changed. No new AI product launched. No patient got a better diagnosis. No new capability appeared overnight.
What did change is the language around expectations. When tech leaders call a goal achieved before anyone agrees what the goal is, it makes it very hard to hold anyone accountable for whether AI is genuinely delivering on its promises.
For anyone following AI in health, education, or public services, the useful question is not "has AGI arrived?" It is: what specific problem does this tool solve, for which people, and what evidence supports that claim? Those questions have real answers. "AGI" does not.
