Nvidia's CEO announced the tech industry’s holy grail on an earnings call Wednesday, then knocked it down in the same breath. Jensen Huang said the company had “achieved AGI,” and almost immediately called the milestone “senseless.”
He’s right, and that’s the whole problem.
Artificial general intelligence is supposed to be the finish line of the AI race. But there’s no consensus on what it means, let alone how we’ll know when we’ve actually got there, which makes achieving it equally arbitrary. You can declare victory whenever you like. Nobody can check your work.
What Huang actually said
Asked about OpenAI’s pursuit of AGI, Huang said that when it comes to Nvidia, “for many tasks, we could say that we’ve already achieved AGI.” No definition followed. No benchmark. Just a shrug: “I think of all of those milestones and all those, you know, they’re kind of senseless at this point.”
He pointed to AI moving past simple prompt-and-response toward autonomous agents that learn new skills and improve themselves “recursively.” What matters, Huang said, is that AI is “doing productive and useful work” and “generating profitable tokens.” More compute, more tokens, more profit. “This is the exact phase where we’re at. Which is the reason why everybody’s leaning in.”
That last part is the tell. The AGI claim is decoration. The token math is the business.
He’s said it before
This isn’t Huang’s first declaration. In March, on the Lex Fridman podcast, he said flatly, “I think we’ve achieved AGI.” He didn’t specify what he meant then either.
Fridman offered his own oddly specific definition: an AI system that can “essentially do your job,” meaning start, grow and run a successful tech company worth more than $1 billion. Huang backed off. “The odds of 100,000 of those agents building Nvidia is zero percent,” he said.
So AGI is here, except when you attach a testable standard to it. Then it isn’t.
Everyone’s definition is different, and that’s on purpose
Tech leaders have spent years mining the term’s fuzziness for a grab bag of definitions and benchmarks, all circling the same nebulous idea: AI that matches or surpasses human intelligence across a broad range of domains. Never mind that “intelligence” doesn’t have a universally agreed-upon definition either.
OpenAI was founded with the explicit goal of building AGI, and its charter still leaves enormous room for interpretation, defining it as “highly autonomous systems that outperform humans at most economically valuable work.” Sam Altman has conceded the point himself, admitting last year that AGI is “not a super useful term.”
Then there’s the version OpenAI worked out with Microsoft, which is reportedly financial: systems that can generate at least $100 billion in profits. Meanwhile chief research officer Mark Chen estimated OpenAI is “80% of the way” to AGI, and Altman said that by the end of the year the company would have something he would call AGI.
Eighty percent of the way to a destination with no coordinates.
The rebranding hasn’t helped
Nobody’s hiding the vagueness. Anthropic CEO Dario Amodei has called AGI “imprecise,” even a “marketing term,” and prefers “powerful AI.” Others have coined their own labels for broadly similar ideas. In theory the distinctions mean something. In practice they bleed together.
Meta talks about “personal superintelligence.” Microsoft says “humanist superintelligence.” Amazon goes with “useful general intelligence.” Google DeepMind’s Demis Hassabis describes arriving at the “foothills of the singularity.” And OpenAI cofounder Ilya Sutskever, who reportedly led employees in chants of “feel the AGI,” now runs a company called Safe Superintelligence.
New terminology hasn’t made the meaning more tangible. It’s made it easier to move.
Why the talk won’t stop
So long as AGI stays poorly defined and carelessly used, the whole thing is senseless. Unless, of course, you want a handy tool for hyping up progress, in which case an undefined milestone you can claim on an earnings call is exactly the right instrument.
Expect more of it from the industry, Huang included. Maybe an AGI will eventually show up and tell us what AGI actually means.
