Nvidia’s Jensen Huang Claims AGI Milestone, But Its Impact Remains Unclear
During Nvidia’s recent earnings call, CEO Jensen Huang made waves by declaring that the company has “achieved AGI,” a goal that many in the tech world have been chasing for years. However, he quickly downplayed this milestone as “senseless.”
Huang’s skepticism is valid. The concept of artificial general intelligence lacks a clear definition, and there’s no agreement on how to recognize its arrival, making the pursuit of it feel arbitrary. When asked about OpenAI’s quest for AGI, Huang suggested that Nvidia has already reached this level for many tasks, though he didn’t specify what that means. He remarked, “I think of all of those milestones and all those, you know, they’re kind of senseless at this point.” He emphasized that AI is evolving from merely responding to prompts to becoming autonomous agents that can learn and improve themselves. What truly matters, he argued, is that AI is “doing productive and useful work” and generating profits, with increased computing power leading to more output and, ultimately, more revenue. “This is the exact phase where we’re at. Which is the reason why everybody’s leaning in.”
This isn’t Huang’s first assertion of reaching AGI. Back in March, he stated on a podcast, “I think we’ve achieved AGI,” without clarifying his definition. The host suggested a specific interpretation: an AI that can effectively run a billion-dollar tech company. Huang later tempered his claim, stating that the chances of 100,000 such agents building Nvidia are “zero percent.”
Over the years, various tech leaders have exploited the ambiguity surrounding AGI, offering a range of definitions that revolve around the same elusive idea: AI that can match or exceed human intelligence across various fields, despite the lack of a universally accepted definition of “intelligence.”
OpenAI, a company dedicated to developing AGI, defines it as “highly autonomous systems that outperform humans at most economically valuable work.” However, CEO Sam Altman has admitted that this is not a practical standard, calling AGI “not a super useful term.” Adding to the confusion, OpenAI has a financially motivated definition of AGI in collaboration with Microsoft, which involves systems capable of generating at least $100 billion in profits. Recently, OpenAI’s chief research officer estimated that the company is “80% of the way” to AGI, while Altman claimed they would have something he would label AGI by year’s end.
The ambiguity surrounding AGI and its criteria is well-known, with tech leaders acknowledging it even as they make predictions based on it. Anthropic CEO Dario Amodei has referred to AGI as “imprecise” and even a “marketing term,” opting instead to discuss “powerful AI.” Others have coined their own terms for similar concepts, leading to a muddled landscape. Meta refers to “personal superintelligence,” Microsoft talks about “humanist superintelligence,” and Amazon uses “useful general intelligence.” Google DeepMind’s Demis Hassabis has suggested we are at the “foothills of the singularity,” while OpenAI cofounder Ilya Sutskever, who once led chants of “feel the AGI,” now heads a company called Safe Superintelligence.
Despite the introduction of new terminology, the meaning remains elusive. As long as AGI is poorly defined and used carelessly, the conversation feels pointless—unless it serves as a tool for generating hype. Expect the industry, including Huang, to continue the AGI dialogue. Perhaps one day, an AGI will emerge and clarify what AGI truly means.
