Photo: SHOX ART / Pexels
Huang defines artificial general intelligence through an economic lens, sparking industry-wide debate about what the term actually means
Jensen Huang thinks we’ve crossed the AGI finish line. During a March 2026 appearance on the Lex Fridman podcast, the Nvidia CEO declared, “I think we’ve achieved AGI,” a statement that immediately set off the kind of internet firestorm you’d expect when someone claims to have solved one of computing’s greatest open questions.
A billion-dollar benchmark #
Huang’s version of artificial general intelligence isn’t about an AI that can do everything a human can. It’s about an AI that can start and run a technology company valued at over $1 billion. By this logic, if an AI agent can autonomously navigate the complex, multi-step process of building a billion-dollar business, it meets the bar for general intelligence. Huang pointed to AI agents like OpenClaw as examples of systems that could theoretically generate that kind of short-term value.
He wasn’t entirely bullish without caveats, though. Huang acknowledged that the probability of 100,000 AI agents independently building something resembling Nvidia was, in his words, zero percent.
The numbers behind the narrative #
Nvidia posted $215.9 billion in revenue for fiscal 2026, a figure that reflects the company’s near-monopolistic grip on the AI hardware market.
By August 2026, Nvidia’s own AI coding agent, called AVO, achieved a perfect 100% score on the ARC-AGI-3 benchmark. That benchmark is specifically designed to test reasoning and generalization capabilities.
The definition wars #
Traditional definitions of AGI typically require an AI system to match or exceed human cognitive abilities across a wide range of tasks, not just business operations. Huang’s framing shifts the goalposts toward measurable economic outcomes: revenue generation, autonomous task execution, and company building.
This isn’t the first time Huang has made bold predictions about AGI timelines. He has consistently emphasized the potential of agentic AI, systems that can autonomously plan and execute complex multi-step tasks, as the bridge between current AI capabilities and true general intelligence. His March comments represent a natural escalation of that narrative.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our