Nvidia CEO Jensen Huang Says AGI May Already Have Arrived

Nvidia CEO Jensen Huang has made a striking claim about the state of artificial intelligence, saying he believes artificial general intelligence (AGI) has already been achieved. Huang made the comment during an interview with Lex Fridman in March 2026, responding to a question about how far away AI is from reaching a broad human-level capability.

However, Huang’s definition of AGI is important. His comments were based on a relatively specific benchmark involving whether AI could perform the work required to create and operate a highly successful technology company. That is considerably narrower than the broader idea of AGI as an AI system capable of reliably performing essentially any intellectual task at a human level.

Huang’s argument reflects how rapidly AI agents and autonomous software systems have developed. Modern AI can write and debug software, analyze large amounts of information, use tools and complete increasingly complicated multi-step workflows. Nvidia has also been emphasizing the transition from experimental AI toward systems that perform economically valuable work at scale.

Still, Huang’s statement should not be interpreted as scientific confirmation that AGI has been achieved. There is no universally accepted definition or test for AGI, and experts disagree about what capabilities would constitute genuine general intelligence. Some current systems remain unreliable on long-running tasks and can struggle with planning, consistency and situations requiring broad real-world understanding.

The debate is becoming increasingly important because Nvidia sits at the center of the AI infrastructure boom. The company reported $96.2 billion in quarterly revenue and expects approximately $108 billion in revenue for the following quarter, demonstrating how rapidly demand for AI computing continues to expand. Huang’s AGI prediction therefore comes at a moment when the economic impact of AI is becoming increasingly significant.

Whether AGI has truly arrived remains an open question. What is much clearer is that AI systems are becoming increasingly capable of performing complex cognitive work, and the boundary between conventional software and autonomous AI agents is rapidly disappearing.