NVIDIA CEO Jensen Huang recently asserted that artificial general intelligence (AGI) has already been achieved, a claim predicated on a narrow definition of the technology’s capabilities. Huang made the statement during a conversation with podcaster Lex Fridman, framing AGI as the ability of an AI to generate a billion dollars in revenue, even if fleetingly.
Fridman initially posed a hypothetical scenario, asking Huang if an AI could start, grow, and run a technology company valued at over a billion dollars within the next five to twenty years, given the emergence of agentic AI tools. Huang responded affirmatively, stating, “I think it’s now. I think we’ve achieved AGI.” However, this assessment hinges on the condition that the AI only needs to reach a billion dollars once, not sustain a long-term business.
Huang illustrated his point with an example of a simple web service—an application that goes viral, is used by billions of people at a modest per-use cost, and then ceases to operate. He drew a parallel to the dot-com boom, suggesting that many of those early websites were no more complex than what current AI agents could create. “You said a billion,” Huang told Fridman, “and you didn’t say forever.”
This definition sharply contrasts with broader conceptions of AGI, which envision a transformative technology capable of reshaping the global economy. According to the IBM definition, AGI represents an AI system that can match or exceed human cognitive abilities across any task. The Wikipedia entry for Artificial General Intelligence defines it as a type of AI that matches or surpasses human capabilities across virtually all cognitive tasks. Huang himself acknowledged the limitations of his vision, stating that the probability of 100,000 such AI agents building a company like NVIDIA is “zero percent.”
The assertion comes as the AI industry faces increasing scrutiny over its rapid growth and substantial capital expenditure. A recent report from IDCA estimates that the global AI market could exceed $1 trillion in revenue. Companies like OpenAI, DeepMind (Google), Anthropic, IBM, and Microsoft are investing heavily in agentic systems and multimodal models, driving the pursuit of AGI. Aggregate forecasts suggest at least a 50% chance of achieving several AGI milestones by 2028, with some experts estimating a 10% chance of machines outperforming humans in all tasks by 2027 and 50% by 2047.
Despite the optimism, challenges remain. A 2025 MIT Technology Review article highlights the difficulty of creating AI models that can rival human intelligence across all domains. The article points out that even advanced AI models struggle with tasks easily mastered by humans. The road to AGI, as described by Anthropic co-founder Dario Amodei, involves achieving “Nobel Prize-level domain intelligence” and the ability to seamlessly transition between different interfaces, including the physical world.
At Google I/O 2025, DeepMind CEO Demis Hassabis and Google co-founder Sergey Brin suggested AGI could arrive around 2030. Other industry surveys predict AGI development between 2027 and 2032. However, the definition of AGI remains a central point of contention, as evidenced by Huang’s recent comments.
