During the Q2 2026 earnings call on August 27, Nvidia CEO Jensen Huang stated that AGI has already been achieved for many practical tasks. He dismissed traditional benchmarks as senseless, suggesting that the real metric for AGI should be the ability of AI systems to generate productive economic output, which he termed profitable tokens.
Nvidia is implementing a workforce model that integrates over 400,000 AI agents to work alongside approximately 40,000 human employees. These agents are intended to handle various tasks across business functions, operations, and engineering.
Regarding hardware, the company has discussed deploying more than 1.5 million chips in a single generation to power AI factories. These specialized facilities are designed to produce intelligence as an industrial output.
Huang's perspective differs from OpenAI's formal definition, which views AGI as systems that outperform humans in most economically valuable tasks. While OpenAI's definition is tied to its corporate governance and contractual changes, Nvidia's approach focuses on the commercial application and hardware demand.
