Nvidia CEO Jensen Huang claims company has achieved AGI

During Nvidia's latest quarterly earnings call, chief executive Jensen Huang casually declared that the chipmaker has already achieved artificial general intelligence (AGI) across various tasks. However, Huang quickly downplayed the significance of the milestone, labeling such industry targets as "senseless" at this stage. He argued that the real focus for the sector should be on deploying autonomous agents that can recursively learn and generate profitable tokens, rather than chasing arbitrary benchmarks.

This is not the first time Huang has claimed the milestone. In March, during an appearance on the Lex Fridman podcast, he made a similar assertion, though he later walked it back when Fridman defined AGI as an AI capable of launching and running a billion-dollar company. Huang admitted the chances of autonomous agents building another Nvidia were "zero percent." The episode highlights a broader industry trend where tech leaders use highly divergent definitions for the concept. For instance, OpenAI officially defines AGI in its charter as autonomous systems outperforming humans at most economically valuable work, while its financial agreement with Microsoft reportedly ties the term to systems generating at least $100 billion in profits.

Other AI firms have introduced their own terminology to bypass the ambiguity. Anthropic CEO Dario Amodei prefers "powerful AI," Meta uses "personal superintelligence," and Amazon refers to "useful general intelligence." Meanwhile, OpenAI chief research officer Mark Chen recently estimated his firm is "80% of the way" to AGI. For AI practitioners and developers, this linguistic drift means that marketing claims about reaching AGI should be largely ignored. Instead of waiting for a single breakthrough moment, developers must focus on practical utility, building systems that deliver measurable productivity gains and generate profitable tokens in real-world applications.