Jensen Huang Declares AGI Has Arrived, Critics Cite Nvidia's Financial Stake

Jensen Huang, CEO of Nvidia, declared on Sunday that human-level artificial intelligence already exists. As proof, he pointed to OpenAI's GPT-6 Astra model. The statement came just four days after Huang struck a more cautious tone at the G20 summit in North Carolina, saying the world was 'practically' on the threshold of AGI. Commentators were quick to point out that it is Nvidia's own chips that power the systems Huang is now calling fully intelligent.

A Shifting Position

Huang's position on AGI shifted over the months, with each successive statement sounding more confident than the last. In March, on the Lex Fridman podcast, he said flatly: 'I think it's now. I think we've achieved AGI.' In August, during Nvidia's second-quarter fiscal 2026 earnings call, he refined that: 'For many tasks, we can say we've already achieved AGI.' On Tuesday, at the G20 summit in North Carolina, he softened his tone again, telling attendees the world would reach AGI 'within the next two years' and that 'we're practically there already today,' adding the caveat: 'That either means a whole lot or it means nothing at all.' By Sunday, the caveats were gone, and Huang directly credited OpenAI's GPT-6 Astra model with achieving AGI.

The Seller's Interest

Critics point to an obvious conflict of interest. Nvidia supplies the hardware that runs every major large language model, and has a direct financial interest in sustaining the belief that AI infrastructure spending is worthwhile. The more spectacular artificial intelligence's capabilities appear, the easier it is to justify billions more in orders for graphics chips. Huang himself has proposed his own, nonstandard definition of AGI, based not on classic cognitive tests but on whether AI systems generate 'profitable tokens', meaning real, productive economic output.

The Clause That Disappeared

AGI declarations are not purely a rhetorical skirmish, they also carry real contractual consequences. Microsoft's 2019 agreement with OpenAI included a clause under which a formal declaration that OpenAI had achieved AGI would end Microsoft's access to future models and close out the existing partnership. In April, the two companies rewrote that agreement: payments between them are now set to continue through 2030 regardless of how far the technology advances. The financial trigger tied to an AGI announcement is gone, which some commentators read as a sign that the AGI label itself has become more of a marketing tool than a technical claim with real legal consequences.

If we had a country of geniuses in a data center, we would know it... we don't have that right now, that's very clear - Dario Amodei, CEO of Anthropic

Of course I fully expect AGI to be achieved at some point. But anything said before that moment is a premature declaration - Gary Marcus, cognitive scientist

A Divided Field

Huang is not alone in making these claims. Partners at Sequoia published a piece in January titled bluntly '2026: This Is AGI', and physicist Mark Gubrud, who coined the term back in 1997, also believes the threshold has been crossed. On the other side stand far more cautious voices, such as Anthropic's Dario Amodei, whose definition of AGI requires a system capable of replacing 'a country of geniuses' in a data center, and cognitive scientist Gary Marcus, who calls every earlier declaration premature. Huang's position also contradicts OpenAI's own formal definition, under which AGI means systems that outperform humans at most economically valuable tasks.

The Scale of Nvidia's Business

Behind the declarations lies the scale Nvidia's AI business has reached. The company has reported deploying more than 400,000 AI agents alongside roughly 40,000 human employees, and shipping more than 1.5 million chips within a single generation of the processors powering so-called AI factories. Those numbers show why every statement from Nvidia's CEO about artificial intelligence's capabilities is also read as part of the company's sales strategy, not merely as an assessment of the state of the technology.

What It Means for the Market

For Polish companies and investors tracking the AI boom, the dispute over the definition of AGI carries practical weight beyond academic debate. Valuations of technology companies, including Nvidia itself, rest heavily on the belief that AI capability will keep accelerating at a pace that justifies current infrastructure spending. If AGI declarations turn out to be primarily a marketing tactic from hardware vendors, that could affect how the market judges the real pace of technological progress in coming quarters.

The question that remains open despite Huang's declaration is simple: how much real mental labor can systems like GPT-6 Astra actually take over in a repeatable, reliable way. That practical criterion, not the AGI label, will ultimately determine whether the current wave of investment in data centers and graphics chips pays off.