NVIDIA CEO Jensen Huang says human-level AI has arrived. His company, Nvidia, supplies the hardware behind it and has a financial stake in convincing the world to keep spending on AI.
On Sunday, he credited OpenAI’s GPT-6 Astra with reaching artificial general intelligence, or AGI. Broadly, that means AI capable of handling intellectual work at human level. There is no universally accepted test.
Four days earlier, at a G20 meeting in North Carolina, Huang had sounded less certain.
“In the next couple of years, we are going to achieve essentially what people call AGI… In fact, I would argue that we’re practically there today… It either means a lot or it doesn’t mean anything,” he said.
That ambiguity extends to the contracts funding AI.
AGI Hype is Not Showing
OpenAI’s Microsoft deal once reportedly tied AGI to systems capable of generating about $100 billion in profit. An October 2025 revision required an independent expert panel to verify OpenAI’s declaration.
The companies rewrote their agreement in April. OpenAI’s payments to Microsoft now continue through 2030 regardless of technological progress. The financial arrangement no longer needs that verdict.
Meanwhile, Sequoia partners published “2026: This is AGI” in January. Physicist Mark Gubrud, who used the term in 1997, also says it has arrived.
The disagreement partly reflects different expectations. Anthropic chief Dario Amodei described a much higher threshold in February, months before Astra’s release.
“If you had the country of geniuses in a data center, we would know it… We don’t have that now. That’s very clear,” he said.
For investors, Huang’s Sunday claim leaves a practical question unanswered: how much human work can these systems reliably take over?
Cognitive scientist Gary Marcus remains unconvinced.
“Of course I fully expect that AGI will be achieved someday. But everything said before then is a premature declaration,” he noted.
Bitrue