Nvidia's CEO Declared 'AGI Has Arrived.' Not Everyone Agrees.

The quick version

OpenAI released a new flagship AI model called GPT-6 Astra. Nvidia's CEO, Jensen Huang, celebrated on X with the line "AGI has arrived. Congratulations @OpenAI team." OpenAI's president echoed him, saying the company is "moving into the AGI era." Prominent AI skeptics immediately pushed back, noting there is no shared definition of AGI and Astra hasn't passed any objective bar for it.

What happened

"AGI" stands for artificial general intelligence — loosely, an AI that can do most intellectual tasks about as well as a capable human, rather than being good at just a few narrow things. There is no official checklist for it, which is part of the problem.

Huang's post framed the progress as a timeline: "From ChatGPT to o1 to Astra in 4 years." He also used the moment to talk about hardware, noting Astra was trained on more than 100,000 of Nvidia's Grace Blackwell systems and that another 400,000 GPUs are coming online. Nvidia sells the chips that basically every large AI model is trained on, so a bigger, more capable model is also a bigger sales story for Nvidia.

OpenAI president Greg Brockman reshared the post approvingly. AI researcher Gary Marcus, a longtime critic of hype, disputed the claim directly, arguing there is "no evidence and no definitions" behind calling this AGI.

Why it matters

When the head of the company that profits most from AI training declares the goal has been reached, it is worth reading with a little skepticism. The claim also matters commercially and politically: "we have AGI" affects investor expectations, regulation debates, and how much money keeps flowing into data centers.

What's next

Expect the definition fight to continue. Astra will be measured against benchmarks and real-world use over the coming weeks, and whether the "AGI" label sticks will depend less on marketing posts and more on what the model can actually do consistently.