Jensen Huang: AGI Has Arrived! Nvidia Powers OpenAI Astra, 400,000 GPUs to Be Deployed Soon

Nvidia CEO Jensen Huang has declared that the era of Artificial General Intelligence (AGI) has officially begun, crediting this historic milestone to OpenAI’s latest flagship model, Astra—which was trained on Nvidia chips. This statement once again cements Nvidia’s status as the dominant force in AI infrastructure and implies that even larger-scale compute expansion is on the horizon.

On Sunday, Huang posted on X to congratulate the OpenAI team, stating "AGI has arrived," and revealed that Astra was trained using over 100,000 Nvidia Grace Blackwell NVLink72 chips. He also mentioned, "400K GPU coming soon," sending a strong signal that Nvidia’s compute supply is about to see a significant boost.

This statement carries direct significance for the market. Nvidia’s data center business has become its core growth engine, and Huang’s hint at adding 400,000 new GPUs indicates that order demand from leading AI clients continues to expand.

OpenAI released Astra last Thursday, describing it as “the world’s most intelligent model, aligned with human values,” and stated that it can “tackle the most demanding professional tasks with unmatched speed, accuracy, and judgment.” OpenAI President Greg Brockman told the media at the launch, “Welcome to the era of AGI.”

Brockman further predicted that in the future, people looking back will consider AGI to have emerged “around this time, and I think it might just be this model.” “Personally, I truly believe we have arrived,” he said. Astra is rolling out to users this week.

OpenAI defines AGI as “highly autonomous systems that surpass human performance in most economically valuable work,” positioning Astra as a major research breakthrough that fundamentally shifts the boundaries of tasks that can be entrusted to AI.

However, not everyone agrees with this declaration. Prominent AI researcher and critic Gary Marcus directly asserted that Huang was “jumping the gun.”

“Huang neither provided evidence nor definition, which, to me, looks like an attempt to impose corporate will on a scientific issue,” Marcus wrote on his Substack. “Declaring victory without a definition only sows more confusion.”

Marcus then listed his own ten criteria for AGI, pointing out that Astra meets only one or two. “By conventional definitions, Astra still falls short,” he said.

Even OpenAI’s own CEO, Sam Altman, has been rather cautious regarding the term. In a recent “Sources” podcast episode, he remarked that AGI is “a term with an extremely vague definition. I was going to say it’s more like a meaningless marketing slogan.”

No matter how the debate over AGI’s definition unfolds, Nvidia’s position as the core compute supplier to leading AI companies is indisputable. In a funding announcement this March, OpenAI called Nvidia “the cornerstone of our infrastructure,” adding, “Our training clusters and most of our inference stack continue to run on Nvidia GPUs.”

Cutting-edge AI companies such as Meta, Anthropic, and Google are also highly reliant on Nvidia’s high-end chips. This demand has translated into stellar financial performance for Nvidia—the company posted quarterly revenue of $96.2 billion in August this year, doubling year-over-year; data center business revenue, which includes AI chips, reached $89 billion and is the main growth driver.

Huang’s post on X, revealing that the next batch of 400,000 GPUs will be deployed, further strengthened expectations of Nvidia’s accelerating compute expansion. From ChatGPT to o1 to Astra, each leap in the past four years has been built on the relentless advancement of Nvidia’s chip technology.