Jensen Huang: AGI has arrived! OpenAI Astra is backed by NVIDIA, with 400,000 GPUs about to go online

Author: Xu Chao, Wall Street Journal

NVIDIA CEO Jensen Huang announced the official start of the era of artificial general intelligence, attributing this historic milestone to OpenAI's latest flagship model, Astra—this model was trained on NVIDIA chips. This statement once again places NVIDIA in a dominant position in AI infrastructure and suggests that a larger-scale expansion of computing power is imminent.

On Sunday, Huang posted on the X platform congratulating the OpenAI team, stating, "AGI has arrived," and revealed that Astra was trained on over 100,000 NVIDIA Grace Blackwell NVLink72 chips. He also mentioned, "400K GPUs are about to go online," sending a strong signal that NVIDIA's supply of computing power will further surge.

This statement has direct implications for the market. NVIDIA's data center business has become its core growth engine, and Huang's forecast of adding 400,000 GPUs indicates that order demand from leading AI clients continues to expand.

Astra Release: OpenAI Calls It the "Smartest" Model

OpenAI released Astra last Thursday, positioning it as "the smartest model in the world, aligned with human values," claiming that this model can "perform the most demanding professional tasks with unparalleled speed, accuracy, and judgment." OpenAI President Greg Brockman stated to the media at the launch event, "Welcome to the AGI era."

Brockman further predicted that in the future, when people look back at history, they will believe AGI was born "around this time, and I think it might be this model." "Personally, I do believe we have arrived," he said. Astra will start rolling out to users this week.

OpenAI defines AGI as "a highly autonomous system that outperforms humans in most economically valuable work," and positions Astra as a significant research breakthrough, marking a fundamental shift in the boundaries of tasks that can be entrusted to AI.

Ongoing Controversy: Disagreements on the Definition of AGI

However, not everyone agrees with this declaration. Renowned AI researcher and critic Gary Marcus directly pointed out that Huang is "too hasty."

"Huang has neither provided evidence nor a definition, which to me seems like an attempt to forcefully occupy a scientific issue with corporate will," Marcus wrote on his Substack, "Announcing victory without a definition only adds to the confusion."

Marcus then listed his own 10 criteria for defining AGI, pointing out that Astra meets only one or two of them. "By common definitions, Astra still falls short," he said.

Even OpenAI's own CEO Sam Altman has a rather cautious attitude towards this concept. He recently stated on the "Sources" podcast that AGI is at best "a term that is extremely vague. I would say it’s more like an irrelevant marketing term."

NVIDIA: The Core Support of AI Infrastructure

Regardless of how the definition debate of AGI unfolds, NVIDIA's position as a core computing power supplier for leading AI companies is undisputed. In a funding announcement in March this year, OpenAI referred to NVIDIA as "the cornerstone of our infrastructure," stating that "our training clusters and most of our inference stack continue to run on NVIDIA GPUs."

Meta, Anthropic, Google, and other leading AI companies also heavily rely on high-end chips produced by NVIDIA. This demand has translated into NVIDIA's impressive financial performance— the company reported quarterly revenue of $96.2 billion in August this year, more than doubling year-on-year; among which, data center business revenue reached $89 billion, with AI chips being a major source of growth.

Huang's post on X stated that the next batch to go online will be 400,000 GPUs, further reinforcing external expectations for NVIDIA's expansion momentum in computing power. From ChatGPT to o1, and now to Astra, the technological leaps in four years have all been built on NVIDIA's continuously upgraded chip foundation.