Technology

Nvidia Chief Declares AGI Milestone as OpenAI Launches New Model Astra

By Truman Dickerson, Georgia Hennessy

Nvidia CEO Jensen Huang publicly praised OpenAI on Sunday following the release of the company's latest AI model, Astra, declaring that artificial general intelligence, or AGI, has been achieved. In a post on X, Huang wrote, "From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team," also noting that the model was trained using Nvidia's chips.

OpenAI introduced Astra on Thursday, describing it as the world's "most intelligent and aligned model," capable of handling complex professional tasks with high speed, accuracy, and judgment. The model is set to roll out to customers this week. During a press call, OpenAI president Greg Brockman said, "Welcome to the AGI era," and suggested that history may eventually mark this period, and possibly this very model, as the moment AGI was realized. "For me personally, I do think we're there," he said.

AGI generally refers to AI systems capable of matching or exceeding human intelligence across a broad range of tasks. OpenAI itself defines it as "highly autonomous systems that outperform humans at most economically valuable work."

This is not the first time Huang has made such a claim. In March, when podcaster Lex Fridman proposed a definition of AGI as an AI system able to launch and grow a billion-dollar company, Huang responded, "I think it's now. I think we've achieved AGI." Other tech leaders have made similar declarations. In January, Elon Musk said humanity had entered a "technological singularity," a point where AI surpasses human intelligence. In May, Andreessen Horowitz cofounder Marc Andreessen said the AGI threshold had already been crossed months earlier, claiming several advanced models were "as smart as a person."

However, not everyone agrees with these assessments. AI researcher and critic Gary Marcus criticized Huang's comments, arguing that he provided no evidence or clear definition of AGI. "Declaring victory without a definition simply muddies the waters," Marcus wrote in a Substack post, adding that by his own ten-point definition of AGI, Astra meets only one or two criteria. Researchers behind the ARC Prize benchmark, which OpenAI cited when announcing Astra, also declined to label the model as AGI, calling its performance a major advance but noting that the benchmark's controlled testing environment doesn't capture the complexity of real-world scenarios.

OpenAI CEO Sam Altman has previously downplayed the significance of the term, telling the "Sources" podcast that AGI is "a very poorly defined term" that borders on being an "irrelevant marketing term."

The implications of achieving AGI remain uncertain, with potential applications ranging from advances in medicine and science to highly capable personal AI assistants, alongside concerns about its impact on employment.

Meanwhile, Nvidia continues to benefit from the growing demand for advanced AI chips. Major AI firms, including OpenAI, Meta, Anthropic, and Google, rely heavily on Nvidia's hardware. In a March funding announcement, OpenAI called Nvidia "the foundation of our infrastructure," noting that its training and inference systems primarily run on Nvidia GPUs.

This demand has driven significant revenue growth for Nvidia, which reported $96.2 billion in quarterly revenue in August, more than double the figure from the same period last year. Its data center division, which includes AI chips, generated $89 billion of that total. In his Sunday post, Huang also hinted at further expansion, stating that "400K GPUs" are expected to come online soon, referring to the graphics processing units essential for training and running AI models.