Nvidia CEO Jensen Huang made conflicting statements about artificial general intelligence (AGI) development during recent public appearances. According to reports from The Times of India, Huang credited OpenAI's GPT-6 Astra with reaching AGI on Sunday, describing it as AI capable of handling intellectual work at human level. However, four days earlier at a G20 meeting in North Carolina, Huang had sounded less certain about AGI's arrival. As reported by The Times of India, he stated that while the company was "practically there today," he added "It either means a lot or it doesn't mean anything." In March, Huang had previously defined AGI as an AI system capable of starting, growing, and running a tech company worth more than $1 billion, telling podcaster Lex Fridman that "I think it's now. I think we've achieved AGI." Huang's latest declaration ties AGI to four years of development powered by more than 100,000 Nvidia Grace Blackwell NVLink72 chips, representing the company's infrastructure behind frontier model training.
Nvidia's position as a major supplier of AI hardware creates a financial incentive for the company to promote AGI development. According to reports from The Times of India, Huang's company supplies the hardware behind AGI systems and has a financial stake in convincing the world to continue spending on AI infrastructure. The CEO's statements come as OpenAI's Microsoft deal previously tied AGI to systems capable of generating about $100 billion in profit, though an October 2025 revision required independent expert panel verification. The companies rewrote their agreement in April, with OpenAI's payments to Microsoft now continuing through 2030 regardless of technological progress. Nvidia reported $96.2 billion in quarterly revenue in August, more than double the amount reported in the same period a year earlier, with its data center business generating $89 billion in revenue. The company has also announced 400K GPUs coming online next, referring to graphics processing units used to train and run AI models.
OpenAI unveiled Astra on Thursday, calling it the world's "most intelligent and aligned model" and capable of performing "the most demanding professional work with unmatched speed, accuracy, and judgment." As reported by The Times of India, OpenAI president Greg Brockman said on the Thursday call: "Welcome to the AGI era." OpenAI's training fleet and the majority of its inference stack continue to run on Nvidia GPUs, with the company calling Nvidia "the foundation of our infrastructure" in a March funding announcement. OpenAI's rollout begins with limited organizations, then expands to ChatGPT Plus, Pro, Business and Enterprise users, with Astra also becoming available through OpenAI's API, Microsoft Azure and Amazon Web Services' Bedrock. Nvidia CEO Jensen Huang tipped his hat to OpenAI on Sunday, taking to X to congratulate the team on Astra's release and making the notable statement that "AGI has arrived." Huang noted "From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team." Industry expert Gary Marcus remains unconvinced, writing on his Substack Sunday evening that "Huang gave no evidence and no definitions, which feels to me like an effort at a takeover of a scientific question by corporate fiat." Marcus noted that "By conventional definitions, Astra still falls short," and attached his own 10-point definition of AGI, emphasizing that "everything said before then is a premature declaration."
Nvidia CEO Jensen Huang identified cybersecurity as the next major use case for artificial intelligence, highlighting the expanding applications of AI beyond data centers and generative AI. Speaking at the Goldman Sachs conference, Huang emphasized the growing diversity of AI applications across multiple industries. For investors, Huang's Sunday claim leaves a practical question unanswered: how much human work can these systems reliably take over. According to reports from The Times of India, the disagreement partly reflects different expectations about what constitutes AGI achievement. Sequoia partners published "2026: This is AGI" in January, with physicist Mark Gubrud, who used the term in 1997, also claiming AGI has arrived. Elon Musk said in January that humanity had "entered the technological singularity," which he described as a point of no return in which AI accelerates beyond human intelligence. Marc Andreessen said in May that the AGI threshold had been crossed roughly three months earlier, arguing that several frontier models were already "as smart as a person." However, Sam Altman told the "Sources" podcast that AGI is, at best, "a very poorly defined term" and "an irrelevant marketing term." What AGI would mean for everyday life remains unclear, potentially leading to breakthroughs in medicine and science while putting more jobs under pressure. The critical number for Nvidia investors may be the planned 400,000 GPUs, with the company moving from a training fleet exceeding 100,000 chips to hundreds of thousands more, reinforcing the view that frontier AI remains extremely compute-intensive and supporting demand for Nvidia's systems and high-margin data-center ecosystem.
