Skip to content
Tech News
← Back to articles

Jensen Huang declares AGI has arrived, again, and again he's selling GPUs

read original get MSI GeForce RTX 4070 Ventus 3X Graphics Card → more articles
Why This Matters

Nvidia's Jensen Huang has declared AGI's arrival for the second time this year, this time on the back of OpenAI's GPT-6 Astra launch — a model trained on more than 100,000 Nvidia Blackwell GPUs, with 400,000 more coming. The pattern highlights how AGI declarations from vendors tend to coincide with product cycles, while benchmarks show Astra trailing Claude in some areas. For buyers and investors, it's a reminder that 'AGI' remains an undefined marketing term rather than a measurable milestone.

Key Takeaways
Worth a Look

MSI GeForce RTX 4070 Ventus 3X Graphics Card — If Jensen's AGI hype has you curious about running AI models yourself, an Nvidia GeForce RTX 4070 card is the accessible way in — CUDA support means it works with virtually every local LLM and image-generation tool. MSI's Ventus 3X cooler keeps it quiet while you tinker, and it's a great gaming card when you're done experimenting.

See MSI GeForce RTX 4070 Ventus 3X Graphics Card on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.

In context: Tech giants have repeatedly described Artificial General Intelligence (AGI) as a benchmark marking the point at which generative AI can do virtually anything at least as well as a human. Few agree on where to draw that line, when the technology will reach it, or even whether it is just an investor buzzword.

Nvidia CEO Jensen Huang has declared the arrival of AGI for the second time this year. Last time, he cited the popularity of AI agents such as OpenClaw; this time, he made the declaration in response to the launch of GPT-6 Astra. In both cases, he sounds like he's just trying to sell more Nvidia chips.

Huang and other major AI players such as OpenAI CEO Sam Altman have made numerous predictions about AGI, the supposed point at which generative AI will perform a broad range of intellectual tasks at least as well as a human. Altman and SoftBank's Masayoshi Son previously predicted that AGI would arrive by 2030, but Huang and OpenAI president Greg Brockman think it's already here.

– Jensen Huang (@JensenHuang) September 6, 2026

Brockman declared the start of the "AGI era" upon last week's launch of GPT-6 Astra, OpenAI's latest generative AI model. However, benchmark results immediately raised doubts. While Astra's near-perfect scores on several tests are impressive, it actually trails Claude in other areas.

Notably, Brockman claimed Astra's cybersecurity capabilities let it prevent incidents like GPT-5 Sol's unauthorized breach of AI company Hugging Face. Huang's praise of Astra is similarly tied to his company's products. The Nvidia CEO noted how OpenAI trained its new model on over 100,000 Nvidia Blackwell GPUs and that 400,000 units are set to come online soon.

Huang's last AGI claim came in March, when he pointed to OpenClaw's success while envisioning a world where agents can design influencers or run digital communities autonomously. The CEO also claimed that an AI could run a $1 billion company, just not Nvidia.

However, Huang has also claimed that AGI shouldn't be the industry's true goal. During an earnings call for the quarter ending August 2026, the CEO described the milestone as senseless, stating that AGI has already arrived for certain tasks.

His most vital claim was that generative AI was already productive and generating profitable tokens. Huang stressed the need for more compute – that is, more Nvidia GPUs – to increase profits from those tokens.

Astra is currently rolling out to OpenAI's premium and business subscribers. Plans for free-user access remain unannounced.