In a post on X on Sunday, Huang congratulated the OpenAI team and wrote, “AGI has arrived.” He also said Astra was trained using more than 100,000 Nvidia Grace Blackwell NVLink72 systems.
“From ChatGPT to o1 to Astra in 4 years,” Huang wrote.
Huang said another 400,000 Nvidia GPUs were “coming online next”, highlighting the computing capacity being added to support the development and operation of AI models.
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years.
AGI has arrived. Congratulations @OpenAI team.
400K GPUs coming online next.
— Jensen Huang (@JensenHuang) September 6, 2026
AGI refers to AI systems that can understand, learn and perform a wide range of tasks at a level comparable to humans, rather than being designed for a specific task. OpenAI defines AGI as “highly autonomous systems that outperform humans at most economically valuable work”.
Anthropic’s Claude, on the other hand, refers to AGI as a hypothetical form of AI that can understand, learn, and apply knowledge across a broad range of tasks at a level comparable to, or exceeding, human cognitive ability — rather than being limited to one narrow domain.
It should also be noted that there is no universally accepted definition of AGI.
OpenAI’s model
OpenAI released GPT-6 Astra on September 3 and said it can perform tasks involving computer use, software engineering, cybersecurity, science and professional work. The company is making the model available to paid and enterprise customers.
OpenAI President Greg Brockman also linked Astra to the company’s AGI ambitions, saying, “Welcome to the AGI era.” He said he personally believes AI has reached AGI and suggested that the development of Astra could later be viewed as a point when AGI was created.OpenAI defines AGI as “highly autonomous systems that outperform humans at most economically valuable work”. However, there is no universally accepted definition of AGI.
Not everyone agrees
AI researcher Gary Marcus disputed Huang’s claim, arguing that the statement did not provide a clear definition or evidence for declaring AGI achieved.
The debate reflects the broader uncertainty over how AGI should be measured. While OpenAI and some industry leaders view increasingly capable AI models as evidence that AGI is approaching or has arrived, researchers differ on what capabilities should qualify as AGI.
Nvidia chips
Huang’s comments also underline Nvidia’s role in OpenAI’s AI infrastructure. OpenAI has said that most of its training fleet and inference infrastructure continues to run on Nvidia GPUs.
The scale of Astra’s training also points to the growing computing requirements of frontier AI models. Huang’s reference to another 400,000 GPUs coming online suggests Nvidia expects further demand for its AI chips as companies expand their computing capacity.
Leave a Reply