Jensen Huang says GPT-6 Astra trained on 100K+ Grace Blackwell GPUs
Jensen Huang says Astra was trained on NVLink72 clusters and that AGI has arrived, with the poster speculating that labs not using Nvidia platforms may fall behind.
Jensen: "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." As I mentioned before, frontier labs that use NVDA platforms for training may have huge lead over the ones that dont. I wonder if Google's Gemini is lagging behind for the same reason? Anthropic's claude could be the next one who suffer if they double down on none NVDA solution.
This post quotes another post: “Jensen Huang: AGI has arrived”