Street Calls of the Week
Investing.com -- Chinese AI developer DeepSeek said that it had spent just $294,000 to train its R1 artificial intelligence model.
The cost estimate, revealed by the Hangzhou-based for the first time, was published in a peer-reviewed article in the academic journal Nature on Wednesday. The paper revealed that DeepSeek used 512 Nvidia H800 chips to train the reasoning-focused model over a period of 80 hours.
The H800 chips used by DeepSeek were specifically designed by Nvidia for the Chinese market after the U.S. banned the export of more powerful H100 and A100 AI chips to China in October 2022.
In supplementary information accompanying the Nature article, DeepSeek acknowledged for the first time that it does own A100 chips, which it used in preparatory stages of development. "Regarding our research on DeepSeek-R1, we utilized the A100 GPUs to prepare for the experiments with a smaller model," the researchers wrote.
This disclosure stands in stark contrast to statements from OpenAI CEO Sam Altman, who indicated in 2023 that "foundational model training" at his company cost "much more" than $100 million, though OpenAI has not provided detailed figures for its releases.