DeepSeek AI's covert use of Nvidia's powerful H100 chips has ignited controversy within the tech industry.
The startup is said to be using 50,000 Nvidia H100 GPUs, despite US export restrictions that limit their availability to China.
According to Scale AI CEO, Alexandr Wang, DeepSeek workers are unable to discuss the use of these chips publicly due to US regulations.
The revelation has caused a stir, with the news of DeepSeek's operations impacting stock markets. Nvidia's shares plummeted by 16.9%, and other tech stocks followed suit, dragging the Nasdaq composite down by 3.1%, marking its largest loss in over a month.
Although it was initially thought that DeepSeek was using Nvidia’s lower-performing H800 chips, Wang confirmed that the startup is leveraging the banned H100 chips.
"DeepSeek has about 50,000 H100s, which they can't talk about due to export controls," Wang stated during an interview with CNBC. He also emphasized that future access to these chips could be limited by US regulations.
Elon Musk seemed to support Wang’s claim, responding with “obviously” to a post referencing Wang’s interview.
While using these chips could slightly diminish the appeal of DeepSeek's model, its performance remains impressive. In a landscape traditionally dominated by U.S. companies, Chinese startups like DeepSeek and Qwen are now competing at a high level and potentially surpassing US models in terms of cost-effectiveness.
This raises questions about the future of U.S. regulation in the AI space, especially with the potential of Chinese innovation continuing to challenge global AI leadership.
COMMENTS
Comments are moderated and generally will be posted if they are on-topic and not abusive.
For more information, please see our Comments FAQ