But, DeepSeek researchers wrote in a report last month that the DeepSeek-V3 used Nvidia’s H800 chips for education, spending less than US$ 6 million.
The state that the cards used were less effective than the most cutting-edge Nvidia items Washington has tried to keep out of China, as well as the fairly low education fees, has since been refuted, which has prompted US tech executives to question the effectiveness of technical export controls.
Little is known about the company behind DeepSeek, a small Hangzhou-based startup founded in 2023, when search engine giant Baidu released the first Chinese AI large-language model.
Since then, numerous Chinese tech companies, both large and small, have released their own AI models, but DeepSeek is the first to be praised by the US tech sector as achieving performance levels that match or even surpass those of cutting-edge US models.