DeepSeek accused of getting around US sanctions to buy NVIDIA chips

0comments
Nvidia's logo in white and green against a green background.
Chinese AI company DeepSeek is riding a wave of momentum in the U.S. as the firm's DeepSeek-R1 model can supposedly challenge other open-source large language models (LLMs) like ChatGPT-4. However, DeepSeek has the advantage of providing responses at a lower cost while also requiring less computing power. The DeepSeek-R1 AI chatbot took over the number one spot as the most popular free app on Apple's App Store from its aforementioned rival ChatGPT.

A report on Friday says that DeepSeek was able to get around U.S. restrictions on chips used for AI and bought advanced Nvidia semiconductors. People familiar with the situation say that DeepSeek purchased the chips through third parties located in Singapore. Considering that DeepSeek was able to develop its LLM model at a fraction of the cost that OpenAI and Google spent to develop ChatGPT and Gemini respectively, U.S. analysts wonder whether China is far ahead of the U.S. when it comes to this burgeoning technology.


Before you reach this conclusion, in a test among 11 AI platforms, DeepSeek finished only 10th in accuracy answering questions correctly only 17% of the time. 30% of the questions handled by DeepSeek were answered with false claims and more than half of the time-53%-the Chinese AI chatbot gave a vague or not useful answer to news-related queries.

Both the White House and FBI are investigating whether DeepSeek used intermediaries in Southeast Asia to get around U.S. restrictions preventing sales of Nvidia's advanced AI chips in China. Santa Clara California-based Nvidia said in a statement that its partners must act within the law. Earlier this week, the chip maker said that it didn't believe that DeepSeek violated the law.


On the other hand, President Donald Trump's pick to head the Commerce Department, Howard Lutnick, said that it is okay for DeepSeek to compete with U.S. AI companies. However, if they do want to go up against U.S. AI firms, Lutnick said that they need to stop using America's own tools (such as the Nvidia chips) to compete with us.

Some statistics indicate that Singapore might have been involved in helping DeepSeek obtain Nvidia chips. The percentage of Nvidia's revenue derived from the country rose from 9% to 22% over two years.

DeepSeek did admit to using 2,048 of Nvidia's H800 GPU chips to train its V3 model and its R1 model was likely trained on a more powerful machine that probably required the use of more advanced Nvidia GPUs that can't be sold in China.
Create a free account and join our vibrant community
Register to enjoy the full PhoneArena experience. Here’s what you get with your PhoneArena account:
  • Access members-only articles
  • Join community discussions
  • Share your own device reviews
  • Build your personal phone library
Register For Free

Recommended Stories

Loading Comments...
FCC OKs Cingular\'s purchase of AT&T Wireless