The Chinese AI brought down the market
February 3, 2025
On January 20, Chinese IT startup Deep Seek released the reasoning-oriented open source DeepSeek-R1 model. Last weekend, the neural network attracted the attention of the community, which led to sales in the stock and cryptocurrency markets at the beginning of the week. The chatbot has skyrocketed in the ranking of free apps on the App Store in the United States, surpassing ChatGPT among others.
The main reason for the hype is the cheapness of the model. Only $5.6 million was spent on the development of DeepSeek, compared with hundreds of millions from American competitors. In addition, the product requires a much smaller number of GPUs and can run on gaming GPUs.
At the same time, in a number of tests, the Chinese neural network beat the latest Llama 3.1, GPT-4o and Claude Sonnet 3.5. Independent experts tested the AI for the accuracy of answers, the ability to solve complex problems, mathematics and programming.
DeepSeek was also surprised by the fact that the creators of the model managed to circumvent US export restrictions. Even Telegram founder Pavel Durov drew attention to the application.
The success of the chatbot has led to a sell-off in the technology sector of the stock market, including Nvidia. The tech giant's shares fell by 15% during the trading week. The correction also affected the crypto market, especially the direction of AI tokens - a drawdown of 21% in seven days.
Despite the severe blow, DeepSeek has pushed companies to accelerate their development. On January 28, Alibaba introduced the DeepSeek killer, and OpenAI added the o3-mini model to ChatGPT.
The hype around DeepSeek has not been without scandals. At first, the company was suspected of stealing data from OpenAI, and later researchers discovered a leak of user data from the application.
The neural network is also criticized for traditional Chinese censorship.
The main reason for the hype is the cheapness of the model. Only $5.6 million was spent on the development of DeepSeek, compared with hundreds of millions from American competitors. In addition, the product requires a much smaller number of GPUs and can run on gaming GPUs.
At the same time, in a number of tests, the Chinese neural network beat the latest Llama 3.1, GPT-4o and Claude Sonnet 3.5. Independent experts tested the AI for the accuracy of answers, the ability to solve complex problems, mathematics and programming.
DeepSeek was also surprised by the fact that the creators of the model managed to circumvent US export restrictions. Even Telegram founder Pavel Durov drew attention to the application.
The success of the chatbot has led to a sell-off in the technology sector of the stock market, including Nvidia. The tech giant's shares fell by 15% during the trading week. The correction also affected the crypto market, especially the direction of AI tokens - a drawdown of 21% in seven days.
Despite the severe blow, DeepSeek has pushed companies to accelerate their development. On January 28, Alibaba introduced the DeepSeek killer, and OpenAI added the o3-mini model to ChatGPT.
The hype around DeepSeek has not been without scandals. At first, the company was suspected of stealing data from OpenAI, and later researchers discovered a leak of user data from the application.
The neural network is also criticized for traditional Chinese censorship.