The application of domestic big model company DeepSeek surpassed ChatGPT on the free list of Apple's China and the United States app stores, ranking first, attracting widespread market attention and causing a significant impact on Nvidia's stock price. This incident not only highlights the breakthrough in DeepSeek technology, but also reflects the intensity of competition in the AI field and its dependence on underlying hardware. This article will conduct in-depth analysis of DeepSeek's success, Nvidia's reaction, and the impact of this incident on the market.
The application of the domestic big model company DeepSeek topped the Apple app store free APP download rankings in China and the United States, surpassing ChatGPT.
This achievement not only attracted widespread attention from the market, but also had a significant impact on the stock price of Nvidia, the leader in AI chips. Nevertheless, a Nvidia spokesperson gave a high praise for DeepSeek, calling it an outstanding advance in the AI field, noting that DeepSeek's work demonstrates how to utilize test time scaling technology, widely available models, and calculations that comply with export controls. Resources to create new models.
A Nvidia spokesman added that the reasoning process requires a large number of Nvidia GPUs and high-performance network support. Nvidia currently masters three expansion rules: continuous pre-training and post-training, and new test time scaling technology. This statement also means that the GPU used by DeepSeek is fully in line with the export control requirements of the United States.
Testing time scaling technology is a topic that has been discussed in recent years, pointing out that if an AI model that has completed full training invests more computing resources to help its inference when predicting, generating text or images, the model will output better answers. This technology has been applied to OpenAI's o1 and other models.
On January 27, the three major U.S. stock indexes closed mixed, with Nvidia's stock price plummeting 16.86%, closing at $118.58 per share, falling to its lowest point in the past 10 months. Nvidia's total market value was US$2.90 trillion, with a evaporation of US$590 billion in one day, setting a record for the largest single-day evaporation of individual stock market value in history. Nvidia's plunge also caused the founder Huang Renxun's net worth to shrink by US$21 billion.
Under the leadership of Nvidia, the US semiconductor chip sector suffered a collective setback. DeepSeek is a big model company under the quantitative giant Huanfang Quantitative, and officially released the inference model DeepSeek-R1 on January 20. The R1 has gained widespread attention for its outstanding performance, open source nature, significantly reduced selling price and training costs. The pre-training fee for R1 is only US$5.576 million, and it will be completed on a 2,048 Nvidia H800GPU cluster for 55 days.
Points:
The DeepSeek app tops the ranking of Apple's free app downloads in China and the United States.
Nvidia evaluates DeepSeek as an outstanding progress in the field of AI, and reasoning requires a large number of GPUs.
Nvidia's stock price plummeted 16.86%, evaporating US$590 billion in one day, setting a record for the largest single-day evaporation of individual stock market value in history.
DeepSeek's success not only proves the competitiveness of domestic big models, but also indicates an intensified competition in the AI application market. The sharp fluctuations in Nvidia's stock price reflect the market's sensitivity to the development of AI technology and related hardware needs, and also warns that all links of the industrial chain need to pay close attention to technological development and market changes.