Recently, Chinese startup “DeepSeek” has caused a stir in the stock market with the launch of their artificial intelligence (AI) model R1, which boasts low costs. According to Reuters analysis, this phenomenon has not changed the fundamental dilemma of the Chinese Communist Party in the high-tech field – the US export ban on high-end chips remains a major obstacle.
DeepSeek claims that its R1 model, with a cost advantage of 20 to 50 times, has become a low-cost alternative to products from OpenAI and Meta. This news has triggered a strong global market response, leading to a 17% plunge in the stock price of chip giant Nvidia on Monday, wiping out $593 billion in market value, setting a record for the largest single-day market value loss and dragging down US tech stocks across the board. However, on Tuesday (January 28th), market sentiment eased slightly, and tech stocks began to rebound. Nvidia’s stock price rose by about 5% in pre-market trading, and chip-related companies such as Oracle and Marvell Technology also saw a recovery.
According to The Wall Street Journal, doubts are widespread in the market about DeepSeek’s claims of low costs. Stacy Rasgon, an analyst at Bernstein, commented, “DeepSeek did not create ‘OpenAI’ with $5 million.”
Joshua Buchalter, an analyst at TD Cowen, pointed out, “The ‘DeepSeek Moment’ urges investors to act first and ask questions later.”
Atif Malik, an analyst at Citigroup, also questioned DeepSeek’s claim of not using high-end GPUs for model refinement.
Reuters analysis shows that DeepSeek’s success reveals the deep-seated bottleneck in technological innovation in China. Since the US implemented the chip export ban in 2023, China’s development in the high-tech field has been severely restricted. While DeepSeek’s R1 model boasts its cost advantage, it is more due to efficient resource utilization and clever technical design, rather than a true breakthrough in reliance on high-end chips.
In a rare interview, DeepSeek founder Liang Wenfeng said the company’s biggest challenge is not funding but the ban on advanced chips. Despite some progress made by Chinese companies such as Huawei and SMIC in chip manufacturing, these technologies still cannot meet the demands of artificial intelligence technology.
Reuters further pointed out that although DeepSeek’s R1 model has achieved cost advantages, the chips it uses still rely on outdated technology and cannot handle larger scale and more complex AI application scenarios.
CEO of OpenAI, Sam Altman, commented on DeepSeek, saying, “DeepSeek’s R1 model is indeed impressive, especially in terms of pricing, but we believe that greater computing power is the key to success.” He emphasized that OpenAI will continue to focus on executing their research plans and believes that driving improvements in computing power is crucial for the future development of AI technology.
Furthermore, experts believe that DeepSeek’s technology does not yet possess the reasoning capabilities which represent future AI trends like OpenAI’s O3, indicating that there are still significant technological limitations to overcome.