日经BP精选：DeepSeek冲击的真意：盲目扩张只会造就更多“鱼脑AI”

Core Insights - The article highlights a significant turning point in the development of generative AI, particularly large language models (LLMs), marked by the emergence of DeepSeek, a Chinese AI company that has developed a high-performance, low-cost LLM. This development has led to a notable decline in the stock prices of American tech companies like NVIDIA in late January 2025, drawing parallels to the "Sputnik Shock" experienced by the U.S. after the Soviet Union launched the first artificial satellite [1]. Group 1 - DeepSeek's development of a low-cost, high-performance LLM is seen as a major disruption in the AI industry [1]. - The term "DeepSeek Shock" is coined to describe the impact of this development on American tech companies, similar to the historical "Sputnik Shock" [1]. - The training cost for DeepSeek's model is reportedly in the hundreds of millions of yen, but there are emerging cases of models being developed at significantly lower costs, such as the one released by a Stanford University research team in late January 2025 [2].