[NEWS] The Emergence of China's Generative AI "DeepSeek" Causes NVIDIA Shares to Plunge 17% Temporarily

On the 27th, the application developed by Chinese AI company DeepSeek reached the top of the free download rankings on the U.S. app store, marking a significant shift in the competitive landscape of the AI market.

Following this news, NVIDIA's stock price temporarily plunged by 17%, symbolizing the market's reaction. This decline resulted in a loss of approximately $600 billion (about ¥92 trillion) in market capitalization, marking the largest drop in NVIDIA's history.

On the same day, U.S. stock markets saw widespread selling of AI infrastructure-related stocks, and Japan's Nikkei 225 index continued to fall on the morning of the 28th, with the decline exceeding 600 points at one stage. This risk-off sentiment also extended to the cryptocurrency market, causing Bitcoin to drop nearly 7% from $105,000 to around $97,600.

This market reaction highlights concerns that DeepSeek's ability to develop high-performance AI models at low cost could reduce demand for high-end chips produced by companies like NVIDIA. Additionally, shaken investor confidence in NVIDIA's existing technology further accelerated the stock's decline.

The rise of DeepSeek has further strengthened the presence of Chinese companies in the AI industry. This development raises questions about how effective U.S. government restrictions on technology exports to China will be. On the same day, former President Donald Trump announced his intention to impose tariffs on semiconductor chips imported from abroad, adding a new layer of complexity to the AI market.

The tariff policy aims to promote domestic semiconductor production in the U.S., but it also raises concerns about increased product prices, which could burden both consumers and industries.

The R1 model by DeepSeek, with its 671 billion parameters, leverages MoE (Mixture-of-Experts) technology to activate only 3.7 billion parameters necessary for specific tasks.

This technology significantly reduces computational costs, achieving comparable performance at approximately one-tenth the cost of other major AI companies. Notably, it optimizes memory usage and GPU time required for training, improving resource efficiency.

Additionally, DeepSeek's "V3" model has outperformed AI models like OpenAI's "GPT-4o" in key benchmarks, including solving complex mathematical problems, generating programming code based on prompts, and providing accurate multilingual answers to various queries.

The emergence of DeepSeek has intensified competition in the AI industry, profoundly impacting established players in the U.S. technology market. The shock to NVIDIA and the escalation of international competition highlight both new possibilities and risks within the AI sector.

DeepSeek's success in delivering low-cost, high-performance AI models has the potential to reshape the trajectory of AI technology development. As a challenger to the U.S.'s technological and capital dominance, DeepSeek's future developments will be closely watched.

Image: Shutterstock

Related Articles

OpenAI Unveils Reasoning AI "OpenAI o1"

OpenAI to Announce AI Agent "Operator" Soon?